Design of Large Diesels for Operational Reliability

Abstract

This paper considers the conventional approach to the design of large diesel engines, based, as it generally is, on experimental validation by limited engine and/or rig running. It is concluded that to ensure the attainment of a specific degree of reliability the extent of testing required would be economically prohibitive. Consequently, the amount of testing undertaken is largely irrelevant and the failure rates to be expected in service are a matter for intuitive judgement and good fortune.

An attempt is made to establish more realistic design criteria for engines manufactured in small batches and designed to remain operational for long periods of time (15-20 years, typically) with infrequent major overhauls (say, every 20 000 h). It is shown that for structural integrity this may be achieved if the criterion is met, where is the mean strength derived from ideal specimens, is the mean load obtained from a consideration of the engine manufacturing and operational tolerances, f is a factor taking into consideration deterministic errors in and , and r is a statistical risk factor characteristic of the component and application. Typical values for f and r are derived for a medium speed diesel and the detailed application to a typical critical engine component described.

The economic implications of design on the basis of failure rate, rather than arbitrarily chosen safety factors, are discussed in terms of design economy and better understanding of the factors likely to influence service reliability.