FOQA erroneous detections
Thread Starter
Join Date: Jul 2006
Location: USA
Age: 53
Posts: 14
Likes: 0
Received 0 Likes
on
0 Posts
FOQA erroneous detections
I just received an email from my company telling me l exceeded the stablized approach criteria. Foqa registered a descent rate greater than 1500 feet below 500 feet. It never happened during the flight, when l read the email l thought it was a joke, my first officer called me right after he read the email, we both were confused. Can any engineer or expert tell me how this can happen? Thanks
Join Date: Jul 2009
Location: Not far from a big Lake
Age: 82
Posts: 1,454
Likes: 0
Received 0 Likes
on
0 Posts
Can any engineer or expert tell me how this can happen? Thanks
It may be someone else's problem.
Join Date: Jun 2014
Location: Village of Santo Poco
Posts: 876
Likes: 0
Received 0 Likes
on
0 Posts
I just received an email from my company telling me l exceeded the stablized approach criteria. Foqa registered a descent rate greater than 1500 feet below 500 feet. It never happened during the flight, when l read the email l thought it was a joke, my first officer called me right after he read the email, we both were confused. Can any engineer or expert tell me how this can happen? Thanks
It's not entirely impossible to momentarily hit a high descent rate down low (or to have an erroneous VS indication), which is why I'm surprised that your flight got tagged based on that one parameter. Usually FOQA doesn't flag a flight unless a number of parameters were exceeded.
So, I guess my question would be what more there is to the story here...
VSI, rate of change of pressure altitude, or radio altimeter ?
Wind speed, gusts, both pressure altimeters on the same pressure setting ?
Compare with previous flights in the same location.
Data source, recording rate, accuracy, analysis software ?
Aircraft configuration, change in flap / gear, yaw / roll (airflow - pressure crossflow between sensors) ?
Wind speed, gusts, both pressure altimeters on the same pressure setting ?
Compare with previous flights in the same location.
Data source, recording rate, accuracy, analysis software ?
Aircraft configuration, change in flap / gear, yaw / roll (airflow - pressure crossflow between sensors) ?
As far as I am aware, those flight data monitoring systems are highly customizable by individual airlines and their trigger values are subject to change as needed. If a certain setup shows a lot of false warnings, it may well be adjusted not to raise unwanted false alarms any more. This has happened more than once already. It also may find it hard to differentiate between a bog standard Cat I ILS or a more taxing approach, say, with a glideslope angle of 5° that obviously requires sink rates greater than standard, even more so when correcting a slight deviation like 3 whites/1 red. It is is one of the analysts jobs to filter such unwanted alerts out and not tarnish the system by pestering the crews with heaps of unwarranted envelopes.
That said, the other side of the coin is that such a system may take note of exceedances that happened only for a fraction of a second and were immediately corrected. It has also been said that several crews that have been invited for a reenactment of the flight phase in which the system fired a warning were rather surprised at the situation in retrospective; it did not seem all that interesting to them while it occurred.
It is hard to deliver a diagnosis here. It may well be that the collected data is erroneous; that theory might be verified by checking several of the individual aircrafts flights around the one in question for occurrence of the same finding. It is possible as well that we are talking about an approach here that has a certain tendency or even an approval for higher sink rates while still being formally stabilized. Also, there have been findings from that system that looked weird in isolation, but a synoptic view with more data from that flight showed it to be perfectly normal.
That said, the other side of the coin is that such a system may take note of exceedances that happened only for a fraction of a second and were immediately corrected. It has also been said that several crews that have been invited for a reenactment of the flight phase in which the system fired a warning were rather surprised at the situation in retrospective; it did not seem all that interesting to them while it occurred.
It is hard to deliver a diagnosis here. It may well be that the collected data is erroneous; that theory might be verified by checking several of the individual aircrafts flights around the one in question for occurrence of the same finding. It is possible as well that we are talking about an approach here that has a certain tendency or even an approval for higher sink rates while still being formally stabilized. Also, there have been findings from that system that looked weird in isolation, but a synoptic view with more data from that flight showed it to be perfectly normal.
