Retrieving FDR data from a plane that hasn't crashed
 

Good afternoon all,

Apologies first of all if this is in the wrong forum, or if the question has been asked before. I'm trying to find out how easy (or difficult) it is to retrieve data from a flight data recorder with the unit still in situ. That is, would it be feasible to download the data from an aircraft during its turnaround time? I suspect it may be marginally easier on newer aircraft, but probably isn't easy in general?

Why am I asking? I'm a pilot and aviation enthusiast, but I'm also a data scientist and I'm looking to use large volumes of data to analyse flight operations in near real time to highlight issues while the plane is still on the ground that may result in an incident if it's allowed to take off. Things that the crew may have dismissed as not significant enough to warrant grounding the plane at the time.

I am just trying to find out how feasible accessing the data would be because if it's difficult and time consuming then obviously the airlines won't go for it, and so that's a non-starter, and my model will need to focus purely on getting the FDR data from crashes only into my algorithms.

Search online for 'quick access recorders' (qar).

In most cases it is very easy . However reading and viewing the data can take longer , and needs a lot of skill , knowledge and experience . I recently downloaded a QAR and had the results in less than an hour !

Cheers guys. I'll look into the QAR.

Google ACARS while your at it.

There are several approved hand held units that you simply plug into the DFDR and it automatically downloads the data retained within the DFDR. From memory, about 20 to 30 minutes to download.

Just reread your question. the interpretation of data will take time.
Not practical on a turn around.

Pretty sure this sort of thing already existed and is in use. Called Flight Data Monitoring, or FDM if I'm not mistaken. Wireless QAR sends the data once the aircraft has landed, engines shutdown and usually a door open. This is then automatically analysed and any important problems are highlighted. Things like a flap or gear over speed or a heavy landing that the crew were unaware of and didn't report. These are then reported to the maintenance provider and the appropriate inspection must be then carried out before further flight. One system I know of is called Aerobytes. Sure there will be more thou.

See the below thread for some more related info

http://www.pprune.org/engineers-technicians/592001-moqa-fdm-maintenance-predictive-analytics.html#post9708597

It's totally practical on a turnaround because it's just a card swap.

PN

As has already been said Martyn, pulling the data is quite easy.

ACARS and related systems will already do much of what you're talking about in terms of spotting immediate "threats" to aircraft systems integrity.

The possibility of spotting trends and heading off maintenance downtime is a bit of a "complex systems" problem, and various labs around the world are working on that at the moment. I was on the fringes of one such project but backed out - not because it was a bad idea, but because I just didn't have time and good people elsewhere were working the problem already. It is very much a case of monitoring trends and interrelations in massive data sets, and as such a continuous task, rather than discrete during a turnaround.

Whilst you're thinking about this, also chat with both the airline flight data people, and your union, about their views on the confidentiality of data. There are a few pitfalls there.

G

Have a look around the now defunct EASA Operators FDM forum
https://essi.easa.europa.eu/ecast/in...ge_id=888.html
There could be some useful info in the conference proceedings, or at least some other contacts.

http://www.airinc.ca/airgroup/ai

http://www.airinc.ca/uploads/assets/...chure-2010.pdf

This is the best value unit out there. Can interrogate 90% of the DFDR's out there.

On the 787 you can dump the FDR to laptop via an ethernet port in the flightdeck. Quick and simple.

On a more routine basis we dont have a WQAR so download the CPL files (continuous parameter logging), to laptop each day. Once back in the office, you plug your laptop into the ground network, and they magically find their way to the right person at the other end.

Most airlines however will only investigate a hard landing if the crew report it on arrival. It is at that point that any recorded data is accessed via built in access terminals (if available) and the engineers will make a decision about what level, if any, of inspection is required.

Best of luck in your quest, but if you focused on MONEY then airlines might listen. I would also suggest you reconsider your "...not significant enough to warrant grounding the plane.." comment for two reasons. Firstly, engineers work according to the manufacturer's manuals and have to either fix and sign off or defer any defect. Judgment and discretion are rapidly being taken from engineers. Secondly, a single failure of a monitored system will rarely bring down an aircraft. But if someone could use "big data" to predict system failures, then they will have a place in the industry. Furthermore, of things don't break then it will be harder for multiple failures to occur. And yes, we are already working on data projects like this.

PM