OFFSHORE DATA BASE SHOWS DECLINE IN RIG ACCIDENTS

Alain Bertrand, Laurence Escoffier
Institut Francais du Petrole
Rueil-Malmaison, France

Institut Francais du Petrole (IFP) has compiled statistical figures on offshore accidents for risk, safety, and reliability studies.

IFP calls this data base "Platform." It provides a body of essential data for accidents concerning both mobile and stationary offshore drilling rigs.

Historical accident data bases are a basic implement for risk assessment of safety and reliability. IFP has built this data base with all available information from 950 accidents since 1955.

The data base includes all kinds of offshore drilling, production, and accommodation platforms, both fixed and mobile, that had an unplanned and unforeseen work stoppage of at least 24 hr. This criterion eliminates many minor incidents and contributes to the homogeneity of the data gathered, an indispensable condition for obtaining valid statistics.

The data base contains information available to the public from the general and specialized press, official reports, and insurance accident listings.

Platform consists of an accident synthesis data sheet (in English) for each of the 950 accidents bound into four volumes, an ASCII file on a diskette for downloading into a relational data base management system, and a user guide explaining the codes.

Table 1 shows the number of accidents in the data base by 5 year period according to type of drilling unit.

IFP compiles an ongoing file for each accident containing all the information available both on the accident itself and on the characteristics of the rig involved. This file is continually updated as new, and generally more detailed, data are received. A synthesis data sheet is established for each accident, describing the sequence of events of the accident.

There are 25 important parameters in the computerized data base. This breakdown allows statistics to be correlated according to the following:

Information about the rig or platform at the time of the accident: name, type of rig, number of former names, constructor, age, flag, and activity
Location of the accident: hemisphere, geographic zone, country or province, geographic particularity, and water depth
Information about the accident itself: date, type, weather conditions, preliminary factors, sequence of events, aggravating factors, impact of the accident, number of casualties, crew evacuation, crew rescue, pollution, and amount of oil spilled.

Data have been entered either as uncoded facts, or in codes that enable a succinct analysis to be made of the accident. For instance, there are codes for 25 different possible activities under way at the time of the accident, including such things as storm watch, bunkering, ballasting, development or assessment drilling, exploration drilling, production, accommodation, jacking up, maintenance, positioning, raising wellhead, under tow, and self-propelled displacement.

The data base codes also include:

74 different kinds of events making up the sequence of events
72 possible preliminary factors that placed the rig in a dangerous situation
196 countries, provinces, or states (for Canada or the U.S.)
30 major geographic zones (e.g., Gulf of Mexico, Persian Gulf)
139 different constructors of platforms, rigs, drillships, or barges.

The parameters retained for the computerized data base are facts available for most of the accidents and are useful for statistical purposes.

More details on each accident can be found in the accident synthesis data sheets, including unexpected details that are unique to the particular accident. The computerized data base has fewer details, but it enables statistics to be manipulated and lists to be drawn up. Tables 2 and 3 give examples of lists of accidents to drilling rigs working in the North Sea in 1990 and early 1991 and of drillship accidents since 1985, respectively.

STATISTICAL RESULTS

Platform's detailed coding system takes into account all events, even rare ones. The task of compiling overall statistics and determining broad trends is complicated because of the variety of data entered. This leads to results that are difficult to classify in a limited number of general categories.

However, the probability of occurrence of an unusual event may be calculated; for instance, the risk of anchoring because of a broken tow line (5 x 10-5 probability by year and by rig) or sinking because of gas bubbling that rendered the sea less buoyant than normal (2 x 10-4 probability by year and by rig).

Platform data base can furnish statistics on other unusual events, such as the dropping of heavy equipment on the deck, loss of equipment at sea, or broken tow lines.

In studying accident data, it is important to be able to assign a general category to the accident for statistical purposes (e.g., blowout, collision, and structural damage). For instance, the pie charts in Fig. 1 compare the most frequent kinds of accident that occurred from 1955 through 1990 to jack ups, submersibles, semisubmersibles, and fixed platforms.

However, it is also important to analyze the sequence of events making up each accident. The events occurring most frequently in drilling platform accidents have been found to vary according to the kind of platform involved:

Fixed platforms: fire and explosion
Submersibles: gas blowout
Barge rigs: capsizing
Jack ups: structural damage to legs or mat (often because of the seabed)
Semisubmersibles: gas kick.

The frequency of the main kinds of accidents occurring to the fleet as a whole since 1955 is shown in Table 4.

PLATFORM ACTIVITY

Table 5 shows the major activities that rigs and platforms were engaged in at the time the accident occurred, for all types of platforms.

Further statistical analysis for jack ups reveals that 76.6% of accidents occur during towing of the jack up (30.9%), during exploration drilling (27.1%), or during jacking up operations (18.6%). Accidents to fixed platforms occur during production 53% of the time.

Table 6 shows the improvement in the reliability of jack ups and mobile rigs from 1955 through 1990.

