MARINE SAFETY INVESTIGATION

REPORT 183

Independent investigation into the fire in the hold ofthe Netherlands flag general cargo vessel

off the coast of Western Australiaon 28 July 2002

Marion Green

AU S T R A L I A N G OV E R N M E N T – D E PA R T M E N T O F T R A N S P O R T A N D R E G I O N A L S E RV I C E S

Department of Transport and Regional Services

Australian Transport Safety Bureau

Navigation Act 1912Navigation (Marine Casualty) Regulations

investigation into the fire in the hold of the Netherlands flag general cargo vessel Marion Green

off the coast of Western Australia on 28 July 2002

Report No 183

November 2003

ISSN 1447-087XISBN 1 877071 42 0

Readers are advised that the Australian Transport Safety Bureau investigates for the sole purpose ofenhancing transport safety. Consequently, Bureau reports are confined to matters of safety significanceand may be misleading if used for other purposes.

Investigations commenced on or before 30 June 2003, including the publication of reports as a resultof those investigations, are authorised by the Executive Director of the Bureau in accordance with theNavigation (Marine Casualty) Regulations 1990, made pursuant to subsections 425(1)(ea) and 425(1AAA) of the Navigation Act 1912.

Investigations commenced on or after 1 July 2003, including the publication of reports as a result ofthose investigations, are authorised by the Executive Director of the Bureau in accordance with theTransport Safety Investigation Act 2003 (TSI Act). Reports released under the TSI Act are notadmissible as evidence in any civil or criminal proceedings.

It is ATSB policy to publish and widely distribute in full such reports as an educational tool toincrease awareness of the causes of marine accidents so as to improve safety at sea and enhance theprotection of the marine environment. Reports on serious marine casualties are also provided to theIMO.

Australian Transport Safety BureauPO Box 967Civic Square ACT 2608 AUSTRALIA

Phone: 02 6274 64781800 621 372

Fax: 02 6274 6699E-mail: marine@atsb.gov.au

ii

iii

CONTENTS

Summary . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .1

Sources of information . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .2

References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .2

Marion Green . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .3

Fixed firefighting installation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .4

Narrative . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .5

Cargo loading . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .5

Vessel’s call at Fremantle . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .6

The fire . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .7

Port of refuge – Albany . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .9

Cargo discharge . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .11

Comment and analysis . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .15

The investigation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .15

Cocoa beans . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .15

Shipping cocoa beans . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .15

Ventilation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .16

Fumigation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .17

Information provided by the shipper . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .17

The fire . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .18

Self-heating or spontaneous combustion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .19

Phosphine as a possible source of ignition . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .20

Cigarettes . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .21

The cargo light . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .22

The ship's staff response to the fire . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .22

The port authority response . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .24

Albany Port Authority emergency procedures . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .25

Authority in fighting ship fires in port . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .26

Conclusions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .27

Recommendations . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .29

Submissions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .31

Marion Green . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .33

Figures

1. Marion Green at Albany, 30 July 2002 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .v

2. CO2 room, showing banks of gas bottles for CO2 flooding of cargo holds . . . . . . . . . . . . . . . .4

3. Cargo in No. 1 hold, covered with Kraft paper and showing timber dunnage and fumigationsachets of aluminium phosphide . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .6

4. Smoke issuing from No. 2 hold access hatch . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .7

5. Position of Marion Green when the fire alarm activated at 1130 on 28 July 2002 . . . . . . . . . .8

6. No. 2 hold aft hatch panels partially opened . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .10

7. Smoke from No. 2 hold after opening hatches . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .10

8. Albany fire brigade tackling fire in No. 2 hold . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .12

9. Smouldering cargo at aft end of No. 2 hold showing burned cavities and channels . . . . . . . .12

10. Discharging damaged cargo onto the wharf . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .13

11. Narrow ventilation channel between stacks of cargo in No. 1 hold. Note that some layers arealmost touching . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .17

12. Cigarette ends discarded around No. 1 hold access . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .21

13. Electric cable for cargo light hanging into No. 2 hold from hold access . . . . . . . . . . . . . . . . .23

14. Burned end of cargo light cable in No. 2 hold . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .23

iv

v

FIGURE 1:Marion Green at Albany, 30 July 2002

vi

Summary

On 28 July 2002, the Netherlands flag general-purpose cargo vessel Marion Green, of 11 894 gross tonnes, carrying 6 000 tonnes ofcocoa beans, was off the coast of WesternAustralia on passage from Fremantle toAdelaide.

At 1130 that morning a fire alarm was activatedby the smoke detection system in No. 2 cargohold. After a brief inspection through the openhold access by the mate and second mate,during which they saw some flames on top ofthe cargo, all openings were closed and thedischarge of CO2 from the fixed firefightinginstallation was started. By 1715 that afternoon,86 bottles of CO2 had been released into thehold.

At 0700 on the following day, a slight increasein hatch cover temperature was recorded. Themaster was advised by the vessel’s managers todischarge the remaining 11 bottles of hold CO2

and to divert to Albany as a port of refuge.

Marion Green berthed in Albany at 1615 on 29 July. Additional bulk CO2 was delivered tothe ship from Perth and this, too, wasdischarged into the hold over the next few days.On the morning of 31 July, the after panels ofNo. 2 hatch were opened for an inspection.Flames were seen on the top layers of cargo andthese were doused by the fire brigade afterwhich the hatches were again closed and moreCO2 discharged into the hold. At 1400 thefollowing day, 1 August, the hatch covers wereonce more opened and, after further flare-upshad been doused by the fire brigade, stevedoresbegan discharging the cargo into sand bungs onthe wharf. By 10 August all the cocoa beancargo had been discharged and Marion Greensailed for Adelaide.

The report concludes that the investigation wasunable to determine, exactly, the cause of thefire, but four distinct possibilities wereexamined. These were:

• Self-heating of the cargo due to fungalgrowth

• Ignition caused by the flammable character-istics of the phosphine used for fumigatingthe cargo

• Cigarette ends discarded in the hold duringloading of the cargo in Makassar

• A cargo light that had been left in the hold onsailing from Makassar

It also concludes that:

• the vessel’s ‘no smoking’ policy was notproperly enforced during cargo loading

• insufficient CO2 was released into the hold inthe early stages of the fire

• inadequate information on the hazards ofshipping cocoa beans was provided to theship’s staff and

• the response to the fire, once the vessel wasalongside, lacked co-ordination and a clearunderstanding of who had the responsibilityand authority for dealing with it.

The report recommends that:

• the shipowners enforce a strict ‘no smoking’policy in the vicinity of cargo operations

• the shipowners ensure that ship’s masters areprovided with all relevant information on thehazards of carrying organic cargoes and theirfumigation

• shippers, stevedores and ship’s officersensure that adequate ventilation channels areprovided when stowing such cargoes and

• deck watchkeeping officers log the isolationand stowage of all electrical equipment fromthe holds on completion of cargo operations.

1

Sources ofinformation

Master and Chief Officer of Marion Green

The Harbourmaster, Albany Port

Fire & Emergency Services Authority ofWestern Australia

Australian Quarantine and Inspection Service(AQIS)

United States Department of Agriculture

Transport Information Service, GermanInsurance Association

IPCS Inchem – Chemical Safety Informationfrom Intergovernmental Organisations

UK P&I Club

References:Thomas’ Stowage

‘Fire Aboard’ by Frank Rushbrook CBE, 3rd

Edition 1998, Brown, Son & Ferguson

Certain reproductions of chart sections in thispublication are reproduced by permission of TheAustralian Hydrographic Service.

© Commonwealth of Australia 13 October 2000. Allrights reserved.

Other than for the purposes of copying thispublication for public use, the chart information fromthe chart sections may not be extracted, translated, orreduced to any electronic medium or machinereadable form for incorporation into a derivedproduct, in whole or part, without the prior writtenconsent of the Australian Hydrographic Service.

2

Marion Green

Marion Green, is a Netherlands flag, multi-purpose, tweendeck, general cargo vessel,registered in Rotterdam, Holland. At the time ofthe incident it was owned and operated byGenchart BV of Rotterdam, (since changed toBeluga Genchart BV) owners of the ‘GreenFleet’. The ‘Green Fleet’ consists of some eight(and, at the time of the incident, two morebuilding) identical ‘box-shaped’ vessels whichare employed in worldwide tramping.

