For reasons of economy, this document is printed in a limited number. Delegates are kindly asked to bring their copies to meetings and not to request additional copies.

I:\FP\53\5-1.doc

INTERNATIONAL MARITIME ORGANIZATION

IMO

E

SUB-COMMITTEE ON FIRE PROTECTION 53rd session Agenda item 5

FP 53/5/1 6 October 2008 Original: ENGLISH

MEASURES TO PREVENT EXPLOSIONS ON OIL AND CHEMICAL TANKERS

TRANSPORTING LOW-FLASH POINT CARGOES

Extracts from the reports of investigation into the �Chassiron�, �Panam Serena� and �Bow Mariner�

Note by the Secretariat

SUMMARY Executive summary:

This document conveys the relevant parts of the reports of investigation into the Chassiron, Panam Serena and Bow Mariner

Strategic direction:

5.2

High-level action:

5.2.3

Planned output:

5.2.3.4

Action to be taken:

Paragraph 3

Related document:

FP 52/21, sections 15 and 20

1 The Sub-Committee will recall that the Committee, at its eighty-third session, agreed to refer the reports of investigation into the Chassiron, Panam Serena and Bow Mariner casualties to the Sub-Committee for consideration in the context of the work on incidents of explosions on chemical and product tankers. In this context, FP 52 noted that the analyses, as well as the full investigation reports on the aforementioned casualties, are now available to Members in the GISIS module on Maritime Casualties and Incidents. 2 The relevant parts of the reports of investigation into the Chassiron, Panam Serena and Bow Mariner to be considered by the Sub-Committee have been reproduced in annexes 1 to 3, respectively, for ease of reference. Action requested of the Sub-Committee 3 The Sub-Committee is invited to note the above information and take action as appropriate.

***

FP 53/5/1

I:\FP\53\5-1.doc

ANNEX 1

EXTRACTS FROM THE REPORT OF INVESTIGATION INTO THE �CHASSIRON� EXPLOSION ON BOARD THE OIL TANKER �CHASSIRON� 1 Extracts from the Circumstances 1.1 On 12 June 2003, the Chassiron called at Bayonne from Donges (France) to unload her cargo consisting of 3 parcels distributed as follows:

.1 cargo tanks 1 (P & S): domestic heating oil; .2 cargo tanks 2, 3, 4 and 5 (P & S): gas oil; and .3 cargo tanks 6 (P & S): unleaded mogas (98 octane).

1.2 The ship left Bayonne to Donges on the next day to take on an identical, but differently distributed cargo load. After the departure, the tank washing operations started on tanks 1 (P & S) and 6 (P & S), when there was a very loud whistling sound immediately followed by an explosion and fire in cargo tank 6. One person died and the deck of the vessel was ripped open from the bridgehouse to the manifold and the bulkheads of cargo tanks 5 and 6 were severely damaged. 2 Extracts from the Main particulars of the ship 2.1 Construction 2.1.1 The Chassiron is a French flag, double-hull IMO II combined petroleum products/chemical tanker. She was built in 1999, in the Netherlands, and is one of a series of four similar ships built by the same yard for the same owner/operator, PETROMARINE, which run a fleet of nine ships of the same type and two small oil tankers. She is owned by the consortium Dakar. 2.1.2 The cargo area is arranged with 12 cargo tanks and 6 segregations, plus a slop tank of 184 m3. The cargo tanks are separated by vertically corrugated bulkheads which, combined with the fact that the deck stiffeners are outside the tanks, means that the tanks sides and top are flush and therefore easy to clean. 2.1.3 Stripping of the cargo tanks is carried out through the cargo pumps by means of compressed air or inert gas at a pressure of 6 bar. 2.1.4 The tanks are equipped with a thermal fluid heat exchange system which enables the cargo to be maintained at a temperature of 65°C for the carriage of products which require heating. 2.1.5 The vessel is fitted with an inert gas system comprising a Smit Sinus nitrogen generator and a 5,000 l storage tank at 8 bar.

FP 53/5/1 ANNEX 1 Page 2

I:\FP\53\5-1.doc

2.2 Safety certificates 2.2.1 When the accident happened all her safety and pollution prevention certificates were valid. Her target factor in the context of the Memorandum of Paris was 20 and she had never been detained. 2.2.2 The last annual safety inspection was conducted on 8 January 2003 by the Bordeaux Ship Safety Centre. 2.2.3 The safety management certificate (ISM Code) was valid until 22 June 2005. 2.2.4 The Bureau Veritas carried out an annual survey on 13 December 2002, in Bordeaux, in compliance with the provisions of the Survey Guidelines under the harmonized system of survey and certification (resolution A.746(18)). There was nothing to report, the vessel was in satisfactory condition and no reservations were made which could have had any bearing on the accident. 2.2.5 The insulation report of 5 November 2001 revealed no anomalies. 3 Extracts from Determining and commenting on the causes of the accident 3.1 Possible causes of the accident 3.1.1 Risk analysis shows that an explosion could not have occurred without the formation of an explosive atmosphere (ATEX) and the presence of an ignition source within the said atmosphere. 3.1.2 For a given mixture, we can therefore consider that there is an auto-ignition temperature limit (for a given pressure) and an auto-ignition pressure limit (for a given temperature). 3.1.3 The presence of an ATEX 3.1.3.1 As cargo tanks 6 (P & S) had contained unleaded grade 98 mogas, they were full of vapour after unloading. 3.1.3.2 The normal operation of the pressure venting valves during unloading, and the opening of the small inspection hatch for inspection of the tanks after unloading, meant that air was able to enter the tanks. Not enough air entered the tank to bring the air/mogas mixture below the lower flammable limit, but sufficient for there to be a very flammable mixture in the tank when the tank was cleaned. 3.1.3.3 Moreover, as the mogas vapour is heavier than air, it tends to build up at the bottom of the tanks. This difference in density resulted in a gradient of unleaded grade 98 vapours with concentrations decreasing progressively from the bottom to the top of the tanks. 3.1.3.4 The following table summarizes the possibility of there being an air/unleaded grade 98 mogas ATEX and an air/gas oil ATEX after tanks 5 and 6 had been loaded and unloaded.

FP 53/5/1 ANNEX 1

Page 3

I:\FP\53\5-1.DOC

Volume (V1)

of the vapour space after tank loading (m3)

Presence of an ATEX in the vapour space after the tanks are loaded

Presence of an ATEX in the vapour space after the tanksare unloaded

Tank 6 port

16 No, because the concentration of mogas vapour in the vapour space was 44% well above the UEL1 (7.6%).

Unloading the tank results in dilution of the air in the tank. Volume V1 increases and as a result dilutes the mogas vapours (by a dilution factor of about 40%). The average mogas vapour content is around the LEL2 after unloading and it can be concluded that there was an ATEX during tank cleaning operations.

Tank 6 starboard

20 Idem above.

Idem above.

Tank 5 port

20 No, because the temperature of the gas oil (31.7°C) was lower than the flash point (> 55°C). The vapour space contained gas oil vapour which could ignite if heat was applied.

