June 5, 2015 | Written by: Ravi Seshadri
Last Friday I was with a mathematician-friend for dinner and as we finished scooping up the last traces of the heavenly home-baked spaghetti, we arrived at our common interest: math in airline operations; and my friend ventured to show me around the beauties of a state-of-the-art flight planning system that she had helped design. A math buff myself I was aware that a modern flight planning system came with quite a bit of sophistication, but the level of complexity and panache with which this actually handled it left me a bit dazed I must admit.
If you are wondering what this system does, it computes several variables and comes up with the best path (aka Flight Plan) to take for a flight on a given day and type of aircraft. The specific Flight Plan that we held in our hands that day had seen more than 130,000 computations for a six-hour flight mission of an Airbus A330 aircraft – which is more than 6 computations for every second of the flight duration.
While that sounds like a lot of work the real heavy lifting is about not only taking into account the correct physics of the mission: aircraft performance, weather; but also route restrictions, economics: fuel, time-based costs, overflight costs; involving loads of non-linear, non-continuous equations that described the airline’s dynamic state (i.e. not just where in airspace is the aircraft located at the moment but also how it got to that point). And use this to compute how much fuel is needed to accomplish the mission.
And we know fueling up an aircraft is not quite similar to filling up the car tanks before setting off on that long road trip: fuel on-board is additional weight like everything else in the aircraft, and will cause extra fuel burn. Which means, fuel that is not needed for a mission would rather not be on-board else we burn some of it (typically 3% for an A330: every 1KG of fuel on-board burns an extra 30g) – causing economic waste as well as avoidable carbon emissions.
Now with all this math mumbo jumbo you would expect that fuel load calculations are quite thoroughly done with the flight planning system, and every drop of fuel not necessary for a safe and successful mission is squeezed away. Well, oddly enough, that is often not the case! Anybody who has been with an airline’s Flight Operations longer than a week will tell you, flights routinely land and park after their missions with quite a bit of fuel in their tanks.
How do we unravel this: we have some of the world’s smartest math software computing fuel loads and still we have flights landing with avoidable, extra fuel on-board?
The answer to this lies in the “discretionary” decisions that pilots and flight planners take: a few routine aspects of a flight that are not fully within the ambit of the flight planning system, and are hence left to the discretion of planners and pilots on the day of operations.
Depending on variables such as wind direction and how far the runway is from the gate, the time spent by an aircraft in taxiing can vary. The flight plan gets published a few hours ahead of departure so it does not precisely determine the taxi times at departure and arrival. So planners and pilots come up with a guesstimate for taxi-out and taxi-in time in minutes. And yes, almost always they err “on the side of caution” with these estimates.
The operational flight plan for a long-haul could be published several hours ahead of a mid-flight weather event, and therefore might not have accounted for enroute weather developments. So planners/pilots typically guess what detours they might need to take enroute and at what point in the flight, and often come up with simplistic fuel load estimations for the additional flight time and burn.
Quite often a flight is asked by the navigation control tower to ‘hold’ at a point during its approach while descending into an airport. This could be caused by congestion in the airport’s airspace, a technical event on ground that delays landings etc. While this is something that could happen anywhere anytime, historical patterns can be quite helpful is determining where is it more likely to happen and at what time of day. However pilots and planners rarely have access to such intelligence and routinely resort again to a guesstimate of hold time.
Choice of an alternate (to land in case flight cannot land at destination airport due to inclement weather) is often determined by the airline’s own guidelines and are usually the same for a given destination. This again determines how much fuel to carry as the aircraft should have enough to take a detour from destination (where it could not land) and fly out to the alternate. The static guideline-driven process might not take into account latest available weather data for the alternate to decide on its suitability. By examining historical accuracy of weather forecasts for alternate planners can make more informed choices of alternate for a specific flight and time of day. And thus make a more prudent decision on the extra fuel needed.
So how big is this issue?
From our work with major industry-leading airline clients across the world we observe that typically 0.5% to 1% of fuel is routinely wasted from carrying excess fuel onboard. Now, for a 100-aircraft airline the annual fuel bill is about USD 1 Billion, and this avoidable fuel burn from carrying excess onboard translates to USD 5-10 Million.
Industry-wide, potential savings from smarter decisions on discretionary fuel (like the ones above) is USD 1 Billion, at 0.5% of IATA’s 2014 estimated annual fuel spend at USD 204 Billion; not mentioning the several thousand tons of carbon not emitted.
Is this attainable?
Do we have an intelligent system that spots patterns from crunching historical data together with real-time weather/traffic feeds, come up with statistically grounded predictions and recommend to pilots/planners on the discretionary fuel loads?
The answer thankfully, is YES!!
More about this in my next, coming soon. Okay, I am not going to wait till my next dinner appointment with my math friend to share this.