Marion Green, was built in 1999 at the shipyardof Schelde Scheepsnieuwbouw BV, atVlissingen in Holland. The vessel has since thenbeen maintained in class with Lloyd’s Registeras �100 A1,�LMC, UMS, NAV 1, Ice Class1A.

It is a vessel of 142.81 m length overall (132.0 m between perpendiculars), with amoulded depth of 13.3 m and a maximum beamof 21.5 m. It has a gross tonnage of 11 894 tonnes and a summer deadweight of 17 050 tonnes at a summer draught of 9.689 m.

All accommodation and machinery spaces areaft of the forward engine room bulkhead whichis located at frame 39. Forward of the engineroom bulkhead are two fully box-shaped cargoholds, the forward of which, No. 1 hold,(between frames 107 and 151) has a cubiccapacity of 7561 m3 and the aft hold, No. 2,(between frames 39 and 103) a capacity of 12 292 m3. The vessel can be configured with afully flush tweendeck, consisting of loose steel,‘pontoons’ each of 17.75 x 6.52 x 0.90 m,which can be positioned individually at threedifferent heights in the holds. These pontoonscan also be used to form vertical bulkheads atintervals of six metres. The ship can be readilyconverted from either a tween or eventripledecker, to a full container vessel, or fittedwith any combination of tweendecks andbulkheads, thus making it extremely versatile.

The vessel, which has a total container capacityof 962 TEUs1 (215 on deck and 446 underdeck),is fitted with three deck cranes each of 60 tonnelift at 16 m radius.

Marion Green is powered by a single 8-cylinder, four-stroke, Wärtsilä 8L46B dieselengine developing 7 800 kW (10 460 shp) anddriving a single shaft with a variable-pitchpropeller, giving the ship a maximum speed of16 knots. The ship is also fitted with a 750 kWbow-thruster unit.

At the time of the incident, the ship had acomplement of 13. This consisted of the master,mate, second and third mates, chief engineer,second engineer, bosun, cook, four able seamenand a wiper. The master, mate, second mate andchief engineer were Dutch nationals, the thirdmate was Belgian, and the second engineer wasBritish. All the ratings were Filipinos. All of theship’s complement were appropriately qualifiedin accordance with the requirements of theSTCW95 (Standards of Training, Certificationand Watchkeeping, 1995) convention.

The master had first gone to sea in 1966 aftercompleting his training at a pre-sea trainingcollege. He had served on Shell tankers,Nedlloyd container ships and general cargovessels from the 1970s to the 1990s beforejoining Green Fleet in 1999. For most of thetime since joining Green Fleet he had sailed asmaster.

The mate had been at sea for 12 years sincestarting as an apprentice. He had served in theoffshore industry, on reefers, small coasters,general cargo vessels and heavy lift ships beforejoining the Green Fleet in 1999. It was his sixthvoyage on this type of vessel. Neither themaster nor the mate had any previousexperience in the carriage of cocoa, or otherbean, cargoes.

At the time of the incident, all ship’s certificatesrequired under international shippingconventions were valid. In addition, the vesselhad been issued with a ‘Document of

3

1 Twenty-foot Equivalent Units.

Compliance with special requirements for shipscarrying dangerous goods’ by the NetherlandsShipping Inspectorate in June 1999. From July2002 Marion Green, as a general cargo vessel,was required to hold an International SafetyManagement Certificate. Such a certificate wasissued to the vessel on 12 February 2002 by theNetherlands Shipping Inspectorate.

Fixed firefighting installationMarion Green is fitted with a Unitor fixed CO2

firefighting installation for gas flooding either,or both, Nos. 1 & 2 holds and for the engineroom. The major part of the installation iscontained in a CO2 compartment situatedbeneath the main deck on the port side aft ofNo. 2 hold (between frames 26 and 39) fromwhere the release of the CO2 can be controlled.An additional remote release cabinet for thebank of engine room bottles, only, is situated onthe starboard side of the main deck, just off themain cross-alleyway within the accommodation.There are a total of 154 CO2 bottles, each of67.5 litre capacity and containing 45 kg of gas.

Of these, 57 bottles are arranged for release intothe engine room, as a complete bank, in lessthan two minutes, by pneumatically operatedvalves. The other 97 are arranged for manualrelease and are dedicated to the holds. Releaseof those CO2 bottles for the holds is donemanually after selecting the appropriate numberof bottles for the hold to be flooded. It ispossible, by use of a changeover valve, to alsorelease the engine-room bank into the holds.

The CO2 is discharged into the holds throughnozzles arranged on a ring-main around the topof each hold. In No. 2 hold, there are eight suchnozzles, four on each side. These same nozzles,under normal circumstances, are also used towithdraw samples of air on a continuous basisfor passing through the Autronica smokedetection system. This system, driven by smallfans, constantly samples air from both holdsdrawing the samples through a photo-electriccell which, if it detects obscuration due tosmoke, will sound an alarm on the fire detectionpanel situated on the after bulkhead in thewheelhouse.

4

FIGURE 2:CO2 room, showing banks of gas bottles for CO2 flooding of cargo holds

Narrative

Cargo loadingOn 17 July 2002, Marion Green arrived at theIndonesian port of Makassar at the beginning ofvoyage 6 of 2002, to load a cargo of6 096 tonnes of Sulawesi cocoa beans in burlapbags (96 156 bags of approximately 63 kg each)for the Brazilian port of Ilheus. Upon the ship’sarrival alongside in Makassar, moveable steelpontoons were arranged as a vertical bulkheadin No. 2 hold at frame 79 (the aft end of No. 9pontoon position – 19.56 m from the forwardend of the hold) thus partitioning the hold. Theforward section was for two items of machineryto be loaded later in Australia, while the after,major, part was for the loading of cocoa beans.

Prior to commencing loading the bagged cocoabeans, the vessel received 42 m3 of woodenplanks and pallets for use as dunnage, 800bamboo mats for placement over the dunnageon the tanktops and 2 200 kg of Kraft paper.The paper was to be used to line the holds forabsorption of any condensation which mightform during the voyage, due to changes inambient temperature, on the sides of the holdsand the undersides of the hatches

On the 16th July, the day before the ship’sarrival, all the dunnage, bamboo and Kraftpaper had been fumigated with methyl bromide.The cargo was similarly fumigated for a periodof seven days leading up to the loading and, inaddition, on 17 July, the ship’s empty holds(Nos.1 and 2) were sprayed throughout with a 5 per cent solution of pyrethrins.

Shortly after 0100 on 18 July, the holds wereagain opened. At 0230, after preparing thedunnage and other stowage materials, two gangsof stevedores started loading the cargo ofbagged cocoa beans. 48 156 bags were to beloaded into No. 1 hatch and 48 000 into No. 2

hatch. The vessel’s own cranes were used forloading the bags, slung in cargo nets, into theholds where the stevedores arranged theirstowage. Each sling contained five tiers of fivebags each.

Cargo loading continued, with either three orfour gangs working from 0800 on 18 July, until2200 on 21 July. The weather throughout wasfine, with sunny and cloudy periods, but withhigh ambient temperatures, usually above 30°C,and high humidity.

On a number of occasions during this period,the mates on deck watch had cause to draw theattention of the stevedores to the ‘NOSMOKING’ signs painted on the ship’s structurearound the holds, as a some men had beenobserved smoking in the vicinity of the holds.

On 19 July, the vessel’s liferafts underwent theirannual checks. During this process, the drinkingwater in the liferafts was replaced and theremoved sachets were taken by the stevedoresworking in the holds for some refreshment intheir hot, humid working conditions.

Loading of the 6 000 tonnes of cargo wascompleted at 2145 on 21 July, whereupon thetop layers of cocoa bags were covered byabsorbent paper and the hatches closed.Following checks to ensure that all openings toboth holds were able to be thoroughly sealed,both holds were then finally fumigated by thedistribution of aluminium phosphide, in sachets,around the holds. The aluminium phosphide,when exposed to atmospheric moisture,generates phosphine (hydrogen phosphide) gas,a powerful insecticide. The quantity ofaluminium phosphide required to provide forfive days of ‘in transit’ fumigation of the cargohad been calculated by the fumigationcontractors before it was placed in the holds.This process was completed at 0030 on 22 Julyand both hatches were then closed. Shortlyafterwards, the pilot boarded and Marion Greensailed for the Western Australian port ofFremantle.

5

Vessel’s call at Fremantle, WAThe voyage from Makassar to Fremantle wasuneventful. The hatches and all other openingsto both holds remained sealed throughout thevoyage, while the phosphine fumigation of thecargo took effect. All the hold ventilation fansremained off.

