Aviation AI Tech for Greener Flights

One of the most enriching aspects of a lived conference is often found outside the formal sessions in what’s affectionately known as the “Hallway Track.” It’s where spontaneous conversations (on potentially tangential topics), unexpected collaborations, and invaluable networking occur. Interestingly, there’s even a conference dedicated to this informal yet vital element of in-person events – an “Unconference.”

One of the most interesting discussions I had was at such an unconference during the World Aviation Festival 2023 (WAF23), at the cocktail reception hosted by PROS. In addition to conversations with customers, prospects, colleagues, friends, and industry acquaintances, I also connected with individuals unrelated to PROS in the aviation industry. Bearing my Chief AI Strategist title, our discussions naturally focused on AI (which is already a focal topic at WAF23) and the automation of complex airline operations. However, I wasn’t expecting to hear about an AI solution that makes flying greener by reducing 60% of the warming effects of flights from Tim Winter of SATAVIA.

Climate Impacts of Contrails

If you look up and see a plane against the blue sky, you will likely recognize the trails of white clouds left behind. These condensation trails (a.k.a. contrails) are a result of engine exhaust condensing moisture in the cold upper atmosphere. While contrails are mostly ice crystals (not CO₂), they act like natural cirrus clouds that can linger, spread, and trap outgoing radiation. But contrails are completely artificial, and they can double the warming impact of aircraft engine emissions. Yet, they can be avoided with the help of AI.

I had recently learned of a Google effort using AI/ML to predict regions in the atmosphere that are conducive to contrail formation. Subsequently, by rerouting just 1.7% of flights to avoid the humid regions of the atmosphere, we could cut contrail warming by 59% while burning only 0.014% more fuel.

2 Approaches, Same Goal

Google is tackling this contrail prediction problem with a purely empirical approach. Using the global flight path data, they can reliably find and label contrails on satellite image databases. Then for those regions and times where contrails are observed, they can examine the weather database to get the precise atmosphere conditions where contrails would develop. These are the training data that go into the supervised learning model that learns to predict the presence of contrail from measured atmospheric conditions.

Meanwhile, SATAVIA is basically trying to accomplish the same thing, but with a completely different approach. Using atmospheric physics and climate science, they can predict the formation of contrails with much less data. This is because the ML model does not need to learn the complex physical mechanism of contrail formation from the data directly. These mechanisms are already specified by known physics from first principles.

From Prediction to Action

Aside from purely academic exercises, prediction alone is rarely the end goal in real-world applications. We need to turn predictions into prescribed actions. And with an accurate prediction of when and where in the sky contrails would likely develop, the prescribed action is obviously to avoid those regions. In most flights, airline pilots already do this to avoid regions with high turbulence. This is just one more data point to facilitate pilots in planning their routes. And studies have shown that pilots don’t even need to deviate much from their normal routes, as the conditions for contrail formation are quite specific and narrow.

As a citizen of Planet Earth, I believe that climate impact is everyone’s responsibility. But as a global frequent flyer, I have a greater sense of urgency on the climate impacts of flights. So I am quite interested to see how these AI solutions are adopted by airlines around the world.

Coincidentally, I will be at the Aviation Fest Asia 2024 (AFA24) next week, which is again hosted by Terrapinn. I will also be giving an AI workshop, and the following is what I will cover. It’s not about Green AI tech nor is it about sustainability in aviation. But it is the foundation for airlines to be able to invest in a greener and more sustainable future of aviation.

Customer-Centric Airline Retailing: Powered by AI
Abstract:
In the realm of airline operations, achieving customer-centricity without compromising profitability has long been a formidable challenge. Despite the advent of revenue management (RM) techniques, ensuring exceptional travel experiences for individual passengers remains a human-intensive task, particularly during disruptions. However, AI that can automate intricate workflows that were once exclusively within the purview of human intelligence is now available. This paradigm shift involves predicting customer preferences, optimizing pricing and inventories, and tailoring personalized offers. The result is an elevated customer experience while improving revenue, and even cutting operational IT costs. This workshop will take you on a transformative journey that delves into the fundamental components of an AI-powered airline retail system. Join us and witness how AI is poised to redefine the future of travel with a focus on customer-centricity.

Conclusion

Just as WAF23 was my first WAF experience, this is also my first AFA experience. I am eager to learn more about how airlines in Asia are transforming their business and operations with AI. I look forward to seeing you there. Come and talk to me. Maybe we can have a super interesting conversation in the “Hallway Track.”