NTSB to probe Fedex/Southwest close encounter at Austin
Pegase Driver
Join Date: May 1997
Location: Europe
Age: 74
Posts: 3,693
Likes: 0
Received 0 Likes
on
0 Posts
Good expalnations but everyone seems to remains in his mindset that because the Fedex saved the day there was no " luck::, "fate" , "coincidence"etc, involved in the fact the 2 aircraft did not collide a few hundred feet above the runway ( or on the runway)
I still differ.
Both aircraft at any time did not know exactly where the other was, so it was 100% up to the controller to ensure separation. .
.The controller , after having made the error to line up the SW with a CAT III 3NM out, then made a series of wrong assumptions and did no seem to know either where the SW was. ( aborting or taking off) But he did not ask and did not issue any separation instructions to any of the two, The Fedex intervened but , as far as the ADS-B tracjetories shows, it remained above the runway centre line with the SW climbing below. The 150 feet or so vertical separation in the end was not due to any coordinated action.
I would happily remove my "luck" asumption if I had for instance heard the following from the Controller :
Fedex , side step right 737 taking off below. SW maintain runway centre line Fedex going around 1/2 mile behind you .
Or if the Fedex , noticing the Controller was not responsive had called the SW by saying :Fedex is going around passing XXX ft climbing.and side stepping right.
Looking forward with great interest to read the NTSB investigation report.
I still differ.
Both aircraft at any time did not know exactly where the other was, so it was 100% up to the controller to ensure separation. .
.The controller , after having made the error to line up the SW with a CAT III 3NM out, then made a series of wrong assumptions and did no seem to know either where the SW was. ( aborting or taking off) But he did not ask and did not issue any separation instructions to any of the two, The Fedex intervened but , as far as the ADS-B tracjetories shows, it remained above the runway centre line with the SW climbing below. The 150 feet or so vertical separation in the end was not due to any coordinated action.
I would happily remove my "luck" asumption if I had for instance heard the following from the Controller :
Fedex , side step right 737 taking off below. SW maintain runway centre line Fedex going around 1/2 mile behind you .
Or if the Fedex , noticing the Controller was not responsive had called the SW by saying :Fedex is going around passing XXX ft climbing.and side stepping right.
Looking forward with great interest to read the NTSB investigation report.
Last edited by ATC Watcher; 16th Feb 2023 at 19:22.
A relevant trip down memory lane to the 1982 crash of Air Florida Flight 90 at KDCA. Included in the accident report findings:
29. The local controller erred in judgement and violated ATC procedures when he cleared Flight 90 to take off ahead of arriving Eastern Flight 1451 with less than the required separation and jeopardizing. [sic]
30. Eastern 1451 [B727] touched down on runway 36 [now runway 1] before Flight 90 lifted off; the separation closed to less than 4,000 feet [possibly as close as 1,500 ft], in violation of the 2-mile preparation [sic] requirement in the Air Traffic Control Handbook.
Included in the probable cause was the captain’s failure to reject the takeoff due to anomalous engine instrument readings. Would an even worse disaster have occurred if he had rejected the takeoff? Weather was 2,800 ft RVR, moderate snow, braking action poor on a snow-covered runway.
Like the Austin incident, the KDCA local controller had no ASDE, no visual contact with the aircraft on the runway (assuming this also to be the case at Austin), and a landing aircraft on an ILS approach at or inside of 3 miles.
29. The local controller erred in judgement and violated ATC procedures when he cleared Flight 90 to take off ahead of arriving Eastern Flight 1451 with less than the required separation and jeopardizing. [sic]
30. Eastern 1451 [B727] touched down on runway 36 [now runway 1] before Flight 90 lifted off; the separation closed to less than 4,000 feet [possibly as close as 1,500 ft], in violation of the 2-mile preparation [sic] requirement in the Air Traffic Control Handbook.
Included in the probable cause was the captain’s failure to reject the takeoff due to anomalous engine instrument readings. Would an even worse disaster have occurred if he had rejected the takeoff? Weather was 2,800 ft RVR, moderate snow, braking action poor on a snow-covered runway.
Like the Austin incident, the KDCA local controller had no ASDE, no visual contact with the aircraft on the runway (assuming this also to be the case at Austin), and a landing aircraft on an ILS approach at or inside of 3 miles.
... but take the "WRONG TIMING Reconstruction" clip posted earlier, and see for yourself at 2:12 what would have happened if SWA acted upon the "SouthWest abort" (at 2:05 doing 50kts): he would have come to a standstill in front of the aiming-point with its tail sticking up 42ft feet (AGL 18L), right in the area where the 6-second interval ADSB, FDX's lowest point reported was +/-60ft (AGL 18L) but in reality could have been even a tad lower, and with MLG and tail in G/A attitude, even another 'few' feet lower....