No, but the tank was not degassed and the gas oil vapours could auto-ignite in the event of an external source of ignition being present.

Tank 5 starboard

24 Idem above. Idem above.

Notes:

1 Upper Explosive Limit (UEL) 2 Lower Explosive Limit (LEL)

3.1.3.5 We can, therefore, positively assert that there was an air/unleaded grade 98 mogas ATEX in tank 6 (P & S) and that its varying concentration (higher towards the bottom of the tank) fell within the flammable range. 3.2 How the explosion developed 3.2.1 The observations made by the BEAmer (Bureau enquête � accidents / mer) investigators, the analysis of the explosion made by the INERIS (Institut National de l�Environnement Industriel et des Risques) specialists, as well as the statements taken from the crew, led them to the conclusion that the most likely scenario for the explosion was as follows:

FP 53/5/1 ANNEX 1 Page 4

I:\FP\53\5-1.doc

− phase 1: the presence in tank 6 (P & S), during tank cleaning operations, of an air/unleaded grade 98 mogas ATEX the fuel concentration of which was variable (higher towards the bottom of the tank) but within the flammable range.

− phase 2: an initial explosion of the air/unleaded grade 98 mogas ATEX, which,

taking into account the extremely serious damage observed, must have occurred in tank 6 (S), the ignition source being of mechanical or electrostatic origin. The explosion blasted the deck plating over tank 6 (S) into the sea, breached the transverse bulkhead between tanks 6 (S) and 5 (S) and the longitudinal bulkhead between tank 6 (S) and tank 6 (P). The explosion propagated into tank 6 (P); tank 5 (S) apparently did not explode, but judging by the damage incurred, suffered blast damage from the explosion in tank 6 (S)

− phase 3: a second explosion of the air/unleaded grade 98 mogas ATEX in

tank 6 (P) caused by the heat of the first explosion. The explosion then propagated into tank 5 (P). The plating over tank 6 (P) was torn from the deck (it was not thrown into the sea but remained hanging over the port side of the vessel against the sideshell plating). The transverse bulkhead between tanks 6 (P) and 5 (P) was breached.

− phase 4: A third explosion caused by auto-ignition of the gas oil vapours in

tank 5 (P). The blast of the explosion projected the bulkhead separating tanks 5 (P) from 6 (P) into the after part of tank 6 (P).

3.2.2 Most of the transverse bulkhead between tanks 5 (S) and 6 (S) was pushed into tank 5 (S) by the force of the explosion but part of it remained in place. 3.2.3 After studying the damage, notably the way in which the bulkheads separating the tanks buckled, it is possible to assert that the explosion in tank 6 (S) was more powerful than that in tank 6 (P), which was, in turn, more powerful than the one in tank 5 (P). 3.2.4 All three explosions followed the deflagration regime (where combustion of the mixture propagates through a subsonic wave). Tank 5 (S) is not thought to have exploded but to have been damaged by the blast of the explosion in tank 6 (S). 3.2.5 In a closed space, the propagation of a deflagration releases high temperature, high pressure gases. Taking into account the mechanical strength of the structures, the pressure necessary to break the deck plating over tank 6 (S), as calculated by digital simulation (EFFEX modelization software) is about 1.5 to 2 bar. 4 Extracts from the Conclusions 4.1 Up to now it has not been possible to determine unequivocally the origin of the ignition source which caused the explosion. 4.2 The air/unleaded grade 98 mogas ATEX in tank 6 (S) only needed a few microjoules energy to ignite. 4.3 According to the analysis of the accident, the damage sustained was the result of the domino effect of a series of successive explosions (three in all) in a deflagration regime.

FP 53/5/1 ANNEX 1

Page 5

I:\FP\53\5-1.DOC

4.4 The observations made by the BEAmer investigators and the INERIS specialists favour the hypothesis that the first explosion took place somewhere near the bottom of tank 6 (S) (in all likelihood near the cargo pump), followed by a second explosion in tank 6 (P) caused by the propagation of the heat of the first explosion and a third and final explosion due to the ignition of the gas oil vapour in tank 5 (P). 4.5 The noise heard just before the explosion, which was described as a whistling sound, could have been the noise made by turbulent combustion in a small confined space and, as such, would have been the "initial" event characteristic of the explosion. It could also have been due to a rise in pressure inside in the tank, the noise being made by gases escaping through the small inspection hatch just before the explosion, or again, it could have been due to friction between moving parts. 4.6 Bearing in mind the low flash point of the unleaded grade 98 mogas and its temperature (25.2°C) before unloading commenced , it can be affirmed that the weather conditions had no influence on the formation of an air/unleaded grade 98 mogas ATEX during unloading. 4.7 Tank cleaning was carried in the usual way. The pumpman was very experienced as regards tank cleaning but human error cannot be excluded (e.g. dropping a tool). 4.8 As regards fire fighting, the destruction of the fire line on deck and the absence of sectioning valves on the engine-room fire main meant that the fire fighting system was not immediately available (it was necessary to wait for the damaged section to sectioned off by means of a plug). Further sectioning valves should be installed so that the fire main in the engine room remains available for use in the event that other sections of the system become unserviceable. 4.9 Finally, as a preventive measure, the use of electrostatically non-insulating paints or coatings for tank surfaces should be preferred. 5 Extracts from the Recommendations 5.1 Reduce the risk of an explosive atmosphere forming in the tanks 5.1.1 It is advisable to prevent air from entering the tanks during discharging and tank washing. It would seem difficult, however, in practice, to keep the vapour concentrations in the tank atmospheres permanently above the upper explosive limit. 5.1.2 Thus, opening the small tank inspection hatches for visual inspection after discharging as well as during tank washing operations are practices which need to be reappraised. Loading, discharging and washing operations, should be carried out in closed operation mode (closed loading, closed discharging, closed washing), the only air entering the tanks being due to the normal operation of the pressure venting valves.

FP 53/5/1 ANNEX 1 Page 6

I:\FP\53\5-1.doc

5.2 Reduce the possibility of ignition sources appearing 5.2.1 Operational level 5.2.1.1 There is a risk of sparks being produced by discharge of static electricity or of an increase in temperature (creation of hot spots) due to friction between moving parts during tank washing operations. 5.2.1.2 Tank cleaning operations should be kept to a strict minimum insofar as, on modern ships, the quantity of residual cargo in the tanks after discharging and stripping is very small (a few tens of litres at the most ). 5.2.1.3 Arrangements could be sought between owners and charterers, within the framework of charter-parties, to limit the numbers of tank washings according to the type of product carried or to merely rinse out the tanks. 5.2.2 Equipment level 5.2.2.1 Installations should be permanently monitored to ensure that their electrical continuity and equipotentiality is maintained. Sounding pipes should go right down to the bottom of the tanks, particularly where ships not equipped with inert gas systems are concerned. 5.2.2.2 Inspection and checks of mechanical installations should be reinforced and particular care should be taken with tightening and locking the assembly and fixing nuts on submerged pumps and their drive units, as well as any other equipment in the tanks 5.3 Inerting of tanks loaded with volatile petroleum products with a flash point below 60°C 5.3.1 International regulations (1974 SOLAS Convention, as amended, regulation II-2/4.5.5) impose inert gas systems only on tankers of 20,000 tonnes deadweight and upwards. 5.3.2 Tankers of less than 20,000 tonnes deadweight, including chemical tankers, should be required to be fitted with inert gas systems for the protection of cargo tanks, in light of recent accidents and improvements in ship technology and operational procedures. Consequently, regulation II-2/4.5.5 of the 1974 SOLAS Convention, as amended, should be amended to take this matter into account.

