The Allplane Podcast #59-Opticlimb & Opticruise with François Chazelle, SITA for Aircraft
Today we continue exploring the key role that existing technology, and software in particular, can play in making aviation more sustainable and, generally speaking, more efficient.
In the second episode of this 4-part series produced in collaboration with SITA for Aircraft, François Chazelle, will explain how, as Head of Sales Support for OptiFlight solutions at Safety Line, a French startup acquired by SITA in 2021, he helped develop a family of software solutions that crunch large amounts of data from multiple sources to come up with recommendations that help airlines and the planet save millions of dollars and a huge amount of carbon emissions.
I found it really interesting that François, now Vice President Commercial, Middle East & Africa at SITA for Aircraft overseeing the OptiFlight family of products, operates at the intersection of three major themes that are defining our time: big data, machine learning and sustainability.
While I was aware of the potential of algorithms to squeeze even more efficiency gains from airliner operations, in this chat François provides a comprehensive overview of how this is done and, also, of what are the barriers preventing a much widespread adoption of this technology.
But, as usual, is it best to hear it from the man himself…without further ado, let me welcome François to the podcast!
Download this episode from:
Apple Podcasts / iTunes, Spotify, Google Podcasts or Stitcher
Things we talk about in this episode:
François background in aviation
Safety Line, its start, growth and acquisition by SITA in 2021
What is OptiFlight, the different software solutions it contains
How technology can help pilots fly more efficiently during key moments of the flight
What are the obstacles that prevent a more general adoption of optimization software by airlines
Which future flight optimization products can we expect
Resources:
Transavia, Safety Line’s launch customer
Podcast Music: Five Armies by Kevin MacLeod
Link: https://incompetech.filmmusic.io/song/3762-five-armies
License: http://creativecommons.org/licenses/by/4.0/
Interview transcript
Hello François, Bonjour! How are you?
Bonjour! Very good. Thank you.
Thank you very much for coming today on the podcast! You are, right now, a Senior product Manager at Sita for aircraft. But before that you were the Chief Commercial Officer of a company that Sita acquired in July 2021 that was called The Safety Line. And you are active in a very, very interesting area that I hope we will be able to speak about in depth today, which is basically technology to optimize the flight of aircraft and make them more sustainable, and help airlines burn less fuel. So first of all, like all the guests that come here on the podcast, I will ask you to introduce yourself because I think you have a very, very interesting career in the field of aviation technology. So can you tell us a bit about yourself and how you ended up at SITA?
Yeah, thanks a lot for the introduction, Miquel. I started as an aerospace engineer, my career started in Moscow where I worked for five years, and stayed 10 years mostly selling services to airlines. And then I moved on to Airbus, where I was still in sales, selling aircraft first in Eastern Europe and then eventually ended up heading the sales team for Airbus corporate jets. And at some point, I decided to move on and do something else. A friend of mine that I knew for 30 years had started Safety Line and needed a sales director. And so I really jumped in and went into this startup adventure. And, you know, the acquisition by SITA in July is a marker of success of this venture. And we're very proud to now be part of Sita for Aircraft and be able to leverage their international network to sell our fuel saving solutions to the whole world.
It was a startup, how long were you there? And how did you start? You started in France and from the very beginning working on software and technology products to optimize flights…So what's the story? Can you tell us very briefly about this venture, because it is a really, really interesting story.
Actually the founder of Safety Line, which was founded 10 years ago, we just marked the 10th anniversary Saturday, was an aircraft accident investigator for the French authorities. So he had good knowledge of the flight data that he used to analyze these accidents and incidents. And he also had experience for two years as a commercial pilot with Japan Airlines. So, as a pilot, he saw that, you know, there could be better ways to fly, more optimal ways to fly. And he thought of putting those two together and using the flight data, not just for safety purposes, but to improve flight efficiency. So initially, the first products that he came up with were more safety focused, a safety management system that we're still selling today. And one of our largest customers is Airbus Helicopters, also Group ADP, the Paris airports group. So some big customers and a lot of smaller ones as well. And it's only in 2013, that we moved to the efficiency side with T flight really first with research projects, because the idea was to see how we could use big data and machine learning to better optimize the climb phase, which is the most complex phase of the flight, but also the one where you can achieve the most savings. And there was a lot of research initially with our data scientists for several years, before we finally got a customer on board, Transavia airlines, in 2014, they did some first trials 2016, they committed to 10% of their flights, using Opticlimb solution. So it was still experimental, but they were already a paid customer. And it's only in 2018, that we also convinced transit Netherlands who shares the same procedures with Transavia France, to get on board as well. And at that point, we were able then to extend to 100% of the fleet of both customers. So really our first, you know, full customer for Opticlimb came in 2018 after many years of research and trials. Since then, we've accumulated a lot of customers, we now have 22 Airlines on board in just that short time. And despite COVID we did sign up some new customers during the COVID crisis, because fuel and cost savings stay a focus. In these times actually airlines are feeling pressure to demonstrate that they're really short and emissions conscious. And we're there to help them do that.