At 0300 on 27 July, the Fremantle pilot boardedand Marion Green was secured alongside No. 2berth, North Quay, at 0435. After arrival, thesecond and third mates, wearing gas masks, setabout opening all the hold ventilation flaps. Thedeck was then cleared of personnel and the holdventilation fans were started. Ventilation of theholds continued for approximately 30 minuteslonger than the required six hours, until about1300 that day, when the mate then entered theholds testing for traces of phosphine. Testing forphosphine was carried out using ‘Dräegertubes’2 supplied by the shipper of the cargo. Thetests showed that the level of phosphine hadreduced to a zero reading.

The tweendeck at the forward end of No. 2 holdwas then opened and the hold prepared forloading two large heat exchangers, also destinedfor Brazil, which were to be stowed forward ofthe temporary steel pontoon bulkhead across thehold at frame 79. The ship’s lifting gear wasprepared and, during the afternoon, stevedores,assisted by the ship’s crew loaded the heatexchangers into the hold. The task took approxi-mately 45 minutes and, on completion atapproximately 1700, the heat exchangers werecovered with tarpaulins and the tweendeckpontoons replaced. When the hatch had beenclosed, a 20-ft container containing spare partsfor the heat exchangers was then loaded ontothe top of the hatch as deck cargo.

That evening the ship took bunkers. Later thatnight the pilot boarded and, at 0145 on 28 July,Marion Green sailed for Adelaide. After sailing,the hold ventilation was switched off because ofheavy rain showers.

6

FIGURE 3:Cargo in No. 1 hold, covered with Kraft paper and showing timber dunnage and fumigation sachets of aluminiumphosphide

2 Test equipment to which can be fitted the appropriate glass tubes containing reactant chemicals designed to show up the presence of aspecific gas.

7

The fireAt about 0930 that morning, the mate wentdown to the holds to measure the humidity anddew point. He also again tested for the presenceof phosphine, with a negative result. While atthe access to No. 2 hold he noticed that the holdwas ‘soaking wet’. Moisture was running downthe bulkheads and dripping from the hatchcovers. He measured the relative humidity asbeing 93 per cent. By 1015 he had taken therequired readings in No. 1 hold. The relativehumidity, at 82 per cent was significantly lowerthan that in No. 2 hold. Having taken thenecessary readings, he restarted the holdventilation fans then made his way to thewheelhouse.

At 1130 that same morning the ship was at theapproximate position 33° 50'S, 114° 49'E andmaking 16 knots, when a fire alarm wasactivated by the smoke detection system in No. 2 cargo hold. At about the same time,smoke was observed on the main deck comingfrom the No. 2 hold port side ventilation ducts.Hold ventilation was stopped and the mate wentdown to the deck with the second mate to

investigate. They noticed that the access hatchwas warm and there was smoke issuing fromthe upper hold access opening. This informationwas passed to the master in the wheelhouse,who re-activated the fire alarm. Two fire suitsand four breathing apparatus sets were broughtto the hold access from the fire locker on themain deck and from the wheelhouse.

The ship’s crew closed all ventilation openingsand ran out fire hoses on deck while the chiefofficer and second officer donned the fire suitsand breathing apparatus. They opened thestarboard aft access hatch to No. 2 hold and sawflames on the top layer of bags of cocoa beans,not far from the hold access opening. A lot ofsmoke was coming from the hold access hatch.They played a fire hose onto the area where theflames were just visible before closing theaccess hatch and informing the master of thesituation.

The master instructed them to makepreparations to flood the hold with CO2 from thefixed fire-fighting installation. Crewmembersstarted cooling the hatch covers with two firehoses.

FIGURE 4:Smoke issuing from No. 2 hold access hatch

8

FIGURE 5:Position of Marion Green when the fire alarm activated at 11.30 on 28 July 2002

115ºE

Position of Marion Greenat 1130 on 28 July 2002

ACT

NSW

NT

Qld

WA

SA

Vic

Tas

NT

NSW

ACTLocation ofincident

9

At 1150, the mate started to manually releasethe CO2, one bottle at a time, into the hold.Initially, 25 bottles were released, after which itwas decided to wait for the CO2 lines to depres-surise and to don breathing apparatus again, asthere were several leaking connections in theCO2 room. The vessel’s managers in Rotterdamwere advised of the situation.

At about 1210, one vent flap on the forward endof No. 2 hold was opened in order to measurethe oxygen content of the issuing air/gasmixture. It was found that the oxygen contenthad decreased only slightly and, consequently,another 25 bottles of CO2 were progressivelydischarged from about 1220. At 1245, the holdaccess was opened again to check theeffectiveness of the CO2 flooding. The flamesappeared to have been extinguished. However aproposed entry by the mate and second matewas cancelled on account of the thick smokeand the fear of a ‘backdraft’. The entrance wasclosed again, as was the forward vent flap.

At 1300, another 14 bottles were discharged anda fire watch was set up to monitor the temper-atures of the hatch covers and coamings every15 minutes. At 1500, a further situation reportwas made to the ship’s managers and theAustralian Search and Rescue co-ordinationcentre (AusSAR) was also notified.

By 1715 that day, a total of 86 bottles of CO2

had been released into the hold and the temper-atures of the hatch covers dropped steadily.During the night, the hold remained sealed andtemperatures of the hatch covers weremonitored, and recorded, on a regular basis. Asthe temperatures were dropping, thetemperature observations were changed fromevery 15 minutes to hourly. Checks weremaintained in the engine room on the forwardbulkhead, being the aft bulkhead of No. 2 hold.These temperatures, however, were found to besteady.

Port of Refuge – AlbanyAt 0700 on the following day, an increase inhatch cover temperature of about 2°C wasrecorded. The mate and chief engineer opened

one of the drains on the starboard side of thehatch coaming to again test the oxygen level.However, they were unable to take a reading asthey were confronted by a continuous flow fromthe drain of a brown, mud-like, substance. At0835, having again contacted the vessel’smanagers, the master was advised to dischargethe remaining 11 bottles of hold CO2 and todivert to Albany, approximately 100 nauticalmiles distant, as a port of refuge. Agents for thevessel, who were appointed in Albany, informedthe harbourmaster of the situation and, at about0900 that morning, the harbourmaster contactedthe ship by VHF. The ship’s master explainedthat there was a fire in No. 2 hold, that CO2

flooding had been used and that the fire seemedto be under control. The temperatures, however,were not going down.

The vessel proceeded towards Albany and, at1450 that same day, the harbourmaster, as pilot,boarded the ship.

At 1615 on 29 July, Marion Green wasalongside at Albany. Shortly after berthing ameeting was held between the harbourmaster,the master, a representative from Customs andthe ship’s agent. As the vessel’s stock of CO2

had been severely depleted, the agent arrangedfor a bulk tank of 14 tonnes of CO2 to bedelivered from Perth. At 1730, representativesof the local fire brigade arrived on board toevaluate the situation. The hatch coversremained closed and temperatures weremonitored throughout the night on a two-hourlybasis. At 2200 that evening, more hoses wererigged and boundary cooling was again appliedto the hatch covers.

This situation remained unchanged throughoutthe following day, Tuesday 30 July. A surveyor,appointed by the P&I Club requested, throughthe master, that permission be given to open thehatch, but this was refused, the harbourmasterrequiring a further 24 hours of monitoring. Laterthat day the bulk CO2 arrived from Perth andwas placed on the main deck at the port aftercorner of No. 2 hold. It was connected to theship’s CO2 flooding system through the adjacentaccess to the CO2 room.

10

FIGURE 6: No. 2 hold aft hatch panels partially opened

FIGURE 7:Smoke from No. 2 hold after opening the hatches

At 0545 on the morning of 31 July, the temper-atures of the hatch covers appeared to be risingagain and boundary cooling was resumed. Laterthat morning, a decision was made to open thehold and ascertain the situation

At 0930, the access hatch was opened, followedby the two aft panels of the cargo hatch coverwhich were lifted into a partially open position.Much smoke issued from the hold and someflames were observed through the thick smokeon top of the cargo. The fire brigade startedhosing down the top layers of bags of cocoabeans and the flames were eventuallyextinguished. After a short period with no moreflames appearing, the hatch covers were fullyopened.

As no decision had been reached at that stageon what to do with the damaged cargo, thehatches were again closed and more CO2 wasinjected from the bulk tank on the main deck.