The accident data in each column are normalized according to the size of the corresponding fleet. That is, the data are presented as the number of mobile rig accidents/number of mobile rigs, the number of jack up rig accidents/number of jack ups, and the number of jack up rig accidents caused by seabed problems/number of jack up rigs. Seabed accidents include damage to legs and mats.

In all three cases, there is a relative decrease in the number of accidents over time. Jack up rigs are shown to have fewer accidents in general than the whole mobile fleet.

BLOWOUT ACCIDENTS

Table 7 is a simplified summary of the factors that have been most often implicated in blowout accidents to all rigs over the years. As may be expected, the breakdown, absence, or manipulation of the blowout presenter (BOP), casing, drillstring, or diverter are the principal preliminary factors in blowout accidents.

The weather factor is included for the sake of comparison. A chronological analysis shows that the importance of the BOP, drillstring, and diverter has increased while that of the casing has fluctuated. The influence of weather conditions has become practically nil, probably because of increased rig strength, better resistance to adverse weather, and safety precautions during bad weather.

Most blowouts occur during exploration drilling (54.8% of all blowout accidents). Another 30.4% occur during development drilling, and 14.8% occur during production operations.

Table 8 analyzes the various types of blowouts that occurred during each kind of operation for the industry as a whole. Gas blowouts are the predominant cause of blowout accidents, but note the relatively large number of oil blowouts during production operations.

The relative vulnerability of jack ups and semisubmersibles to blowout accidents was calculated. The percentage of the number of blowout accidents and the percentage of the average number of active rigs of each type was compared to the same number for the entire fleet of mobile rigs. The relationship of these two percentages defines the statistical vulnerability of each rig category to blowouts.

A value of 1 indicates that the category of rig has the same vulnerability as the fleet of mobile rigs in general. A value greater than 1 means that the type of rig is more vulnerable to accidents than the fleet in general, and a value less than 1 indicates that type of rig is less vulnerable to accidents than the rig fleet in general.

Table 9 shows that there is a chronological tendency for the coefficient of vulnerability to increase. This reflects the complete modification of the mobile rig fleet.

Between 1960 and 1980 the percentage of other kinds of active mobile rigs (barges and submersibles) dropped from 50% to 5%.

The number of semisubmersibles increased from 5% to 25% and jack ups from 45% to 60%.

Also, jack ups and semisubmersibles undertake the riskiest drilling operations.

Note the good standing of jack up rigs.

The coefficient of vulnerability is always less than 1 (the coefficient of the entire mobile rig fleet) and also less than the coefficient of semisubmersibles, which seem to be more susceptible to blowout accidents.

This result, confirmed in other studies, may be explained by the position of the BOP on the surface for jack ups and on the seabed, thus less accessible, for semisubmersibles.

SEVERITY OF ACCIDENTS

The number of accidents is one way to measure the impact of accidents over a certain time frame. Platform further refines this concept because accidents are assigned a coefficient of severity according to the following guidelines:

Large rig: repaired on site (1), lengthy repairs in dry dock (2), or destruction or scrapping (3)
Smaller rig (such as a barge or submersible): lengthy repairs in dry dock (1) or destruction or scrapping (2).

This weighted characteristic gives a better idea of the impact of the accident. The factor of accident severity increased regularly until the period 1980-84, during which there is a marked rise followed by a decrease (Table 10).

These figures must be compared with general statistics available on world offshore oil and gas production activity, expressed in billion tons of oil equivalent over the same period. The figures show that drilling activity developed rapidly during this key period.

A comparison of the two columns establishes the relative impact of offshore accidents.

After a slow decrease from 1955 to 1970, the impact drops more quickly. There is a slight peak in the impact of accidents, however, for the period 1980-84 when drilling activity and the high price of oil brought about full utilization of the rig fleet.

Table 11 shows the predominant methods of personnel evacuation and rescue in different geographic locations. Out of about 950 accidents analyzed in the data base, 290 include information about personnel evacuation.

Life rafts and survival capsules were little used (5.1% and 6.2%, respectively). In the final stages of rescue, when men are lifted off rigs, survival capsules, lifeboats, or from the sea, helicopters were often used (21.5% alone, 25.5% with the help of nearby vessels). However, most rescues were carried out by ships (72.7% alone, 76.7% with helicopter assistance).

Supply vessels that are usually on constant standby near the rigs provided much of the rescue assistance. Table 11 shows the importance of helicopters in North Sea rescue operations.

These few examples illustrate the scope of the data base. Platform data base opens broad areas for further research by numerically identifying trends in accidents.

BIBLIOGRAPHY

Bertrand, A., and Escoffier, L., "IFP Databanks on Offshore Accidents," Proceedings of the 6th Euredata Conference on Reliability Data Collection and Use in Risk and Availability Assessment, Siena, Italy, Mar. 15-17, 1989, pp. 114-28.
Bertrand, A., and Escoffier, L., "Accident Database Enhances Risk Drilling, Production Assessment," Offshore, September 1989, pp. 33-37.
Bertrand, A., and Escoffier, L., "La banque de donnees Platform de l'Institut Francais du Petrole: un outil de base pour des etudes de risque," Revue de I'IFP, Vol. 46, No. 2, March-April 1991, pp. 263-76.