I am definitely not an expert on FOQA, but viewing this incident as a piece of an information system output, it is entirely possible that the wrong record was identified to your flight.
Join Date: Jul 2008
Location: England
Posts: 82
Likes: 0
Received 0 Likes
on
0 Posts
I had an e-mail last year saying I had busted the landing gate and landed, they noted I had filed an ASR, my reply was that on that date I was at home enjoying a Beer so it was more than likely someone else. They had got it wrong and apologised profusely.
Join Date: Oct 2006
Location: Over the Pacific mostly
Posts: 1,043
Likes: 0
Received 0 Likes
on
0 Posts
Those systems do record erroneous data from time to time...., a couple years ago I had a notification that the flap 30 speed had been exceeded......, we were at .81 and FL390 with the flaps up (of course) when it registered the overspeed...., go figure!
Join Date: Aug 2005
Location: fairly close to the colonial capitol
Age: 55
Posts: 1,693
Likes: 0
Received 0 Likes
on
0 Posts
I would just add to the above that you should request to see the entire record. In our experience, instantaneous readings can (and do) go momentarily out of bounds - particularly when ADIRU is the source or when the bounds are triggered by an input that monitors analogue state changes.
Remember, computers make very fast, very accurate mistakes.
Remember, computers make very fast, very accurate mistakes.
Normally, "instantaneous" readings, (data spikes or very short-term valid excursions) are filtered by the design of the FOQA Event usually by requiring a certain amount of time, say 3 seconds for example for a high-rate-of-descent Event, to ensure that that's what was going on, and it wasn't just short, transient, ("going through the red light") value. Excursions occur all the time, and, particularly if the approach is rapidly re-stabilized, our program leaves such excursions out of the data. We call crews, we do not send emails and the discussion is a welcome one as the crew is often as curious about what FOQA shows as we are about learning from the occurrence.
Also, I am a very strong believer in pilot-run FOQA programs, or at least having active line pilots watching the data on a regular basis and doing the crew contacts under the direction of the FOQA Program Manager. I know this is may seem unrealistic to many but in my experience, (running FDM since the late 80's and flying the line), avoids many of the pitfalls like the one being cited here.
FOQA Programs need to be tailored to the airline's operation and designed and implemented to reflect the air carrier's safety culture. To reinforce something stated above, there are as many ways to do flight data monitoring as there are programs and no two are alike.
Also, I am a very strong believer in pilot-run FOQA programs, or at least having active line pilots watching the data on a regular basis and doing the crew contacts under the direction of the FOQA Program Manager. I know this is may seem unrealistic to many but in my experience, (running FDM since the late 80's and flying the line), avoids many of the pitfalls like the one being cited here.
FOQA Programs need to be tailored to the airline's operation and designed and implemented to reflect the air carrier's safety culture. To reinforce something stated above, there are as many ways to do flight data monitoring as there are programs and no two are alike.
Join Date: Apr 2013
Location: At home
Age: 64
Posts: 140
Likes: 0
Received 0 Likes
on
0 Posts
Have you ever seen a radar altimeter just go nuts for a few seconds. I have seen it and it caused the same thing and I got a call from our FOQA gate keepers, (pilots) and he said forget it. He just had to call and later he followed up that the next day the captains radar altimeter failed and had to be replaced.
It happens.
It happens.
Join Date: Jun 2006
Location: Australia
Posts: 1,186
Likes: 0
Received 0 Likes
on
0 Posts
If ever you get called up for tea and bikkies over an alleged CVR transcription event and the transcription is shown to you on paper, make sure you demand to hear the tape from which the transcription was made. in other words don't take what was written down as what was really said.
