Today’s new aircraft and engines have integrated data acquisition and transmission systems that can actively monitor and store data on the performance of an engine or airframe component. This generates tremendous amounts of data for a fleet of aircraft, typically measured in petabytes. A petabyte is one quadrillion bytes, which is a lot of data.
Collecting and transmitting this data requires a number of sensors on board the aircraft, data storage devices to collect and store the data, and transmission capabilities for either real-time or post-landing download, depending on the criticality of information and the design of the integrated hardware and software on board the aircraft.
The newest aircraft models, including the Bombardier C Series, Embraer E2, Boeing 787, and Airbus A350 have the most advanced health management systems for airframes, while the new technology Pratt & Whitney GTF, CFM LEAP, and GEnx and GE9X, and Rolls-Royce Trent 1000 and Trent XWB are leading the way in engines.
But what happens next? A massive amount of data are collected that need to be analyzed and turned into useful information for customers. It is in the analytics behind big data that will enable the data collection systems to prove their worth.
Big data will tell us when a component is going to fail, and the condition of a number of variables that could impact that component at and leading up to the time of failure. Analyzing conditions that lead to failure can then lead to a set of “early warning” criteria that a component may soon fail, providing the operator an opportunity to fix the fault before a failure occurs to avoid lost revenue. Much like monitoring exhaust gas temperatures in an engine can determine when an overhaul is due, the additional data from an engine or airframe can provide new inferences and guidelines for maintenance activities.
How big is big data for an aircraft or engine? The first health management systems monitored only a few key elements of an engine or airframe. Today, the technologies that are driving the Internet of Things (IoT) have improved in capabilities at lower cost. As a result, we may now have 5,000 or more parameters to analyze on an aircraft at any given point in time.
It may take quite some time before the impacts of changes in each parameter are known well enough to develop predictive models that will produce tangible benefits for new aircraft models.
Simulation is one software tool that has been shown promise in this regard and has resulted in “Digital Twin” technology from GE Aviation. Digital Twin technology is, in its simplest form, a virtual model of a physical product. By pairing the virtual and physical worlds, analysis of data and monitoring information enable advanced analytics that can predict failures and reduce maintenance costs.
The concept of a digital twin has been around since 2002 but has only recently become feasible with the availability of monitoring data gathered through sensors and connectivity that form Aircraft Health Management Systems and collects tremendous volumes of data for analysis.
Essentially, a digital twin is an advanced simulation of a physical entity, collecting data that mirrors the physical experience of an aircraft in a simulation model, capturing the operational characteristics and conditions to enable prediction of future behavior. A digital twin for an aircraft engine utilized on an Emirates aircraft based in Dubai may show quite different results than one for the same type of engine used by Delta based in Atlanta.
Operational considerations, including temperatures, pressures, operating in an environment with a lot of sand in the air, or near ocean water, will have a difference on wear and tear on an engine and its components. Those environmental factors, along with performance data from engine sensors provide a richer database for analysis and refinement of simulations to mirror the performance of a given engine based on its utilization, environment, and operational history.
The ability to accurately predict engine or airframe component behavior is essential to carry out the mission of health management systems, which is to provide maintenance and money-saving ideas. GE Aviation is leveraging its Predix software suite, which provides an integrated platform for storing, analyzing and creating digital twin simulations across a variety of industrial applications, including Aerospace. That cloud-based environment includes a number of robust applications and analytics routines for building customized applications, as well as a library of tools to create and deploy machine learning models that detect anomalies, direct controls, and predict maintenance. The digital twin models created within the Predix environment enable analysts to determine correlations and cross-correlations between variables and to more rapidly understand, predict, and optimize the performance of an engine or aircraft.
The Bottom Line:
Health management systems and the Big Data they gather need to be analyzed and structured into useful information that is actionable to improve performance, reduce downtime, and predict failures before they interrupt operations. Digital twin technology is a logical approach to the problem, simulating physical operations to predict future maintenance events and helping to optimize engine and aircraft performance.
While collecting the data is important, the ability to quickly and accurately analyze the data and turn data into actionable information that provides a payback is key. The MRO business is changing, from borescopes and wrench turners to simulation modelers and logistics systems that will determine when a part needs changing and having it at the right place at the right time to minimize costs and downtime. IT is changing the way we think about MRO.