You are on page 1of 84

BTE Publication Summary

Structural Failure of Large Bulk Ships
Report
Although it is widely known that the risk of failure is high in old ships, old bulk
ships continue to be used. One reason is that until mid 1992 it was financially
viable to maintain an old ship in service rather than dispose of it for demolition.
Even though after mid 1992 it did not pay to keep an old ship in service, it was
even less attractive to purchase a new ship. The BTCE's statistical analysis of
voyage data identified ship age, flag state, commodity carried and voyage route
as important factors influencing the risk of failure. Using this information, the
BTCE has developed a technique for predicting the risk of failure of individual
ships.

Subject
Series
Date
A to Z

Search
Results
Print
Exit

GO BACK

STRUCTURAL FAILURE
OF LARGE BULK SHIPS
B U R E A U OF T R A N S P O R T

AND

COMMUNICATIONS

ECONOMICS

I R E P O R T 8 5

Australian Government Publishing
Service,Canberra

0Commonwealth of Australia1994
I SSN 1034-41 52
ISBN 0 644 33027 9

This work
is copyright. Apart any
from use as permitted under
Copyright
the
Act
7968,no part may be reproduced by any process without prior written
permission from the Australian
Government Publishing Service. Requests and
to the Manager,
inquiries concerning reproduction rights should be addressed
Commonwealth Information Services, Australian
Government Publishing
Service,GPO Box 84,Canberra,A C T 2601.

Produced by The Australian Government Publishing

Service

FOREWORD
The question of what to do about substandard
bulk ships has
become much
more prominentin recent years,The number of bulk ships suffering major
Too lives have
structural failures imore
s than is reasonably acceptable. many
been lost.

SSWgd SkUdi@% have been made in search of the reasons for ship The
losses.
fT!p9f r@.Qent of Australian investigations was that of the House of
Repwoentatives Standing Committee on Transport, Communications and
lRfPl%trUGture:,This broad ranging study considered the commercial and
r@gulatery environment in which bulk ships operate.I t made an important
contribution to understanding the main issues.
Other investigations have
failure.
focused on the technicalof ship
aspects
of
The Department of Transport and Cemmunications requested the Bureau
Transport and Cornmun.jc_a_ticm
Economics (BTCE)to undertake a statistical
analysis to $,S!& if there wet% any factors unique
to Australian bulk trades that
th@ cisk ef structural failure
ofships carrying
our exports. This study
might
in an attempt
to uncover
by th,e B E E also takes a deeper look at the evidence
fuffh@r insights into the issues.
a.@j

Th@ -Study team was lead by Neil Gentle assistedby Stephen Wheatstone and
JQ6gphink 8alrrri,Dr Trevor Breusch of the Department of Statistics at the
Australian National University

provided

valuable toadvice
the team on statistical

methods.
Leo Dobes
Research Manager

iii

CONTENTS Page iii FOREWORD ABSTRACT xi SUMMARY xiii CHAPTER 1 INTRODUCTION Method The data CHAPTER 2 FACTORS CONTRIBUTING FAILURE Vessel condition Commodity Weather conditions Operating conditions Conclusion CHAPTER 3 CHAPTER 4 RELATIVE IMPORTANCE OF THE FACTORS Results Qualifications regarding method the Policy experiment 1 2 3 TO STRUCTURAL 5 5 11 14 16 16 DIFFERENT 17 18 21 22 ECONOMICS OF DEMOLITION AND PURCHASE OF N E W SHIPS Demolition of old ships Purchase of new ships 25 25 31 CHAPTER 5 CONCLUSIONS 37 APPENDIX I DATA ON FAILED SHIPS STRUCTURALLY 41 V .

Page APPENDIX II ASPECTS OF STRUCTURAL APPENDIX 111 RELATIONSHIPS BETWEEN VARIABLES AFFECTING SHIP FAILURE 61 CHARACTERISTICS O F SHIPS DEPARTING FROM AUSTRALIAN PORTSBULK 65 APPENDIX IV FAILURE 45 APPENDIX ANALYSIS V LOGIT 71 REFERENCES 75 ABBREVIATIONS 79 vi .

1 ~ ~ ~ 4.11 Route 3. 1990 quarter 29 vi i . according to inspection threshold level 22 120 000 dwt Voyage chatter rates and demolition price afor bulk carrier 28 age 9 9 age 10 age 13 and failure rate and failure rate 15 the Cape of Good Hope 19 ~ l l l 4.6 Commodity and failure rate 11 2. Brazil Europe. a function as of the number of voyages.10 Weather 14 2.2 Break-even to North second chatter ratea 120 for0oO dwt bulk carrier.9 Size and failure rate 13 2.1 Age and failure rate 2.3 Flag and failure rate 2.4 Proportion of ships that bewould included in inspection.FIGURES l ~ Page 7 2.2 Flag and 2.4 Classification society and 2.2 Estimates of failure probability 20 3.8 Commodity and size 2.5 Classification society and failure rate 10 2.3 Confidence limits for effect of flag 21 3.7 Commodity and 12 2.1 Load line zones around 3.

3 Shear 11.6 Typical 11.2 Hogging 11.3 4.5 4.Page 4.7 World bulk ship failures.2 Australian routes and ship size 67 IV.3 Australian routes and flag 69 v.4 4. 1989 to 1992 55 IV.1 Australian routes and 66 IV.6 Number of voyages Brazil to North Europe required for a 120 000 dwt bulk carrier to break with a even discount rate of 10 per cent 30 Secondhand prices and bulk carrier 31 time charter rates a 120for 000 dwt Net present value of retaining120an old 000 dwt purchasing a new 120 000 dwt ship ship Internal rate of return of retaining an old 120 000 dwt and purchasing a new 120 000 dwt ship and 33 ship 34 11.1 Typical logistic curve 72 viii and 47 sagging 47 stress shell forces exerted dense by and non-dense of a typical bulk carrier cross-section fatigue curves ship age cargo 49 50 52 .4 Side 11.1 46 Flow chart of the factors contributing to ship failure 11.5 Structural 11.

1 Ship age 61 111.1 Australian routes and age ship IV.1 x2 analysis 27 32 1989 toOctober failure. 42 44 abbreviations of iron ore Hope.by port of departure voyages rounding Cape of Good the 57 proxies in theusedanalysis 59 11.3 Classification society 63 111.3 Australian routes and flag weather and possible 64 conditions 64 ship 66 size 68 70 ix .2 Causes of failure 111.1 Relative importance bulk ships used for different in structural factors failure Of of 18 4.TABLES Page 3.2 Flag 62 111.3 Data for analysisof new ship purchase 1.2 Flag 11.2 Operating 4.7 Route IV.4 Commodity carried 63 111.1 Data demolition 4.5 Size of ship 64 111.1 Bulk ships experiencing December 1991 structural 26 analysis costs a for panamax dry bulk carrier 1.6 Voyage 111.2 Australian routes and IV.

Using this information. commodity carried and voyage route as important factors influencing the of risk failure. Although higher freight would rates be required to warrant investment in new ships. increased freight rates would also Improve the attractiveness of maintaining old inships service. Currently. Clearly. voyage route. One option to reduce the risk of structural failure is to improve inspectionsby port state surveyors. Evidence suggests these countries are being more successful at detecting unsound ships.BTCE analysis suggeststhat premiums set on the basis of age.and history of ownership.it is financiallymore attractive to operate old ships ready for scrapping than to invest in new ships. flag state. The BTCE’sstatistical analysisof voyage data identified shipage. making continued Operation On risky voyages commercially unattractive. theB T C E has developed a technique for predicting the risk of failure ofindividllal ships. commodity carried. freight rate increases on their own Cannot alter the balance In favour of improved ship safety. and a thorough structural inspection is not normally pcsslble. xi . the 93 per centOf ships that failed wouldhave beenidentified in an inspection of one in three ships. If the BTCE’s technique had been applied over study period. fiag state and classtflmtlon society could in reducing be instrumental the number of bulk ships that suffer from structural damage. However. port state surveyors have only limited time ts assess a ship.ABSTRACT Between October 1989 and December 1991 50 dry bulk ships worldwide (including nine departing from Australia) suffered structural failure. Australia and other countries now are more vigilant in their inspection programs. (This is the same proportion of ships as currently inspected by the Australian Mahti’me Safety Authority). are already adjusting premiums to reflect the risk associated with different classes of ships. Insurance Companies have a fjfldneisrl stake in reducing the risk of They failure. Insurance companies could use similar techniques to assistthem to set Premiums that more accurately reflect the ofrisk failure. current flag state. Potentially risky ships would thenattract high premiums.

flag state. Clearly. although this is indeed condition before a large-scale renewal of thebulk fleet can be expected. freight rate increases on their own cannot reduce the number of bulk ship failures. i t was Even though after mid 1992 it did not pay even less attractive to purchase a new ship. The technique used (logit analysis) allows the relative importance of the possible factors to be estimated. However.Many of these ships sank the andevidence of thecause of failure was lost with them. Some commentators have proposed that higher freight rates would improve the a necessary financial attractiveness ofnew ships.port state surveyors have only limited timeto assess a ship. is in old ships. to keep an old shipin service.One reason is that until mid 1992 i t was financially viable to maintain an old ship in service rather than dispose of i t for demolition.and a thorough structural inspection is not normally possible. Improvements are being implemented in Australia and elsewhere. The route taken was found to be an especially important factor.commodity carried and voyageroute were all found to be important factorsin structural failure of bulk ships. be towould improve port state One option to reduce the risk of structural failure control inspections. During same period the 50 dry bulk ships (including the nine departing from Australia) failed throughout the world. increased freight rates would also improve the attractiveness of maintaining old ships in service. The BTCE has used statistical techniques to analyse voyage data to identify the most important factors associated with the risk of failure ofbulk dryships. Ship age.SUMMARY Between October 1989 and December 1991 nine dry bulk ships suffered structural failure after departing from Australian ports. Shipsmost arelikely to fail where there is a combination of these factors. but. old bulk Although it is widely known that the risk of failurehigh ships continueto be used. There is some evidence that this is having an effect in the detection of unsound ships. xiii .

Potentially risky ships would attractpremiums. Statistical analysisshowed that. flag or classification society is often considered to be a good indicator of poor ship condition. current'flag state and history of nwngr&ip. Even greater certainty couldbe achieved by inspecting more ships: A recent change in ownership. Nevertheless. However.These factors could not be included in the BTCE analysis. inspection of that proportion ofbulk ships in the sample of voyages @n@/ys@g/ woy!g have identified for inspection 93 per cent of ships that gu$g@$&qqlly fail@. The BTCE analysis suggests that pr@miume @g?! the basis of age. @Iz@ and f!?g stale of ships between the varicws route5 anglyggd. the dlfferanowsdid net in&@$ gmy particular factor likelyto increase the riskof fallure.BTCE Report 85 The flag state of thewasship found not to have as great an influence onof risk #g&#@ frequently suggested. high making continued operation on risky voyages commercially unattractive. the BTCE method used in conjunction with knowledge of recent change of ownership.na dgnlflcant difference was found in the risk of failure between individual Australlan ports ofdeparture. On the basis of a method developedby the BTCE. Ofparticular lnter@!s{ ig wk@tR@f{h@@ g@ MnjqFe factors associated with ships departing from Australian bulk p@fls that could increase the riskof failure. Similar techniques could assist insurance 'companiesto set premiums that more ~ ~ u r a t e lreflect y the risk of failure. flag state and classification society could be instrumental in reducing thenumber of bulk ships that suffer from structural damage. @/€h@u~hthere were gignificant differencesin age. route. I F ) § M ~ ~ R E ~ g~mpgnieshave a financial stake in reducing the risk of failure.When allsw@n@g was made for commodity carried and voyage route. flag or classification society could identify an even larger proportion of ships at risk of failure. commoditycarried. a ship registeredwith a flag state with a high @g@@ rate was found to have the same risk of failure as a ship five yean older regi6itgred with @flagstate with a low casualty rate. xiv . They t&jn$ 8 mu& pore active rolein adjusting premiums to reflect the risk on@ss~a@d with diffe~e~t Glasses of ships. Can the results help identify the ships at risk? The Australian Maritime Safety Authority currently inspects 34 per about cent of all ships visiting Australian ports.

Feamleys (1992a) reported that the age average of drybulk ships was 11. Combination bulk ships are able to cany either dry or liquidbulk commodities. Over the same period. 1. For example. 17-1 8).Lloyd’s Register of Shipping 1992a). The ageing of both of these fleetshas aroused concern. Classification societies were originally developed to carry outofsurveys ships’ hulls on behalf of insurance underwriters. Accordingto HORSCOTCI.pp. A further three ships suffered severe structural damage but were able to reach a port. 3. 1 . many sank.1 to 14. as the old (bloyd’s majority of ships have that been lost since 1989 were over 15 years Register of Shipping 1992a). 2.ship is either lost atsea or it must interrupt voyage its to undergo repairs.4 years in July 1991. another 41 dry bulk ships throughout the world suffered structural failure. p.CHAPTER 1 INTRODUCTION Six dry bulk ships carrying iron were ore lost at sea between October 1989 and December 1991 after loading at Western Australian ports (Daily Commercial News 1992. Over the same period. in resulting the ageing of the bulk ship fleet. Classification societies. eventually contributing to a loss of structural integrity. HORSCOTCI considered that the underlying cause of the in the decline safety of bulk ships was the p o o r state of thebulk ship market. the average age of combination ships increased from14. That the is.’ Although some were able to reach a port for repairs.3 years.*such as Lloyd’s (Lloyd‘s Register of Shipping 1992a). Nippon Kaiji Kyokai (NKK 1992) and the American Bureau of Shipping (ABS 1992). low freight rates have meant that old ships cannot be replaced profitably. Throughout this report a ship is defined to have suffered a structural failure if the damage is ofa severity that prevents the ship from completing a voyage.carried out technical investigations in response to losses of bulk ships.7years in January 1992. to establish the physical causes of failure. Australia’s House of Representatives Standing Committee on Transport. Their has role since expanded and now theycarry out these responsibilities on behalf of owners and as agents for some flag states (HORSCOTCI1992a. X). Communications andInfrastructure (HORSCOTCI) examination of ship safety took a much broader approach.compared with 11. including the commercial and regulatory environment of the industry (HORSCOTCI 1992a. Maintenance effort is reduced as owners seek to reduce costs.