***

FP 53/5/1

I:\FP\53\5-1.doc

ANNEX 2

EXTRACTS FROM THE REPORT OF INVESTIGATION INTO THE �PANAM SERENA�

EXPLOSION ON BOARD THE CHEMICAL TANKER �PANAM SERENA� 1 Extracts from the Summary 1.1 The Bahamas registered chemical tanker Panam Serena, which was built in Turkey, entered service in June 2003, exploded and caught fire at Porto Torres, Sardinia in Italy on 01 January 2004. The catastrophic damage caused resulted in the vessel being declared a constructive total loss (CTL), two crew members were tragically killed and one was injured. 1.2 The Panam Serena had arrived at Porto Torres on 31 December 2003 with a cargo of Benzene and Cut C6 (C6). The Benzene discharge had been completed and the vessel was close to completion of discharge of the C6. All cargo tanks were loaded upon arrival at Porto Torres, except No.4C tank which was washed, clean and dry. 1.3 On 01 January 2004, as the cargo deck watch was changing, the vessel was shaken by the first in a series of violent explosions, which resulted in an intense fire amidships, within the cargo tank area of the vessel. The seaman on duty and the seaman taking over the deck watch were outside on the main deck and were both tragically killed in the series of explosions. The Chief Mate, who was resting in his cabin at the time, was injured. 1.4 Due to the catastrophic damage caused to the vessel it has been extremely difficult to identify the exact cause of the initial explosion; however a number of possible causes were identified and are covered within the report. 2 Extracts from the Narrative of events 2.1 The vessel loaded its cargo of Benzene (2,091 tonnes) and C6 (6,300 tonnes) at Rotterdam and Dunkirk, without incident, for a full discharge at Porto Torres. The vessel had undergone a satisfactory Bahamas Flag State inspection while loading at Rotterdam, no deficiencies had been noted. All cargo tanks were utilized for loading the cargoes except No.4 Centre tank. This tank was not required for the quantity of cargo which was being carried and some repairs were required to the tank coating. These repairs were completed during the sea passage from northern Europe to the Mediterranean. The sea passage was uneventful and the vessel arrived on the morning of the 31 December 2003, being all fast at Berth No.18, Platform B in Porto Torres at 08:50. 2.2 Upon arrival the vessel berthed, was made fast by the dock workers without incident and the usual port arrival formalities were observed. The shore gangway was placed on board between the vessel and quay, port clearance was arranged via the Harbour Masters Office and the cargo surveyor attended on board. Cargo measurement and sampling were then completed. The cargo and vapour return pipelines were connected by the terminal staff in preparation for the discharge of the cargo.

FP 53/5/1 ANNEX 2 Page 2

I:\FP\53\5-1.doc

2.3 The connection between the quay and the vessel by the terminal personnel is a part of the usual vessel arrival routine at Porto Torres. This is a requirement of the terminal procedures, there is some doubt that the connection of the bonding cable was made or made correctly upon the vessel's arrival. 2.4 The vessel started discharging the C6 at about 16:00 and started discharging the Benzene at 18:00 on the 31 December 2003, the discharge operation was proceeding in a routine manner up until the time of the first and reportedly, the most violent, in a series of approximately four explosions (approximately 11:55 on the 01 January 2004). 2.5 The watch system adopted on board the Panam Serena for cargo operations at Porto Torres was satisfactory, ensuring that there were sufficient crew members on duty at all times and that they were adequately rested in between duty periods. 2.6 Up until the time of the first explosion the discharge operation was proceeding in a routine manner without any problems, the discharge of the three cargo tanks containing Benzene (No.1C, 3P & 3S) had been completed at 06:45 on the morning of 01 January 2004. The same morning at 09:30, the terminal staff had connected a fresh water hose from the jetty to the ship, for the purpose of providing fresh water to the vessel. The fresh water was used in order to flush out the hazardous / toxic cargo from the cargo discharge lines prior to disconnection; this was the usual terminal practice at Porto Torres. 2.7 The time of the initial explosion on board the Panam Serena could be placed quite accurately by all members of the crew because they have a common reference time with respect to the change of the watch at 12:00 and the mid-day meal. Members of the crew were either: preparing to end their watch; take over the watch; taking their meal break early or intending to eat their meal later. The split meal times were to ensure continuity of personnel coverage on duty and are common practice on most vessels around meal times; the arrangements are usually made by mutual consent. 2.8 Statements by the crew members described the sensation of the Panam Serena being shaken by a sharp jolt, as if the vessel had been hit or rammed by another ship, immediately followed by a single and very loud explosion. The initial explosion was quite separate and distinct from the series of (approximately three more) explosions which followed the initial explosion; these explosions occurred in succession between, approximately one minute and a few minutes later. 2.9 The vessel immediately listed heavily to Starboard at this time, with many of the crew fearing the vessel would capsize and after a brief attempt by some crew members to fight the intense fire, the crew made their way towards the stern of the vessel, in accordance with the Masters orders in preparation to abandon ship. The starboard list stabilized as the series of explosions ended. 2.10 From the terminal's perspective the discharge operation had proceeded quite normally until shortly before the initial explosion on board the Panam Serena, when the terminal were experiencing some problems on another vessel which had just arrived at Porto Torres on the morning of 01 January 2004 and was berthed close by to the Panam Serena at berth No.13, on Platform B. The terminal personnel on duty and monitoring the discharge operation of the Panam Serena were summoned by their manager to assist their colleagues on the other vessel which had just arrived. Therefore, there were no terminal personnel in attendance on the berth for the Panam Serena, at the time that the vessel exploded.