We will see one day what the NTSB report has to say on this aspect, but my take is that, SWA not acting upon the "SouthWest abort" (un)willingly / (un)knowingly, was a very "lucky" contributory factor. The T/O acceleration decreased moment after moment the closing-in speed of FDX and by the time FDX caught up (laterally) with SWA, the vertical separation, although still shockingly close, was more than enough to avoid any metal scratching!
The controller , after ...., then made a series of wrong assumptions and did no seem to know either where the SW was. ( aborting or taking off) But he did not ask and did not issue any separation instructions to any of the two, The Fedex intervened but , as far as the ADS-B trajectories shows, it remained above the runway center-line with the SW climbing below. The 150 feet or so vertical separation in the end was not due to any coordinated action.
I would happily remove my "luck" assumption if I had for instance heard the following from the Controller : Fedex , side step right 737 taking off below. SW maintain runway centre line Fedex going around 1/2 mile behind you .
I would happily remove my "luck" assumption if I had for instance heard the following from the Controller : Fedex , side step right 737 taking off below. SW maintain runway centre line Fedex going around 1/2 mile behind you .
FDX transgressing the TWR-SWA "controlling-being controlled" relation, which is, although not unprecedented, very unusual, made TWR believe that SWA708 did the "SouthWest abort" call, so at that point not much more TWR could do and the very tense moments were over. About 15 seconds later, he started to clean up the mess with the "SW708 roger, turn right when able".
Another 5 seconds later, confusion started kicking in again, with SWA (unknown to TWR, somewhere around or beyond VR speed) responded with "Negative". What would you be thinking at that moment (in low viz)... "OK, so the guy has aborted T/O but doesn't want to exit the RWY, so what's wrong, does he need RFS to check the overheated breaks, or what else is going on". And by the time this quite normal puzzle started to settle in his mind, SWA must have been popping up on his ASR radar and then TWR was back on the game and issued FDX the 080 heading.
BTW: if somebody replies "why do you think this "made TWR believe that SWA708 did the 'SouthWest abort' call", then my answer is already here: "Did you carefully analyse each detail of the audio recording and compare the "Roger" with other previous "Rogers's" by this ATCO and heard that the intonation of this specific "Roger" speaks for itself.... (but let the NTSB report be the final judge)
I’m interested to see what part the enhanced vision system (EVS) played as it related to FedEx’s actions. On one hand I tend to believe the FedEx crew, through the use of their enhanced vision system had good SA and his “Southwest Abort” call may have been triggered by seeing the Southwest jet on his takeoff roll, via his EVS. On the other hand, if he had good SA from this EVS system, why didn’t he jink right, or offset from Southwest’s flight path which was runway heading? So perhaps FedEx didn’t see the whole picture with this EVS. This will be a very interesting case study, and I look forward to the findings.
I wonder if the EVS gave much of a heads-up of the SWA on the rwy....
Anybody in for an educated guess on the probable moment the TOGA was initiated, based on these ADSB data?
Anybody in for an educated guess on the probable moment the TOGA was initiated, based on these ADSB data?
Pegase Driver
Join Date: May 1997
Location: Europe
Age: 74
Posts: 3,693
Likes: 0
Received 0 Likes
on
0 Posts
@ DIBO. Thanks for that , any chance of having the same ADS-B chart for the SW ? when it actually got airborne and what was its rate of climb would be interesting .
Join Date: Jul 2009
Location: France
Age: 62
Posts: 98
Likes: 0
Received 0 Likes
on
0 Posts
ADSB data of SW (https://globe.adsbexchange.com/?icao...23-02-04&leg=3)
Don't know how accurate the time stamps are (If receiver time not much, if GPS based time should be very accurate).
If the time stamps are correct, SW rotated about when Fedex climbed out over the dep end of rw. Initial climb looking shallow at appr. 500 fpm.
Don't know how accurate the time stamps are (If receiver time not much, if GPS based time should be very accurate).
If the time stamps are correct, SW rotated about when Fedex climbed out over the dep end of rw. Initial climb looking shallow at appr. 500 fpm.
So no question of GPS-based timing - the ADS-B receiver setups I've encountered typically output data with timestamps to about 0.2 s resolution (although they output said timestamps with the seconds value to 3 decimal places, which is daft). With crowd-sourced flight trackers like FR24, I don't know whether they use (or even get) timestamps in the data from their feeders - if not, then add in Internet latency as another source of inaccuracy.
Pegase Driver
Join Date: May 1997
Location: Europe
Age: 74
Posts: 3,693
Likes: 0
Received 0 Likes
on
0 Posts
Thanks Spornrad, Asuming the time stamps are correct and match, , then it would indicate that take off occurred at around 40.50 and the Fedex had already overtaken the SW and was already around 600ft above and well in front. .
The lowest vertical separation should have occurred between 40: 33 and 40:39 between 100 and 200 ft when the SW was either lined up or just started to roll.
But it does not really fits the audio R/T .
The lowest vertical separation should have occurred between 40: 33 and 40:39 between 100 and 200 ft when the SW was either lined up or just started to roll.