Where data were not available for variables that might be considered importantto structural integrity (for example. the problem in enforcing compliance with the standards. from mid 1989 to mid 1992. Because of the small number of failures. An especially difficult statistical problem involved the number small of failures compared with the overallnumber of voyages undertaken by bulk ships. However. one ship was lost for every 940 were lost through structural failure. lay Insurance companies were alsoas slow seento react to increased bulk carrier losses (HORSCOTCI 1992a. Of particular interest are the following questions: Why have ships carrying iron ore from Western Australia experienced more losses than ships carrying other Australian States? Are the risksto iron ore ships trade route? Are the risks to iron ore ships departing from other major other commodities.out of approximately 47 000 voyages by large bulk ships 30(over 000 deadweight tonnes) throughout the 50 ships world. conventional statistical failure. techniques can identify the factors that influence ofthe riskbut cannot 2 . The insurance industry hasin the past relied solely on inspections by the classification societies to ensure that the ships they insured were safe. For example. and suchgrain.BTCE Repot? 85 H O R S C O T C I also found that there was a general failure in regulatory arrangements: unsafe ships were being classified as structurally sound and allowed to continue operating. The standards were adequate. p. failure had Although HORSCOTCl’sfindings are relevant to bulk shipping as a whole. HORSCOTCI concluded that international economic conditions and regulatory been major factors in the loss of bulk ships in recent years. Reports of engineering studies of structural failure were usedto identify major likely physicalcauses. HORSCOTCI found that insurance companies are now taking a more active interestin the previous record of ships and their owners before agreeing to extend insurance. voyages. The major problemin analysing structural failure is that ships frequently sink after suffering structural so evidence failure. as coal from departing from Australia dependent on the departing from Australia greater than for thos exporting countries? METHOD The B T C E adopted a statistical approach to help isolate factors that may contribute to the risk of structural infailure bulk ships. quality of maintenance) proxy variables were used. there still remains the possibility that other may factors have contributed to structural failure of ships departing from Western Australia.of the physical causes of failures is limited. xvii). that is.

l Chapter l estimate the relative strengths of the factors or the interactions between t Logit analysis was used to overcome these problems. flag or classification society are frequently suggested but information was not available on as strong indicatorsa sub-standard of ship. known failures examined (see appendix were Changes to name. 3 . However. obtained. These data allowed a detailed examination of route and ship to the ship at the time of the voyage. for voyages overMaythe period from 1990 to May 1992. South Australia and from the major Africa and the United States of America. particularly the continued operation of sub-standard ships. characteristics relevant full period for which casualty data Voyage data were not purchased for the of. the voyage sample approximately 29 000 voyages was considered a reasonable compromise 60 per of the between costs and effectiveness of the analysis. these variables. or for allbulk ship movements. it useful to consider structural failures in the wider context The BTCE also found of commercial pressures on bulk ship operators. and cent included I).To this extent the analysis does not capturefullthe set of variables that help determine the risk of failure. India.owner. THE DATA The BTCE purchased Lloyd’s voyage data for bulk ships departing from bulk exporting countries of Brazil.

CHAPTER 2 FACTORS FAILURE CONTRIBUTING TO STRUCTURAL Understanding the interaction between factors contributing to ship failure is clearly important in the development of an understanding of the failure mechanisms. commodity. Further details of aspects of structural failurein appendix are provided II.Lloyd’s Register of Shipping . Some cargoes. For purposes of analysis. These factors are examined below. Loss of side shell plating frequently features in ship failures (Lloyd’s Register Shipping of 1992a. the failure was preceded by water being takenin one or more holds. may accelerate the deterioration of protective coatings and thus promote corrosion(ABS 1992. to corrosion is likely to weaken structures to the pointwhere they are unable withstand the stresseswere they designed for. VESSEL CONDITION In nearly all structural failures bulk of ships.NKK 1992). Corrosion All ships corrode. and operating conditions. point to corrosion and metal fatigue as major causal factors leading toin the deterioration integrity aof ship’s structure. Common methods are the application of protective coatingsand the use of cathodic protection systems. Lloyd’s (Lloyd’s Register of Shipping 1992a) and the American Bureau of Shipping (ABS 1992).among others. If protective coatings and other anti-corrosion systems are not maintained. the likely factors influencing risk of failure can be classified into four broad categories: vessel condition. Preventive systems are therefore normally included in the design and construction of ships to limit the onset and progression of corrosion. weather conditions. I t is also common design practiceto use thicker plates to allow forsome wastage of metal through corrosion over the of the life vessel. However. such as coal.the events that are the immediate precursor to failure have generally a long gestation period.

However.l Because of this. N K K 1992). The effects of corrosionand fatigue in combination with poor maintenance may build up over several years before the risk of becomes failure unacceptably high. Age With poor maintenance. fatigue failure is not normally a part of design considerations.1 illustrates the influence of age on the risk of failure bulk of ships. Unloading procedures bulk for ships often require the of heavy use grabs. exposure to riskas measured The figure indicates that ships older than 14 years experienced many more failures than ‘expected’ and those less 10 years thanold failed less often than ‘expected’. through corrosion or physical damage. higher riskof failure than Figure 2. the level of stress at which the will faistructure l -see appendix II for further after it has endureda history of cyclic stresses detail). Cyclic stresses applied to steel can eventually result in the failure of the metal. corrosion resultingin reduction of metal thickness at points of high stress loadswill lead to an increasein stress within the structure.failure usually occurs 1after to 10 million stress cycles. if the cyclic stress is less a specific than value (the endurance limit) no failurewill occur. and that risk of failure is proportional to the by the number of voyages. Coal cargoes canbecome hot enough for sulphur acid. and bulldozers and pneumatic equipment used to remove adhering and inaccessible cargo. theinitial damage from corrosion and fatigue can spread until the structure is seriously weakened. can easilydamage the structure. Unchecked.also reduces the endurance limit. which can be as heavy 35 as tonnes when empty (Lloyd’s Registerof Shipping 1992a). Older may ships thus be at newer ships. it can act as points of stress concentration. ‘Expected’failure rateswere estimated under the assumption age thathas no influence on the riskof failure. For mild steel.I t may not require much loss of metal to increase stressesa to level whichexceeds the endurance limit. the resultant corrosion can leadto serious wastageof metal in critical structuralcomponents of the ship. Ships in the 10-to 14-yearage group failed about as asoften would 1.BTCE Report 85 1992a.acid condensingon the compounds and moisture in the coal to form The inner surfaces of the hold attacks the structuralcomponents of the ship.Above this limit. thus creating conditions favourable to the initiation of fatigue cracks. If the damage is not repaired. 6 . Roughening of the surface of the metal. Metal fatigue Ships are designed so that stressesin the ship structures are below the metal fatigue endurance limit (that is. The grabs.

In the early 1970s the classification societies design compared changed the method of determining the thickness of the steel components used to construct the ship’s structure. 24 years old also have a fundamental differencein their structural Ships over with newer ships. leading economical designs with reduced plate thicknesses. with age being a factor. have less tolerance 7 . shipbuilders could estimate stresses in ship structures more accurately. so structural redundancy could be reduced.The improved accuracy of stress calculations reduced the need for allowances for to more unknowns. However. thenumber of actual failures of ships 24over 111. Figure 2.Chapter 2 l failures Expected failures Actual Note Expected failures are calculated by is proportional tonumber the of voyages. toPrior this period sizes were determined by rules based on experience. significantly higher than the ‘expected’ number of failures (appendix table The ‘expected’ number of failureswas less thanone. This may result be due this age~groupin the data sample Ifused. in fact associated with failures. the thinner plates to wastage through corrosion. ships in this age group do have an elevated risk of failure the number actual of failures would still be relatively small. and in fact there were no to the relatively small number of shipsin actual failures. years oldwas not Surprisingly.1).The in appendix analysis table 111. These rules generally incorporated sizeable safety factors to reflect the lack of knowledge of the behaviour ofWhen ship structures. that age isis.1 Agerate and failure assuming that risk Maritime Information of failure BTCE be expected on the basis of voyage numbers.1 shows that this aisstatistically significant result.The The fact that there are none is not inconsistent relatively small size of these may older alsoships be a factor. sophisticated computer analysisbecame available in the early 1970s. Source BTCE estimates based on Lloyd’s Services data.

the This is in fact observed for the actual failures shown in figure 2. were flags ranked according to casualty experience of all ships in their fleets.BTCE Report 85 Flag state Flag state2has attracted considerable attention as apossible indicator of the risk of ship failure (DCN 1992). If this is so. so that the remaining ships 24 over years will be more likely to be those registered with the other flag categories. The categories were determined using loss ratios foreach flag state calculated by the U K Institute ofLondon Underwriters (Commissionof European Communities 1993)for the period 1987 to 1991. Flags were then classified into four categories (A to D).12per cent Figure 2. Similarly older ships tend to be not registered with flag states in the A category. Poorly maintained ships registered with D flags are likelyto have been scrapped before they reached 24 years old. The category containing flagswith the largest loss rate was category D while categoryA contained the flags with the smallest loss rate. The world averageloss ratio for the period was reported as 0.44per cent C >0. the data to test this explanation were not available. so that each category contained approximately one-quarter of the world bulk fleet in terms of gross registered tonnage. to be a contrary tendency for ships 24 years over old.The D category had 2. not just bul carriers. then it can be expected that ships registered with states with poor maintenance standards would be likely to exhibit greater than average failure rates. 13). The country in which a ship is registered and which undertakes the responsibility for the (HQRSCOTCI 1992~1. Note thatlossthe ratio reflects the losses of all ships. Owners are considered likelyto register their ships with flag states whose standards are compatiblewith the owner’s approach to maintenance.The loss ratio is calculated as the ratio of the total gross tonnage ofa year the losses fo divided by total registered grosstonnage of the fleet. then the proportion of bulk ship failures should be less for Athecategory than forD category. 8 . Unfortunately.2 shows the proportion bulk of ship voyages in the BTCE database in each age group in each flag category. implementation of international conventions relating to that ship p.23and 10. expressed as a percentage. in the fleet of each flag state. There appears to be a clear tendency for older ships (15to 24 years) to be registered with flag statesin the D category.3.29per cent.44per cent B >0.12and 10. A possible There appears explanation is that ships older 24 than years and still in service are those most likely to have beenwell maintained throughout their lives. The four groups were: D >0.23per cent A 10.as follows. To test this hypothesis. If flag is an indicator of risk of failure bulk of ships.

0 - r g Q 20- L 0- 19-24 15-19 10-14 5-9 >24 Age (Years) Note See text for definition of flag category. See text for definiti ofonflag category. Expected failures are calculated by assuming that risk of failure is proportional to number the of voyages. BTCE Figure 2. Source BTCE estimates based on Lloyd's Maritime Information Services data. 2.3- 5 80- 0 L P) n v 8 60- 0) m >.3 Flag and failure rate 9 l .2 Flag and age Information BTCE Flag Expected failures category Actual failures Notes 1. 0 > 40- C . Figure 2. Source BTCE estimates based on Lloyds Maritime Services data.Chapter 2 100h .

failure is proportional 2. BTCE Figure 2.BTCE Report 85 loo Classification society category 1 0-4 5-9 20-24 10-14 -19 15 Age (years) >24 Note See text for definition of classification society category.5 Classification society and failure rate 10 . Source BTCE estimates based on Lloyd's Maritime Information Services data. Source BTCE estimates based on Lloyd's Maritime Information Services data.4 Classification society and age BTCE Classification society category Expected failures Actual Notes 1. See text for definition of classification society category. Figure 2. Expected failures are calculated by assuming that risk of to the number of voyages.

there is little associationbetween classification societyand risk of failure bulk of ships.However. B and C categories are combined ainto single category.When the A.4). However.33 were carrying iron ore.6 Commodity and rate failure BTCE 11 l . the but this effect of flag couldbealso effect of flag becomes marginally significant. ships classified Classification societieswere categorised into four categories A to D using the bulk ships aremore likely to be same procedure as for flag.6 shows that ships carrying iron ore fail much more frequently than would be expected i f commodity had no Other 0Phosphate Bauxite & alurnina Coal grain 0 m Iron ore Expected failures failures Actual Note Expected failures are calculated by assumingof failure that risk is proportional to the number of voyages.2).As with flag.Chapter 2 more bulk ship failures thanbe would expected if flag had no influence on of risk failure. 1992b). This is confirmed by the statistical analysis in appendix table 111. this association of flagwith risk of failure is not statistically significant for the four-way classification shown in figure 2.in contrast to the flag analysis. Classification society Classification societieshave also been criticised for not adequately inspecting by them (Stephens 1992).5 indicates that. of discussed as above.3. related to the influence age. Figure 2.3 (see appendix table 111. Source BTCE estimates based on Feamleys (1992a. COMMODITY Of the 50 bulk ship failures worldwideduring the period October 1989 to December 1991. older inspected by D category societies compared with A category societies (see figure 2. Figure 2. figure 2.

ships less than 50 000 dwt exhibited fewer failures than expected.5 illustrates how stresses in the ship's structure are influenced by the commodity carried.3 to 0.9 indicates that bulk ships over80 000 dwt failed more often than would have been expected if size had no effect on failure whereas rates. The higher failure rate of ships carrying iron ore could therefore also to theirbeage.BTCE Report 85 influence.8cubic metres per tonne. and iron pellets 0. However. size or type ofcommodity carried. consideration of the mechanisms of failure.7 Commodity and age 12 on Lloyd's Maritime Information data. Appendix figure11. i t is not clear which is the Because iron ore tends more important causal factor.This is a muchhigher density than other majorbulk commodities and can havea marked effect on the stresses imposed ona ship's structure. 100.related Commodity and size Figure 2. grain or other commodities. BTCE ore . Iron ore has a stowage factor of0.8 show that iron ore ships are likely than bulk ships carrying coal. Commodity I""""1 Coal&grain 0Other E Iron 10-14 15-19 20-24 5-9 0-4 ore >24 Age (years) Source BTCE estimates Services based Figure 2.5 cubic metres per tonne (ACA 1992). as discussed in appendix II.7 and 2. to be carriedin larger ships. suggests that commodity is undoubtedly important.4shows that thecommodity carried is statistically significant in explaining ship failures. to be older and larger Figures 2.The analysis in appendix table111.

0 - r - n 80-100 50-80 30-50 100-150 >l50 Size ('OOOdwt) Source BTCE estimates based on Services data.8 Commodity and size BTCE Expected failures failures Actual Note Expected failures are calculated by assuming risk of that failure is proportional to number the of voyages. Figure 2. Lloyd's Maritime Information Figure 2. Source BTCE estimates based Lloyds on Maritime Information Services data.9 Size and failure rate BTCE 13 .Commodity m Coal& grain 0Other m Iron ore 0 > c 40 .

Figure 2. Weather zone category 0Bad Good 0- Expected failures Actual Note Expected failures are calculated by assuming that risk of failure is proportional to the number of voyages.ondata weather conditions experienced during successful voyages are not available. likely weather conditions in zones transited by ships in the as a proxy for actual weather conditions.BTCE Report 85 WEATHER CONDITIONS Structural failure occurs mostly during bad weather conditions.Of six ship failures discussed in the Department of Transport and Communications submission to the HORSCOTCI inquiry into ship safety. For this reason. Bad weather database were used zones weredefined as those whereat least 1 in 1000 waves exceeds 14 metres in height during season the in which the voyage was undertaken. Source BTCE estimates based on Lloyds Maritime Information Services data.10 indicates that bulk ships traversing zones likely tohave bad weather if weather conditions were conditions fail more often than could be expected irrelevant.10 Weather and failure rate 14 BTCE . Figure 2. five experienced heavy seas and gale-force winds at the time of (DTC failure1992). Large waves pounding an already weakened structure can initiate a failure that might nothave occurred under calmer conditions. Although information on the weather conditionsat the time of the failures is generally available.