FP 53/5/1 ANNEX 2

Page 3

I:\FP\53\5-1.DOC

2.11 Drug and alcohol tests were performed on all members of the crew by the Italian Authorities following the casualty, including the two seamen who were tragically killed, were completely negative. During the Italian Police investigation on board the vessel, no alcoholic beverages of any description were found anywhere on board, the police search included store rooms, recreational areas and crew cabins. The police findings were in line with Company policy, which prohibited drugs and alcohol on board the vessel. 2.12 There was a safe smoking room provided on board the vessel, located within the vessels accommodation and situated close to the galley, this room was utilized by crew members on board the vessel who smoked. There was no indication that any crew member was smoking on board the vessel in an unauthorized area. 3 Extracts from the Analysis 3.1 The experience of the crew; the majority of the crew were very well experienced in all types of tanker operations, especially chemical tanker operations and it is unlikely that the casualty was the result of crew error, misconduct and/or negligence during the course of the discharge operation. 3.2 The catastrophic damage caused to the vessel indicates that the explosions took place inside the cargo tanks. The eyewitness evidence obtained by both the Italian Police and the Bahamas Approved Inspector who attended the scene immediately following the casualty, place the location of the first explosion in the amidships area of the vessel or slightly forward of amidships. This was in the vicinity of the vessel's cargo discharge manifold, within the cargo tank section of the vessel. The witnesses clearly described the initial explosion and fire as taking place at some distance away from the vessel's accommodation. The majority of the crew were located within the accommodation and witnessed the initial explosive damage, together with the early stages of the fire. There is substantial evidence proving that further explosions took place within other cargo tanks, some of which still contained cargo and were closer to the vessel's accommodation, as the emergency situation quickly escalated. 3.3 The majority of the cargo tanks had been discharged and were empty of cargo at the time of the first explosion, however the empty tanks were still full of potentially volatile vapour both from the small amount of residual cargo remaining within the tanks and due to the fact that vapour had been returned to the vessel via the vapour return line (VRL) from the terminal, throughout the discharge operation. A VRL is often utilized in hazardous chemical cargo loading and discharge operations, in order to retain the hazardous cargo vapour within a closed cycle, returning the vapour from the shore to the ship, as in this incident, or visa versa. There were only three cargo tanks (6P, 7P and 7S) still being discharged with a relatively small quantity of C6 cargo remaining in each tank. 3.4 The cargo pumps fitted to the Panam Serena were Marflex deepwell pumps, i.e. each cargo tank was fitted with its own discharge pump. The Marflex pumps are designed to extract the maximum amount of product from each tank and are fitted with a main discharge line, as well as a narrower stripping line. Upon nearing completion of discharge when the bulk of the product has been discharged from a tank, the valves are set to the stripping mode, the main discharge line is purged with inert gas or air and the final quantity of product is discharged ashore via the separate and smaller stripping pipeline, minimizing the cargo residue remaining within the tank.

FP 53/5/1 ANNEX 2 Page 4

I:\FP\53\5-1.doc

3.5 The connection of the bonding cable is usually the first operation to be performed after the gangway has been placed on board, when the vessel has received port clearance, prior to the connection of the cargo hoses and start of the discharge operation. However, no members of the terminal personnel on duty at the time can remember who made and checked the connection or state categorically that they were the person who made the connection. Although a number of terminal personnel stated that they were sure the connection had been made, probably by someone else. 3.6 Some members of the ships staff stated that the connection of the bonding cable was not made between the terminal and the ship. It should be noted that the industry recommendations are that Bonding Cables should not be used between the terminal berth and the vessel. 3.7 There is sufficient evidence to demonstrate that it is possible for a large static or electrical charge to have accumulated within the structure of the Panam Serena during the course of the discharge operation. Many of the circumstances which were contributory to previous accidents are also evident in the case of the Panam Serena, including:

.1 the dangers associated with the loading, carriage and discharge of refined liquid products, which tend to be "Static Accumulators". Charge generation and separation occur when the liquid moves in contact with other materials, such as piping etc. The risk is increased during the early stages of loading and when "stripping" the tanks during discharge, when the tanks are at their lowest level;

.2 the dangers associated with the introduction of impurities into a liquid product,

such as water. Static is generated through friction with the water droplets, producing a high voltage at the liquid interface. Water, was used to flush the lines of hazardous cargo upon completion of discharge. The water hose had been connected from the terminal to the vessel for this operation;

.3 the release of air and / or inert gas into a liquid can generate a strong electrostatic

charge, by bubbling action and agitation of the fluid. This was a standard practice required within the operating procedure for the deepwell pumps fitted on board. The vessel was fitted with a small supply of nitrogen in bottles, it has not been ascertained if air or nitrogen was utilized from the ship or shore supply during this operation;

.4 within the Italian Criminal Courts report on the casualty, great emphasis was

placed upon the correct connection of the bonding cable by the terminal. However the guidance (International Chamber of Shipping (ICS) Tanker Safety Guide (Chemicals) and the International Safety Guide for Oil Tankers and Terminals (ISGOTT)) on this subject is that, a ship/shore bonding cable is not effective as a safety device and may even be dangerous! A ship/shore bonding cables should therefore not be used. ICS and ISGOTT acknowledges that although the dangers associated with ship/shore bonding cables are widely recognised, attention is drawn to the fact that some national and local regulations may still require them to be used. The terminal procedures at Porto Torres required the Panam Serena to be fitted with a bonding cable supplied by the terminal, to try and ensure electrical continuity between the terminal and the ship. This cable was probably not connected, or if it was connected it is possible that it was not correctly connected upon the vessels arrival;

FP 53/5/1 ANNEX 2

Page 5

I:\FP\53\5-1.DOC

.5 the terminal was utilizing a bonding cable within their procedures, attempting to achieve electrical continuity between the terminal and the vessels which berthed alongside. There was no indication within any reports that "insulation flanges" were used within the discharge hose string and in view of the terminal policy for electrical continuity, the use of insulation flanges, would seem unlikely. Insulation flanges are generally used where the terminal policy is to insulate the vessel from the terminal, in order to create electrical discontinuity;

.6 Benzene / C6 vapour is heavier than air and it is quite possible that towards the

end of the discharge operation that volatile vapour had accumulated around the vessel. A VRL was in use, returning cargo vapour under pressure, to the vessel from the shore tanks. One of the seamen who had been on deck duty and was killed in the accident, had been wearing a gas-vapour mask, commonly used on chemical tankers. This may indicate the presence of gas vapour around the deck area or that an access to a cargo tank was being opened for operational reasons. Cargo tank 6P was stripping and the crew were in the process of preparing the fresh water hose for line flushing. The good weather conditions prevailing at the time would have contributed to any accumulation of gas vapour around the vessel; and

.7 no mention has been made within the terminal personnel statements with respect

to any cathodic protection fitted to the jetty; if fitted, cathodic protection is another source of difference in electrical potential between vessel and terminal jetty.

3.8 Examination of the bonding cable by investigators determined that it was partly corroded internally and not well maintained, this corrosion would have affected its electrical continuity, even if it had been connected between the terminal and the vessel correctly. The examination of the bonding cable also determined that it had suffered heat damage as a result of the fire, due to the transmission of heat along a length of the cable from the metal clamp, which was usually used to connect the cable to the vessel on the terminal berth. 3.9 Further investigative work is still required to establish if there was a problem with one of the deepwell cargo pumps. There was some evidence to suggest that this may have been the case, however the overwhelming evidence within the witness statements, with respect to the location of the initial explosion and fire is not consistent with the theory that a cargo pump problem caused the initial explosion. 4 Extracts from the Conclusions 4.1 The most probable cause of the initial explosion was due to a static or electrical discharge of sufficient strength to create an ignition source within a volatile environment which had developed on board the vessel. Igniting an air / Benzene and / or C6 vapour mixture, which being heavier than air, had accumulated within the vicinity of the vessel. While the majority of the cargo had been discharged, the vessel's tanks were full of Benzene and C6 vapour, which had been returned to the vessel from the shore reception tanks throughout the discharge operation. 4.2 The guidance issued by ICS and ISGOTT with respect to the recommended precautions concerning electrical continuity, the use of bonding cables and/or electrical insulation (including, insulation flanges) between the jetty and the vessel was disregarded by the terminal operator.

