The derivation of airworthiness performance climb standards

Abstract

From the first, civil airworthiness requirements have included climb performance among the safety criteria. Hitherto climb performance standards have been empirical, and magnitudes have been chosen by reference to current aircraft types regarded as satisfactory. A weakness of this empirical type of requirement is that no method is provided for modifying the standards to meet new operating procedures and aircraft design features. To overcome this difficulty, a more rational basis for deriving the climb standards is proposed. The conception is introduced of a 'datum' performance, below which conditions predisposing to an accident exist, and the level of safety judged by an 'incident rate' which is the frequency with which the operational performance of aircraft falls below this datum. A standard is chosen so that when the aircraft type complies, the incident rate will not exceed some tolerable value. To derive such a standard, account must be taken of the various conditions such as weather and airframe state which affect performance. The standard need only be framed in terms of some of these conditions; the effect of others may be included on a statistical basis by providing an appropriate 'performance margin' over the datum. It is shown how the treatment of the conditions affects the form and efficiency of the standard. The margin appropriate to a given incident rate is obtained from the distribution function of the climb performance; this function is, in turn, derived from the distribution functions of the conditions treated statistically, and their effect on climb performance in a given aircraft configuration. The effect of engine failure is included by taking account of the probability of engine failure and the associated loss in performance. To simplify the treatment of changes in aircraft configuration, the flights are divided into stages, such as take-off climb, in which the configuration, except for the incidence of engine failure, is sensibly constant. It is then shown that the required standard (datum plus margin) for any stage may be specified in terms of a single case (i.e., number of operative engines); the case chosen is that found to be dominant in incident causation. Numerical examples are given of the derivation of standards by the method described.