A further analysisof Australian routescompared with other routesis given in appendix IV.11 are those on more whichthan one ship failed during the analysis period.0 - r 0 a 20- n" Expected failures failures Actual Notes 1. Figure2. The category 'other routes' combines all the routes on which no ships failed or onlyone ship failed. The routes in figure 2. This suggests that the combination of the carriage and of iron areas of ore rough sea can result in dangerously high stresses in ship structures.11 Route and failure rate Maritime Information BTCE 15 .=l m ' c Other 40- . Expected failures are calculated by assuming that riskof failure is proportional to the number of voyages.11 supports the hypothesis that structural failuresmay occur more often onsome routes than can be attributed to chance alone. 2. Because of this i t is easier to test for an association between particular routes and risk of failure than between bad weather zones risk and of failure. 'Other' routes are wiroutes th zero or one failure. 100- Route 0WA-Nth h +-a 80 - WA-Japan c P) 0 m P) v) 60- e! - India-Mediterran Sth Af rica-Asia Brazil-Europe 0 L Q Europe 0Brazil-Asia . Most of these routeswhere morethan one ship failed are principally iron ore routes (South Africa to Asia and the 'other' category are the exceptions).Chapter 2 l Route Voyages are more easily classified according to route rather than of the areas the sea through which they pass. Figure 2. Source BTCE estimates based on Lloyd's Services data.

commodity and flag may be important individually in or particular combinations. which moderates its roll motions (Lloyd’s Register Shipping of 1992a). For example. the analysis is complicated by apparent interactions between the factors. Nor are data available on the withspeed which ships pass through rough seas.BTCE Report 85 OPERATING CONDITIONS H o w a ship is operated duringvoyage the can have an important influence its crew failure risk.~ CONCLUSION The analysis so far suggests that age. For the loading of iron ore. Some informationwas available for the Western Australian iron ore ports for the last quarter of 1991. Appendix II has further details on loading methods. weather and can all contributeto the risk of failure. size. flag. after allowing for route and commodity. the important parameters arenumber the of passes made during the loading procedure and whetheris loaded the ship in alternate holds orhomogeneously (some cargo in each hold). 3.The Australian Coal Association (ACA)in its submission to HORSCOTCI commented that ‘insufficient training in seamanship and ship operationmay result in vessels being held on course at dangerously highspeeds in heavy seas. For this it was reason not possible examine to the influence crew of qualifications on the risk of ships failing.iron ore tendsto be carried in older ships and on routes that appear to present higher than average risk of failure. disasters’ on Although informationis usually available about crew qualifications for ships that fail. Statistical analysis of portof loading (appendixII) showed no significant link between specific ports and the probability of a ship failing.During that period approximately 50 per cent bulk of ships were loaded in alternate holds and of passes was 2. commodity. However. similar information is not available for ships that successfully complete voyages. are they appear to be a universal problem.An “analysis of loading speed for iron ore ships failed to detect any significant link to ship failure. size. Factors such as age. The average number ro . If poor loading practices a problem. Data for actual averagespeed of loading o’r number of passes were not available for all the ships in the databa~e. Alternate hold loading is often preferredbecause it allows faster loading and raises the ship’s centre of gravity.8.route. The HORSCOTCI (1992b) inquiry into ship safety identified training as animportant factorin ship losses. 16 the remainder were loaded homogeneously. Another important aspect of ship operations is that of loading and unloading. Investigations of several recent losses of bulk carriers would suggest that failure to speed reduceor to divert around bad weather may have been the principle cause of these (ACA 1992).

This chapterseeks to assess the importance of the different factors and their interactions in the failure bulk of ships.confirm the Table 3. All of the variables listed in table 3. theycan take a value of1 if they areto be included in the estimation of failure probability 0ori f they areto be excluded.the logit equationwill estimate the probability of failure of a ship registered in a flag statewith a good casualty record1 carrying coal or grain on a route other than those listed 3. except age. Using logit analysis.setting the flag 2.CHAPTER 3 RELATIVE IMPORTANCE OF THE DIFFERENT FACTORS The analysis in chapter 2 and appendixIll suggests that the major factors influencing the risk of failure age. These results are generally wellknown. route. sizeand flag state. the kgit equation dummy variible to1 is equivalent toa flag state represented by table 3.1. in principle. Because of the small number of failures.1 illustrates the most useful of the results significant role played by route.age and commodity in failure risk. nor cannot be used can it be usedto measure their relative strengths. The technique can be to used estimate the probability of failure. Variables thatadd little to the resultscan be progressively dropped until the most useful combination for the particular application remains. are commodity. also significant. the interactionbetween variables to be explored. If all dummy variables are set to 0. in table 1. the effects of the different variables on risk of failure can be obtained by examining results for different combinations of variables. but little analysis appears to have of the different factors in contributing to the been done on the relative strength risk of structural failure in bulk ships. The results obtained. Nor have the interactions between different factors been fully investigated. conventional statistical analysis to assess the interaction between these different factors. The flag state is but has less of an effect on failure commodity risk orthanroute. See chapter 2 for a discussion on the categorisation of flag Instates.1.1. Logit analysis provides an appropriate method that allows the relative strength of the various factors to be measured and. in category D as defined in chapter 17 . That is. A brief discussionon logit techniques is presented in appendix V. for are dummy variables.

Of the 30 failures includedin the database.1 illustrates the load line regions by defined the IMO in the vicinity of Cape the I 18 zero . iron coal’and The probability grain.BTCE Report 85 TABLE 3.7).04 0. Commodities other than iron by setting the failing is estimated ore.when was examined in combination with other variablesin the logit analysis size found to be unimportant. RESULTS Iron ore is’shown by the logit analysis to be less important as a factor influencing structural failure of bulk ships than the in chapter analysis 2 might suggest. 16 occurred on these three the of Good routes.66 0. Hope or in the Southern Indian International Maritime Organisation (IMO)regulations limit the maximum load that the ship can carry (thesummer load line).The size effect detected may thus be related to the commodity or the route.7. This is because the routes included in the results are predominantly to some iron ore routes. 13 were either in the vicinity of Cape Ocean (appendix figure 11. is a. Figure 3.82 1. Other routes were tested These three routes all involve a transit of the Cape of Good little to the results. Iron ore shipsto be tendlarger than ships carrying in chapter 2 and appendix Ill other bulk commodities.61 0.x2 value is for covariites1 1 7.27 3.31 0.1 RELATIVE IMPORTANCE FAILURE O F BULK SHIPS DIFFERENT OF FACTORS IN STRUCTURAL ~~ ameter Variable ~~ Standard Intercept Iron ore Other commoditiesa Age South Africa -Asia Brazil -Asia 0. Hope.54 0.19 2. This means that the effect of iron ore is represented extent by the route variables. Brazilto Asia.72 0. the winter load and lineother load lines effectively reduce the load the shipis permitted to carry in recognition of additional weather hazards likely to be encountered.46 0. The analysis reveals that three routes and WesternAustralia to North Europe make large contributionsto the but they added probability of failure ofbulk ships. of coal or grain voyages ore and ‘other commodities’ variables to Source BTCE estimates based on Lloyds Maritime Information Services data. Of these 16 failures.94 1.92 0. - South Africa to Asia.8.4 Note Log likelihood ratio for intercept and covariites 353.54 Western AustraliaNorth Europe - 1 Flag error -1 1. Although when ship sizewas examined on its own in chapter 2 and appendix Ill i t appeared to have some significant associationwith risk of failure.54 2.88 0.

1 may not be appropriate for large bulk vessels. The current regulations allow vessels rounding Cape the of Good Hope to load to the summer load line irrespective of the season. However. lnfernafional Convenfion on Load Lines 7966 (IMO) currentty under reviewby IMO subcommittee on Stability. Load Lines and Fishing Vessels 19 . Figure 3.* - 2.The IMO is presently conducting a review of these regulations.1 Load line zones around the Cape of Good Hope of Good Hope. with the elevated risk of failurebulk forships transiting the Cape of Good Hopeidentified by this analysis.Chapter 3 BTCE SOUTHERN WINTER SEASONAL Z O N E Source Hydrographer of the Navy (1987). the region boundaries shown in figure 3.

Those with a good casualty record arein A.1 c c 0 5 m $ (good ore Iron 0. in the estimates. The method used to calculate the confidence limits shown isin appendix V. Figure 3. The two curves for iron ore voyages in figure 3. I I I I I 15 Age (years) 10 I I I 20 . That is.1 .those B and C categories.000010.2 Estimatesof failure probability 3. This be can measured An important issueis the likely error by the confidence limits for the estimated probabilities. carrying iron ore from with a Western Australia to northern Europe.2illustrates the probability of failure for ships undertaking three t of voyages. 20 the .2 differ onlyin flag category. The flag stateswith the bad casualty record are inthose the D category in chapter 2. Source BTCE estimates based on Lloyd's Mariiime Information Services data. The 'coal or grain' voyages do not transit southem Indian Ocean. ships registered with a 'good' flag state are not as likely to fail as early in their livesas others. I I .001 a 0.3 90 per cent confidence limits for two iron the orevoyage curves in figure 3.000001 1 1 I I I 5 I I I I I I . The group with the highest failure probability are those ships registered with a flag state3with a poor casualty record.2 is that the probability of failure an iron of ore is equivalent to an ship registeredwith a flag statewith a poor casualty record with a flag statewith a identical voyage of a ship five years older registered good casualty record. 24 Note Iron ore voyages are on the Western to northern Australia Europe route. One way of interpreting figure 3.4 The marginal separation of twothe confidence intervals for ships 10 older than years indicates that the logit equation is better at predicting failure rates for BTCE '1 e! -mflag) 0. Figure shows the 3.0001 0. 4.? 0.01.BTCE Report 85 Figure 3.Voyages of bulk ships registered flag statewith a good casualty record carrying coal or grain over routes that do not transit Cape the of Good Hope face the lowest risk of failure.

were undertaken by approximately 3000 individual ships. 1 . This is not surprising as most of the ifailures n the occur older ships. That is. Although the confidence interval is large. Appendix V gives details of how the confidence interval for thechange in age required can be estimated.Chapter 3 l BTCE '1 Western Australia to North Europe route 0. Figure 3.change the in age to compensate for the effect of flag is significantly different from zero.3 Confidence limits for effect of flag these older ships. on average each ship completed about tenduring voyages the two-year period coveredby the database. confidence intervals also indicate that as that aremost at risk of failingThe IMO regulations the ship ages the policies of the flag state on enforcingwould become increasingly more important. [ 1 1 1 1 1 1 10 15 Age (years) 1 1 1 20 1 1 1 24 Source BTCE estimates based on Lloyd's Maritime Information Service data. The mean effect for iron ore ships (see figure3. However. QUALIFICATIONS REGARDING THE METHOD The logit procedure is based on the assumption that each observation is independent of all other observations. ~ A different way of considering the effect of flag is to estimate the change in age required to exactly compensatefor the effect of registering a ship with a flag state with a bad casualty record.2)is about five years. the almost 29 000 voyages in the database. This differs from an examination of confidence intervals because only two variables are changing (flag and age). all other variables being held constant.3 to about 9 years.The separation of the two confidence intervals provides support for the view that the influence of flag is an important factor for the older ships . This large number of repeated voyages by 21 .0001 1 1 1 .The estimated confidence interval is from 1. 1 5 1 1 1 .

0. the essential issue is the ability of the model logit to correctly identify ships with a high probability of failure. The number of ships identified for inspection using this procedurecan then be compared with the number of failed ships that wouldhave been inspected. Appendix V contains more detail on this point5 POLICY EXPERIMENT Although the factors in table 3. . Breusch of the Australian National University Department of Statistics.0008 probability Source BTCE estimates based Services data.4 Proportion of ships that inspection. 0. 34 per cent of all The Australian Maritime Safety Authority currently inspects ships trading to Australia.4 shows how the proportion of failed ships includedin the inspection group declines as the threshold for inspe is raised. Figure 3. The test was based on setting a threshold probability of failure. Suppose that with an estimated probability of in failure excess of a threshold level only ships would be subject to inspection. of on l l l .0012 failure Lloyd's would Figure 3.0004 Threshold l l l 0.1 were all shown to be statistically significantin ship failures. l l -" inspected . The qualifications in this section were suggested in a personal communication from Dr T.BTCE Report 85 individual ships is likely to result in more optimistic confidence limits than might otherwise be the case. 22 . One method of testing the model is to measure how well it can identify the ships that actually failed. compared with the internationally accepted "- Total o ~ 0 l . according to level Maritime Information be included in inspection threshold 5.

0005. 23 .0003.0005 probability of failure.93 per cent of failed ships would have beenincluded in the group of ships inspected.Chapter 3 l inspection target of25 per cent (Stephens 1993). Thorough inspections that would be certain to detect structural deficiencies ar time consuming and may not be possible under operating conditions by faced port control surveyors. as such recent change of ownership. This approach on its own could therefore capture a large proportion of structurally deficient ships in the inspection net. respectively. Althougha large proportion of ships that subsequently failed would have beenincluded in an inspection system based on BTCE theapproach. Nevertheless. that was not includedin the analysisbecause of lack of data. These are the proportions which wouldbe inspected i f thresholds of 0. it But would be better if used in conjunction with other information.it is not certain that their potential for failure would have been detected.At a threshold of0.only 25 per cent of ships would have beeninspected. Used in this way the results of the BTCE analysis couldcomplement present methods of selecting ships for inspection. including only 80 per cent of failed ships.4).0003 and 0. it is important to be aware that no inspection system is foolproof.were used to determine whether a ship would be inspected (seefigure 3. At the higher threshold of 0.

and on its condition.value of the cash and selling both of these choices can then be 10 compared.As a result. conditions have not generally been conduciveto expenditure on maintenance by likely have suffered afrom lack of shipowners. For convenience it is assumed that the decision period one year.) For purposes of analysis a typical 120 000 dwt iron ore bulk carrier trading between Brazil and Europe was chosen as a representative ship (tables 4. crew port charges.many old ships are to maintenance or have been operated beyond their design life. Major assumptions made were: All loan capital has been offpaid . is That is.1 and 4. 25 .CHAPTER 4 ECONOMICS OF DEMOLITION AND PURCHASE OF NEW SHIPS l Age is clearly an important factor in the failureof bulk ships. bunkers and agency out offees. the BTCE analysed some of the factors involved in scrappage and purchase decisions. However. A voyage charteris a contract to carry a cargo on a single voyage specified between two ports or areas.p.2). such ascosts. actual number could be less. at the h the beginning of the year the shipowner wit is facedchoice of demolishing the as many voyages as possibleoneformore year ship or of committing the toship it for demolition at the end ofThethe present year. In order to gain some insight into the reasons for the continued of oldoperation and possibly substandard ships. The ship is employed on a voyage charter basis. Some old ships of safely undertaking long have been well maintained and are perfectly capable in the marketin which bulk ships operate ocean voyages. 25). the freight (Packard 1986.The planning horizon for keeping the ship in service will depend on the prospects for finding employment for the ship.The shipowner has to defray all operating expenses. per cent(A discount flow under rate is ustoedcalculate the present value of all costs and revenues.’ 1.The maximum number of voyages the representative ship make couldin the twelve months at normal operating speeds and port times but the is eight. DEMOLITION OF OLD SHIPS Owners of old ships are continually with the faced choice of disposing of a ship or of keeping i t in service.