FP 53/5/1 ANNEX 2 Page 6

I:\FP\53\5-1.doc

4.3 The vessels crew did not check and confirm with the terminal that the bonding cable was in good condition and correctly connected in order to ensure the safety of the vessel. While there was a clear terminal responsibility, with respect to the application of national and local requirements, the master, officers and crew had a duty to ensure the safety of the vessel and those on board. 5 Extracts from the Recommendations 5.1 The Panam Serena due to her size was not required to be fitted with a nitrogen inert gas system, such systems are not mandatory on Chemical tankers of this tonnage. However the owners / managers, following this incident have fitted nitrogen inert gas systems to all subsequent vessels of this size and class. There is an obvious cost implication with respect to this action, which the owners have decided to accept in order to enhance safety. The responsible and expert industry bodies are expected to submit their views and proposals to IMO, on the requirements for all chemical tankers to be fitted with Nitrogen inert gas systems. 5.2 There is a clear need for agreement on International Standards to be adopted with respect to the precautions required to minimize the risks associated with static, electrical charge generation and discharge. The safety precautions applicable with respect to shipping as an international industry should not be subject to differing national and local regulations, with respect to such a fundamental safety matter.

***

FP 53/5/1

I:\FP\53\5-1.doc

ANNEX 3

EXTRACTS FROM THE REPORT OF INVESTIGATION INTO THE �BOW MARINER�

EXPLOSION AND SINKING OF THE CHEMICAL TANKER �BOW MARINER� 1 Extracts from the Summary 1.1 At 1805 on Saturday, 28 February 2004 the chemical tanker Bow Mariner caught fire and exploded while the crew was engaged in cleaning residual Methyl Tert Butyl Ether (MTBE) from cargo tank number eight starboard. The ship sank by the bow at 1937 in position 37-52.8N/074-15.3W, about 45 nautical miles east of Virginia (United States). Of the 27 crewmembers aboard, six abandoned ship and were able to make it to an inflatable life raft and were rescued by the Coast Guard. An unknown number of other crewmembers abandoned ship to the water. The Coast Guard and Good Samaritan vessels recovered three of these crewmen from the water, one deceased. The other two died before reaching a hospital. 18 crewmen remain missing and are presumed dead. The vessel�s cargo of ethyl alcohol (3,188,711 gallons) was released, along with the vessel�s heavy fuel oil (192,904 gallons), diesel fuel (48,266 gallons) and slops (quantity unknown). 1.2 The cause of this casualty was the ignition of a fuel/air mixture, either on deck or in the cargo tanks, that was within its flammable limits. The ignition source could not be precisely determined. 2 Extracts from the Jurisdiction The Bow Mariner exploded and sank off the coast of Virginia, within the United States Exclusive Economic Zone (EEZ), resulting in the total loss of the ship, loss of life and severe pollution; therefore, this casualty is a very serious casualty as defined in the Code for the Investigation of Marine Casualties and Incidents (IMO resolution A.849(20)) (hereinafter the �Code�). In accordance with section 4.11 of the Code, Greece, the Philippines and the United States are substantially interested states, and Singapore was initially the lead investigating state. Accordingly, the Coast Guard made immediate notifications to Greece, the Philippines and Singapore. On 29 February 2004 the Maritime and Port Authority (MPA) of Singapore verbally designated the United States as lead investigating state in accordance with article 7 of the Code. 3 Extracts from the Finds of fact 3.1 Vessel data 3.1.1 Vessel history The Bow Mariner, which is a Singaporean flag, was constructed at the Brodogradiliste Split Shipyard, in what was then Yugoslavia, between July 1981 and October 1982, under the name Atlas Petros. Odfjell Tankers Asia II PTE Ltd. (hereinafter �Odfjell�) purchased the vessel in 2000. Ceres Hellenic Shipping Enterprises Ltd. (hereinafter �Ceres�) operated the ship.

FP 53/5/1 ANNEX 3 Page 2

I:\FP\53\5-1.doc

3.1.2 Ceres Hellenic Ship Enterprises, Ltd. 3.1.2.1 Ceres is a ship management company based in Piraeus, Greece. The company manages a fleet of 39 vessels. 3.1.2.2 A review of the Coast Guard histories of Ceres vessels that have called on United States ports in the last five years revealed few significant deficiencies or violations. One exception was the detention of the Bow Power in Philadelphia, Pennsylvania, in February 2004 after a boarding revealed the cargo tanks were not inerted below eight percent oxygen as required; however, the requirements for inerting tanks on the Bow Power were different from the Bow Mariner because the Bow Power was built after 1 July 1986. The condition was corrected by inerting the tanks. 3.1.3 Vessel description The Bow Mariner was a single side, double bottomed chemical and oil tanker capable of carrying 28 different cargoes with double valve segregation. Two longitudinal bulkheads and ten transverse bulkheads, forming nine center tanks and nine pairs of wing tanks, subdivided the cargo space. The number nine center tank was subdivided by a longitudinal bulkhead to form two tanks, which were designated as slop tanks. The center tanks had external stiffening members and the wing tanks had internal stiffening members. The tank bottoms were sloped to facilitate drainage to the sumps. All tanks were constructed of mild steel and were coated with zinc or epoxy. The total capacity of all cargo tanks was 47004.8 cubic meters at 98 percent full. 3.1.4 Inspection history The Bow Mariner was classed as a +1A1 Tanker for Chemicals and Oil by Det Norske Veritas (DNV) on 21 September 2002. Survey records for the five-year period before the explosion revealed the number, type and frequency of deficiencies was typical for a vessel of this age, service and route. There were no outstanding conditions of class on 28 February 2004. 3.1.5 Wilden pumps The Bow Mariner carried three Wilden air-driven, double-diaphragm, positive displacement pumps, used to pump liquids in hazardous environments. One large pump was intended for useas an emergency cargo pump and two smaller pumps were part of the spill response kit. On 20 February 2004 Captain Kavouras requisitioned 14 a new Wilden T4/SPPB/TF/STF pump and additional spare parts, including several diaphragms. He made the following statement in the requisition remarks block: �On board exist tow air operated double diaphragm pums for OPA equip. The one of them is new and in good working condition the other is very old and is out of orders.� 3.1.6 Inert Gas System (IGS) 3.1.6.1 Chemical tankers built before 1 July 1986, carrying flammable cargoes other than crude oil or petroleum products, are not required to have an IGS. However, the Bow Mariner was constructed with an inert gas generator, a nitrogen generator and a bottled nitrogen storage unit.

