The competitive ability of jet engine maintenance companies depends mainly on turn around time and overhaul costs. Both airline and maintenance companies need the best possible accuracy regarding the prediction of emerging costs and time of engine maintenance process to secure their operation. Estimating the deterioration status of engine modules prior to disassembling is one of the greatest challenges for the maintenance process. In a pilot project a Bayesian belief network (BBN) has been developed to determine the deterioration condition of the General Electric CF6-80C2 first stage high pressure turbine (HPT) nozzle guide vane (NGV). The aim of this paper is to extend the used BBN techniques to the HPT first and second stage rotor blades and the second stage vanes. Thereby, its objective is to prove the successful application of the developed method for constructing a BBN for component hardware forecast.

The BBN is composed of following parameters: component repair history, region, on-wing cycles, airfoil material, thrust rating, engine wing position and customer segment. Performing statistical data analysis and combining these parameters with expert knowledge result in component specific BBNs. These nets provide a moderate forecast accuracy of 59 percent for the first stage rotor blades, 65 percent for the second stage rotor blades and promising 89 percent for the second stage NGVs.

The paper concludes that a BBN has very good qualities to forecast the hardware condition of HPT components impressively shown by virtue of the nozzles. Therefore, it is worth to transfer the developed method to other modules in order to accurately predict the degradation of the components in an unconventional way.

This content is only available via PDF.
You do not currently have access to this content.