Big data analytics and predictive maintenance are hot topics in maintenance IT today. But what sort of data is really needed to do predictive maintenance, and how do the organizations best equipped to analyze it obtain the data?

Richard Brown, a principal at aviation consultancy ICF International, says a full predictive-maintenance system benefits “from all data that you can get.” But Brown lists three especially essential data sets.

First comes reference data to show the normal behavior of aircraft systems and to identify failure patterns. Reference data includes design and test data, maintenance data and old operating data.

Graph: Evolution of aircraft maintenance approaches

Next comes operating data to understand what is happening with actual aircraft, either in real time during the flight or transmitted on the ground.

Finally, there are maintenance manuals that enable techs and engineers to act on failures and fix issues promptly.

Before any data-sharing agreements, Brown says data is owned by the organization that generates the data. Design and test data and repair instructions are owned by OEMs, or by MROs in the case of Designated Engineering Representative (DER) repairs. Operating data is owned by aircraft operators.

As all these kinds of data can be duplicated easily, many kinds of data-sharing agreements are possible. ICF is now seeing many engine and airframe OEMs obtaining sharing agreements with operators that give data to the OEMs and prevent the operator from sharing data with other organizations. “This makes it difficult for component OEMs or MROs to gather the data,” Brown notes. These data sharing-agreements can be part of airplane health management or other airline-OEM agreements. “Typically, a lot of integrated support contracts require the company performing the work to access the data.”

There are a number of challenges in executing data sharing. First is the quality and availability of the data itself. Much data gathering is still manual and on paper, and not all required data is downloaded from aircraft. So there are data gaps even before sharing.

Then there is the confidentiality and security of data. “Operators see their data as sensitive and don’t want others to get access to it,” Brown says. “They might not want their data to be used to improve the services of other airlines.”

Costs of data also matter. Data-sharing agreements have to be negotiated, and the organization trying to access the data must pay a cost. This organization may have only a limited understanding of the actual value of the data it will receive. In the past, Brown says ICF has seen OEMs offering health management services at limited costs in order to access data.

Nevertheless, progress is being made. For example, avionics systems are being installed that automatically transmit relevant data through on-ground systems using cellular networks or other systems. And GE has launched Configuration Data Management with Capgemini to ease data sharing between airlines and MROs and during lease returns.

Most companies now say they have secure systems and are working on ways to keep sensitive data secure. But Brown says operators still see a tradeoff between the benefits of data sharing and its costs and risks. And airframe OEMs still regard the costs of obtaining data as investment costs.