FPSOs: New Storage Tank Explosion Frequencies

The fatal explosion in the pump room on the Brazilian FPSO Cidade de São Mateus in February 2015, demonstrated the potential for major accidents on this type of unit. Although there has never been an explosion in the oil storage tanks of an FPSO, there is a potential for such an event and offshore operators should assess the risks of such events to design against them. It is therefore imperative that risk assessments of FPSOs address this type of event. For this, they need to estimate its annual likelihood of occurring. An early estimate of the frequency of cargo tank fire/explosion was 4.4 x 10-4 per FPSO year or one chance in 2,300 for each year of operation*1. This was based only on experience with trading tankers in port, which is not appropriately comparable with conditions for loading and offloading FPSOs at sea. A frequency of 2.2 x 10-3 per FPSO year can also be found in a risk assessment textbook*2. A recent study by the International Association of Oil & Gas Producers*3 combined this with an assumption that FPSOs had half as many tanks as trading tankers, and estimated a storage tank explosion frequency of 8.8 x 10-4 per FPSO year. Both these sources used tank explosion experience on trading tankers during 1972-86, which is now severely outdated.

Using up-to-date data, in conjunction with a systematic consideration of the differences in risk between trading tankers and FPSOs, a new frequency estimate for storage tank explosions on FPSOs has been developed by DNV GL*4.

The DNV GL study selected data from large oil tankers, defined as those over 80,000 dwt, as these are comparable in size to FPSOs. Currently, there are roughly 1,200 such ships in service around the world. The IHS Fairplay casualty database was used as the basis for the analysis. Data was selected for the period 1980-2013, in all reported locations worldwide, including at sea, in port, at anchor and in construction, repair and scrapping yards.

The analysis covered events categorized as ‘fires/explosions’, as available accident descriptions do not normally distinguish rigorously between the two, and hence it is difficult to separate them. Events occurring in tank spaces have been identified from the details provided by IHS. This excludes events in the engine room, pump room, other equipment rooms, superstructure, transfer equipment or forecastle area, even if they subsequently escalate to affect cargo tanks. 

In total, 88 tank fire/explosions were identified. The exposed population of oil tankers over 80,000 dwt during 1980-2013 is estimated to be 40,097 ship-years, based on the IHS database. The overall average frequency of tank fire/explosion is therefore estimated as 88/40097 = 2.2 x 10-3 per tanker year. Table 1 gives the breakdown of these events by severity.

Of the 88 events, 17 resulted in total loss of the ship and 31 others were considered serious, sufficient to require repairs before the ship could continue trading. 

Figure 1 shows the fire/explosion frequencies in each year. Although there are many random variations year-to-year, it is clear there have been fewer accidents in recent years, and hence, the average frequencies may be pessimistic if applied to current operations. The reduction in frequency since the mid-1980s may be due to the progressive introduction of inert gas systems, segregated ballast tanks and double hulls. By 2010, these were fitted to the entire fleet, but as there have been so few accidents since, it is not yet possible to obtain robust results from this period alone. Using the population of double hull tankers alone, the frequency is 3.9 x 10-4 per tanker year, with an uncertainty range of roughly a factor of 2 higher or lower.

The DNV GL analysis considered the reasons why explosion frequency was much lower on double hull tankers than on the single hull tankers that were in service up to 2010. Double hulls are mainly intended to prevent oil spills and this may help prevent tank explosions originating from these. However, they introduce explosion hazards in the double hull spaces. The research found that the explanation is not related to the double hull as such, but to the fact that most double hull tankers are relatively new, and newer tankers tend to experience fewer accidents than older vessels.

Therefore, although the trend shows steady improvement over the last decade, it may be optimistic to expect the frequency to continue reducing. Recent regulations have resulted in a very high proportion of the fleet being new tankers, whose accident frequency is likely to increase as the vessels age. So, it is possible that the trend will reverse, as it did following the dip in accidents shown in Figure 1 shows around 1995. 

Oil storage tanks on FPSOs are broadly similar to cargo tanks on trading tankers. Although some FPSOs are new-build vessels, others are in fact converted single hull trading tankers, taken out of service due to recent double hull requirements. A systematic comparison of differences between trading tankers and FPSOs demonstrates that many of these differences depend on the specific design of the FPSO, as well as the specific service of trading tankers. Some would tend to increase the explosion frequency while others would reduce it. Making use of statistics on the causal breakdown of trading tanker accidents, the study concluded that most changes would have effects that are relatively small and less than the uncertainty in the original estimate. Overall, there are no strong reasons to make a major change in the explosion frequency from trading tankers. The study recommended that, until a more comprehensive analysis is available, it was preferable to avoid arbitrary adjustments and use the trading tanker data unmodified to estimate the frequency on FPSOs. The frequency of storage tank explosions on new-build FPSOs is therefore taken as 3.9 x 10-4 per FPSO year. his implies an explosion on average every 2,600 FPSO-years. Cumulative experience with FPSOs to date is less than this, so the frequency is consistent with the absence of explosions to date. The trading tanker data suggests that almost half of serious casualties occur during repair and tank cleaning. If the scope of the FPSO study excludes these operations, this would justify making a further reduction in the frequencies by a factor of 2. Individual FPSOs may differ substantially, and need to be studied in detail to quantify the effects of their differences from trading tankers.

(As published in the September 2017 edition of Maritime Reporter & Engineering News)