At 1500 that afternoon the vessel was requiredto shift ship to the next berth. On completion ofthe shift, steam was noticed coming from theport side coaming of No. 2 hatch. A small areaof paint was scorching on the hatch cover andthere was some noticeable deformation of thehatch panel. It was immediately evident that theintensity of the fire was rapidly increasing andthe fire brigade was again called. At 1600, twoappliances attended on the wharf and more CO2

was injected into the hold. Substantial boundarycooling was applied to the hatch covers andcoamings, by both the fire brigade and theship’s crew, throughout that night.

At approximately 1730 that evening, a tanker,Scottish Bard, arrived at Albany and was due toberth close to Marion Green. The port, however,was closed due to the fire.

On the following morning, Thursday 1 August,the temperatures of the hatch covers, beingmonitored by the ship’s staff, appearedrelatively stable. The weather had deterioratedovernight and the day became windy with rainshowers and drizzle. This assisted in keeping

the hatch covers cool. No attempt, however, wasmade to open the hold again before 1400 thatafternoon, when a meeting of involved partieswas convened by the harbourmaster todetermine the future course of action.

Cargo dischargeIn spite of various evident conflicts of interest(see p25), it was eventually agreed that therewas no real option but for the cargo to bedischarged and, at 1600 that afternoon, on 1 August, the hatches were partially opened.Several spot fires were again visible through thesmoke. Both the fire brigade and ship’s crew, allwearing BA sets, attacked the fires with severalhoses.

Sand dams, into which the cargo would beplaced for final dousing, were constructed onthe wharf adjacent to the ship and arrangementswere made, under the auspices of AQIS, for thedamaged cargo to be trucked to the Albany tipwhere it would be buried as landfill.

At 1835 on 1 August, in consultation with thefire brigade, the discharge of the cargocommenced. The port remained closed.

On the following day, 2 August, the spot firesappeared to be under control. At 1120, thetanker Scottish Bard was allowed to berth at No. 2 wharf to discharge its cargo of petrol. Thedischarge of burning cargo from Marion Greenwas suspended while Scottish Bard dischargedits cargo.

Over the next few days all the bags of cocoabeans in No. 2 hold were discharged bystevedores using the ship’s cranes and grabs.Any spot fires or areas of smouldering cargowere doused during the discharge. On 6 Augustthe ship’s agent was informed by AQIS that ithad reviewed its decision regarding burial of thecargo at the Albany tip and it could bereclaimed where appropriate. Approximately 44 per cent of the cargo from No. 2 hold,relatively undamaged, was salvaged and movedto a cargo shed on the wharf. Flare-ups in the

11

12

FIGURE 8:Albany fire brigade tackling fire in no.2 hold

FIGURE 9: Smouldering cargo at aft end of no.2 hold showing burned cavities and channels

13

cargo occurred frequently during the discharge,in the hold, on the wharf and in the cargo whichhad been moved to the shed. The fire brigaderemained in attendance at the vessel, dealingwith these fires, until they were finally stooddown at 2020 on 7 August.

After completion of the discharge of cocoabeans and the cleaning out of residue in No. 2hold, Marion Green sailed for Adelaide at 1550on 10 August.

FIGURE 10:Discharging damaged cargo onto the wharf

14

Comment andanalysis

The investigationA marine investigator from the ATSB arrived onboard Marion Green on the morning of 30 Julyand, while conducting the investigation over thenext five days, was also able to observe thesubsequent events and the measures taken todeal with the fire during that period. The masterand chief officer were interviewed, copies of theship’s logs and all other relevant documentationwere obtained and samples of cocoa beans weretaken from both Nos. 1 and 2 holds for laterexpert examination. Evidence concerning theresponse of the port authority and the local firebrigade was obtained. Details of the cargo, itsloading in Makassar and its fumigation werealso obtained, as was information which hadbeen provided to the master by the shipper ofthe cargo.

Owing to the nature of the incident, it wasinevitable that much of the physical evidencerelating to the source of ignition was destroyedas the cargo, a large proportion of it burned, wassoaked by water from fire hoses on severaloccasions before its discharge. In addition, theaction of the grabs during the dischargedamaged many unburnt bags and caused thecargo to finish up as an homogenous heap onthe wharf. The pattern, however, in which thefire spread through the hold could be observedthrough the smoke on those occasions when thehatch covers were opened for fire-fightingoperations.

Cocoa beansCocoa beans are the seeds, contained in acucumber-like fruit, of the cacao tree,Theobroma Cocoa Linné. (Theobroma, from theGreek, literally means ‘food of the gods’) It is amember of the Sterculiaceae family. The cocoabean (there are usually about 30–40 seeds in

each pod) consists of the seed coat whichencloses the cocoa kernel. The cocoa kernel isthe principal component for the production ofcocoa products.

The history of cocoa and chocolate dates backto the time of the Incas, Mayas and Aztecs ofCentral and South America. It was at one timethe currency used by those peoples. It wasintroduced to Europe in 1527. During the 16thand 17th centuries, the Spaniards establishedcacao tree plantations in their colonies inCentral and South America. The Dutch latertook trees to other countries including theircolony of Indonesia. Indonesia is not one of thelargest of cocoa producing countries, unlikeBrazil, the destination for the cargo aboardMarion Green. At the time of the incident, aworldwide shortage of cocoa existed andcompanies in Brazil were having to importcocoa beans in order to be able to fulfil theircontracts.

Cacao seeds are the source of commercialchocolate, cocoa, and cocoa butter. Cocoa ismade by removing most of the fat from thebeans, then roasting and grinding them. The fatthat is removed is called cocoa butter. Cocoabutter is used in confections and in themanufacture of pharmaceuticals, soap, andcosmetics. Cocoa butter has been described asthe world's most expensive fat and cacao beanscontain a great deal of this fat, some 40–50 percent, in addition to small amounts of mildlystimulating alkaloids, including caffeine.Consequently, cocoa beans are a very high valuecargo. The value of the cargo aboard MarionGreen on 28 July 2002 was approximatelyUS$10.5 million.

Shipping cocoa beansCocoa beans are usually shipped in burlap(made from jute or sisal) bags, each bagweighing from 60–65 kg. Owing to the value ofthe cargo, new, high quality bags are usuallyused. Cocoa beans require particulartemperature, humidity and ventilationconditions. Due to its high fat content, the cargois very apt to self-heat and there may be a risk

15

16

of cargo fire on contact with flammablesubstances.3 The cargo requires to be stowed ina dry place, normally below decks. The cargoloses value, however, if allowed to become toodry and brittle; around 7 per cent moisturecontent being acceptable.

The temperature of the cargo is frequently hotterthan the ambient temperature in areas throughwhich the vessel may pass. Too little ventilationresults in hold temperatures rising above thedew point with the consequence that conden-sation forms on the ship’s shell plating, tanktops and hatches. Excess moisture content willcause mould growth and, if excessive, will rotthe cargo. The beans themselves release largeamounts of water vapour during extendedvoyages after which the moisture content of thebeans may be observed to have decreased by 1to 3 per cent. The vapour given off may containacetic acid which can cause corrosion.4 Inaddition to the moisture content, the beanscontain certain enzymes that bring about post-fermentation which is prevented only by goodair circulation around the cargo.

To keep the bags dry should condensation form,the cargo aboard Marion Green was loaded ontop of dunnage which had been made intowooden frames, not unlike conventional pallets.The wooden frames were then overlaid withbamboo matting which was in turn covered witha layer of Kraft5 paper. Other wooden frameswere placed up the ship’s sides as the cargo wasloaded and, again, covered with a layer of Kraftpaper. Finally, once the loading of each hold hadbeen completed, the cargo was completelycovered with a layer of Kraft paper to protectthe top bags from any condensation which coulddrip from the hatch covers and coamings.

VentilationGood ventilation is essential, shippers requiringa minimum of 20 air changes per hour. Since

the beans continually release water vapourduring the voyage, this vapour must be removedto reduce the risk of condensation forming inthe event of cooler ambient conditions, and therisk of mould growth due to high relativehumidity in the hold. No. 2 hold of MarionGreen is ventilated by means of two supplyfans, port and starboard, ducted through theafter bulkhead and discharging into the hold onfour levels. At the forward end of the hold thehold atmosphere is extracted by two extractionfans drawing from the four levels. The capacityof the fans is sufficient to provide 20 airchanges per hour when the holds are empty;considerably more than sufficient when theholds are loaded.