FP 53/5/1 ANNEX 3

Page 3

I:\FP\53\5-1.DOC

3.1.6.2 The inert gas generator produced hot inert gas in a water-jacketed combustion chamber. Sulphur oxides were washed out and the gas cooled in the cooling/washing section of the combustion chamber. The gas was further cooled and scrubbed in the washing tower, which had a demister at the top to remove water. Inert gas pressure was maintained by one of two centrifugal blowers, feeding combustion air, and a pressure control valve in the gas outlet line. The gas was cooled even further in an R-22 refrigeration unit, which lowered the temperature and removed additional water. The inert gas generator was situated in the non-hazardous area of the vessel and was equipped with non-return devices comprised of �block and bleed� valves, supplemented by a non-return valve and deck isolation valve. These valves were upstream of a pressure vacuum breaker that protected the cargo tanks from overpressure or vacuum conditions. The IGS main ran forward to the number one cargo tanks, with individual branch lines serving each cargo tank and the manifold. The IGS was also capable of providing fresh air for gas freeing using a mode selector switch on the local control panel. In this mode the burner is not ignited. 3.1.6.3 The IGS was initially capable of producing nitrogen, which was stored in an eight cubic meter storage tank in the IGS room, for use as a cargo pump purging medium or cargo padding. Inert gas was drawn downstream of the refrigeration unit by a compressor, discharged to a CO2 stripper and then stored in a compressed gas accumulator tank. However, several years before the explosion, the nitrogen generation system was disabled with the approval of DNV. The tank remained in the IGS room but was not used. 3.1.6.4 The bottled nitrogen storage, located in the forecastle, was used to pad sensitive chemical cargoes when requested by the cargo shipper or receiver. The storage room could accommodate up to 48 compressed gas bottles, all of which could be connected to a manifold with two regulating valves to reduce the supply pressure based on whether the nitrogen was being used for purging or padding. Two supply outlets were located outside the forecastle bulkhead and were connected to the nitrogen main deck line with a flexible hose. 3.2 Drug and alcohol testing Sentara Norfolk General Hospital tested five of the six survivors for drugs and alcohol by blood test shortly after arrival and well within eight hours of the casualty. The drug test results for one crewman, Electrician Bactat, were not recorded. Another crewman, Messman Tagle, was not tested for drugs or alcohol for unknown reasons. The results of all drug and alcohol tests taken were negative. 3.3 Fatigue 3.3.1 96-hour work/rest histories were collected for all six survivors. Deck officers below the chief officer stood four-hour watches in a three-watch system while at sea, and a modified watch system in port. Deck ratings stood a similar watch schedule, except the watch rotated periodically so that each crewman stood an equal share of undesirable watches. Engineering officers below the chief engineer were day workers and did not stand a watch. Engine ratings stood four-hour watches in a three-watch system. 3.3.2 When not on watch the crew performed their routine duties, which varied according to their position and rank. The crew had 15-minute coffee breaks at 1000 and 1500, and ate lunch at 1200 and dinner at 1700. Other than watch standers the crew secured from work and ate their meals together at the appointed times.

FP 53/5/1 ANNEX 3 Page 4

I:\FP\53\5-1.doc

3.3.3 Based on the 96-hour work/rest histories the Bow Mariner was in apparent compliance with the rest standards of the STCW Convention. However, because the three senior officers did not survive, a complete fatigue analysis could not be accomplished. Of particular concern are reports from several survivors that the chief officer routinely stayed in the CCR throughout cargo operations, taking only short breaks for meals and an occasional nap in a chair. The survivors stated this was routine aboard Ceres ships they had sailed on. 4 Extracts from the Analysis 4.1 The International Safety Management (ISM) Code 4.1.1 The ISM Code requires that every company develop, implement and maintain a Safety Management System (SMS), including a safety and environmental protection policy; instructions and procedures to ensure safe operation of ships and protection of the environment in compliance with relevant international and flag state legislation; defined levels of authority and lines of communication between, and among, shore and shipboard personnel; procedures for reporting accidents and non-conformities; procedures to prepare for and respond to emergency situations; and procedures for internal audits and management reviews. 4.1.2 Ceres Safety, Quality & Environmental Protection Management System (SQEMS) In accordance with the ISM Code, Ceres established a Safety, Quality & Environmental Protection Management System (SQEMS). Documentation of this system includes three basic manuals:

Safety & Quality Management Manual (SQMM) Company Operating Procedures Manual (COPM) Fleet Operating Procedures Manual (FOPM).

4.1.3 Audits 4.1.3.1 Ceres received a Document of Compliance (DOC) from the American Bureau of Shipping (ABS) on 12 December 2002, based on an audit completed on 24 October 2002. Three non-major Corrective Action Requests (CAR) were documented. The company completed its first annual external audit on 16 February 2004 with one non-major CAR documented. 4.1.3.2 The Bow Mariner underwent an internal audit from 5 to7 June 2003 while the vessel was under the command of a different master. Twenty-five observations were recorded, including five pending from the previous year. One non-conformity report was issued listing ten record keeping errors, including one pertaining to failure to complete an enclosed space entry permit and another for failure to record training. The non-conformance was closed on 6 July 2003. 4.1.4 ISM compliance 4.1.4.1 Despite this apparent documentation of full compliance with the Code, there are numerous indicators that the Ceres SQEMS was not fully implemented or functional aboard the Bow Mariner:

FP 53/5/1 ANNEX 3

Page 5

I:\FP\53\5-1.DOC

.1 the cargo tanks were not inerted as required by section 1.10.1 of the Cargo & Ballast Operations Manual and section 1.5 of the IGS Manual;

.2 the procedures for tank cleaning in section 1.13.9.1 of the Cargo & Ballast

Operations Manual, Tanker Safety Guide, Chemicals and Dr. Verwey�s Tank Cleaning Guide were not followed;

.3 procedures for confined space entry in section 16.8 of the FOPM and section 1.10.6

of the Cargo & Ballast Operations Manual were not followed; .4 the failure of one of two required blowers on the IGS, a critical safety system, was

not reported as a non-conformance as required by section 19.1.1.2 of the SQMM and section 1.10.1 of the Cargo & Ballast Operations Manual;

.5 monthly fire and boat drills were not conducted as required by SOLAS

regulation II-2/19 and section 7.4 of the FOPM; .6 training was scheduled and recorded in the Minutes of Safety Committee

Meetings but not conducted. Survivors reported that the training scheduled for February was not conducted, and that the only training given was on the ISPS Code, which was not scheduled or recorded in the Minutes of the Safety Committee Meeting for February;

.7 Second Assistant Engineer Aguilar joined the Bow Mariner in Port Said, Egypt

on 5 February 2004, and was making his first voyage on a Ceres vessel. Under section 2.1.2 of the FOPM, he was required to have an overlap of one voyage leg, but not less than 72-hours, with the departing engineer. Instead the two men crossed on the gangway and had no overlap or sharing of information. Ceres officials later stated the departing second assistant engineer was discharged for cause so an overlap was not possible. However, the required overlap could have been met by retaining the discharged second assistant engineer until the Bow Mariner arrived at Kali Limenes, Greece on 7 February 2004;

.8 none of the survivors went through indoctrination or familiarization as described

in section 2.1.2 of the FOPM, stating that �there was no time.�; and .9 officers below the grade of chief officer and chief engineer had not read

applicable portions of the SQEMS. Second Assistant Engineer Aguilar reported that he was sternly rebuked when he asked Chief Engineer Athanasiou about administrative procedures.