4 97 72 108 138 1991 Q1 Q2 Q3 Q4 7. Source Quarterly figures are averages of monthly figures (1989a.70 7.8 3. new building pricesand demolition prices are those published in Lloyds Shipping Economist (1989a.00 5.0 101 71 70 81 Q2 Q3 Q4 5.20 5. A panamax ship is usually in the range 50 000 dwt to 80 000 dwt bulk ship able to pass Panama through Canal.00 7. fuel costs.4 74 1992 1993 Note The representative ship ais120 000 dwt dry bulk carrier.20 6.23 3.8 5.00 5.9 3. To convert panamax the ship costs to those appropriate toa 120 000 dwt ship. published in Lloyd's Shipping Economist Operating cost.00 7.3 5.80 7. b. Price for demolition in the Far East. a.3 3.3 3. the technical and management and miscellaneous costs reported by Lloyd's were increased by 40 per cent.0 5.70 4. 1989b).1 3.4 4.1 DATA USED FOR DEMOLITION Voyage ANALYSIS Demolition Quarter price Year Fuel charter price rate (US$/tonne) (US$million)"(us$/tonne)b Q1 1988 Q1 Q2 Q3 Q4 7.00 7.00 4.7 4.8 5.3 73 79 70 59 1989 Q1 Q2 Q3 Q4 8.BTCE Report 85 TABLE 4.40 6.40 4. and lubricating oils).6 5. insurance and communication).00 8.3 5. This increase was based on the results a of previous Bureau 2.6 5. running repairs and maintenance. Operating costs include crew. the 26 andis the largest size .90 6. Price forhigh viscosity oil in Rotterdam.40 4. management and'miscellaneous (management fees.6 74 95 84 101 1990 Q1 Q2 Q3 Q4 8.80 6. technical (stores and supplies.3 69 82 90 92 Q1 4.80 5.60 3.2 5.20 7. 1989b). Lloyds Shipping Economist reports costs for a Panama2 vessel but not fora 120 000 dwt ship. The ship for which the Lloyd's Shipping Economist costs are estimated is manned under an open registry by Indian officersand Korean ratings.

study of bulk ship costs (BTCE 1988).4 25.4 .4 42.7 14.0 38.3 49.0 20.6 105. The more voyages costs for ships undertaking few voyages undertaken during the year the less is the inoverstatement costs.6 49.5 72.3 49. Ship manned under an open registry with Indian officersand Korean ratings.The costs of sailing theto the ship breaker’s yard is deducted from the demolition price.0 46.0 Q4 53.8 31. insurance and communication.7 102.0 38. Operating costs are incurred for full year the evenif the ship is not earning revenue.0 1990 25. of monthly figures published in Lloyds Shipping Source Quarterly figures are averages Economist (1 989a).1 55.8 8 3.0 34.0 Q4 46.0 28.7 28.5 .3 Management and Technicalb miscellaneous c 17.0 Q1 20.3 27.4 21.0 a.6 Q1 57.2 33.1 Q1 28.Crew costs tend to be independent of ship size.8 Q2 Q3 18. Technicalcosts are for stores.0 Q2 Q3 17.8 54.4 46.4 25.7 1993 33.4 12.9 Q2 Q3 33.8 1989 22.5 Q1 Q2 26.0 1992 30.4 1991 25. supplies. This is probably an unrealistic assumption and would overstate during the year.4 46.4 23. so the crew costs for 120 000 thedwt ship were assumed to be the same as those reported panamax for the ship. running and maintenance.0 28.4 Chapter 4 l TABLE 4.5 Q4 42. lubricating oil. Costs are also spread evenly throughout the year. This assumption may not would bea reasonable exactly represent the actual receipt of but revenue approximation for the payment of costs.4 54.8 Q1 17.6 28.8 22.0 42.0 Q4 38. b. Revenue is received in twelve equal payments throughout the year.5 105.4 6 .2 OPERATING COSTS FOR A P A N A M A X DRY BULK CARRIER (US$‘000) Year Quarter Crewa 1988 22.8 83.7 104.0 4 91.0 Q1 17.5 17. The ship is demolished in Asia.4 Total 16.8 22.6 Q3 Q4 27. repairs c.0 Q2 Q3 22.0 49.8 42.5 28. Management and miscellaneous costs are management for fees.9 38.2 126. 27 the .

the owner is assumed to keep maintenance costs down so as to maximise the short-term benefits of retaining the ship for an extra y trading.1 Voyage charter rates 120 000 dwt bulk carrier 28 price for a m ~ r ~ m ~ .-g li -. c e" \ L " " * 3. the following additional assumptions are made: Maintenance costs are assumed to be nodifferent than fora 10-year-old vessel.1 indicate that there is some correlation between demolition pricesand voyagecharter rates. i f the market and regulatory systems operated adequately. Even though an old ship can be expectedto require additional maintenance. which indicated that it is only recently that insurance companies have begun to arrange their own inspections of ships and to adjust premiums to reflect risk.- 5 . The results in figure 4. and demolition Figure 4.The ship on which this analysisis based is likely to be considerably older. is downward there pressure on charter rates. This is consistent with the HORSCOTCI (1992a)report.4"" 8 4- Demolition price(US$ mil) * . Thatis. ' I 0 1 r 1 ~ 1 1 m 1 ~ 1 1 r u ~u um I uuuuuuu 1991199019891988 I 1 1 1 1 1 ~ 1 r 1 N uuuuu l 1 1 1 um u du I 1 1 1 r N uuu 1992 I Source Lloyds Shipping Economist (1 989b).BTCE Report 85 The operating costs reported Lloyd's by Shipping Economist(1989a) are for a 10-year-old ship. As fewer ships are demolished. -- -. alternative option of retaining old inships service becomes more attractive. Insurance costs are assumed to be thesame as for a 10-year-old vessel. the BTCE 8Voyage charter rate (US$/t) 76. but no allowance is made for the highermaintenance and insurance costs which could be expected on an old ship.One possible explanationis that if demolition prices decline in response to conditions in the scrap steel market.

there is an increased supply ofto ships availab breakers with a resultant downward pressure on demolition prices.Thus.The break-even charter rate for the representative bulk ship for three voyages during the following months twelve was estimated to be US$6.A typical result of this analysis is shown in figure 4.2). second quarter 1990 29 .Chapter 4 An alternative explanation could be that as charter rates decline due to conditions in the freight market.80 per tonne (figure 4. if the shipowner could be sure of three or more voyages i t would pay to retain the iship n service.The charter bulk ship to break evenfor up to eight rates required for the representative voyages completed during the year were estimated.2 Break-even charter frate or a 120 000 dwt bulk to North Europe. for from each thequarter first quarterin 1988 to the first quarter in 1993.00 per tonne for the second quarter of 1990 (Lloyd's Shipping Economist1990). as a function of the carrier. individual shipowners will take into account both sets inofmaking prices their decisions. Irrespective of which explanation better describes causality in the market. 1989b). 8 Economist Figure 4. 2o 1 BTCE 1 2 3 4 5 6 Number of voyages 7 Source BTCE estimates based on Lloyds Shipping (1 989a. The voyage charter rate for carrying iron between ore Brazil and North West Europe averaged US$7. Brazil number of voyages. l Both explanations are probably partially correct. The number of voyages expected to be undertaken in the forthcoming 12 months is a key parameterin deciding thebest course of action.2 forsecond the quarter of 1990.

8 Q223.0 Q134.00 1989308.0 329. This assumes degree a of optimism on the part of the period the shipowner. Required at intervalsof30 months.0 1993492.0 16.5 29.5 Q427.7 Q3 36. Source Quarterly figures are averages of monthly figures (1989a.0 17.6 Q432. and operating costs have to tended The ship is dry-docked at intervals of 30 months at prices reported in Lloyds Shipping Economist (1 989a).0 329. Q2 32.5 Q128.7 469. b. 1989b) indicate that time charter rates have trendeddownwards increase. Five years old.3 DATA FOR ANALYSIS OF NEW SHIP PURCHASE Time Second- Year Quarter charter rate handprice (UWOOO/day) (US$millionp 1988 196.60 12.0 Dry dock (US$OOO)” Q 10. a.0 424.2 Q427.5 Q3 Q4 28.BTCE Report 85 TABLE 4.0 350.20 17. as the datain figure 4.50 14. of the period from 1988 to early1992 it was profitable to retain an old For most ship.0 280.0 Q1 21.0 1991 3. published in Lloyds Shipping Economist Any money left over after paying for operating costs and interest payments is used to reduce the principal of the loan.The net present value estimated at a discount 32 .7 Q332.6 454.7 Q324.40 13.7 Q328.3 41Q2 33.80 1990 350.0 9.8 16.3 10.0 424.0 1992 439.7 9.10 17. provided that more than three voyage charters could be during arranged the 12-month analysis period.8 11. 1989b).0 Q1 29.20 399.4 and in Lloyd‘s Shipping Economist (1989a.0 Q1 32.3 14.9 15. Price estimated from costs (1 989)for a panamax reported by Lloyds Shipping Economist bulk carrier.40 13.2 Q230. Operating costs and time charter ratesassumed were not to change over of the analysis.0 466.0 280.8 13.30 15.0 Note The representative ship ais120 000 dwt dry bulk carrier.2 Q435.5 Q128.5 378.30 17.0 280.0 308.7 13.0 367.5 Q230.80 15.

BTCE New ship - \ \ l l l l ship Old L - *r*r\ * \ 1 l I I I \ \ \ \ - I I \ ' \ ' ' 4 Discount rate = 10 per cent Old ship makes 4 voyages per annum Source BTCE estimates (1989a.6. Although the net present value of was the not old large ship (negative in late 1992). 8) made the following observation aattime when the prospects for new bulk carriers began to become doubtful: [Elxamples such as these artare ificial because always they look at single vessels In practice. Figure 4.-.5 Net present value of retaining an OM 120 000 dwt ship and purchasing new a 120 000 dwt ship 33 . Figure 4. ' .p. The results suggest that nor to purchase a new one.1989b).It is not surprising.6 shows that the internal rate of return for the oldwithship four voyage charters is higher than that of the new ship for almostof all the period analysed. it is verycommon for the as isolated entities rather than part of a fleet. \ \ l l l l I I .Chapter 4 l rate of10 per cent is compared for the old ship (with four voyage charters) and compared are the new ship in figure 4. in these circumstances. After199early 0 ship was far higher than that obtained the net present value aat10 per cent discount rate fornewtheship became negative.it was superior to that of the new ship. in late 1992 it was not profitable to retain an old vessel.l * 4 ' -4 based on Lloyds Shipping Economist Figure 4.5and similarly the internal rates of return in figure 4. Lloyd's Shipping Economist(1990. modern second-hand vessels to produce negative purchase of new buildings and 1 p \ * -. The retention of an old ship gave a positive return until early 1992. that there so were many old ships offered for charter contracts.5shows that until early 1990the net present value of purchasing a new by retaining an old ship.

6suggest thatit is not currently worthwhile considering the purchase a of new ship. A 400 per cent increase over 1992 rates is necessary to provide a proper reward to shipping companies to replace aging tonnage . since it encourages shipping companies to be content with a level of remuneration which is too low to support the renewal and growth of their business.5 and 4. In a fleet with a well spreadage profile. but in fact is most unhealthy. or This totally is paid a situation to which the industrybecome has habituated over the years. these ships are initially subsidised by the strong net cash flow produced by older vessels on which the debt has been largelydown. Market conditions have clearly worsened1990 since .BTCE Report 85 BTCE - I 0 -20 1 Old ship makes 4 voyages per annum Source BTCE estimates (1989a.In the last quarter 1 of992 an increaseof 80 per centin freight rates would have improved the internal rate ofof return retaining an old ship.1989b). from -20 per cent to 34 . an increase in freight rates makes also retention ofaged vessels more attractive. although less than thatby estimated Sanderson.' The results 'inthis chapter indicate that the increase in freight rates required to make i t attractive to replace aging ships..However. probably in the order of 80 per cent. was substantial. Sanderson (1993)commented in the Daily Commercial News that '[wlith steadily depressed freight rates and sharply rising carrier costs the necessity fora substantial risein freight rates is the biggest single issue facing the shipping industry.6 Internal rate of return of retaining an old 120 000 dwt ship and purchasing new a 120 000 dwt ship cash flow after debt service in the early years. based Lloyd's on Shipping Economist Figure 4.The results in figures 4. ..

and ithis from the world but fleet.of better insurance and regulatory arrangements the scope of this paper. I t is feasible to set premiums according to the risk to the add costs of employing old associated with particular voyages. rather tothan Australia coming where they would be subjected to a higher level of inspection. Sanderson (1993)reports that improvements in control by the port state are having an effect on which countries substandardto ships are call at. the results in chapter 3 suggest that it may be possible to be moreeffective. Now that insurance companies have suffered large losses from marine insurance. more than inan freigincrease ht rates is requto ireremove d old tonnage from the world fleet. leaving the potentially hazardous conditions. However. such as carrying iron ore Cape aroundof Good the Hope.He reports thatsome ships are now sailing in ballast from the Pacific to South Africa and the Atlantic. greater attention is being focused on the level of premiums and Companies now seem to be charging premiums that better reflect the risk s likely to assist in removing substandard ships (HORSCOTCI 1992a). This would ships on risky voyages. 35 pr . Old ships could thenbecome uncompetitive on risky voyages. them to those ships better tosuited is beyond Full consideration.Chapter 4 32 per cent. the results in this and the previous chapter suggest that a close alignment of insurance premiums with risk would assist in discouraging continued operation of unsafe ships. Clearly. Increased regulatory control of ship standards one isapproach that has been r y to reduce the number of implemented in Australia andin other countries tto unsafe ships.

route to North Europe via South Africa poses especially high failure risks. does suggestis that the classification societies in the database have been equally successful (or unsuccessful) at detecting sub-standard ships. t may be appropriate for the IMO current review of load lines to consider changes for this region. given the enhanced risk of failure associated with this area of the iocean. the analysis suggests that flagas isimportant not as other with a flag with a bad casualty factors. ship failures includedin the database. the important than flag state. to have a higher Although ships departing from some specific ports appeared than average probability of failure.route. carrying iron ore on routes from South Africa and Brazil to Asia also average risks of structural failure. for Australian exports of iron ore. Of the 30 bulk These routes all transit same the region in the Southern Ocean. ships adequately. This does mean not that classification societies are providing an adequate standard What of it inspection. Classification societies have been frequently criticised for failing to inspect in the continued operation of sub-standard ships.but. and In particular.The results suggest athat ship registered record has same the probability of failure a ship as from a ‘good’flag that is five years older.13 occurred in this region. resulting However. International load line regulations allow shipsto travel around theCape of Good Hopeloaded to summer load lines. the analysis of voyage records failed any statistical link to establish between failure risk and classification societies. Weather conditions in the SouthernOcean can be very severe. bulk ships travel was found to bemuch of greater statistical The route on which importance to the risk of structural failure than previously suspected. that and commodity (especially iron ore) are major indicators of increased risk of structural failure.corrected for the effect of 37 . the failures tended to bewithassociated the routes identifiedas having high failure When rates.CHAPTER 5 CONCLUSIONS The statistical analysis of Lloyd’s voyage records indicates age. Although flag state is frequently mentioned a significant as indicator of high failure probability.

in therole control of ship quality. there are The best that can be said is that the cargoes carried by tankers and to be low density cargoes compared with iron ore. ship on a particular voyage that a p has robability of failur above e a pre-determined In practice. the more likely a sub-standard structurally deficient shipwill be caughtin the inspection net. but they are notThethe economic total solution. or of classification society toarebe said I t would have been pointers to the possibility ofa ship being sub-standard. which in conjunction with information not able to be modelled (such of ownership).The model can be used probability of failure for each ship in terms of its characteristics and Aroute. including Australian ports. The larger the number of ships inspected. Tankers and container ships have different types of structure. inspections by port state surveyors means that there can still be no certainty will be prevented from sailing. But the differences support the conclusions regarding the riskiness of Ocean the Southern routes. an iincrease rates of older sub-standard ships more attractive.of flag. The sizes and ages of the ships employed on routes departing from W Australia differ from those employed on routes from Eastern Australia. ports. Recent change of ownership. Freight rates need to be higher than they are at present before ship operators are likelyto commit themselvesto a substantial number ofnew n freight rates would make alsothe continued use orders. Insurance companies can play an important There is evidence that insurers now are taking a more active rolein monitoring 38 . Unfortunately the datato the did not include these variables. the building environment in which bulk shipping operates does not favour new ships. Although the analysis focused on dry bulk ships. would allow inspections to be directed towards as recent change the ships most at risk. developed. available Bureau useful to examine this statistically. an interesting question is whether the results would carry over into other ship types. However. Although an inspection planmay ensure almost all ships that have a high in practice the limited time available for probability of failure are inspected. so they may be ships tend more likely to have similar risks to dry bulk carriers employed in the carriage of coal and grain. had no statistical link with structural failures. sub-standard ships Improvements to inspections by the port state are part of the tosolution the problem of sub-standard ships. The differences do not indicate that any specific factor uniqueto Australian conditions increases risk of structural failure.BTCE Report 85 route.some simple decision rules could be threshold would be inspected.A logit model developed 'from the data provides a tool that could be used to assist in the to estimate the selection of ships for inspection.some butsimilarities. Clearlyin freight an increase would not provide a solution to the problem of sub-standard ships.

old ships could have premiums a voyage set by on voyage basis. to reflect the high level of risk faced by ships transiting this region while carrying iron ore. For example. and are If they also transit the registered with flag stateswith high casualty rates.If the Ocean then the voyage was intended to carry iron ore through the Southern premiums could be increased significantly. example very old ships could to be carry used iron ore Better quality ships become would more competitive on the more risky routes. It might be thought that an insurance company adopting this policy would lose a companycharging high business to a less discriminating company. Consideration should be to given amending load line regulations for the area in the vicinity of the Cape of Good Hope. Ships facing would therefore tend to be used on routes for which they are better suited. on known safe routes.A more effective means of reducing failures would be for insurance premiums to closely reflect the degree toofindividual risk ships and possibly voyages. freight rates would ultimately reflect the premiums high failure risk would bemuch less competitive and charged. then provide incentive for all to companies In summary.Chapter 5 the quality of vessels they insure. carry iron ore.if the whereas voyage was to carry grain on a safer route then the premium would be lower. unless it could beshown that their conditionwas of a high standard.the major conclusions that can be drawn from the analysis High risk ships are those that are 15 years over old. But premiums for old ships carrying iron ore on dangerous routesto would be able provide lower premiums for good quality ships than the less discriminating to the insurer. The better quality and lower risk ships would be attracted company that discriminated against risky ships. Southern Ocean they face an even higher level of risk. insurers to more accurately reflect these factors (O’Brien should adjust their premiums 1 992). Knowing that old ships are more at that old ships carrying iron ore on specific routes face elevated risks. The other companies would soon find they were carrying a greater proportion of risk. Continuing this scenario. Enhanced port state inspections would tohelp decrease the risk of structural failure. 39 . Market forces would adopt similar practices.