I can see you are at the intersection of several macro trends, you are in big data, machine learning and sustainability. So those are like three three big, big, big topic themes of our time…!
Yeah, absolutely. And the specificity of having a fully aviation dedicated activity. All our customer facing staff are our aerospace engineers, like myself and our founder. Some of them are pilots like a few others in the team. So we're really focused, and we understand the aviation community's needs. And so anyone who's speaking to customers is not with an IT background or with an aviation background. And then we get the IT and data science teams involved in the back end.
Yeah, actually, I was amazed at a recent Airbus summit that took place just a few days ago, that they mentioned some figures about how much extra efficiency you could squeeze out of flight operations. I think the numbers they mentioned were something in the order of between 5 and 10%, something like that. So in the big order of things that's quite a big number, if you multiply by all the planes that are operating out there.
Absolutely, we would agree with those figures, 10% potential, and with Opticlimb we can contribute to significant parts of that, as well as with the other flight solutions we are addressing the other flight phases.
So can you tell us about the offering? How does it work? I mean, is this piece of software, is this a cloud based solution? How does this work in practice? So if I'm an airline, and I wish to optimize my operations, what's the way to go? What does it mean in practical terms?
So again, it's really about using the historical flight data to build machine learning performance models for each individual tail, which will allow us to predict the fuel burn for each towel for each flight, looking at many different scenarios, and so that the first job will be to share that one year of data with us, for each for each aircraft. And we'll build those models, and then we'll need to receive the five plans for each flight. So that's in combination with 40 weather forecasts, we can really simulate different scenarios before the flight, predict the fuel burn, and then recommend the best scenario to the pilots saying, Okay, if you follow these speeds, these altitudes, these are the savings that you can achieve. So it's really about using predictive models so that we can help make it happen in the cockpit.
So basically, you use the past data to build a sort of profile for each aircraft, and for each route that that aircraft is flying, or you have some flexibility here in terms of the itinerary?
No, the performance doesn't depend on the route. So it's really a performance model for each tail that we take into account, you know, the whole network that will be flying over a year since we take one year of data to build these machine learning performance models.
And is there much variation? for example, with different pilots? do you have this on an aircraft by aircraft basis? but cannot the same aircraft flown by different pilots have a very different performance in terms of fuel consumption or flight patterns?
Yeah, so I mean, if you look at the way you drive your car, for instance, if you drive more aggressively, you'll consume more fuel. The great thing about the climate phase is that it's fully automated, meaning that the pilot will enter the time speeds into the flight management computer before the flight, and then it will automatically be flown. So it's really not pilot dependent. In this case. It's about entering the right speeds before flight preparation, and we will, you know, enter really customized speeds for each flight with different speeds, and different speech changes at different altitudes than what would normally happen with your financial computer.
And then you mix this past data with the real world. I don't know if you can call it real time, but present time data of the weather forecast for a specific flight.
Yeah, exactly. I mean, you can actually say the future because it's a forecast
True!
Yeah. But yes, it's a four day forecast, meaning that we have a weather forecast at a given point in the trajectory at a given time. So yes, it is the future if you know what the weather will be at that point at that time. And for climate, for instance, we'll look at wind and temperature forecasts for every 1000 feet, which is really much more inputs than then you would be able to enter into the flight management computer otherwise,
And what are the sources? I guess, obviously, the flight data is from the airline, what about the weather forecasts? do you have partners that file these very detailed forecasts?
Yeah, so weather is really a specialty. We work with a company called Made Safe, in Toulouse, they are a spinoff of Méteo France. And so because they are a spin off of Méteo France, they have access to National Weather forecasts from other National Weather providers. And so the deal that we have with them is that they are the ones who provide the best weather for a particular region. They are the specialists that we count on to provide us with the best weather source for a given week. anywhere we're optimizing a flight
So, this is global, for any region in the world, it can be France, Indonesia, Brazil, wherever…
Absolutely. They provide global weather forecasts. And we're a big user, because for each trajectory that we optimize we will probably run 10s of 1000s of scenarios. So you can imagine the number of weather points that we are requesting from them on a continuous basis
Then, when you have combined the sources of data, run it through your algorithms, your models, what does the airline get? And how does it get an actionable set of recommendations? And how are these recommendations then translated into specific actions at the operational level, or at aircraft level?
