[116] Data-driven prototype of turbo engine blade monitoring
C Ying, J Lacaille and C Berthou Safran Aircraft Engines, Moissy-Cramayel 77550, France. Tel: 33 1 60 59 32 74; Email: charles.ying@safrangroup.com
Commercial aircraft engine high-pressure turbine (HPT) blades are subject to high temperature and pressure conditions. In order to prevent in-flight breakage, companies schedule inspections to capture HPT blade crack propagations. The time interval between two inspections is chosen to ensure that unobserved crack propagation will be revealed prior to breakage. Nevertheless, practical experience shows that, in addition to nominal usage, time spent in sandy environments or aggressive driving accentuate HPT blade deterioration. Thus, time-based inspections are expected to be suboptimal with respect to maintenance costs, for instance. The idea developed in this paper is to use operational, production and maintenance data combined with a data-science approach and machine learning techniques to provide a condition-based maintenance model of deterioration. Objectives are set to maintain crack coverage performance while reducing the rate of inconclusive inspections.