APPENDIX I DATA O N STRUCTURALLY FAILED

SHIPS

41

P
N

TABLE 1.1 BULK SHIPS EXPERIENCING STRUCTURAL FAILURE, OCTOBER 1989 TO DECEMBER 1991

Size

Destination Departure
Year
Name
no. Lloyds
Flag built GRT
date

m

Failure

port

TVpe

Ademontasa
6807747
26 271 BC
BC
AI Taludi
7041 0 4 1
24 956
Alexandre P
6803222
54
566 oc
Alexita
60 846
BC
7227229
Algarrobo
89 178 oc
7327665
8 0 1 0453
74 729
BC
Amazon
731591 1 112 306
Atlas Pride
OB0
6916366
44 276
oc
Azalea
734301 1
Berlisa
80 174
BC
70 872
BC
Blooming Orchard 7027435
71 17084
Cape North
41 565
OB0
29 966
BC
Continental Lotus 6717899
20
966 BC
7328542
Elounda Day
35 104 BC
7329596
Entrust Faith
64 967
BC
7389637
Gallant Dragon
70371 55
Juliana
32 521 BC
7372892
OB0
71 739
Kashee
King William
7330234
42 236 BC
Kiwi Arrow
7909865
26 191 BC
Manila Transporter 7533018
67
624BC
Marmara S
7021
302 58 785 BC
Me1 Gui Hai
BC
7002306
21 508
7343059
BC
Melete
35 516
8 0 1 5726
BC
Mineral Diamond
75 330
BC
36 330
7233723
Mineral Star
7039452
51 506
BC
Orient Pioneer
55 084
Pacific S
71 17474
OB0
Pan Dynasty
BC
21 567
6902951
Pankar Indomitable7205740
39 219 BC
oc
Pasithea
7045607
80 225

Apr-90
Feb-90
15-Mar-90
D=-90
18-Sep-90
Aug-90
Aug-91
22-Mar-90
May-91
Aug-91
Jan-90
21 -Jaw91
23-Dec-90
27-Nov-91
23-0ct-90
Nov-90
May-91
Aug-91
30-Apr-91
7-JuI-91
D=-91
Nov-90
24-Aug-91
17-Apr-91
25-Jan-90
7-Jan-90
Jun-90
4-Oct-89
Jan-91
4-Aug-90

1968
1970
1967
1972
1973
1981
1973
1969
1975
1970
1971
1967
1973
1973
1976
1971
1973
1974
1981
1976
1970
1969
1975
1982
1973
1971
1971
1968
1971
1971

Lib

port

Cargo

Barytesa
Zhenjiang
New Orleans
Ballast
Ningbo
San Nicolas
Iron ore
Dampier
Gijon
Port Cartier
Europoort
Iron ore
Nor
Iron ore
Lib
Kawasaki
Huasco
Chiba
Tubarao
Iron ore
Nor
Angra dos ReisOil
Ras Tanura
Lib
Iron ore
Bremerhaven
Krs
Narvik
Nor
Tubarao
SE Asia
Iron ore
Tubarao
SE Asia
Iron ore
Cht
Iron ore
Ponta do
Bremerhaven
Ubu
CYP
Iron ore
I nd
Motmugao
Geno
Potash
Vancouver
Pan
SE Asia
Iron ore
G rc
Puerto Ordaz
Bremen
Iron ore
Pan
Tubarao
Kakogawa
Dutch Harbour
lnchon
Ballast
CYP
Iron ore
Rotterdam
Port Hedland
CYP
G bi
Hampton Roads Ballast
Setubal
Steel
Bah
Taichung
Cape Town
Port Talbot
Iron ore
Phi
Dampier
Iron ore
Trk
Sepetiba Terminal Las Palmas
New Orleans
Bauxite
Chr
Zhenjiang
G rc
Dampier
Port Talbot
Iron ore
Iron ore
Ymuiden
HKg Dampier
Coal
Hampton
Roads
Belem
CYP
Lib
Tubarao
Kaohsiung
Iron ore
Oil
Trk
United Emirates
Japan
Phosphate
Krs
Tampa
Kwangyang
G rc
Ironore
Polt Cartier
Oxelosund
Ironore
G rc Walcott
Port
Wakayarna
Mta
Pan

(dW
55118
41 300
94 532
122 544
135 466
140 832
248 604
78 571
154 489
1 4 0 440
85 180
53 346
38 250
65 533
123 126
65 455
138 673
79 304
38 695
115 960
121 552
37 326
72 063
141 028
66 350
108 504
103 480
36 650

77 996
155 407

3

93

8

TABLE 1.1
Size

BULK SHIPS EXPERIENCING STRUCTURAL FAILURE,OCTOBER 1989 TO DECEMBER 1991 (CONT.)

Destination Departure
Name
Lloyds no.
Petingo
Protektor
Pythia
Rokko San
Rollon
Salvia
Scandinavian Pine
Shensi
Shou An Hai
Silirnna
Snestad
Sonata
Starfish
Tao Yuan Hai
Theanoula
Tribulus
Vallabhbai Pate1
Vasso
Vulca
Walter Leonhardt
Note
BC
dwt
GRT
OB0
OG

Year

Failure

GRT
date Type

6702662
38 997 BC
6704957
43 218
BC
64 355
7235343
BC
7118753
71 877
OB0
BC
7301764
38 611
7027386
82 014 OB0
7409023
34 157
Wood
7522320
88 675
BC
7340980
62 811
BC
32 508
7525944
BC
71241 80 28 656 BC
6829719
25 597
BC
7007100
BC
28 147
7389675
64 920
BC
7341336
35 1 0 0
BC
68 619
7917850
BC
7391563
62 563
CB0
6801705
34591
BC
19 699 BC
6814049
6608725
23 570
BC

9-Jut-90
1 l-Jan-91
Oct-90
15-Jun-90
Jan-91
9-Feb-91
Jan-90
15-May-90
Mar-91
12-May-90
Sep-90
13-Nov-91
8-Apr-91
23-May-90
Apr-91
Feb-90
Jan-90
4-Apr-91
31-Dec-89
18-Feb-90

built

port
Flag

port

1967
1967
1973
1971
1972
1970
1976
1977
1974
1978
1972
1969
1970
1977
1974
1981
1977
1967
1968
1966

Van
Sng
Cyp
Pan
Grc
Krs
Nor
Pan
Chr
Lib
Nor
Pan
Pan
Chr
Cyp
loM
Ind
Bah
SVC
Cyp

Saldanha Bay
Port Cartier
Port Walcott
Unknown
Tubarao
Huasco
Unknown
Dampier
Hampton Roads
Morrnugao
Cape Town
Kirkenes
Port Walcott
Puerto Ordaz
Trombetas
Seven Islands
Tubarao
Saldanha Bay
New York
Tampa

Cargo
China
Oxelosund

Dunkirk
Richards Bay
Las Palmas
Pohang
Unknown
Yokohama
St Michael's
Italy
Barcelona
Bremerhaven
Swinoujscie
Port Kembla
New Orleans
Rotterdam
Mizushima
China
Busan
Antwerp

(dW

Ironore
80 580
Ironore
80 185
Ironore
120 143
Ballast
150 900
Ironore
67 826
Ironore
153 256
Unknown
41 203
Ironore
169 999
Grain
119500
Ironore
69 165
Grain
62 503
Ironore79681
Ironore
56 277
Ironore
122 734
Bauxite
72
063
Ironore
127 907
Ironore
113 925
Ironore
57 181
Scrap steel 42 245
Phosphate 42 805

Flag abbreviations are explained
in table 1.2.

Bulk carrier

Deadweight tonnes
Gross registered tonnage
Oil or bulk ore carrier
Ore
carrier
b
Wood Woodchip carrier
a. Sulphate of barium.
!%L
Sources ABS (1 992);AMSA (1 992);DTC (1 990,1991 a, 1991 b, 1991c, 1991d); Lloyds
ofLondon (1 992a, 1992b, 1992c, 1993b); Lloyd's Register
G.
of Shipping (1992b,1992~);
NKK (1992).
%

8

p
0

-

BTCE Report 85
TABLE 1.2 FLAG

ABBREVIATIONS

Flag code

Flag code Country

Countw

~~

Bah

Bahamas

Chr

China

Sng
Cht

South
Korea
Krs
Lib

Liberia

Singapore
Taiwan

Mta

Malta

Nor

Norway

CYP
Gbi

Cyprus
United Kingdom

Pan

Panama

Phi

Philippines

G rc

Greece

svc

St Vincent

Trk

Turkey
Vanuatu

Isle
loM
of

Peoples

Republic

Man

Ind

India

HKg

Hong
Kong

Source Lloyd’s Register Shipping
of
(1992b).

44

Van

in the majority of reported ship losses to structural due failures. The events that started the failure may have processoccurred years earlier. Most engineers agree that beam failure is the most likely mechanism causing ships to break in two. Figure 11.11. Shell platingis the steel plating forming the outer side of the and hull. Bulkheads are 'the vertical partition walls which subdivide the interior of a ship into compartments' (D'Arcangelo 1969. Survivors reported that their ships ibroke n two and sank rapidly.3). or has largein the cracks side shell.1 shows the major factors which can contribute to ship structural fai MODES OF FAILURE Ferguson (1991.bottom 2.p. when the length ofwave is roughly equivalentto the ship's length.p.Figure shows extreme hogging and sagging conditions. 7-9) and Nippon Kaiji Kyokai (NKK 1992.Over 50 per cent of the ship structural failures inanalysed these studies were lost. 45 the . 594).A ship which is missing large areas of side torsional stresses due plating. 1. close to the bulkhead (see 11 figure .l APPENDIX II ASPECTS OF STRUCTURAL FAILURE Bulk ship failures are normally the aculmination long slow ofprocess involving a combination of events and circumstances throughout a ship's working life. normally in heavy 2 seas when the ship isin an extreme hogging or sagging condition. the final with 'the loss of side shell platiinng'the cargo holds' (Ferguson stages started 1991.Beam failure often involves some to twisting. The difference between the upward buoyancy forces on full hold in ship hold and downward forces on the acts like scissors slicing the two.The only alternative mechanism is shear failure.pp.appendix 1). I) both agree that. is particularly tovulnerable torsional stresses caused by crossing seas. Shear failure normally occurs at bulkheads2when one hold becomes the adjacent hold overloaded through flooding or poorly distributedwith cargo nearly empty. Beam failure occurs to a ship when i t fails through bending.

CORROSION Corrosion occurswhen a structure reacts chemicallywith its surrounding environment resulting in a deterioration of the structure.1 Flow chart of the factors contributing to ship failure Shear failure is likelyto occur when the hull beam3 cross-sectional area has as been reduced by severe corrosion. However. corrosion. particularlyin the wind and waterline5 region (D’Arcangelo NKK (1992)has documented corrosion 2. Traditionally ships were designed with extra thickness to allow for0.BTCE Report 85 BTCE wind and MODES OF FAILURE Beam failure Shear failure DIRECT CAUSES OF FAILURE Figure 11. 3.15 mm per yearto be lost to corrosion on all steel surfaces over the life of the vessel.8mm of over a two-year period. 5.will iwhich n turn lead to beam failure (AMSA 1993). Failure is usually sudden.07 to 0. Damage to the hold structure. such cracking. such high overload forces usually cause localised failures such a loss of side shell plating. can also increase shear stresses. not ‘frothy’ white water. bulkheads) which is considered in its entiret as y beam. Solid unbroken waves on deck. However. The wind and waterline region is the area of hull thebetween the lightship waterline and the loaded waterline. a 4. The hull beam is the structural body of the ship (including shelldecks plating. which is well over traditional expectations. 46 . pitting and abrasion can leadto much higher rates of 1969). and with extreme overloading due to green watefl on the would probably occur weather deck or major flooding of the holds. andframes.

Figure 11.Appendix II BTCE A W B B Sagging W B W Weight of vessel B Bouyancy forces Figure 11.2 Hogging Hogging and sagging BTCE I Effect on girder I Source Bendall (1 993).3 Shear stress 47 .

where barrier systems such aspaint are employed. plating etc. then the over the ship’s If the life. . anydamaged area of the barrier will result in corrosion.The carriage of iron ore is ‘most the severe cargo in terms of loading on the ship’ (Ferguson 1991). classification societies require scantlings6 ofnew ships to allow for about 20 per cent corrosion. They are based on the likely between interval scheduled maintenance dockings expected at the time of construction. As a result many older ships’ original protection systems are unlikelyto be functioning properly towards the ofend the periods between dockings. corrosion prevention system rate of corrosion can increase considerably. Commodity The type of commodity andthe way a cargo is shipped can also resultin structural failure. Age A ship startsto corrode from the time theis created steel in the steelmill. The available data that can best represent these factors age. Instead.Parker 1992). are commodity carried. Protection systems on older ships were for designed the shorter periods between docking many and have not been updated. Scantlings 48 are the dimensions a ships of frames. Docking every two years was once the normbut every four years is now common. girders. and continues to corrode throughoutits entire life. Dense cargoes (such as iron ore).Ships are designed built and to counter corrosion. The speed and degree of deterioration depend on: ship maintenance and corrosion protection. Irrespective of the 6. A ship’s corrosion prevention system includes control systems such as paint schemes and cathodic protection. and methods and procedures inused loading and unloading cargo. In particular. flagandstate type of cargo carried in the past. Steel is requiredto be replacedwhen this allowance has been used up (AMSA 1993. type of cargo carried. However. age is the best available variable to represent the possible extent of corrosion. the interval between regular planned dockings hasto tended increase. Most corrosion prevention systems are atdesigned the time of building the ship. as well as design allowances for wastage is not maintained. classification society. as underwater paint schemes have improved. Because corrosion occurs progressively over time.BTCE Repolt 85 Present design rules do not includea corrosion rate.