A pilot was accused of being impatient on the radio and a violation filed using the written transcription by a clerk in the DCA (Australia) accident investigation office as evidence. Three R/T transmissions by the pilot was offered as evidence. One was "Request Airways clearance" but written down as "are we clear"
Another was "Ready 30 seconds" which was an authorised terminology in AIP meaning we are ready for take off but need 30 seconds in the lined up position.
The investigating official used that transmission as evidence the pilot was in a hurry and therefore impatient.
A pilot was accused of being impatient on the radio and a violation filed using the written transcription by a clerk in the DCA (Australia) accident investigation office as evidence. Three R/T transmissions by the pilot was offered as evidence. One was "Request Airways clearance" but written down as "are we clear"
Another was "Ready 30 seconds" which was an authorised terminology in AIP meaning we are ready for take off but need 30 seconds in the lined up position.
The investigating official used that transmission as evidence the pilot was in a hurry and therefore impatient.
Thread Starter
Join Date: Jul 2006
Location: USA
Age: 53
Posts: 14
Likes: 0
Received 0 Likes
on
0 Posts
Thanks to all of you
I really appreciate the replies I received, l forgot to mention that my employer is an Asian carrier, having said that, pilots who work or have worked in this part of the world know what l mean, common sense is a very rare commodity here. I had a meeting with my chief and to put it in a nutshell, my conversation could have been more productive if l had it with a wall. Next week l m going to review the data with the foqa team, l requested an hearing. Thanks again to all of you
Last edited by saraceno; 5th Mar 2017 at 13:53.
Join Date: Jun 2014
Location: Mordor
Posts: 336
Likes: 0
Received 0 Likes
on
0 Posts
All I can add is that when I did FOQA, every flight that had been flagged by the computer had to be reviewed by me (the FOQA pilot) and if the parameters didn't make sense, it was immediately binned. Almost 30-50% of flights that have been flagged by the FOQA system were "false positives" and separating the chaff was a big part of my job...
Then again, it was in an European company. I too know first hand how sometimes things work in Asia...
If you get the chance of seeing your flight's record, check for consistency with other parameters, like pitch, thrust, speed, radio altitude change, G/S deviation etc. if you really had 1500ft/min sink rate on approach, the other parameters would show too. Good luck!
Then again, it was in an European company. I too know first hand how sometimes things work in Asia...
If you get the chance of seeing your flight's record, check for consistency with other parameters, like pitch, thrust, speed, radio altitude change, G/S deviation etc. if you really had 1500ft/min sink rate on approach, the other parameters would show too. Good luck!
Join Date: Mar 2017
Location: Bangalore
Posts: 1
Likes: 0
Received 0 Likes
on
0 Posts
From AC120-82, this seems to be a High Descent Rate exceedance. The basic algorithm should be "Descent rate > x fpm for x seconds", so if properly implemented, you can rule out transient instrument errors. Would suggest checking what the monitoring window is for this event and if possible the FPM and altitude plotting for this particular descent. Since your FDM software has the data, I would guess it should be readily available for your review.
This is my first post in this forum, I am an engineer getting started with FOQA implementation for our product.
This is my first post in this forum, I am an engineer getting started with FOQA implementation for our product.
, and, I will add, also with air carriers in Canada.
Join Date: Jun 2001
Location: Rockytop, Tennessee, USA
Posts: 5,898
Likes: 0
Received 1 Like
on
1 Post
I'll observe that over the years these safety monitoring programs always seem to start out as 'totally non-punitive' and sometimes morph into something else. I believe the NTSB has already come out in favor of random CVR auditing and cockpit cameras. I'm sure when these innovations come we'll be similarly assured that any information garnered can never be used for punitive purposes.
Have you ever seen a radar altimeter just go nuts for a few seconds. I have seen it and it caused the same thing and I got a call from our FOQA gate keepers, (pilots) and he said forget it. He just had to call and later he followed up that the next day the captains radar altimeter failed and had to be replaced.