FAA source text, reformatted for reading. Page boundaries and paragraph numbers are retained.
APPENDIX A. HISTORICAL PERSPECTIVE ON THE USE OF STATISTICAL PROBABILITIES IN SYSTEM SAFETY ASSESSMENT
A.1 Concorde Transport Supersonique Standard. The British Civil Aviation Authority (BCAA) applied the concept of proportionally assigning statistical rate goals to categories of accident causes during the design and certification of the Concorde in the Concorde Transport Supersonique Standard in the 1960s. At that time, the BCAA considered the probability of a severe accident to be on the order of one per one million hours of flight (1 x 10-6 per flight hour). The BCAA roughly estimated that 10 percent of those accidents were the result of design systems- related hazards. Based on those assumptions for the Concorde, the BCAA reasoned that probability of a severe accident from design systems-related hazards should be less than 1 in 10 million flight hours, or 1 x 10-7 per flight hour. The BCAA standard defined hazard categories as minor, major, hazardous, and catastrophic, and it assigned qualitatively allowable probability for each category, e.g., probable, remote, and extremely remote. The BCAA also apportioned statistical probabilities to the categories (except the catastrophic category) for use in controlling “statistically controllable” hazards. The standard did not establish a numerical probability for catastrophic failure conditions because, per the overriding fail-safe philosophy, no single failure regardless of probability should foreseeably be allowed to result in a catastrophic failure condition. However, the cumulative probability of all catastrophic failure conditions should be no greater than 1 x 10-7.
A.2 British Civil Airworthiness Requirements. The British Civil Aviation Authority replicated the Concorde airworthiness requirements in the British Civil Airworthiness Requirements (BCAR). During certification of the Concorde, the BCAA recognized that analyzing every hazard for the purpose of assuring that the probabilities collectively were less than 1 x 10-7 was an onerous and somewhat impractical task. To address this problem, the BCAA assumed that there were no more than one hundred systems-related, catastrophic failure conditions and that a direct allotment would be sufficient for certification. Therefore, the BCCA apportioned the allowable average probability per flight hour of 1 x 10-7 equally among the theoretical, one hundred catastrophic failure conditions, resulting in 1 x 10-9 per flight hour as the upper limit average probability per flight hour of a statistically controllable catastrophic failure condition. The 1 x 10-9 per flight hour probability was not applicable for single failure conditions that could lead to a catastrophic outcome.
A.3 FAA AC 25.1309-1. The intent of the BCAR systems guidance was first adopted by the FAA in AC 25.1309-1, System Design Analysis, dated September 7, 1982. The BCAR and previous Concorde standards defined four hazard categories in terms of specific airplane level hazards and the effect of those hazards on the airworthiness of the airplane. AC 25.1309-1 defined three functional hazard categories. The AC defined the