49 . Figure 11. In certain high temperatures can lead to hull degradation. Temperature. hence the cargo does not exert any direct loads onto the side shell. pelletised oreand coal cargoes.Appendix I/ BTCE Dense carao Non-dense cargo 1 Typical cross-section of hold Forces diagram of side shell Source NKK (1992). Davies (1988)notes the effect of ores coming straight out of furnaces and being loaded at well above the recommended temperature of 65°C.4)(figure . This is no force.4 Side shell forces exerted by dense and nondense cargo loading method used.Cargo loaded ata high temperature cause can damage to the hull structure by affecting the metallurgical properties of the steel. the structure itself. iron ore cargoes do not occupy a large volume of thehull. other than that provided by increases thehull stresses since there to counteract the external hydrostatic pressures11.

5 Structural cross-section of a typical bulk carrier Coal. When a ship 'is loaded with lighter cargoes such as 50 .5).BTCE Report 85 Cargo hold Section 'AA' Source NKK (1992). Cargo abrasion also promotes corrosion. Abrasion. which are cooled by the lower temperature of the sea water outside hull.I f the paint system is breached. the The acid then trickles down the side shell frames causing localised corrosion where the frame is welded to the side shell (figure 11. Studies by bothNKK (1992)and Lloyd's Register of Shipping(1992a) indicate that high sulphur coals can initiate corrosion.with cracks appearing within six months of carrying coal. Figure 11. corrosion sets in. Sulphur compounds from the coal and water vapour combine to form acid whichcondenses onthe internal surface of the side shell structures.

owning from 1 to 44 ships each. Data on past owners. or the number of a past ship. The most common stress cycles affecting the ship’s hull are those dueto passing waves. depending on space of a year. the side shell and side in direct frames contact arewith the cargo 11. owners were of also not available.Appendix II l coal or grain. Stress cycles for ships vary from the long-period cyclesloading caused by ship and unloading to the very rapid cycles bycaused machinery vibration. There is no statistically satisfactory a large.4). influence a ship’s seaworthiness and structural integrity. It has been claimed that poor quality ships migrate from traditional flag states and classification societies to ‘flagsof convenience’ and to second-rate registers.4).An abrasive cargo can remove both the paint and (see figure of corrosion. in ship Unfortunately i t was not possibleto study the effect of different owners failures.a ship could experience the time at sea and the wavelength of in. Unvaried. although ultimately the owner is responsible for the condition of the ship. 51 . These cycles occur approximately every 12 to 24 seconds depending on the sea conditions (British Maritime Technology 1986). fluctuating or interrupted loads. because owners with safety and maintenance inspection standards are able to select societies that match their approachto maintenance and the conditionof their ships (HORSCOTCI1992a). alternative measure of owners’ to approach market forces since Classification societies and flag states are both to subject within each group there is competition for ship registrations.Of the approximately 3000 ships in the database there762were different owners. Fatigue failure occurs well below the stress levels at which failure would occur under monotonic7 loading. Ironically iron ore rarely causes any mechanical damage to the protective coating because it normally doesnot reach the side frames (NKK 1992)(figure 11. non-changing load. someTo degree flag state and classification society provide an maintenance. In the up to 3 million cycles. any evide Classification societies and flag states Flag states and classification societies.As the number of cycles or fluctuations ch fatigue failure is likelyto occur is increases. through their respective regulations. diverse group given the relatively numbersmall method to analyse such of failures. the critical stress levelwhiat reduced. theSince seas most waves it operates 7. FATIGUE Fatigue is a phenomenon that leadsto failure under conditions of repeated. The rate at which fatigue normally develops depends upon the type and magnitude of these various stress cycles.

-. 9. For design purposes. Mild steel8 members with low stresseswill experience fatigue failure once they have corroded sufficiently to increase the stress endured during cyclic fluctuations to levels at which fatigue failurebecomes probable.However.h i t from endurance limit. gradually getting lower and lower. the magnitude of the stresswillcycles be small. 52 . Likewise.BTCE Report 85 + BTCE \ \ \ \Typical U) mild steel fatigue curve \ v) 92 Endurance limit typical for mild v) .6 1 o5 etc 1 06 1 o7 Cycles of stress polymer curve \ . particularly ferrousg and titanium alloys.1-5.2. Fatigue failure usually starts with micro-cracks whichgrow over timeuntil the stresses within the structure increase to such an extent that the structure fails (Blodgett 1975. polymers and composites. There is no fatigue limit for most other metals.-.6). because of the large number of cycles. fatigue cracks provide an excellent environment for corrosion to develop.----- Endurance stre_ngt_h typical for polymers " " " " " " " " Typical fatigue 1 oo Typical fatigue curves encountered are not large and their dowavelengths not correspond to the ship's length. (Often. The allowable stress for these materials continuesto fall beyond 108 cycles. the stress at which fatigue failure occurs reaches a minimum at l@ cycles and falls no further. This stress is called the fatigue endurance or limit (see figure 11. the stress that allows strength failure-free operation 107 for cr 108cycles is often called the fatigue to distinguis. 2. their effect be cannot ignored. in expressing the data the terms 8. For steels whereuts is the less than or toequal 1. The steel most commonly used in structure. I t is very hard to separate corrosion failureand fatigue failurebecause they usually occur together. In some materials.4GPa.Sl).f - steel m LL 1 o4 Figure 11.pp.

for torsional loads. Thus. 1975.1-5). 53 . (Sharpe All these influences reduce the ofstrength the ship’s structure. Fatigue strength is also by: reduced Surface condition. The endurance limit or fatigue strength quoted for commercial is usually steelsfor bending using a polished steel round (Shape bar 1989. B3/55-B3/56). 2. the endurance limit is further reduced. The endurance limit for hot rollemdild steel is approximately per cent of the endurance limita polished for mild steel bar. The extent of the which reduction can be sufficient for fatigue a major toproblem be for structures were considered to be fatigue-free.Appendix I1 endurance limit and fatigue strengtht0 are interchanged it and is left to the reader to determine which is appropriate.pp. the lifatigue mit for endurance in old ships is not 210 the MPa for which they were corroded steel structures designed but may actually be less60than MPa. For example. age is a reasonable proxy to measure the fatigue process.” For shipsbuilt in mild steel.58(changefor torsional load) X 0. For corroded steels in salt water. ships were considered tonot suffer from fatigue since they were primarily constructed out of mild steels.for axial loads. B3/55).6shows failure stresses according various materials. 10. In designing structuresto withstand fatigue. Butin reality the endurance limit of mild steel ship structures is aactually lot lower.5(conoded surface) X corrections for notches etc.p. Fatigue can be delayed by maintenance activity to reduce the detrimental effect but every ship’s structure will eventually fail. with corrosion. and 58 per cent of the bending endurance limit.which is thesame designed so the stress (in this case the endurance limit) at 106 level of stress as the fatigue strength cycles for bending loads (Blodgett p. tearsp. as of poor surface condition. Historically. brackets. 50 from per cent d steel. The endurance limitsor fatigue strengths quoted for axial and torsional loads are lower than for bending loads: 85 per cent of the bending endurance limit. and themild steel structures were did not exceed 210 MPa (30 000 psi). Normal bending endurance limit X 0. B3/56).These can occurin places such as frames. there are three major types of All three act upon a fluctuating loadsto consider: axial.down to 10 per cent for the strongest of the endurance limit mil for high tensile steels (Sharpe 1989. Notches and sfress concentrations. ship. repairs. fatigue failure becomes a real possibility as they age. bending and torsional. cracks and 1989.) to the number of load cycles for Figure 11. 11. The normal endurance limit quoted is for a polished steel 80 round bar.

the on available proxy was flag state. such as a ship encountering bad weatheror beginningto break up. Ship designers presumably were aware well that ships constructed from high tensile steels would eventually be at high risk of fatigue failure and so have a finite life. The stresses imposedby waves are discussed in the fatigue section above. Proxies thatmay be usedto measure extreme heavy areas. Further work may be warranted to determine the importanceof training in helping save ships that suffer structural damage during a voyage. and wavelength relationships.New buildings today have up to 90 per cent of their hull constructed in high tensile steels (Parker 1992). It has been argued that poorly trained crews lose ships whilegood crews save ships (HORSCOTCI 1992a). increasingly they are being used more for highly the stressedmembers of the hull in large ships to keep down weight and to reduce the use of extremely thick members (DArcangelo 1969). High tensile steelsdo not have a fatigue endurance limit (Sharpe 1989. seas are route.BTCE Report 85 High tensile steels Although traditionally only small quantities of high tensilewere steels usedin ship building. many experienced heavy weather before and during failure(NKK 1992. During a crisis. p. 63/55). 54 se . Other things being equal. But thisa poor is proxy since crew nationality is not necessarily related to flag. season. p. A strake isa course orrow of shell. CREW TRAINING Poorly trainedcrews andlack of a common language between crew.N K K 1992).deck. and the bilge shear strakes12). isthere a good deal of merit in this argument. Ferguson 1991. crew training and communicationmay well be critical to the safety aof ship. HEAVY SEAS Heavy seas can act as a catalyst for failure. Beresford & Dobson 12. Since there were no data on crew nationality or training standards. officers and land based workersmay result in inefficient ship operations (HORSCOTCI 1992b). Both Lloyd’s and N K K note the concurrence of heavy seas failure and (Lloyd’s Register of Shipping 1992a. bad Of the 50 world fleet failures examined. bulkhead or other plating (DArcangelo 1969. primarily around the bottom and weather deck plating an of side plating adjoining these regions (that is. 600).

Appendix II 55 .

BTCE Repod 85 1989. 56 . i f shear failure is the primary cause then size would not be important.7). Wave-induced load and hammering in rough seas cantrigger cracks and detachments of hold frames. especially those weakened by corrosion and wastage (Lloyd’s Register of Shipping1992a. Loading practices Three loading issues may that affect the risk of failure are: 13.5). According to the N S W Coal Association (1992). 1991b.3). and in the both Cargo loading and unloading procedures can affect ship structures. Deadweight was used as a proxy for size. and there should a strong be correlationwith alternate hold loading (seefigure ll. the loss of the Daeyang Honey in late 1992 occurred duringa typhoon 300 nautical miles east of the Philippines. many papers on structural failure aplaced heavy emphasis ship on loading and unloading practices. when the length ofwave is roughly equivalent to the ship’s length11. Most of the Australian departure failures occurredin the Of the ships which sank. loads exerted on hull the and in damage caused to the hull.N K K 1992). Southern IndianOcean. loading and unloading. The majority of the ships failed in well known bad weatherzones (figure 11. LOADING A N D UNLOADING Recent research into structural failure has focused on the condition of th In the past. As discussed above. Although 150 000 expected size and is not ships in the size ranges80 000 to 100 000 dwt and 100 000 to dwt13 were found to have failure rates significantly higher than lll. ships be Alternatively. under modes of failure.If beam failure is the primary failure mechanism it could be expected that there would a strong correlation between the size of failed and route.1991a. the riskbeam of failure is highin heavy seas when a ship isin extreme hogging or sagging conditions. four experienced gale-force witwinds h rough seas and swell prior to failure. Ferguson (1991) ways the cargo can overloadhullthe structure and N K K (1992)both detail the and cause damage. The statistical evidence is thus consistent conclusive. Bad weather can place extra stress on areas already weakened by corrosion and fatigue.2(see figure ).DTC 1990. 1991c.For example. on cargo characteristics. where storms are prevalent. 1991d).IMO research during the 1980s into causes of ship losses concentrated trimming procedures.but not with size of the ship.there is also a strong correlationbetween ship (appendix table with beam failure but route.

However. if bad loading practices are a problem. Australia Saldanha Bay. cargo loaded too for ballastpumps to keep up.3 Port Walcott. Source BTCE estimates based on Lloyd’s Maritime Information Services data.0 Other 15.and poor adherence to loading procedures. including trimming.3 0.4 0. Table 11. 57 fa . fas or example. Expected failures are calculated by assuming that risk of failure is proportional number to theof voyages.0 0. Includesx2 for successful voyages. bad loading practices. although a 2 analysis of the Lloyd’s voyage showeddata a strong statistical relationship between port of departure14 and failures.0 Total ?3 4 1. 2.8 1. Bad loading practice.7 Sepetiba.2 1. the relationship became marginal when the analysiswas extended to allow for the effect commodity of carried.5at the 0. In the absence of data on port of loading. 11. I f bad loading practices (for example. loading speed. South Africa 1. for iron ore voyages which round Cape the of Good Hope.1.0 0.1 results show that. port of was departure used for port of loading. 3.05level of significance a.6a Notes 1. DEPARTURE Eipected Port failures V O Y A G ROUNDING ES PORT BY OF Actual failures Dampier. can This damage may only become apparent when the vessel is undertakinga voyage whenit is fully 14. hull Loading or unloading operations can damage cause or weakening of the structure which is not apparent go andunnoticed.0 1. Australia 2. Figures may not add to totals due to rounding.Brazil 3. When route as well as commodity werein table used.Appendix I1 TABLE 11.2 Port Hedland.3 4 1 15 3. and homogeneous versus alternate hold loading.5 1 2 2 1 0. then i t is a universal problem.7 Tubarao. Significant x2 for 6 degrees of freedom is 12. and one not that can be attributed to particular ports. Brazil 3.1 x2 ANALYSIS OF IRON O R E THE C A P EGOOD O FHOPE. and not keeping to loading plans) at the port of loading are a major cause of failure then there should be a statistical link between ships that faila few and particular ports of loading. Australia 1. the relationship was found to be insignificant.

hold floors are specifically reinforced to accommodate the high stress produced by falling cargo. 58 . The ship may not be strong enough for some loading practices (Lloyd’s Register of Shipping 1992a). and encounters bad weather.and ship a loaded with iron ore will reach its mass capacity long before it reaches its volume capacity. but there were no data damage from past ports of call could available to analyse this effect. Both homogeneous and alternate hold loading methods were used (49 per centhomogeneous. particularlywith dense cargoes. Department of Transport and Communications ship failure reports have found that both types of cargo distribution have been used amongst ships which have failed. It is for dense cargoes such as iron ore that alternate hold loading and its variants are often used (Lloyd’s Registerof Shipping 1992a). the Not all ships are designed to be operated using all of theabove methods. Although this practice may cause some localised high stresses on the hold floor for a short period until the floor is covered by cargo. A third hypothesis is that the order of loading the holds and the amount placed in each hold are important determinants of structural damage. Very little information is available to analyse loading practices. Low density cargoes such asgrains and coal are usually loaded homogeneously to maximise the of use the ship’s volume capacity. A mismatch between designand loading practice may therefore resultin unwarranted stresses. Hence be the true culprit.Another hypothesis relates ship failures to high speed of loading. 51 per cent alternate) during the threemonths.A X* analysis did not demonstrate any link between rated speed of loaderand ship failure. Iron ore loading rates are intypically the range 6000 to 16 000 tonnes per hour (Lloyd’s of London 1993a). alternate holds (distributed odd-numbered over holds). statistically significant Homogeneous versus alternate loading. Loading speed. Some proponents of loading practices as a cause of structural failure note the practice of dropping cargo from heights in excess of 20 metres onto the hold floor.BTCE Report 85 loaded. or combinations or variations ofabove. Cargo may be loaded in several ways (Lloyd’s Register of Shipping 1992a): homogeneously (evenly distributed over all the holds). Iron orea very has high specific gravity. A M S A (1992)collected some load configuration data on ships leaving Western Australian iron ore ports for threemonths in 1991.

