Predictive maintenance for airlines

Leveraging aircraft sensor data and maintenance logs to help airlines avoid costly maintenance delays and cancellations.

The issue 

In 2016, in the US alone, the cost of maintenance related delays for airlines was well over $0.5B. Almost a third of total delay time is due to unplanned maintenance.*

The PwC solution 

PwC’s predictive maintenance solution can predict 15-30% of maintenance related delays and cancellations, leading up to a 0.3-0.6% improvement in on-time performance.

  • Each aircraft type has unique operating characteristics and certain components that drive frequent and costly delays.
  • Amidst all the white noise of aircraft messages and maintenance activities, airlines need to identify and act on pertinent signals that are highly indicative of potential delays or cancellations (D/C).

PwC uses predictive maintenance to help clients avoid costly delays and cancellations

For aviation companies, delays and cancellations are a huge and expensive problem. Up to 30% of the total delay time is due to unplanned maintenance. PwC uses advanced analytics to rationalize, predict and streamline maintenance, helping aviation clients increase maintenance efficiency, improve the health of their fleet, and reduce delays and cancellations. This predictive maintenance approach can also help improve areas like supply chain optimization, inventory allocation and planning, aircraft reliability improvement and operation and schedule planning. Predictive maintenance can also be applied to other sectors such as railway, automotive, power and utilities.

How it works

PwC’s predictive maintenance solution leverages aircraft sensor data and maintenance logs to help airlines avoid costly maintenance delays and cancellations.

  • We evaluate priority ATA chapters for modeling and use sophisticated analytic filters to identify the key messages that correlate most highly with delays and cancellations.
  • We combine signals from fault messages with signals obtained from text mining of maintenance logs to identify failure indications.
  • We develop a suite of rare event models, using variables from prior analysis of failures, and weigh the models towards the actual incidence of these failures in order to strengthen its predictive power.
  • We work with our clients to conduct field tests. This helps to prove the efficacy of the solution and identify any process and/or organizational refinements needed for full solution deployment.

Our Approach

  • Pilot - 8-10 weeks: We check the readiness of data content and quality, conduct diagnostic analysis to identify key signals for failure prediction, develop prediction model for key ATAs and develop a business case to ascertain the prediction accuracy of the developed model.
  • Trial - 10-12 weeks: We demonstrate how the model works in an operational test setting, prove its accuracy, obtain feedback from users to refine the model and identify technological or organizational challenges. We then create a roadmap for solution implementation.
  • Solution - 4-6 weeks: We deploy a Microsoft Azure based predictive maintenance solution. 

Contact us

Fred E. Cleveland

Managing Director, US Transportation & Logistics, PwC US

Kumar Satyam

Director, Analytics, PwC US

Ajay Singh

Director, Analytics, PwC US

Paul Dibble

Director, Transportation & Logistics, PwC US

Richard Wysong

Director, PwC US

Follow us