To facilitate the movement of air around thecargo, the holds were loaded with ventilationchannels provided both in the fore-and-aftdirection and athwartships. The channels wereapproximately 300 mm wide and extendeddown the full depth of the cargo. At variouslevels, individual bags were arranged as ‘ties’across the ventilation channels to prevent thelayers of bags collapsing together and blockingthe channels.

The cargo in No. 2 hold was arranged with asingle ventilation channel along the centre-lineof the hold and two channels in the athwartshipsdirection, such as to divide the length of thehold into three equal sections.

During the investigation, the stowage of thecargo in No. 1 hold was assessed. It wasobserved that the width of the ventilationchannels in many areas was considerably lessthan 300 mm and, indeed, in some spots hadclosed up completely, due either to improperstowage on loading, or to subsequent movementof the cargo. It was also noted that the Kraftpaper, covering the top layer of cargo, alsocovered the ventilation channels, hence

3 Transport Information Service, German Insurance Association, Berlin.

4 Thomas’ Stowage.

5 A brown paper used extensively in the packaging and manufacturing industries. Available in numerous grades and for numerouspurposes. In this instance the paper used was of a relatively heavy, absorbent grade.

17

restricting any vertical movement of air throughthese channels.

FumigationTo both protect the cocoa bean cargo fromdamage by insect pests and to prevent thetransport of such pests from one country toanother, cargoes of cocoa beans are fumigated,both before loading and for the first few days intransit. During the fumigation period, in thiscase for five days, the holds must remain tightlysealed and all ventilation shut down. Only whenthe fumigant has had sufficient time to penetrateall of the cargo is ventilation restored andmaintained for the remainder of the voyage.

One of the most common fumigants used ishydrogen phosphide, usually referred to asphosphine gas (PH3). Phosphine is a colourlessgas which is toxic to insects, humans and otherforms of animal life. It is very mobile with ahigh vapour pressure which enables it topenetrate to all parts of the cargo. Thiscombination, together with high molecularactivity and toxicity at low doses, accounts forits wide acceptance as a fumigant.

Phosphine is generated by allowing atmosphericmoisture in the surrounding air to react withaluminium phosphide in the form of eithertablets, pellets or sachets. Aluminium phosphideis also referred to as the ‘dry gas’. Once spent,the aluminium phosphide degenerates into agrey-white powder composed of aluminiumhydroxide and some inert ingredients.

On completion of loading of the cocoa beancargo at Makassar, sufficient aluminiumphosphide, in the form of sachets, to provide forfive days ‘in transit’ fumigation was scatteredrandomly across the top layer of the cargo ineach hold. In No. 1 hold, some of these sachetscould be observed to have fallen down into theventilation channels provided between thestacks of bags.

Information provided by shipperNeither the master nor the mate of MarionGreen had any previous experience of thecarriage of cocoa bean cargoes and both had torely on information from the sub-charterersprovided to the ship, in a telex from Genchart, amonth before the incident.

FIGURE 11: Narrow ventilation channel between stacks of cargo in No. 1 hold. Note that some layers are almost touching

18

The information related to the way the bags ofcocoa beans were to be slung to avoid damage,the arrangement for dunnage on the tank-topsand at the ship’s sides to avoid contact with anycondensation and the requirements forventilation, temperature and dew-pointmonitoring.

The information went on to discuss theproblems for the cargo caused by eitherexcessive or too little ventilation, and the effectsof passing through zones of ambienttemperature below, or above, that of the cargo.Examples used were generic in nature,discussing the problem of ‘ship sweat’ (conden-sation on ship’s sides and hatches) for baggedrice loaded in a tropical climate, and ‘cargosweat’ (condensation on the cargo) for steelcoils loaded in a cold climate. No informationwas provided to the ship specifically on thecarriage of cocoa beans, nor on the subject ofthe possibility of self-heating of the cargo due tothe growth of fungi as a consequence of warm,moist conditions.

Information on fumigation and the use ofphosphine as a fumigant was provided to theship by the Pan Asia SuperintendenceCorporation at Makassar. The five-pagedocument described in detail phosphine and itsdetection, the process of fumigation, therequirements for ventilation, the toxic effect thatphosphine has on the human body, precautionsto be taken and medical treatment shouldpoisoning occur. The company also provided thegas masks and gas detection equipment(Dräeger tubes) to be used by the ship’s officersduring the voyage. However, nowhere in theinformation provided was the hazard of flamma-bility of phosphine mentioned.

Further information on the carriage offumigated cargo is contained in the IMO’s‘Recommendations on the safe use of pesticidesin ships’. This document, however, was notcarried on board Marion Green.

The fireThe quantity of wooden dunnage used in theholds was 42 m2, the number of bamboo matswas 800, and the quantity of Kraft paper usedwas 2 200 kgs. The total amounted to asignificant quantity of dry, combustible fuel fora fire in the hold, quite apart from the burlapbags and the cargo itself with its relatively highfat content. There is no question as to the fuelload available for the fire, as these materialswere observed to be burning on each occasionthat the hold was opened. By the time the cargohad been discharged from No. 2 hold, thesmouldering fire had spread throughout the bagsof beans with frequent flare-ups of the timber,paper and burlap.

The issue facing the investigation was thesource, or origin, of the fire.

During the course of the investigation, fourpossible sources were identified:

• Self heating of the cargo, promoted by fungalgrowth

• Phosphine used for fumigation

• Cigarette ends found amongst the dischargedcargo

• A cargo light

In each case, the fire would have started at thetime of, or shortly after, the loading of the cargoin Makassar. In Fremantle, the mate noticed thatthe humidity in the hold was very high. Thiswould indicate that heat was being generated inthe lower part of the hold, driving moisturefrom the cocoa beans. However, no sign ofsmoke or flame was evident at that time. It wasonly after air had been admitted, through holdventilation and opening the hatch covers inFremantle, that the fire gained a hold.

Observations when the hold was first opened inAlbany indicated that the source of the fire wasin the lower tiers of cargo near the starboard aft

19

corner of the hold. The way that the fire hadburned slowly up towards the top layers ofcargo, and then also spread along the bottom ofthe hold to the forward end would be consistentwith smouldering, or slow combustion, havingtaken place during the five day passage betweenMakassar and Fremantle. It continued in thismanner, albeit accelerated, after the vesselarrived at Albany.

Self-heating or spontaneouscombustionCocoa beans are not generally regarded as anhazardous cargo and are not mentioned as suchin the IMDG Code. However, under suitableambient conditions (temperature > 25°C, highrelative humidity and lack of oxygen) and dueto their elevated fat content which, inconjunction with moisture, results inhydrolytic/enzymatic fat cleavage, cocoa beanshave a tendency for postfermentation and self-heating. Some species of fungus, such asAspergillus Fumigatus, participate in the self-heating process. Spontaneous combustion ofbroken raw cocoa beans in bulk usually takesthe form of smouldering cavities or channels,such a fire developing in four distinct phases:6

• first phase: a general biological phase inwhich mesophilic micro-organisms multiplyin a wet spot, raising the temperature to37°C.

• second phase: this involves the highest levelof activity from thermophilic micro-organisms at temperatures of up toapproximately 70°C.

• third phase: the thermophilic decompositionphase which is characterised by exothermicchemical breakdown reactions, in particularby oxidation reactions between unsaturatedfatty acids and atmospheric oxygen (thecocoa beans having a fat content of >50 percent).

• fourth phase: the pyrophoric gas phase, inwhich pyrophoric carbon and gasses (eg

phosphine) are formed. The resultant abruptincrease in temperature gives rise to thesmouldering cavities and channels within thebulk cargo.

Observation of the progress of the fire duringthe period 29 July to 1 August showed clearlythat the fire had spread from the starboard aftercorner of No. 2 hold through such channels,with cavities appearing at the top of the cargo invarious places. This, however, does not in itselfprovide sufficient evidence that this was thesource of ignition for the fire as the fire couldhave spread in this manner, whatever the sourceof ignition.

Samples of the cargo were taken from both No.1 hold (undamaged cargo) and from thedamaged cargo in No. 2 hold. The samples weresubmitted to the Eastern Creek Plant QuarantineLaboratory of the Australian Quarantine &Inspection Service (AQIS) near Sydney, NSW.There, the seeds were examined by a quarantineplant pathologist and placed in a humidenvironment in petri dishes for the culture ofany fungi that may have been present.