4.2 IGS operation 4.2.1 Principle of operation Chemical tankers are fitted with special inert gas generators, like that on the Bow Mariner, because boiler flue gas is too dirty for use with some chemicals. These generators are more expensive to install and operate because they are an entirely separate piece of machinery and burn a high grade of fuel oil. When starting to inert with an empty tank it is first purged by dilution or displacement to remove the air. Dilution is done at high velocity and requires three to five air changes of the cargo tank; displacement is done at low velocity and requires about one

FP 53/5/1 ANNEX 3 Page 6

I:\FP\53\5-1.doc

and a half air changes. Purging continues until the oxygen content in the tank is below eight percent. After loading cargo the IGS is operated until the tanks are pressurized sufficiently to prevent air from being drawn into the tanks. The IGS should be operated throughout cargo discharge to provide a steady supply of inert gas to replace the volume of cargo as it leaves the tanks. A properly inerted cargo tank cannot explode or support combustion. 4.2.2 Regulatory requirements Chemical tankers constructed before 1 July 1986 and carrying flammable cargoes �other than crude oil or petroleum products,� are not required to carry or use IGS. On 28 February 2004 the Bow Mariner was carrying ethyl alcohol and residual MTBE, both non-petroleum, Grade C flammable cargoes; therefore, the tanks were not required to be inerted by international or United States regulations. 4.2.3 Ceres company policy Although exempt by regulation from IGS requirements, the Bow Mariner was required to inert its tanks by Ceres� SQEMS, which also requires that the inability of the system to operate at optimum efficiency be reported to the Ship Manager as soon as possible. 4.2.4 Actual practice These manuals clearly and unequivocally establish that the cargo tanks of the Bow Mariner should have been inerted during cargo discharge in New York (25 February 2004), based on the Ceres SQEMS in effect at that time. According to Second Assistant Engineer Aguilar and Third Officer Ortilano, they were not. In fact, Third Officer Ortilano stated that in his experience aboard the Bow Mariner, both as third officer and able seaman, he had never known the IGS to be used. This was corroborated by former Bow Mariner crewmembers that were interviewed when the investigating officer visited the Bow Transporter, in Singapore. In addition, the master, chief officer, chief engineer and Ceres officials aboard the Bow Transporter confirmed that prior to the Bow Mariner explosion, they operated the IGS only when it was required by the port state or facility. When shown section 1.10.1 of the Cargo & Ballast Operations Manual they appeared to be unfamiliar with it, and at first asserted that the section did not apply to chemical tankers. When the investigating officer pointed out the manual itself was specifically for �Tankers/Chemical Tankers,� they continued to assert that they were not required to inert because of the vessel�s date of build. In fact, the chief officer of the Bow Transporter scoffed at suggestions that the Cargo & Ballast Operations Manual was governing, stating that his �30 years of chemical tanker experience� was all he needed to perform his job. The Ceres official then stated that they were now inerting at all times �� at company expense.� 4.2.5 Condition of the IGS 4.2.5.1 According to Second Assistant Engineer Aguilar and Electrician Bactat, the IGS aboard the Bow Mariner had an inoperable blower for several months before the explosion. Second Assistant Engineer Aguilar stated that shortly after he reported aboard Chief Engineer Athanasiou instructed him to remove the number one (starboard) blower motor, which was �burned up,� and replace it with the number two (port) blower motor. He could not state why this was necessary, since either blower could be used during IGS operation. Electrician Bactat recalled being asked to test the starboard motor but had not been asked to test the circuit to determine the cause of the motor failure. He opined that the motor likely failed because it was dirty. Neither Second Assistant Engineer Aguilar nor Electrician Bactat tested the motor after it was swapped out.

FP 53/5/1 ANNEX 3

Page 7

I:\FP\53\5-1.DOC

4.2.5.2 The vessel�s preventive maintenance system required inspection of the IGS blowers at intervals of 30 months. Records indicate the blower was examined externally and operationally tested on 1 November 1998 and 28 March 2003. 4.2.5.3 A review of records from the Bow Mariner did not uncover a report of the motor failure as required by section 1.10.1 of the Cargo & Ballast Operations Manual, nor was it reported as a non-conformance as required by section 22.1.6 of the COPM and section 19.1.1.2 of the SQMM. There was also no requisition for a new motor. These facts are further indication that Ceres and the senior officers of the Bow Mariner did not consider the IGS a critical piece of safety equipment. 4.3 Tank cleaning 4.3.1 Instructions for tank cleaning 4.3.1.1 Instructions for tank cleaning for the Bow Mariner are contained in the Cargo & Ballast Operations Manual, Procedures and Arrangements (P&A) Manual and Ceres Circulars. The SQEMS also cites Dr. Verwey�s Tank Cleaning Guide and the Tanker Safety Guide, Chemicals as sources of information for tank cleaning procedures. 4.3.1.2 All of the references reviewed during this investigation recognize tank cleaning as the most hazardous operation routinely carried out aboard a chemical tanker. 4.3.2 Actual practice 4.3.2.1 On 28 February 2004, the Bow Mariner was not engaged in a standard procedure for tank cleaning or gas freeing. The master ordered the crew to open all of the cargo tank hatches for the empty tanks once the vessel was at sea. He did not explain these instructions and the crew did not question the order. Since neither he nor the chief officer survived, his intentions could not be determined; however, it appears that he intended to ventilate all cargo tanks at once while cleaning the sumps of residual cargo, and then mechanically ventilate the cargo tanks. 4.3.2.2 After the explosion Ceres officials asserted that the charterer�s Commodity Book

contained instructions for gas freeing the tanks by cleaning the sumps and ventilating the tanks, and that the master may have been following those instructions. 4.3.2.3 These instructions lack detail and do not specify how the contents of the sumps should be removed. There is also no reference to opening the cargo tank hatches to naturally ventilate on deck. In fact, a search of the literature on tank cleaning and gas freeing procedures revealed no reference to a procedure that involved the natural venting of flammable vapors through the cargo hatches as was done aboard the Bow Mariner. The Tanker Safety Guide, Chemicals published by the International Chamber of Shipping and incorporated by reference into the Ceres SQEMS, addresses the subject as follows:

�7.7.1 Safe procedures for gas freeing after tank cleaning

.1 Venting of toxic and flammable gas during gas freeing should be through the vessel�s approved gas freeing outlets, and therefore the exit velocity should be sufficient to carry the vapours clear of the deck. No escape of cargo vapours should occur at deck level before concentration within the tank has fallen below 30% LFL (Lower Flammable Limit) and the relevant Threshold Limit Value. Thereafter, final clearance of the vapour mixture may continue at tank deck level through other larger deck openings.