2 CAUSES OF FAILURE A N D PROXIES USED IN THE ANALYSIS Causes Proxv of failure Corrosion and Fatigue Crew Heavy seas Loading Unloading poor maintenance Age. PROXIES FOR FACTORS Table 11.ports Previous commodities poand rts 59 . Unfortunately.flag and classification society Age Flag and classification society Route. If this were the only reason for failure then there should be some correlation with particular previous types of cargo carried. TABLE 11. deadweight. the data available did not allow the effect of previous ports or cargoes to be tested.commodity.Appendix I1 l Unloadingpractices Poor unloading practices have been hypothesisedamong as the beingreasons why ships suffer structural damage. and sea regions Commodity.2 summarises the CAUSING FAILURE proxies used in the BTCE analysis. or ports visited.

07. cent level from successful voyages (see text).we assume that a 25-year-old ship is just as likely during to faialvoyage as a new ship on its maiden voyage (age has no influence). First.7 0 2 5 12 11 6. the f failurisk re being analysed (ship age) assumed is to have no influence on othe (the null hypothesis). 2.8 35. Includes contribution Source BTCE estimates based on Lloyd's Mariiime Information Services data. it is important to establish if the relationship has statistical significance. a.APPENDIX 111 RELATIONSHIPS BETWEEN VARIABLES AFFECTING SHIP FAILURE Although a relationship between bulk ship failures and various ship and voyage characteristics may be suspected.8 5. In the X* analysis we compare the number of failures null hypothesis) with the actual expected where no relationship exists (the TABLE 111. As anillustrative example. On this basis the expected number of failuresin each age group is assumed to be proportional tonumber the of voyages the ships in this age group have undertaken.1is.5 54.7 0. A chi-square(X2)analysis is used in this appendix to test for this significance.1 we examine the hypothesis thata relationship exists between ship age and ship structural failures. 61 . in table 111.2 0.1 2.2 0.3a ~~~ Notes 1.0 Actual Eipecfed Voyagesfailures failures 435 597 4.8 7.1 5.1 SHIP AGE Age (Years) 4 554 10 335 5 413 5 781 2 079 0-4 5-9 10-14 15-19 20-24 224 Total 30. Numbers may not add to totals due to rounding.5 0 28 30 ~ ~ 2 4. inThat table 111.8 10. Significantx2 value for 5 degrees of freedom and 5 per of significance is 11.

111.84.7 30.07).Logit methodology is discussed in appendix V.0 0.66 16 14 3.C Total ~~ classification 9 860 18 737 597 28 10.0 30 4.3 8. 62 .1 1. value of I f the totalsum of thex2 exceeds the significant x2 value thenw e are satisfied of this result(54.1is less than the significant 7. Includes contribution from successful voyages (see text).6 30. If the pattern of actual failures is similar to the pattern of expected failures the totalsum of thex2 will be less than the significant x2 value andw e have 111.0 ~ Notes 1.81).The number of failures aretoo few to allow valid results to be obtained from two-way classifications. Table result (6.4 6. Numbers m a y not add to totals due to rounding. Significant x2 value for 3 degrees of freedom and 5 per 7. 2. on Lloyds Maritime Information Source BTCE estimates based Services data.0 30 6. the number of actual and expected successful voyages should be compared as well as the actual and expected unsuccessful voyages.34 19. In calculating thex2 score in a contingency table.BTCE Report 85 number of failures recorded.4 2.1a 10. of The tables in this appendix (usually referredto as contingency tables) are limited toone-way classifications.81isand for 1 degree of cent level of significance freedom is 3. Multi-variate classifications are effectively carried out through use of ilogit analysis n (chapter 3).2 FLAG Actual Expected 2 failures failures Voyages Flag Four-wayflag D C B A Total 597 classification 9 860 7 853 5 161 5 723 28 16 4 2 3.1 Tableis an example is greater than the significant 1value 1. The x2 score is a measure of the difference between the expected and actual failures. in this analysis the comparison of actual and expected successful voyages generallylittleadded to the x2 score.7a 8 Two-wayflag D A. However.2 5.3 that a significant relationship exists. a. B.2 is an exampleof this proved that no significant relationship exists.1 0. TABLE 111.

All but one of the tables are based on voyage and ship data purchased from Lloyd's Maritime Information Services.6a Notes 1.49.8 27. Sources Fearnleys (1992a.l 0.81. Includes contribution from successful voyages (see text).Appendix 111 TABLE 111. and 5 per a. TABLE 111. Numbers may not add to totalsdue to rounding. Source BTCE estimates based on Lloyd's Maritime Information Services data.3 0. 63 .2 7. The one exception is table 111. Includes contribution from successful voyages (see text).1 0.3 CLASSIFICATION SOCIETY Society category fipecfed Voyages failures failures D C B A 6 850 7 239 6 808 7 700 Total 597 28 10 Actual 7. Significant of significance is 9.3 and 0. X* value for 4 degrees of freedom and 5 per 2.1 8. 1992b).1 6 7 7 30.3 Notes 1.1 1 .0 0.4. SignificantX* value for 3 degrees of freedom cent level of significance is 7.0 15 911 46. Lloyds of Register Shipping (1 992b). 2. cent level a. Numbers may not add to totals due to rounding.which is 1992b)data.8 3.0 46 77. which allows a more disaggregated based on Fearnleys (1992a.5 2.4 21.0 3 3 4 0.0 30 2 1 . analysis of commodities.6 7.4 COMMODITY CARRIED Average fipecfed Actual failures failures Commodity voyages ore Iron grain and Coal Bauxite alumina Phosphate Other Total 2 3 373 9 332 9.4a 3 55.0 33 1 193 974 1 039 3.

Numbers may not add to totals due to rounding.o 0.3 25.9 13. Significant X* value for 4 degrees of freedom and 5 per cent level of significance 9.0 597 13 520 9 174 589 3 256 2 058 28 14.BTCE Report 85 TABLE 111.0 9.8 0.9 3 6 3 3 2 2 7 7 41.6 5. Includes contribution from successful voyages (see text).7 ROUTE Route Brazil-Asia Brazil-Europe South Africa-Asia India-Mediterranean WA-Japan WA-North Europe Other 258.7a 3 failures failures 30-50 5&80 8&100 100-1 50 >l50 30 Total 30.4 2.7 1.8 2.0 1. Significant x2 value for 1 degrees of freedom and 5 per cent level of significance is 3. 2. 64 .5 15.84.9 45. TABLE 111. 2. Source BTCE estimates based on Lloyd’s Mariiime Information Services data. 2.3 Notes 1. Significant x2 value for 6 degrees of freedom and 5 per cent level of significance is 12.6 0. Includes contribution from successful voyages (see text).1 1 . Includes contribution from successful voyages (see text).3 0.6 3. a.5 11.2 0.2 4 9 3 11 3 7.59.8 0. TABLE 111. Numbers may not addto totals due to rounding.5 SIZE OF SHIP Expected Size (‘000 Voyages dwt) 33. Source BTCE estimates based on Lloyd’s Maritime Information Services data.is a.6 V O Y A G E Weather W E A T H ECONDITIONS R Voyages Good Bad Total Expected Actual failures failures 23 7 30 14 16 30 21 913 6 684 28 597 2 3. Numbers may not add to totals due to rounding.5a 30 Total 30 597 Notes Expected Actual Voyages failuresfailures 646 1 172 771 78 901 289 24 740 0.49.8 28 1.1a Notes 1. a.2 9.2 147.2 16. Source BTCE estimates based on Lloyd’s Maritime Information Services data.

AGE Figure IV.However.l illustrates theage distribution ofbulk ships departing from Australia on major routes. 65 . The routes from South Africa to Asia and Brazil to Asia shown (not in figure lV.11 shows that these routes also have comparable elevated failure patterns. namely 24 and 33 p e r cent. commodity.1)have comparable proportions of ships older than 15 years to the Western Australia to Japan and WesternAustralia to North Europe routes. the Western Australia to Asia route (mainly Chinese destinations) has an even higher proportion of old ships(57 per cent) but recorded no failures during the study period.l).APPENDIX IV CHARACTERISTICS OF SHIPS DEPARTING FROM AUSTRALIAN BULK PORTS There is a substantial difference in the number of structural failuresbulk of ships departing from the eastern Australian seaboard and those departing from the western seaboard. Also of interest is whether bulk ships serving Australian ports differ from those serving in other portsbulk exporting countries.l and table IV. Between October 1989 and December 1991 there were nine failures of ships departing from Western Australianbut ports no failures of ships departing from eastern Australian ports. Figure 2.It is clear that Western the Australia to North Europe route (which has higher a failure rate) involves older (31 per ships cent are older to Japan and Asia route (12per cent than 15 years) than the eastern Australia are older than 15 years). The age distributions are significantly differentbetween the different routes (table IV. The analysis here therefore focuses on these characteristics as well as examining in the ship contextsize of the Australian routes.3 suggests that route. The age difference would partly explain the difference in failure rates. flag age andof the ship are the major factors influencing the risk of structural failure. respectively. Chapter. The larger number of failures ofbulk ships departing from Western Australian ports raises the question of whether differences in the ships serving the two coasts might explain the divergence in failure rates.

Source BTCE estimates based on Lloyds Services data.3 x2 22.l Australian routes Western Australia to Asia Maritime Information BTCE TABLE IV. 66 route.o x2 Western Australia to Asia excluding Japan 1 006 172 175 229 107 Actual 183 140 1 1 006 96 21 Expected 167 464 257 21 348 66.2 2111 18925 21.24 h S 2&24 80- 015-19 010-14 0 5 n v 60- m5-9 m04 m S > c 40- .2 53.30. 1 0 .0 0. and ship age Figure IV. Figures may not add to totalsdue to rounding.2 0.-0 r e 20a 01 Eastern Australia to Asia &Japan Western Australia to Japan Western Australia to Nth Europe Note Asia excludes Japan. See appendix 111 for an explanation of x2. . 1 1 .4 8. Source BTCE estimates based on Lloyds Mariiime Information Services data.BTCE Report 85 loo Age (years) 1 m .l AUSTRALIAN R O U T E S M A NSHIP D AGE (vovages) Aae " 5-9 10-14 75-19 Asia Japan including 31 1 20-24 > 24 Total Eastern Australia to 3 276 67 5 Actual 545 1511 837 Western Australia Japan to 901 145 42 1 157 Actual 320 236 901 18 1 86 Expected 41 230 6 150 94.8 x2 170.1 30.6 1 6 . 'Expected'number of voyages is calculated on the basis number that the of voyages by each age group would be proportional number oto f voyages the by ships in that age group on the Eastern Australia to Asia and Japan 2.7 89.5 115. 3.1 41.1 Western Australia to North Europe 77 13 0 289 Actual 71 87 41 289 6 0 27 48 74 Expected133 0.3 Notes 1.

2 illustrates the sizes of bulk ships on the major Australian routes. Ships departing from Western Australiato Japan and to North Europe are significantly larger than ships departing on the other routes IV. Source BTCE estimates based on Lloyd’sMaritime Information Services data.2 Australian BTCE 1. The route from Western Australia to Asia hasa high proportion of small ships. The eastern Australia to Japan route isused by both coaland grain ships._ E E E 20- Q- Eastern Australia to Asia &Japan Western Australia to Japan Western Australia to Nth Europe Western Australia to Asia Note Asia excludes Japan. ships in the size range 80 000 to 150 000 dwt were found to be overrepresented in structural failures. the next largest being theto South Africa Asia routewith 29 per cent.The combined size distributions of to be tend the ships carrying grain and coal from eastern Australian ports would more balanced compared with the distribution of ship sizes on the other routes. Grain ships on Australian trade routes rarely panamax exceed size. However. which also tend to be older ships. the relationship between size and failure Il). The proportion of ships in this size range on the Western Australia to North Europe route (35 per cent)was considerably larger than any on of the other routes.Appendix IV l SIZE Figure IV. (table These ships are mainly large ships carrying iron ore. the 67 . risk is inconclusive (see appendix 100- -E Size (‘000dwt) > l 50 0100-150 m 80-100 [ 50-80 m 30-50 80- kQ v 0) P) 60- m 0 > 5 40C 0 .’ Coal ships are more likely tobe larger thanpanamax size. In appendix Ill.2). They are the maximum sized bulk ships that can Panama transit Canal. Panamax size refers to ships between 50 000 and 80 000 dwt. routes and ship size Figure IV.

The proportions for the South to Africa Asia and Brazil to Asia routes are both 34 per centand the Brazilto Europe route is 37 per cent.The eastern Australian routes have a higher proportion of voyages of ships registeredwith flags with poor casualty records (category D).4 0 127 167 9.9 to 100-150 > 150 Total 165 544 145 3 276 7 45 32.# 592 386 109. FLAG STATE Figure IV.8 901 901 4 322.4 289 289 1 269. 68 .3 shows the flag state of bulk ships employed on the major routes.5 51 50. - Because the two routes with the highest failure rates Western Australiato Japan and to NorthEurope -have relatively low proportions of ships registered with flags with high casualty rates.7.2 Western Australia Actual Expected x2 116.3 45.7 Notes 1. 3. The evidence here suggests that route might a muchbemore important factor than flag.it appears that flag is not as important a factor indicating risk of failure as is often suggested.The Western Australian routesto Japan and North Europe both have higher proportions of voyages of ships registeredwith flags with low casualty rates (category A). Source BTCE estimates based on Lloyds Maritime Information Services data. 98 48 52. Figures may not add to totals due to rounding.2 4 15 7. Excluding Japan.8 1 006 1 006 466.5 North Europe 22 111 71.3 to 80-100 34 103 As. ‘Expected‘number of voyages is calculatedon the basis that number the of voyages by each age group would be proportionalnumber to othe f voyages by ships in thatage group on the Eastern Australia to Asia and Japan route. There is a significant differencebetween the flag distributions on the different routes (see table IV. See appendix I l l for an explanation of x2.9 Western Australia Actual 153 Expected 357 x2 137.3).BTCE Report 85 TABLE IV. a.4 435 40 3 914. 2.6 134 45 179.1 131 13 1 092.2 AUSTRALIAN R O U T E S A SHIP N D SIZE fvovages) Size (‘000) 3&50 50-80 Eastern Australia to Asia Japan including Actual 1 258 1 164 Western Australia to Japan Actual 128 84 Expected 320 346 x2 174.5 247 150 63. The routes from Brazil and South Africa havea much higher proportion of voyages by ships registered with flags with high casualty rates (category D)than the Australian routes.

respectively. 2. evidence relating size 69 . Asia excludes Japan. The Australian iron ore routes are servicedwithbya ships similar proportion of old ships to other comparable international iron ore routes. chapter 2 for the method of allocating flags to categories. the statistical to risk of failure is not conclusive. The larger size of ships one on of theWestern Australian routes was the only factor where an Australian route showed a significant difference that might influence risk of failure. on the Western Australia to Japan and North Europe routes. CONCLUSION Although there are significant differences between the ships serving the vario routes departing from Australian bulk ports.3 Australian routes and flag BTCE compared with 14 and 23 per cent. although they have similar proportions of older ships. Figure IV. and the flag distribution favours the Australian The analysis routes. However. Source BTCE estimates based Lloyds on Maritime Information Services data.as noted above. discussed in appendix I l l found no port effect.Appendix /V Eastern Australia to Asia &Japan Western Australia to Japan Western Australia to Nth Europe l Western Australia to Asia Notes 1. the differences do not suggest that there are any factors unique to Australian ports that contribute to failure. The difference in failure rates between ships departing from eastern and western seaboards can beexplained by differences in age of the ships and the commodities carried.Flag category A has the lowest casualty rate and D category has the highest casualty Seerate.