The results of the tests showed no Aspergilluscontamination could be isolated from theundamaged sample of cargo from No. 1 hold.However, the tests revealed that a mould,Aspergillus Niger, which is closely related toAspergillus Fumigatus, was present in thesample of cargo from No. 2 hold. The tests wererepeated and, again, the same results wereobtained. Aspergillus Niger was present in thecargo in No. 2 hold and, in the opinion of theAQIS pathologist, this could have had the sameheating effect.

If the condition of the cocoa beans in the holdwas conducive to mould development, then anumber of different fungi, including AspergillusNiger, could take advantage of the favourablegrowing conditions of warmth and highhumidity to proliferate through the stored beans.Once the mould activity is encouraged to

6 Transport Information Service, German Insurance Association, Berlin.

20

develop and spread, so too will there be anincrease in the heat generated by the fungalgrowth.

The conditions necessary to promote fungalgrowth are warmth and moisture.

If the moulds find favourable living conditions,i.e. when the critical water content of 8.5 percent is exceeded at an equilibrium moisturecontent of approximately 88 per cent, themoulds rapidly develop within 3 - 4 days, at theend of which period thousands of spores haveformed on the surface of the cocoa beans7.During this process, considerable heat may begenerated.

The cargo aboard Marion Green was loaded intropical weather conditions where the ambientair temperature was consistently above 30°Cwith high relative humidity. In addition, duringthe loading of the cargo the sachets of water inthe liferafts were replaced with fresh ones.Those removed from the rafts were taken forrefreshment by the stevedores working in theholds. During the discharge of the burned cargoat Albany, several of these sachets, having beendiscarded in the hold, were found amongst theremains of the burned cargo. Some were empty,but some still contained water. Such sachets,discarded during the cargo loading in Makassarmay well have wetted small areas of some ofthe bags of beans and thus provided ideal levelsof moisture for the fungi to flourish.

During the later discharge of the, mostlyundamaged, cargo from No. 1 hold it was foundthat 14 bags of beans had localised areas of wetstaining.

When the mate first entered No. 2 hold after thevessel’s arrival in Fremantle, he noticed that No.2 hold was ‘soaking wet’, with moisturedripping from hatch covers and running downbulkheads. No. 1 hold was not so wet. Thiswould indicate that an area, or areas, of thecargo in No. 2 hold were already hot, and

driving off moisture, before the vessel arrivedand the hatches were opened. During thepassage from Makassar to Fremantle the holdventilation had been off, to maintain theintegrity of the fumigation. In Fremantle,restoring the ventilation and opening the hatcheswould have allowed the ingress of air which, inturn, would have allowed the fire to develop andbecome evident after sailing on the morning of28 July.

Although it cannot be confirmed with anydegree of certainty, there is sufficient evidenceto indicate that moisture, warmth andsubsequent mould growth leading to self-heating and combustion should be considered asone possible source of the cargo fire in No. 2hold.

Phosphine as a possible sourceof ignitionPhosphine is a highly flammable gas which canauto-ignite at ambient temperatures. At concen-trations greater than 1.8 per cent, it formsexplosive mixtures with air.8 To reduce the riskthat this poses, all commercial formulationsavailable contain ammonium carbamate whichreleases ammonia gas and carbon dioxide toreduce the potential fire hazard posed byphosphine. The ammonia in the formulation alsoserves as a warning agent, giving out a garlic-like odour. Nevertheless, great care has to betaken in how aluminium phosphide is applieddue to its inherent problem of flammability. Theammonium carbamate reduces the potentialdanger, but it does not eliminate it.

In February 2003, the UK P&I Club published aBulletin (No. 289), reproduced in Seaways, thejournal of the Nautical Institute, drawing theattention of the maritime industry to the dangerof phosphine explosions. The bulletin noted thata number of such explosions were known tohave occurred. It also noted that the process ofaluminium phosphide reacting with moisture inthe air sometimes produces small quantities of a

7 IBID.

8 IPCS Inchem – Chemical Safety Information from Intergovernmental Organisations.

gas known as diphosphine, which, unlikephosphine, is spontaneously flammable reactinginstantly with oxygen in the air. This is likely tooccur when there is an imbalance between thealuminium and phosphorus, with an excess ofthe latter. Such a situation may arise duringproduction of the tablets if an excess ofphosphorus is inadvertently used during thepreparation.

In describing a recent incident, the bulletin thengoes on to state:

Although not definitively proven, we have beenadvised that it is likely that potentially explosivemixtures of air and phosphine are frequentlyencountered during the first 12 to 24 hours ofphosphine fumigation when the phosphineconcentration in the upper section of the holdreaches a peak concentration. The resulting highconcentration of phosphine then disperses bydiffusion, with the gas diffusing into the lessaccessible lower sections of the cargo. In thisrecent case, the explosion occurred some 12 hours or more after the fumigation had beenstarted and the hatch covers had been closed.Although no source of ignition was identifiedconclusively, it is suspected that defectivealuminium phosphide tablets, containinglocalised excesses of phosphorus, were the cause.

Such tablets could be envisaged as producinglocalised high concentrations of diphosphineleading to a very rapid reaction with oxygen andto ignition.

Aluminium phosphide tablets are routinely usedin fumigation and a very large number ofshipments are fumigated annually without anyproblems. Incidences of explosions are thereforevery rare and, as far as we have been advised,fumigant explosions have only been encounteredwhen companies have used cheaper brands ofaluminium phosphide tablets produced indeveloping countries.

The origin and chemical composition of thealuminium phosphide used to fumigate thecocoa bean cargo aboard Marion Green is notknown, however it cannot be eliminated ashaving been a factor in the fire. This is partic-ularly so if sachets had fallen into confinedspots in the lower tiers of cargo, where therewas potential for the gas concentration to haverisen to extreme levels.

Cigarettes‘NO SMOKING’ signs are prominentlydisplayed, painted in large letters, around thehatches of Marion Green. Nevertheless, during

21

FIGURE 12:Cigarette ends discarded around No. 1 hold access

the loading of the bagged cocoa beans atMakassar, the deck officers on cargo watch hadneed, on several occasions, to reprimand certainmembers of the loading gangs for smokingaround the hatches and hold access spaces.

During the investigation, empty Indonesiancigarette packets and numerous butt-ends ofcigarettes were found in the accesses to bothholds. In addition, it was later reported by thesurveyor appointed by the P&I Club, thatsimilar cigarette packets and cigarette ends hadbeen found amongst the cargo from No. 2 holdafter it had been discharged ashore.

A cigarette end or match, discarded in the holdduring cargo loading, without being properlyextinguished, could well have initiated somesmouldering amongst the cargo and associatedKraft paper, dunnage and burlap bags. Thissmouldering, in the absence of sufficient air, theventilation being shut off, may have developedinto a significant slow combustion fire over thenext few days, developing more rapidly oncethe ventilation had been restored and the hatchcovers opened while the vessel was inFremantle.

Smoking materials, discarded in No. 2 holdduring the loading of the cargo in Makassarmust be regarded as another possible cause ofthe fire.

The cargo lightWhen the after panels of the hatch covers werefirst opened up on 31 July, it could be seen thatan electrical cable was hanging from the upperaccess at the starboard after corner of No. 2hold, down into the cargo of bagged beans. Thiswas close to the area where the fire had beenfirst observed and to where it is believed tohave originated. See fig. 13.

In each hold access space there were a numberof halogen cargo lights, stowed with theircables, and the sockets into which these lightswould be plugged during cargo operations. Allthe plugs in the starboard aft access had beenremoved from their sockets, this being in

compliance with the ship’s routine oncompletion of loading.

The cable hanging into the hold was the sametype as that attached to the cargo lights and theplug on the end had also been removed from itssocket. However, the other end, in the hold, wasfound to have been burned away. Evidencegiven to the investigation indicated that it hadalso been attached to a cargo light which, inMakassar, was hanging on one of the timberdunnage frames against the aft bulkheadadjacent to the hold access. This cargo light,however, could not be located during theinvestigation, neither was it found amongst thecargo either in the hold, or once discharged tothe wharf.

The whereabouts of this cargo light, or thereason for its cable having been left in No. 2hold after the loading of the cargo, could not beascertained. Neither could the point in time bedetermined when its plug had been removedfrom the power socket.

These 1500-watt cargo lights generate consid-erable heat. If this light had been left, switchedon, in the hold at the completion of loading andhad come into contact with either the baggedbeans, the Kraft paper or other combustiblematerial, it may well have provided a source ofignition.