FP 53/5/1 ANNEX 3 Page 8

I:\FP\53\5-1.doc

.2 If portable ventilation equipment is to be used to blow air into a tank, tank openings should be kept closed until work on that tank is about to commence.�

4.3.2.4 Opening of all of the hatches for the empty cargo tanks, as was done on the Bow Mariner, fails to conform to any known customary marine practice. Because the tanks had not been washed or mechanically ventilated the concentration of vapor was very high, and certainly well above the UEL for MTBE. Opening all of the cargo tank hatches permitted vapors to escape at deck level, where the crew was actively working. This exposed them to toxic vapors and increased the likelihood of an explosion to initiate from an accidental spark. MTBE vapors are heavier than air, and there were many obstructions on deck created by the cargo systems, manifolds and midship deckhouse where pockets of vapor could accumulate, despite cross deck wind or the vessel�s forward movement. It also permitted oxygen to enter the tanks, diluting the fuel-rich atmosphere and possibly bringing the mixture within the tanks into the explosive range. There is no evidence that the speed of the gas freeing process would have increased by opening the cargo hatches. 4.4 The ignition source The source of ignition is not known and cannot be determined to any degree of certainty. 4.5 Structural damage, flooding and sinking The Bow Mariner suffered catastrophic structural damage throughout the cargo block in two main explosions, each of which was actually a rapid series of explosions representing individual tanks exploding in sequence. The first tank to explode could not be determined; nor could the sequence of tank explosions. According to the only witness the initial explosion occurred on the port side in the midbody. Underwater video revealed that nearly all of the wing tanks exploded, with most of the side shell missing in the cargo block on both sides. Damage control efforts were not possible and would not have delayed the ultimate sinking. 4.6 Other recent tank ship explosions 4.6.1 This report would not be complete without observing that the Bow Mariner casualty was the worst of four tank ship explosions that occurred worldwide between December 2003 and June 2004. The other casualties were:

24 December 2003 � The tanker Sun Venus exploded while enroute Kobe, Japan from South Korea. The ship was cleaning tanks after discharging a cargo of benzene and ethyl alcohol. Two crewmen remain missing. 2 January 2004 � The tanker Panam Serena exploded in Sardinia, Italy while discharging a cargo of benzene. Two crewmen remain missing. 4 June 2004 � The tanker Ncc Mekka, operated by Odfjell, exploded off the coast of Brazil while cleaning tanks. Two crewmen were killed.

4.6.2 In just six months there were four chemical tank ship explosions resulting in 27 deaths, severe environmental damage and significant economic damage, including the total loss of one ship. Three of the four casualties involved tank cleaning. This abysmal record and alarming trend deserves immediate and decisive action to identify common causes and implement corrective actions before more needless deaths occur.

FP 53/5/1 ANNEX 3

Page 9

I:\FP\53\5-1.DOC

5 Extracts from the Conclusions 5.1 The cause of this casualty was the ignition of a fuel/air mixture that was within its flammable limits, leading to a fire on deck. 5.2 The ignition source could not be determined. Sources that could not be ruled out include electrostatic discharge; mechanical sparks caused by metal-on-metal contact; faulty electrical equipment; hot soot or particles from the funnel; and sparks from changing the batteries of portable electrical equipment in a hazardous location. 5.3 Opening the 22 cargo tanks that previously held MTBE permitted flammable vapours that were heavier than air to accumulate on deck, and diluted the fuel-rich atmosphere in the cargo tanks with oxygen, bringing them into the flammable range. 5.4 The fire was followed by two significant explosions that occurred less than two minutes apart. The first occurred at 1806, and the second occurred before 1808. Each of the major explosions was actually a series of rapid explosions as each of the empty tanks exploded within seconds of one another. 5.5 The explosions caused catastrophic structural damage and led to immediate flooding of nearly the entire cargo block. The ship sank one hour and 32 minutes after the first explosion. 5.6 Contributing to this casualty was the failure of the operator, Ceres Hellenic Enterprises, Ltd., and the senior officers of the Bow Mariner, to properly implement the company and vessel SQEMS. 5.7 The cargo tanks were not inerted during the discharge of MTBE in New York, as required by the SQEMS. The tanks were not required to be inerted by United States law or international conventions because the Bow Mariner was constructed before 1 July 1986. If the tanks had remained closed, the explosion would not have occurred. 5.8. Captain Kavouras� order to open the 22 cargo tanks that had previously held MTBE was a stunningly significant breach of normal safe practices for a tank ship and defies explanation or excuse. 5.9 The entry of the boatswain into the cargo tanks wearing a SCBA was dangerous and ill advised, but did not violate the Ceres� Confined Space Entry Policy in effecton 28 February 2004. That policy has since been revised to prohibit entry under the samecircumstances. 5.10 The failure of Captain Kavouras to properly organize a response to the explosions contributed to the high loss of life. He abandoned ship without sending a distress signal, without attempting to contact a nearby ship, without conducting a proper muster or search for injured crewmen, and without attempting to launch primary life-saving appliances. 5.11 Captain Kavouras and Chief Engineer Athanasiou abandoned ship within 10 min of the first explosion, leaving behind other crewmembers they knew to be alive. Their premature action exposed the crewmen who entered the water with them to the cold water far earlier than necessary, and contributed to the high loss of life.

FP 53/5/1 ANNEX 3 Page 10

I:\FP\53\5-1.doc

5.12 The actions of Third Officer Lugen Ortilano, making his first trip as a licensed officer, were commendable and helped save the lives of himself and five others. The Coast Guard recognized him for his heroic efforts. 5.13 The lack of immersion suits contributed to the high loss of life. There was sufficient time for the survivors of the explosion, who were relatively uninjured, to don an immersion suit before entering the water. If the survivors had immersion suits the probability of rescue by one of the many responding vessels or aircraft, several of which were on scene in less than two hours, was very high. 5.14 The failure of Captain Kavouras to conduct regular and effective fire and boat drills contributed to the high loss of life. 5.15 There is evidence that Ceres and the senior officers of the Bow Mariner failed to follow proper relief and indoctrination/familiarization training for critical crewmembers. 5.16 There is evidence of a lack of cohesiveness between the three senior Greek officers and the Filipino crew, but it could not be determined if this contributed to the casualty. 5.17 The Bow Mariner spilled about 3,188,711 gallons of ethyl alcohol, 192,904 gallons of HFO, 48,266 gallons of LFO and an unknown quantity of slops into the U.S. EEZ. 5.18 There is no evidence that drugs or alcohol contributed to this casualty. 5.19 There is no evidence that fatigue contributed to this casualty; however, it is noted that the practice of the chief officer to remain awake for the entire cargo transfer operation would surely lead to extreme fatigue. 6 Extracts from the Recommendations 6.1 Recommend 46 CFR Part 32.53, Inert Gas Systems, be revised to require the inerting of all cargo tanks carrying flammable cargoes aboard vessels equipped with Inert Gas Systems, regardless of the vessel�s date of build. 6.2 Recommend that Ceres review their internal policies and procedures concerning workforce interaction and cooperation, including but not limited to delegation of appropriate duties to qualified officers. 6.3 Recommend the Commandant approach the International Maritime Organization, the International Chamber of Shipping and INTERTANKO about forming a study group to examine the causes of all tank vessel explosions in the last five years involving tank cleaning to search for common factors. 6.4 Recommend Commandant send a message to all marine safety field units emphasizing the importance of randomly verifying a tank vessel�s compliance with its safety Management System (SMS) for tank cleaning, confined space entry and tests and inspections of equipment.

__________