DGITAL TWINS they are virtual representations of a component, a device or even an entire production line in a factory. Powered by data from sensors installed at their opposite physical number, the digital version can be used to plan maintenance, detect any emerging issues, and simulate the effect of design updates and changes. Aircraft engines in particular are followed by their manufacturers in this way. General Electric, Rolls-Royce and Pratt & Whitney, the three main companies involved, maintain simulations of individual engines at engineering centers on the ground. These suck real-time data from their on-air counterparts.
At Cranfield University in Britain, however, a group of engineers hope to go further. They propose to expand the idea of digital twinning to produce what they grandly call “aware aircraft”. This would involve creating a digital twin of an entire plane, merging its various monitoring systems and interpreting the result using artificial intelligence (AI).
Currently, monitoring systems on airplanes tend to work independently of each other. This means, for example, that if an eye sensor on a valve in the fuel system indicates a problem, engineers can replace that valve, only to find that the problem remains because the real fault lies in the electrical system. Ian Jennions, technical director of Cranfield’s Integrated Vehicle Health Management Center, proposes a “reasoning system” that presents an overview of individual monitoring systems.
Since it’s not practical to put a sensor on every potentially breakable part of an airplane so it doesn’t get too heavy to take off, this will use AI trained to find faults that can be detected from broader measurements such as vibration patterns and temperature changes. Furthermore, this entire plane digital twin is likely to reside on the aircraft itself, rather than relying on large amounts of raw data transmitted to a maintenance base for processing. This will provide faster analysis and avoid transmission bottlenecks.
All of this will allow an aircraft to reconfigure things like battery connections and fuel usage on the fly, so to speak. The reasoning system would also keep an eye on things like the weather the aircraft was operating in, because that can also affect parts wear. Flying through clouds of dust, for example, can accelerate the abrasion of engine parts.
The aircraft would communicate with its pilots (assuming there were any, as some future aircraft, especially freighters, could very well be drones) and also with ground engineers. Such a plane would, in a sense, be “aware” of how its various components were acting and interacting with one another. It is this level of self-awareness that the researchers use as an excuse to describe the system as conscious, a term they say reflects a direction of travel rather than a goal.
This self-awareness, however, would allow the plane to book itself for maintenance when needed, rather than a fixed schedule, and order the necessary replacement parts in advance. The predictive maintenance that such a system would provide should help prevent unscheduled downtime and repairs. Every day a passenger jet is on the ground costs several hundred thousand dollars in lost revenue. Jennions believes that a conscious aircraft could reduce maintenance expenses by around 30%.
This maintenance can also be automated, at least in part. Cranfield, which started as an aeronautics college in 1946, has its own airport, a hangar and a Boeing 737 to try out. Researchers are exploring using drones equipped with optical and thermal sensors to fly around an aircraft in a hangar, looking for external anomalies. Meanwhile, small snake-like service robots can crawl into enclosed areas such as fuel tanks to carry out repairs.
Cranfield’s team hopes to have an entire aircraft digital twin operating by 2024 and a prototype with a degree of self-awareness flying an aircraft by 2035. It remains to be seen whether this system will include a sense of shame for poor cabin service. . ■
To get more out of our mind-expanding science coverage, subscribe to Simply Science, our weekly newsletter.