Yeah, so yeah, there's three, two things. The first one is how to deliver the recommendations to the pilots. And then the second thing is, how do they use these recommendations. So, the first one is that we really wanted our solution to be universal. So we will adapt to whatever the airlines are using to communicate the briefing package to the pilot, some airlines are still working with paper briefing package. So in those cases, we'll send the artifact recommendations and a PDF to dispatch so they can be printed. With each briefing package. A lot of them are using electronic flight bags with either electronic flight folders or weather applications or navigation applications. And we can integrate with each of these types of ESD apps ready to make sure that we are the ones adapting to what airlines are using and we're not forcing an additional app or an additional integration to them, we will adapt to what they have. So that's that's for delivery, the pilots will receive the OptiClimb recommendations in the briefing package or the ESD application, and then, very simply, it will be three speeds to be entered into the flight management computer, they will do this during flight preparation, and then the profile will be automatically flown. And the great thing is that we bring the aircraft to the same point in distance at the same time. So we're still respecting the fight plan, which is having a better trade off between vertical and horizontal speeds, that has a more efficient profile. So because we're sticking to the fight plan, pilots are able to apply our speeds on average 75% of the time, which makes it really a systematic opportunities. So it's both universal because it works out any airline, and systematic because pilots can use it 75% of the time,
Pardon my ignorance, because I'm not a pilot, so some of these small details of flight operations escape me a little bit, but you mentioned that you stick to the flight plan, but what's the degree of flexibility that the system allows in an environment that is controlled by air traffic control? There might be other other constraints as well, that I'm not aware of, but what is the degree of flexibility? And that's maybe more of a general question, what's the degree of flexibility that each aircraft has to adjust its movements in this context, I mean, what other factors can influence these movements of the aircraft.
So air traffic control will be a big constraint for our crew solutions. If you want to take a shortcut that we recommend with LT direct, you will need to request to ATC. If you want to change flight levels as recommended by Optilevel, you also need to make a request to ATC. But you make informed requests that have better chances of being accepted. Whereas for climb, again, because we're complying with a flight plan, there's no requirement to request anything from ATC. However, in some cases, you know, whether you're using Opticlimb or not, ATC will sometime request to impose a particular speed or they might ask for a level of or also the other case where you might not be able to apply our speeds as if there's turbulence and and the pilot will need to adapt the speeds to safety speeds. So those are really the three cases where maybe pilots will not be able to apply all the time. And that's why we have not 100% But you know, an average of 75%. But still, it's really unique and that the application rates are extremely high compared to other best practices for flight phase that is receiving the most
Basically all the action as far as software and technology are concerned is before the flight one here in the flight you are not supplying any new data…
So because we use you know, mostly winds and temperatures and the weather forecasts that we need the the two things that you want to look at in terms of performance are winds and temperatures contrary to weather events, you know, like icing turbulence, etc, once in temperatures are highly accurately forecasted, at least for the next 12 hours. Which means that even if there's no connectivity, you know, using these forecasts, we have something quite accurate for these medium haul flights. For the very long flights, of course, there may be a benefit in having updated weather. And the good news is that when, you know, when airlines have connectivity on board, we are able to piggyback on that connectivity and update our recommendations. So most airlines, you know, will use the non connected version, and that's fine. But for very long flights, and particularly for flight level changes, where it will be interesting to have an update on the weights of the aircraft and the actual position to make more informed recommendations, there will be benefits in using the existing connectivity.
What about post flight? Do you provide some sort of analysis or review of performance?
Yeah, so because we are proposing new best practices that were not existing previously, we've had to develop our own tools to measure the savings for the airlines. So after the flight, we'll analyze the flight data, the QAR data, and basically check whether the pilots were able to follow our recommendations or not, whether it's for opticlimb, optidirector of optilevel, and every time they were able to follow our recommendations, we'll give the airline the the amount of savings that is achieved for each fight with these initiatives.
When you say you give the airline the amount of savings, you mean, you provide a report quantifying this?
Exactly, we will provide a monthly report that will show both the application rates and the savings. And we also give them access to a web interface, which allows them to see the savings for each flight.
And I don't know if you have the numbers now at hand, but what could be the average saving…tons of fuel saved? and how this translates into money and carbon emissions? That it's an increasingly important element to quantify as well.
