Where do FR24 get their aircraft reg linked to scheduled future flight data?
 

Been trying to plot delays for a party of friends just left the Mediterranean headed northwards (on time!) for warmer climes (such is the times!)

I was confident I knew the reg no of their aircraft even before they read it from the hull as they boarded, but FR24 told me it was going somewhere else and another aircraft was scheduled to do their flight.

And so it remained until the aircraft taxied and lit up on live plots in FR24. Same was true about the aircraft which was originally scheduled to do their flight but wouldn't be arriving for a while yet.

Currently the aircraft they are on is now shown by FR24 as going where they are going (that's handy!) but it is also shown some 30 minutes after it became airborne in a second FR24 data record today as still going to its original destination a bit later. It'll be defying the laws of physics if it ever manages such a feat today!

I can work it out what's been juggled, and I was guessing they might do it sometime before boarding, but I just wondered if anyone knew where FR24 gets that particular data and why their algorithms aren't quite up to the job of filtering out the nonsense data the moment it becomes nonsense according to their own better data through other streams!

I don't think that's what the OP is referring to.

FR24 often shows, for a given aircraft registration, flights that are planned hours, or even days, in the future.

I've often wondered, too, where that information comes from, as it often turns out to be wrong - its clearly not from a transponder.

Thanks Dave. Yes that's what I was referring to, coupled with the fact that sometimes for a given aircraft that scheduled data is not updated until the another aircraft is allocated to the flight and doors are shut, and that maybe as happened today, the bad data persists for a couple of hours after it has became physically impossible for the original aircraft to operate on the flight. Today that was because for those two hours it had been heading away from the departure point on a totally different flight!

FR24 is of course no longer a toy. What it means is that any passenger attempting to verify that their aircraft will be on time may need to be quite lateral in analysing where their aircraft may be if not already on the ground. Said pax also has to be quite novel in recognising data that must be discarded and hunting down the correct answer by elimination and deduction!

Aside from where FR24 gets the sometimes superceded data, I am pleased to report how my little delay prediction investigation today did however impress me when seeing how Ryanair has built schedules which not only allow for variable flight times, but can also be juggled at short notice to mitigate variable delays. What was approaching a three hour departure delay at Stansted caused a knock on arrival delay at another destination later in the evening of less than an hour I think. Quite beautiful to behold really, and I got told off for telling my friends they might be delayed because their aircraft was still three hours away when they had been called to board! 'O ye of little faith' they said as they boarded on time!

I guess the size of the fleet helps, but remember it didn't involve using any "spare" aircraft, just well planned use of what I guess is planned-in slack as might be possible at a particular base when an inbound is seen to be very late..

I think to a large extent you make your own luck if you can, TA ;)

Might be worth a closer look at the specific example if you like - we are talking about Athens based* aircraft. Take a look at FR14 31.7 operated by EI-ESZ which arrived in Stansted on time. It was then almost 3 hour delayed (no EU261 claims possible there!) as FR15 back to Athens and had next been pre-allocated (according to Flightradar24) to FR4934 Athens to Billund. Now when FR4934 were called to board there were two FRs on the ground allocated to other destinations, but there was a third very shortly inbound from Memmingen that was due out again two and a bit hours later to Bratislava. That was EI-FTA. It arrived on stand where FR4934 were waiting to board.

EI-ESZ remained allocated to FR4934 (per Flightradar24) right up to EI-FTA startup in its place. EI-FTA took off operating FR4934 on time, and probably simultaneously EI-ESZ ceased to be allocated to FR4934 per Flightradar24 data. I saw a while later that for a period it didn't show as allocated to any other flight 31.7 until it actually started up an hour or so late to Bratislava! Meantime though for those same couple of hours EI-FTA was still allocated for Athens-Bratislava even though it was more than half way to Billund :}

Both aircraft returned to Athens base where they had both started the day* - EI-FTA on time and EI-ESZ actually an hour and a half late as it picked up extra delay in Bratislava but still nowhere near the 3 hour delay it had picked up at Stansted - I am still impressed. It is surely part of Ryanair's base route schedule design plan and quite clever if I am right about it.

*Edit: Actually not quite true - EI-FTA started in Santorini and short-hopped to Athens and returned there 40 odd mins late through Athens last flight (8 sectors yesterday 31.7), but all the same, quite neat!

I don't see it as anything else, an occasionally useful toy I would agree, not something I would use as the single source for making decisions though.

A very interesting area as I have been spending several hours trawling data from FR24 for some work I am doing trying (there must be a better way of accessing data but it would cost). They state

Disclaimer: The information provided on this page is a compilation of data from many different sources including flight scheduling systems, airline booking systems, airports, airlines and other third-party data providers. The data is provided as is, there are no guarantees that the information is fully correct or up to date. Changes and errors may occur. Therefore Flightradar24 cannot be held liable either for the accuracy of the information or for ensuring that the information is up to date at all times. Some of the flights presented may be charter, cargo, ambulance or other types of flights not available for passenger travel.

I assume from the above that apart from the obvious source of transponders they access airline websites or possible their computer system directly. I believe that the Schiphol enthusiasts site uses a webcrawler.

I have generally found that info on passenger flights is much more accurate that than for cargo where there are often duplicate entries (a middle sector of a triangular flight may have the flight number of both the inbound & outbound sector) & Fed Ex & UPS can be hit or miss.

Errors I have found include:
The flight apparently operating from a by nearby airport. Have a look at flight UA1917 in 18th July (the day I was looking at). The Denver QQH - San Diego sector appears to be a wobbly.
- Diversions that appear not to have happened. Have a look at flight DL273 LIS - JFK - SFO on 8th Aug. The first sector apparently diverted to SFO but not according to the replay. - Obviously a glitch on the system.
- Getting the direction of the flight wrong (quite easy when inbound and outbound legs have the same number). An example DL1865 on 18th July. It shows two FLL - DTW sectors but the first one is actually the opposite direction (and the landing time is incorrect).

Unfortunately airline websites do not give historic data so I can't double check more than about 24 hours after the event. I used to use Flightstats to cross check but they now charge - I don't know if its worth taking out a subscription.

Its an interesting area and I'd love to hear from experts.

If you are looking for historical data on US flights, try here: Bureau of Transportation Statistics