4 98.0 100690 1 886. Source BTCE estimates based on Lloyds Maritime Information Services data.5 x2 126 Western Australia Actual Expected 332 Western Australia to 503 Actual 183 Expected 2 B C 361 81 0. 'Expected'number of voyages is calculatedon the basis that number theof to the number of voyages by voyages by each age group would be proportional ships in that age group on the Easternto Asia Australia and Japan route. 70 .7 x2 170 categotf 51 289 52 116 0.1 901 183.8 Flag category A has the lowest casualty rate and D has category the highest casualty rate. in conjunction with commodity carried. See appendix 1 1 1 for an explanation x2.0 1. Notes 1. 2. Figures may not add to totals due to rounding.' 150 99. a.020.BTCE Report 85 TABLE IV.3 AUSTRALIAN ROUTES AND FLAG (voyages) Flag D Eastern Australia to Asia including Japan Actual 294 590 1 080 Western Australia toJapan Actual 162 297 Expected 0.7 403 134. of 3. Excluding Japan. and the condition of the ship as m by age and flag state.See chapter 2 for the method of allocating flag states to categories.3 A Total 1 312 3 276 84. b.5 0.3 to North Europe 66 95 289 9.8 42 26 9.9 130 Asid. The evidence supports the conclusion in chapter 3 that weather conditions on particular routes play a major role in structural failure of bulk ships.6 2 120.

then the The logistic is well to suited the analysis of binary response It is data. Chapter2 highlights the main successfully orit is unsuccessful (that it is. the confidence of the data.Also needed is a tool that allows surveyors more likely to fail. The logit technique statistical tool which a ship that fails coded 0. n the curve iused analysis has the form: Pr (Shipfails)=l/(I+e-’) to be estimated where p is a vector of coefficients variables. the approximately by about 3000 ships. There is very 71 . is based on the of all other observations. Collett One method of examining the usefulness of the model isto calculate the confidence limits for the predicted logistics curve. to predict which ships are is needed. However.APPENDIX V LOGIT l ANALYSIS The analysis of structural failures of bulk ships is essentially an analysis of voyages that have one of two outcomes: the voyage is either completed fails). a permits estimation of the probability of an If event occurring is coded 1 and aas ship that successfully completes a voyage is model represents the probability of failure. 10 completed voyages departing from the five origin icountries n the database.l illustrates a typical logistic curve. and X is a vector of explanatory Figure V. That is. each of the ships. may be affected by an inherent characteristic The logit procedure. assumption that each observation is independent 29 OOO voyages in the database were undertaken However. and regression analysis in general. on average. factors influencing ship failure. The theory of logit analysis is described in several books such as (1991). To understand the factors better and their a technique which reflects the binary response interrelationship with each other. The vector p is estimatedby maximum likelihood methods.

can be computed the as square root of ?’V.30.9 m 0. The linear predictor. where ZaI2is the 100(1-a/2)percentile point of the normal distribution. . The logit proceduredoes not allow for the possible correlation between repeated voyages.70) 0.2- 3 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 l l l l l l l l l l l l l l l l l l l l 0 5 10 25 20 15 30 35 40 45 50 Explanatory variable Figure V.p.40.:’B 7 = P’?.6c - 0 E 05‘5 0.... Unfortunately i t is not possible to estimate the magnitude of the effectto confirm or its direction. at least questionable. In theory.6(fi). the effect of correlation between voyages either can In practice.l Typical logistic curve likely to be some correlation between repeated voyages undertaken by the same ship on the same route but at different dates. 1091)as follows. The assumption of independence between observations is.errors.BTCE Report 85 BTCE 10.. e($. is the estimated covariance matrix of the parameter P. although its existence can affect the accuracy of thecomputed sts31dard errors.8 c S> 0. it is more increase or decrease the magnitude of the standard likely to lead to standard errors that are smaller than those calculated taking account of the correlation.? where V. is no there bias in the estimationof the coefficients. CONFIDENCE LIMITS Confidence limits for the main parameters can be calculated (SAS Institute 1989. therefore.ii(fi))]) 72 and l/{l+exp[-(fi-z.. Although the standard errors are affected.estimates The confidence interval around fi is given byfi k za. The upper and lower confidence limits are then given by: l/{l+exp[-(fi+z.~(fi))]} . is estimated by fi = The standard deviation.

held constant.S. The confidence limit for in age that is equivalent to a change in flag is not same the as the the difference were confidence limits for6. Only the variance oftwothe variables and the covariance between them are needed in the calculation of the confidenceThe limits.In the previous section. The model is given by: Pr (Shipfails)= f(X1P2+.) 1 P2 P2 The variance is approximately: - 1 var(8) =T(v.Appendix V Confidence interval for change in age that different flag implies same risk as a The results in chapter 3 indicate that ships registeredwith a flag with high casualty rates are subject to same the risk of failureas ships registered with flags with low casualty rates but which are 5 years older.. The full covariance matrix is therefore inused the calculation.(6).Pl+ X2P2unchanged.~6. Then: Let 6 = PP a 6-e=(-.1-e)=--(~~+ep. +2evI2+e 2 P2 si a. that exactly is. '12 The standard error6(0)is [var (6)] . following is from notes on logit model predictions suppliedby Dr T.) where f(Z)= l / l(l+e-') The aim is to find a confidence interval fora change in X. Method l -Linear Approximation -5. A symmetric 100(1-a/2) per cent confidence interval can then be formed with end points 6 k ~~. That a confidence interval 8for = -p. the confidence limits calculated on the assumption allthat variables can change. 1993). (pers.There are two methods that can be used. compensates for a one unit change in X.. leaving X.. Australian National University comm. obtained where vu is the covariance betweenand from vb . /p2. Breusch of the Department of Statistics.. th all other variables The problem examined here concerns only two wivariables. 73 .

The confidence limits for change the in age to compensate for a change in flag category is 1. although they are only normal here. and 1.2to 8. symmetric about the point estimate Method 2 is preferred by many users because i t is exact if the coefficient estimates are exactly normally distributed. 96-101). The two methodscould be usedin a similar way to estimate confidence intervals for changes in age to compensate for changes in route or commodity.BTCE Report 85 - Method 2 Fieller’s Method The following statement has probablity contenta/2) lOO(1: The end points of the confidence interval for8 values the are defined by the roots of the quadratic equation: The confidence interval will generally notbe 6 = -p1 /p2.7years usingmethod 2. Neither approach is to be trusted i f the results are not roughly similar.7years usingmethod 1. 74 approx .The results using the two methods are sufficiently similar to allow some trust to be placed in them. Both methods are discussed by Collett (1991.pp.6to 9.

Canberra. Brisbane. part 2.Canberra. on Communications and lnfrastructure Nippon Kaiji Kyokai N e w South Wales United Kingdom A B S 1992. unpublished. UK.in Papers of the Australasian Transport Research Forum 1993. HORSCOTCI. A M S A 1992. -1993.H O R S C O T C I Submission no. HORSCOTCI. Communications and lnfrastructure lnquiry into Ship Safety.pp.18.Submission by the American Bureau of Shipping to the House of Representatives Standing Committee on Transport. on bulk ships loading at Western Australia ports. 1993. Data Canberra. & Dobson H.‘Theaging international bulk fleet: implications for Australia’. and Shipping 75 . Communications and lnfrastructure lnquiry into Ship Safety. A C A 1992. Graduate School of Management.HORSCOTCI Submission no. 27.H. Beresford. Submission by the Australian Coal Association to the House of Representatives Standing Committee on Transport. HORSCOTCI. University of Queensland. Bendall. Communicationsand lnfrastructure lnquiry into Ship Safety.H O R S C O T C I Submission no.39. (eds) 1989.vol.REFERENCES ABS ACA AGPS AMSA BTCE DTC HORSCOTCI NKK NSW UK American Bureau Of Shipping Australian Coal Association Australian Government Publishing Service Australian Maritime Safety Authority Bureau of Transport and Communications Economics Department of Transport and Communications House of Representatives Standing Committee Transport.Lloyds Maritime Atlas of World Ports Places.Submission by Australian Maritime Safety Authorify to the House of Representatives Standing Committeeon Transport. 18. 665-81. 16th edn. A. AMSA.Canberra. Colchester.Lloyd’sof London Press.

BTCE Report 85 Blodgett. BTCE 1988.7985 Ferguson (1991). Communication fromthe Commission. Lincoln Arc Welding Foundation. Departmental Investigation into Certain Aspects Concerning the Loss of the Bulk Carrier Mineral Diamond in the South Indian Ocean on 77/78 April 7997. -1991b.Canberra. BTCE Occasional Paper no. Crisp.Daily Commercial News. Canberra.‘Governmentsto blame for “quality gap”. Departmental Investigation into Certain Aspects Concerning the Loss of the Bulk Carrier Manila Transporter in the South Indian Ocean on 7 August 7997. Commission of the European Communities. DTC. 74 March 7990. Canberra. -A Contributory Cause of Loss?. Brussels. Society of Naval Architects and Marine Engineers. -1991d. Collett.Canberra.Arctic Carrier.DTC. M. Report 28.p. D.‘Liferaft discovery suggests bulker was lost in typhoon’. James F.Canberra. 1991a.Middlesex.DTC. 1992.A Common Policy on Safe Seas. D’Arcangelo. -1991c. 24. Davies 1988.Global Wave Statistics.Maritime Operations. 24 February. 6. 76 .P. Ship Design and Construction.Daily Commercial News.W. Commission of the European Communities1993.27 October. 1969. 93.p. cited in DTC 1988.Report 29. D.28 October. Characteristics and Cost Structure of the Bulk Shipping Fleet. Departmental Investigation into Certain Aspects Concerning the Loss of the Bulk Carrier Melete in theSouth IndianOcean on 24 August 7991. -1990. says ONeil’. Modelling Binary Data. Departmental Investigation Loss of the Bulk Carrier Starfish in the into Certain Aspects Concerning the South Indian Ocean on 8 April 7997.A.Cleveland. British Maritime Technology(ed. Chapman &Hall.AGPS.DTC.Canberra. Design of Welded Structures. DTC. 1.British Maritime Technology. Daily Commercial News 1992.Ohio.Canberra.Report on the Preliminary Investigation into Certain Aspects of the Loss of the Philippine FlagSinga Ship Sea in the South Indian Ocean. 0. Report 34.London.)1986. 1991.I.DTC.Report 32. COM(93)66 final.Feltham.New York. 1975.Report on the Preliminary Investigation into Certain Aspects of the Loss ofthe Panamanian Flag ShipAlexandre-Pin theSouth IndianOcean.

1992. World issues. Lloyd‘s Nautical YearBook 7993. Bulk Trades 7997. Gehling. - Ferguson J. Canberra.Casualty Return. Colchester. -1993b.Official Hansard. Canberra. Communicationsand Infrastructure Inquiry into Ship Safety HORSCOTCI Submission no 34. in Bulk Carriers. -1992b.Fearnresearch. 1992.1991. Colchester.UK.Oslo. 4th edn.Submission by the Department of Transportand Communications. (and earlier issues. M. -1992c. 25-27 November 7992.1990 &1989 annual issues. Taunton. ‘Technology relating to bulk carrier safety’.The Lloyd’s Register of Shipping. Lloyd’s Maritime Information Services 1993. UK. HORSCOTCI 1992a. Bulk Carriers The Safety Issues. Colchester.HORSCOTCI. Lloyd’s of London Press. in Maritime Technology 27st Century Conference.UK. 77 l . UK. toHouse the of Representatives Standing Committee on Transporf.Canberra. Colchester. Unpublished data provided from Lloyd‘s Register and Lloyd’s Voyage Recorddatabases. Ocean Passages for the World. Lloyd’s of London Press. Lloyd’s ofLondon Press. 6-monthly). -1993a. Melbourne.1991. 1992. Lloyd’s Maritime Information Services. Colchester. Maritime Policy Division. Norway (and earlier annual). Register of Shipping. Lloyd’s of London Press. Reference: Ship Safety 4 September 7992. University of Melbourne. Colchester. Colchester. Safety Issues. Lloyd’s of London Press. - Lloyd’s Lloyd’s Register of Shipping 1992a. -1992b. U K (various issues). 1992. Hydrographer of theNavy 1987.Ship Division.Lloyd‘s Shipping lndex. Ministry of Defence. London.Norway Fearnleys 1992a. Ships of Shame.References -1992.Fearnresearch. section VI-3.UK.1990 issues. UK. Inquiry into Ship Safetx A G P S . -1992b.‘Bulk carriers the challenge’. -1992b.Lloyd‘s Ports of the World. 1991. UK. World Bulk Fleet January 7997. R.Oslo.Lloyd‘s List Daily. Lloyd’s of London 1992a. Lloyd’s Nautical Year Book 7992.

19918. 4. 1. 2. no.W. 25-27 November 7992. 32.Lloyd‘s Shipping Economist. T. University of Melbourne. ‘P & I insurance: its current dilemma and future role’. Stephens. Oslo (cited in Gehling 1 992). B.Daily Commercial News. Melbourne.p.BTCE Report 85 -1992c.SAS/STAT User’S Guide. 1 992. no.‘Ratesrise a must to improve shipping’.12. Parker. 1992.1992. in Maritime Technology 27st Century Conference. ‘Do w e really need new ships?’. 7989. -1989b. Japan. p. 1993. 18 March.Fairplay.‘Classification societies the big losers’. 1 1 . 1992. 2.) 1989.Morgan-Grampian Sharpe.Dai/y Commercial News.section Vll-6.Australia O’Brien.Newsletter June 7992. 12. S A S Institute.Kempe’s Engineer Year-Book Book Publishing. 78 . vol.Lloyd‘s Shipping Economist.1990 -1992d.‘Financial no.Casualty Weekly. 17 August. Packard.June. 1992. 15-1 7 March 1992. 2.V.‘Collins pledges govt to “Ships of Shame” fight’.4th edn.London. Lloyd‘s ShippingEconomist 1989a.p.p.p. North Carolina. 6. Cary. 13 (and later issues). S A S Institute 1989. D.’Costs’. 1986.p. J. 3. Sanderson.Study N S W Coal Report on Bulk Carriers Loss. London. -1990. 11. ‘Rates and Prices’. Association 1992. 6-8. vol. Daily Commercial News. (ed.7 September. Gearing: Shipping Industry Remains Vulnerable’ vol.V.vol.Lloyd‘s Register of Shipping.Trading. vol. Version6. 21 (and later issues). C.NKK.Paper presented to ‘International Shipping Seminar’. N K K 1992. -1993. pp.Sea-trading. Colchester. weekly issues.

ABBREVIATIONS ABS ACA AMSA BTCE DTC dwt GPa GRT HORSCOTCI IMO MPa NSW NKK OECD psi uts American Bureau of Shipping Australian Coal Association Australian Maritime Safety Authority Bureau of Transport and Communications Economics Department of Transport and Communications Deadweight tonnage Gigapascals Gross registered tonnage House of Representatives Standing Committee on Transport. Communications and Infrastructure International Maritime Organisation Megapascals New South Wales Nippon Kaiji Kyokai Organisation for Economic Co-operation and Development Pounds per square inch Ultimate tensile stress 79 .