So as an example, just with OptiClimb, airlines can save around 81 tons of fuel per aircraft per year, which translates to around 255 tons of co2 emissions avoided,
That would be what? What type of aircraft? let's say, a midsize aircraft like Airbus A320, Boeing 737, something like that?
So the good news is that this is actually an average conservative figure. Taking into account very simple assumptions is that when the aircraft is bigger, it consumes more fuel, but it also does less flights. So it kind of averages itself out. And of course, it's a conservative figure, when we start working with an airline we will, of course, give them the actual savings. You know, we have an example for instance, with Aerologic, an aircraft cargo operator who operates very heavy aircraft, so they will save more than those 81 tonnes per year. But as a conservative figure, basically, we're assuming that a narrow body can save around 75 kilograms per climb. And we'll do it an average of four times per day. So it's a potential of 300 kilograms saved each day. And likewise, a wide body maybe would save 150 kilograms per day and do two flights per day. So you know, again, it's a rough figure, but it's something you can use as a first approach before you actually measure for each airline. What are the stages that each airline operates, different ways, different temperatures, different route lengths, etc. So we'll, of course, fine tune those figures as we go.
And that translates in in money and carbon terms…
So again, yeah, that rule of thumb is great, because in terms of dollar savings, airlines can save around $1 million per year for every 20 aircraft. So say you have 100 aircraft in your fleet that will be $5 million dollars saved each year. And in terms of emissions avoided, co2 emissions avoided, we're looking at 255 tonnes again, the rough figure 255 tonnes per aircraft per year,
When it comes to applying these technologies, what sort of barriers? Or what sort of challenges are there? Because it seems maybe even too good to be true, right? So you, you have all this data, and you put it to good use, but I'm sure I mean, this is a highly regulated environment, there are lots of things to take into account, what is preventing more widespread adoption of these types of systems?
You know, first of all, as I said, initially, we've been adding on a lot of customers since our first full customer in 2018. But yes, there are sometimes some barriers, particularly with for some larger legacy airlines, sharing the data, sometimes can be an issue, you know, there's usually agreements with the unions that prevent airlines from sharing the data for anything else than safety purposes. And you know, these have been existing for a long time. And so, in those cases, the airlines need to make a case to the unions that this is a useful application for the data. And also that, you know, the data will not be used for other purposes, such as monitoring how each pilot flies or anything like that. So some guarantee Used to be given to the unions that there is no other use of the data then. And for safety and efficiency. That takes time. But it's achievable. Even with the largest airlines, we've been able to onboard some airlines who had these types of agreements in place and with negotiations with the unions, they were able to free the data.
So there is no, let's say, singling out specific pilots for a better use of fuel or something like that?
So the way it works, we don't need to know which pilot is flying which aircraft. But some airlines have asked us in order to give some assurances to the pilots to report the savings on a monthly basis, but without showing time and dates in the web application that management is using, so that they can't directly associate a flight to a pilot. Of course, it will depend on the region in which you operate and, and the practices that exist in each region. But in Europe, for instance, typically, we find ways for the airlines to be able to reassure the pilots that this is not being used against them.
And in terms of hardware, do you need to know specific hardware is everything it's software that can be just accessed through any interface?
Yeah, so it's all software, it's a cloud based solution running on our private servers, based in France, and there is no hardware required. No, we don't charge any setup fees or integration fees. Really, we want to keep it subscription base so that the airlines can re you know, have some strong ROI is looking at what are the savings versus what are the fees?
Basically, you crunch these numbers on your own computers on your own systems. And then you input so you take the input from the airline's mix it with your own, like your secret sauce, and then you deliver the output but everything happens on your side of things.
Yeah, so we don't own the servers, we use third party providers for the servers. But yes, it's a it's our controlled servers are computations that are running on our own servers. And it's very, you know, very heavy calculations. It's not something that could be done on an iPad or or a PC inside the cockpit, you really need this, this computing power. But we put enough power into the computations for the results to come out quickly. So despite the fact that we're looking at 10s of 1000s of scenarios, the airlines will receive the recommended speeds within minutes.
We've been talking often during the conversation about opticlimb, opticruise…those are basically the two main products you have now in the market optic line, obviously for the for the climbing phase of the of the flight, optic cruise covers other parts of the flight, things like speed as well, the itinerary, what else can you manage through Opticruise?