Although possible, however, it would seemunlikely that the holds would have been closedup with the light still switched on and without itbeing noticed. More likely is that it wasunplugged at the time that the other lights wereunplugged, but the fact that the cable still ledinto the hold, instead of being coiled up in theaccess space, went unnoticed.

The ship’s staff response to thefireThe fire was first discovered at about 1130 onthe morning of 28 July with a fire alarm beingactivated on the Autronica detection system, atthe same time as smoke was seen issuing fromthe No. 2 hold vents. The initial response by the

22

ship’s staff was correct in that all ventilationwas stopped and all openings were sealed off.After a brief inspection by the mate and secondmate through the opened hold access andplaying a hose on the flames, the decision wastaken to flood with CO2 and the first of 25bottles was discharged at 1150.

According to the instructions on CO2 release,issued to the vessel by the manufacturers of thefixed firefighting installation, the number ofCO2 bottles required for flooding No. 2 holdwas 77 if the hold was full, and 116 if it washalf full.

No. 2 hold contained 3 048 tonnes of cocoabeans with a stowage factor of approximately1.95 m3 per tonne. Hence the volume of cocoabean cargo was approximately 5 950 m3. It alsocontained the two machinery modules in theforward part of the hold. Allowing anapproximate volume for these two crates as 220 m3 and taking the volume occupied by thesteel pontoons, from the specifications, as 833 m3, the total volume of cargo and pontoons

in the hold was roughly 7 003 m3. The totalvolume of No. 2 hold is 12 282 m3 hence, byvolume, the hold was approximately 57 per centfull.

With this volume of cargo loaded in No. 2, bysimple proportion, it would have been necessaryto release, as a minimum, 110 bottles of CO2 inorder to raise the concentration of gas tobetween 30 and 40 per cent; the level necessaryto smoother the fire.

After discharging the first 25 bottles, releasingthe gas was suspended for a while due to leakson several pipe connections in the CO2 room. At1210, a forward ventilation flap on No. 2 holdwas opened to check on the effect of the CO2

flooding, by analysing the oxygen content of theissuing air. However, as insufficient gas hadbeen released for even a full hold, and as CO2 isheavier than air, it is unlikely that anysignificant effect on the oxygen level wouldhave been recorded at the ventilation flap on themain deck. A large proportion of the gas wouldhave been discharged into the virtually empty

23

FIGURE 13:Electric cable for cargo light hanging into no.2 holdfrom hold access

FIGURE 14:Burned end of cargo light cable in no.2 hold

24

part of the hold, forward of the pontoonbulkhead, and descended to the bottom of thehold. In doing so, it would slowly displace theair upwards towards the top of the hold.

A little more than half an hour later, at 1245, thehold access hatch was opened for an inspection.Acknowledging that the action to be taken isalways a matter of judgment for the master orofficer in charge, in the opinion of the Inspector,this was too short a period to allow before anyhold penetrations should be opened, forwhatever reason.

It was not until 1715 that afternoon that 86bottles of CO2 had been discharged, with theremaining 11 bottles not being discharged untilthe following day. It is unlikely that, at anytime, there was sufficient CO2 within the hold toeffectively stifle the fire which, indeed, was stillactive after the vessel arrived at Albany.

A safety bulletin published by the UK P&I Clubon 13 August 1996 describes at some length themeasures to be taken in the case of self-heating,or fire, in a cargo. The bulletin deals specificallywith bagged fishmeal cargoes, but the principlesinvolved apply to combustion in any bagged,organic cargo. The bulletin emphasises theimportance of sealing, and keeping sealed, theaffected hold. It states:

It is basically undesirable to inject less CO2 thanis recommended in the manual (provided by themanufacturers of the CO2 system) even thoughthis means that only a few cargo spaces can be sotreated.

It should be appreciated that any cargo heatingresults from an oxidation process. This meansthat the oxygen concentration in a hold isdepleted and the concentration of nitrogen, aninert gas, increases. Hence in a sealed hold, cargoheating tends to be self-quenching. It is thereforeof paramount importance that the Master has allnecessary materials on board to allow veryefficient sealing of cargo spaces in order tominimise atmospheric interchange. Very efficientsealing may be a time-consuming operation but

should never be skimped. Technically, providedthat hold sealing is adequate, it would be possiblefor a ship with cargo heating in all her holds, tosail safely across the Pacific Ocean with her CO2

supply exhausted (assuming a sufficient reservefor the engine room). However, such actionwould only be recommended if sealing efficiencycould be guaranteed. Under normal circum-stances where there is obvious progressiveheating a ship would be recommended to a portof refuge to obtain adequate CO2 supplies. Thisoften involves fitting a bulk tank containingseveral tonnes of CO2. If considered necessary,further sealing should be performed whilst theship remains in port.

There is often a tendency for personnel trainedin firefighting to exercise an ingrained urge toget to grips with the seat of a fire with the leastpossible delay. In a well-respected referencebook on fighting ship fires, ‘Fire Aboard’, theauthor states that there is ‘…much to be said forplaying the waiting game’.9 He goes on to say:

If 95 per cent of a cargo can be saved by acertain line of attack and one which involves alsominimal risk of structural damage to the vesselitself, what matters it that final extinction of thefire may take 48 hours or so? …..He would urge,rather that, wherever a fixed extinguishingsystem is fitted in an endangered hold, it beactuated at once and the hold in question be kepttightly battened down until temperature, smokeand other tests indicate that the fire has beenbeaten down or suppressed.

The port authority responseThe ‘waiting game’ was indeed played whenMarion Green arrived alongside in Albany on29 July, although there was considerablesurprise and consternation on the part of theship’s master and officers that the local firebrigade was not in attendance on the wharfwhen the vessel berthed. The brigade, havingnot been advised of the ship’s correct arrivaltime by the agent, arrived to assess the situationapproximately 45 minutes later. Shortly after thevessel’s arrival, a meeting was held between themaster, the harbourmaster, a representative from

9 ‘Fire Aboard’ by Frank Rushbrook CBE, 3rd Edition 1998, Brown, Son & Ferguson Ltd. Glasgow.

25

Customs and the ship’s agent. The master’sinitial consternation was compounded by theimpression which he gained that the meetingwas focussed far more on administrativeprocedures and paperwork than on the measureswhich should be taken to deal with the fire.

Albany is a small port and the unit of the Fireand Emergency Service of Western Australia,based in the town, is also numerically small andhas little experience of ship fires and, indeed,little familiarity with ships. From the time ofMarion Green’s arrival it was evident that therewas a lack of co-ordination in the response tothe incident. The master expected the firebrigade to take charge of the firefighting, whilethe fire brigade expected the master to tell themwhat measures he wanted taken. The firebrigade suggested that the hold should be filledwith high-expansion foam but this was vetoedby the P&I Club surveyor who was concernedabout further damage to the high-value cargo.

The vessel’s owners in the Netherlandsconsulted with the Rotterdam port fire brigadeto obtain advice on how to deal with a fire in acargo of cocoa beans. It appeared, however, thatthere had been little experience in fighting suchfires and the advice forwarded to the master wasto flood the hold with more CO2, to inspect thehold later, then flood with more CO2, until thefire had been overcome.

On the morning of 30 July, the P&I Clubsurveyor requested that No. 2 hold be opened upbut this request was refused by the harbour-master who, correctly, required a further 24hours monitoring of the situation and whowanted to wait for the bulk CO2 to arrive.

On the following morning, 31 July, the holdtemperature appeared to be rising in spite of theearlier use of the bulk CO2. There had been,however, some confusion as to whether the bulkCO2 was actually getting into the hold, sincethere was uncertainty regarding the correctalignment of various valves on the tank and theaccuracy of the contents gauge.

At 1400 that afternoon, a meeting of involvedparties was convened by the harbourmaster, to

determine the future course of action. Nodecision had been taken at that stage as to whatcould be done with the damaged cargo.

The meeting revealed several conflictinginterests. The master and the vessel’s ownerswanted the fire extinguished as soon as possibleand the cargo discharged. The AustralianQuarantine and Inspection Service (AQIS) didnot want the cargo, which was not destined forAustralia, to be landed in Australia. Thisreluctance was shared by the harbourmaster,who did not want 3,000 tonnes of burned,soaked, cocoa beans landed on the wharf. Theharbourmaster also wanted Marion Green offthe berth to enable the waiting tanker, carryingfuel for the Albany area, to berth and, inaddition, there were concerns from theEnvironmental Protection Agency about anypossible environmental impact which mightresult from the incident.