So Opticruise actually includes three solutions. The first one is OptiDirect, which is our shortcuts solution. Basically, pilots are already used to requesting shortcuts to ATC, but they don't really have tools in the cockpit to tell them which shortcuts have chances of being granted, or how much they could save with today's winds and temperatures using that shortcut. And so what we do is we look at historical tracks flown. And we only recommend shortcuts to the pilots that we know have been flown before. So this really increases the chances that ATC might be willing to grant them. And we also tell the pilots how much fuel and time they can save on today's flight based on today's winds and temperatures on these shortcuts. So that really allows pilots to more systematically request shortcuts, knowing they have a better chance of being granted. And, in fact, when we measure the savings, we see that they are indeed able to take a good number of shortcuts that we are recommending, which more than covers the cost of a solution. Then we have Optilevel which is about flight level changes, you know, flight level and flight level changes are only indicated at waypoints in the flight plan. Whereas you could change levels in between waypoints. And there's also a number of constraints that can be cleared when you request to ATC. And so the idea is that when the aircraft gets lighter, it can go higher. When it's higher, it usually consumes less fuel except if the winds are unfavorable. And so that's where you know when a pilot sees that his aircraft is light enough to go up. We will tell him whether it's worth it or not, depending on what are the winds at the flight level he's going to
So you have these two products, any other products or projects that you are working on at the moment?
Yeah, so for cruise we also have OptiSpeed which was requested by the chief pilot of Transavia France, basically, you know, airlines often building some extra time into the schedule just to make sure that they take into account things that can happen longer taxis etc. And sometimes they find that when they align to the runway, they see that they have extra time. And sometimes it's not very useful to arrive in advance at an airport, because you have a slot with your handler with your gates, and they may not be ready to welcome you. So it could actually generate additional costs. And the idea was to see whether we could use that extra time to slow down a little bit and save some fuel. The problem is that depending on the winds, if you slow down too much, there's a possibility that at some point, you're standing still, and you know, your trip fuel will be increased, because you're not going fast enough. So it's really about helping the pilots anticipate what will be the impact on fuel and time of using a lower Mach number to save fuel. And then there's a point where, you know, they can slow down to a certain speed, but going down further, will still spend more time but will not save more fuel, it can be used the other way as well, if pilots need to make up for time, you know, they would usually accelerate, and then maybe find that they actually went too fast. And that's not beneficial either. So we can tell them exactly which Mach number will save how many minutes that they need to recoup. So that's up to speed. In the research phase, we also have OptiDescent, which is now being worked on with our data science team and being transferred to our operations team. For exposure to our innovation partner airlines who are working on this solution, we have three innovations or airlines involved in OptiDescent. And the idea is very different. It's not just about optimizing the speeds, but it's more about anticipating what will be the actual descent path because very surprisingly the descent is often very constrained. And it's really ATC who will impose a particular descent upon you. And it may not be the descent that was in the flight plan or in the charts. So the idea is really to help pilots using historical tracks again, anticipate on what are the likely design paths that he will be asked to do to perform if he's coming from a certain direction at a certain time to a certain destination, so that he can be ready for the possible alternative designs that may be imposed upon him.
22 Airlines already using these systems? Do you commercialize them as SITA for Aircraft or the Safety Line brand is still active? How can people find more about that, basically?
Yeah. So of course, since we were acquired in July, we're still in transition, but the branding is really OptiFlight, you know it's around the product. And so that will continue to to stay we will continue naming the products with this opti branding, which is registered and now property of Sita. And going forward really this is joining the SITA for Aircraft Digital Day of Operations product range. So we are you know, with the OptiFlight solutions becoming one of the Digital Day of Operations solutions.
And where would you send people that would like to learn more about it? Any specific websites, digital channels, social media channels where they can get more detailed information and updates about all the new products that you guys are launching?
Yep, so the historical Safety Line website is still active during the transition phase, but you can also go to the SITA for Aircraft website, and you will already have information about the OptiFlight solutions.
Well, thank you so much for all these round up about this very exciting technology. But you know, we hear so much about new propulsion technology whether it’s electric, hydrogen… that of course, that's very important, sustainable aviation fuel as well. But there's still so much extra optimization that can be squeezed from existing technologies that…yeah, I guess we're gonna see more and more as sustainability becomes more and more important now it's practically on top of the agenda for every airline out there. So yeah, it's interesting to see how big data in this case has these very practical applications.
You're right. I mean, like you said, we don't need to wait for these new technologies, SAF and electric etc. to come. There are opportunities today to save a lot of fuel thanks to machine learning and big data, but also So in the future you know for now the SAF fuels are quite expensive so you will still want to save on those and so our technology will still be applicable for those.
Yeah, of course, for sure. Well, thank you so much, François. It's been great having you on the show today. Wishing you all the best with all these new products you have on the line! Take care.
Alright, thanks for the opportunity. You too, take care!