Eventually, however, it was agreed that therewas no option but to discharge the cargo.

Albany Port Authority emergencyproceduresThe Albany Port Authority is a corporate bodyconstituted under the Port Authorities Act 1999and is responsible to the Minister for Planning& Infrastructure. Under the Act, the PortAuthority is required to have in place a compre-hensive Marine Safety Plan which is to includean Emergency Response Plan. At the time thatthe incident involving Marion Green occurredthis plan was in draft state and had not beensigned off by the Minister. (However such anEmergency Response Plan was in place prior tothe publication of this report.)

Albany Port Authority did have in place, at thetime of the incident, emergency procedures forfire, serious injury, collapse, chemical spillage,explosion or bomb threat. The procedures,however, contained very little and wereessentially directed at the wharf labour force.The only procedures relating to fire instructedSupervisors to clear the area of labour and orderthem to the amenities block, to call the fire

26

brigade, to advise the harbourmaster, to hold aroll call and liaise with the ship’s master.

It is the policy of the Albany Port Authority:

…to achieve Best Practice in the management ofall risks that threaten to adversely impact theAuthority, its customers, people, assets,functions, objectives, operations or members ofthe public. Risk Management will form part ofoperational and line management responsibilitiesand be integrated into the Business PlanningProcess10.

It is to the credit of the Albany harbourmasterthat he did accept Marion Green into the port asa port of refuge. The issue of ports of refuge,following the Erika and Prestige disasters inEurope, is one which is the cause of muchconcern and contention both among countries ofthe European Union and at the InternationalMaritime Organisation.

In terms of safety, the operation was a success,the fire was extinguished without injury or lossof life and the ship was able to resume itsvoyage, undamaged, and with the cargo in No. 1hold intact.

Authority in fighting ship fires inportThe question of who has ultimate authority (andresponsibility) in the case of a fire aboard shipin port, the master, the harbourmaster or thesenior fire brigade officer on the scene, is not asimple one. It is a question which has arisen anumber of times in the past, (eg ATSB ReportNo. 142, Fire aboard the Australian flag tankerTasman), and is one about which there is greatuncertainty in the minds of many on thoseinvolved.

In 1999, the Inspector of Marine Accidentscanvassed the senior officers of the fire andemergency services of all the Australian states

and the Northern Territory on theirunderstanding of the answer to this question.The responses varied enormously, some statingthat the fire brigade had complete authority andothers that the master retained completeauthority.

The response from Western Australia indicatedthat, in the opinion of the Fire and EmergencyServices Authority of Western Australia, themaster or chief engineer of the vessel remainsresponsible for the vessel at all times. Theresponse added, however, that in regard to thelegal aspects, the Authority believed that portsare beyond the Fire District and remain underthe control of the harbourmaster.

Consideration of this issue revealed a number ofanomalies between the various laws of thedifferent states and Commonwealth law. Severalstates have laws clearly detailing the authorityof the fire brigade, whereas the authority of themaster, although traditional, is less clearlydefined in terms of legislation. However,sections 6 and 278 of the Navigation Act 1912imply that ultimate responsibility does rest withthe master.

To further pursue these complex legal anomaliesis beyond the scope of this investigation. In theopinion of the Inspector, however, what is clearis that, whatever anomalies exist, where atrained fire brigade is involved in fighting a shipfire, the fire brigade should be regarded as theauthority on firefighting procedure, yet workingin close co-operation with the ship’s staff andharbourmaster.

In the Port of Albany’s Emergency ResponsePlan, promulgated since this incident but beforethe publication of this report, the roles of theharbourmaster, the ship’s master and theemergency services in such incidents have beenmore clearly defined.

10 Albany Port Authority, Internet www.albanyport.com.au/risk_management_policy

27

Conclusions

These conclusions identify the different factorswhich may have contributed to the cargo holdfire aboard Marion Green and should not beread as apportioning blame or liability to anyparticular individual or organisation.

1. It was not possible to positively identify thesource of the fire in No. 2 hold, althoughfour possibilities existed. These were:

a) Self-heating of the cargo of cocoa beanspromoted by the growth of a fungus,Aspergillus Niger. This may have beeninitiated at a damp patch caused byspillage of the liferaft water, taken by theloading gangs working in the holds, andexacerbated by poor ventilation in an areawhere the ventilation channels did notallow sufficient air flow.

b) Combustion initiated by the decompo-sition of aluminium phosphide intophosphine gas used for fumigation.Combustion is known to have occurred inother instances, particularly when there isan imbalance in the phosphorus contentduring preparation of the chemical.

c) Smoking materials discarded in the holdduring loading of the cargo. There is

considerable evidence of this in the formof empty cigarette packets, cigarette endsand matches found in the hold accessspaces and amongst the cargo once it hadbeen discharged onto the wharf at Albany.

d) A cargo light left in the hold after thecompletion of cargo loading.

2. The discovery of discarded smoking materialsin the discharged cargo indicates that thevessel’s ‘no smoking’ policy around theholds was not sufficiently policed by thedeck watch.

3. With the volume of cargo in No. 2 hold, andthe number of CO2 bottles discharged beforethe vessel arrived at Albany, it is unlikelythat the concentration of gas in the hold wasat any time sufficient to stifle the fire. Thenumber of bottles released was less thanindicated in the instructions provided by themanufacturer of the CO2 system.

4. Information provided to the ship’s staff on theshipping and stowage of cocoa beans, theirfumigation and the associated hazards wasinadequate.

5. The response to the fire, once the vessel hadberthed in Albany, was lacking in co-ordination and there was no clearunderstanding of who had the authority andresponsibility for actions to be taken indealing with the fire.

28

29

MR20030037 That Beluga Genchart BV take measures toensure that a strict ‘no smoking’ policy in thevicinity of cargo operations is enforced on boardtheir vessels.

MR20030038 That Beluga Genchart BV ensure that ship’smasters are provided with all relevantinformation regarding the stowage of cocoabeans, their fumigation and any associatedhazards. As a minimum, the IMO’s‘Recommendations on the safe use of pesticidesin ships’ should be available on board any shiploading a fumigated cargo.

MR20030039 That shippers, stevedores and ship’s officersshould all ensure that, when bagged cocoa beanor other organic cargoes are being loaded,adequate ventilation channels are provided andthat cargo is stacked with sufficient ‘cross ties’to ensure that the channels remain clear duringthe voyage.

MR20030040That Beluga Genchart BV take measures toensure that, on completion of loading cargoes,the deck watchkeeping officer makes an entry inthe deck logbook confirming that all electricalequipment in the holds has been isolated andstowed as applicable.

Recommendations

30

31

Submissions

Under sub-regulation 16(3) of the Navigation(Marine Casualty) Regulations, if a report, orpart of a report, relates to a person’s affairs to amaterial extent, the Inspector must, if it isreasonable to do so, give that person a copy ofthe report or the relevant part of the report. Sub-regulation 16(4) provides that such a personmay provide written comments or informationrelating to the report.

The final draft of the report was sent to thefollowing:

The Master, Marion Green

The Mate, Marion Green

Beluga Genchart BV

The Harbourmaster, Albany Port Authority

The Area Manager, Fire Services of WA

Australian Maritime Safety Authority

Submissions were received from BelugaGenchart BV and from the Harbourmaster,Albany Port Authority. The text of the reporthas been amended where appropriate.

32

Marion Green

IMO Number 9164029

Flag Netherlands

Classification Society Lloyd’s Register

Ship Type Multi-purpose general cargo

Builder Schelde Scheepsnieuwbouw B.V. - Vlissingen

Year Built 1999

Owner C.V. Marion Green

Ship Manager Beluga Genchart B.V., Rotterdam

Gross Tonnage 11 894

Net Tonnage 5 920

Summer deadweight 17 050 tonnes

Summer draught 9.689 m

Length overall 142.81 m

Breadth 21.5 m

Moulded depth 13.3 m

Engine 8 cylinder Wärtsilä 8L46B four-stroke diesel

Power 7 800 kW

Crew 13 – mostly Dutch & Filipino nationals

33

34

ww

w.a

tsb.

gov.

au18

00 6

21 3

72

Independent investigation into the fire in the hold of the Netherlands flag general cargo vessel Marion Green

off the coast of Western Australia, on 28 July 2002

ISS

N 1447-087X

ISB

N 1 877071 42 0

Marion Green. 11.03