But it does not really fits the audio R/T .
Join Date: Jul 2009
Location: France
Age: 62
Posts: 98
Likes: 0
Received 0 Likes
on
0 Posts
https://globe.adsbexchange.com/?icao...amp=1675514458
can replay both planes simultaneously (click on replay speed then on departure "fork" or the taxying SW, beware not to click on the 2nd approach FDX). Timestamps should correspond since receivers are probably the same. Looks like on the RW at least at one point they were exactly on top of each other (However, datapoints for the rolling SWA on the ground are less granular)
can replay both planes simultaneously (click on replay speed then on departure "fork" or the taxying SW, beware not to click on the 2nd approach FDX). Timestamps should correspond since receivers are probably the same. Looks like on the RW at least at one point they were exactly on top of each other (However, datapoints for the rolling SWA on the ground are less granular)
Last edited by spornrad; 17th Feb 2023 at 12:30.
Join Date: Jul 2009
Location: France
Age: 62
Posts: 98
Likes: 0
Received 0 Likes
on
0 Posts
Join Date: Nov 2007
Location: NY
Posts: 36
Likes: 0
Received 0 Likes
on
0 Posts
The rule which should have been applied in this scenario is a simple one. Since the tower controller is unable to provide visual separation they must utilize two miles increasing to three miles within one minute. The rule does not allow for separation to fall below two miles at anytime. This takeoff clearance never had a chance to be technically legal once issued.
Join Date: May 2020
Location: Virginia, USA
Posts: 47
Likes: 0
Received 0 Likes
on
0 Posts
Timestamps are normally added by the receiving software (the actual packets send by the aircraft contain no timing information).
So no question of GPS-based timing - the ADS-B receiver setups I've encountered typically output data with timestamps to about 0.2 s resolution (although they output said timestamps with the seconds value to 3 decimal places, which is daft). With crowd-sourced flight trackers like FR24, I don't know whether they use (or even get) timestamps in the data from their feeders - if not, then add in Internet latency as another source of inaccuracy.
So no question of GPS-based timing - the ADS-B receiver setups I've encountered typically output data with timestamps to about 0.2 s resolution (although they output said timestamps with the seconds value to 3 decimal places, which is daft). With crowd-sourced flight trackers like FR24, I don't know whether they use (or even get) timestamps in the data from their feeders - if not, then add in Internet latency as another source of inaccuracy.
I have sent in a FOIA request to the FAA for the ADS-B data they collected, and that has very accurate timestamps, but I am not expecting a quick turnaround. Note, for both flightradar24 and ADS-B exchange, the timestamp only applies to the long, lat, and baro alt (not the track and velocity). Track and velocity actually come in other messages whose timestamp is not captured. I will also note that one should be careful in trying to extract sub-second accuracy of position based upon the timestamp on the ADS-B message. Here is the U.S. legal spec:
(e) ADS-B Latency Requirements -
(1) The aircraft must transmit its geometric position no later than 2.0 seconds from the time of measurement of the position to the time of transmission.
(2) Within the 2.0 total latency allocation, a maximum of 0.6 seconds can be uncompensated latency. The aircraft must compensate for any latency above 0.6 seconds up to the maximum 2.0 seconds total by extrapolating the geometric position to the time of message transmission.
(3) The aircraft must transmit its position and velocity at least once per second while airborne or while moving on the airport surface.
(4) The aircraft must transmit its position at least once every 5 seconds while stationary on the airport surface.
I suspect that the equipment on the involved aircraft is much better than that, but I don't have any documentation on those specific implementations.
See attached plot, with FR24 datarecords of FDX 40:33 and SWA 40:31. Unfortunately the next SWA position recorded is 11 second later. Therefore I added a calculated approximation of SWA's position at 40:33
As it's all about little seconds difference, I think for the moment nothing reliable can be concluded from the first data&audio syncro attempt. All are hope is now based on the good work by airplanecrazy in getting the FAA recordings. Or wait a year or two for the NTSB report
Here's a plot I produced a few days ago, but didn't post at the time because of timing and other issues in the data (which I haven't yet resolved).
It attempts to show the simultaneous positions of the two aircraft at 10 second intervals (by interpolation of the granular data):
The lowest height recorded for the FedEx, just after the 12:40:30 timestamp, resolves to approximately 40' ±12½', hence my earlier comment about not ruling out ground contact during the GA.
It attempts to show the simultaneous positions of the two aircraft at 10 second intervals (by interpolation of the granular data):
The lowest height recorded for the FedEx, just after the 12:40:30 timestamp, resolves to approximately 40' ±12½', hence my earlier comment about not ruling out ground contact during the GA.
Last edited by DaveReidUK; 17th Feb 2023 at 22:20. Reason: ±12½', not ±25'
Join Date: May 2020
Location: Virginia, USA
Posts: 47
Likes: 0
Received 0 Likes
on
0 Posts
Actual landing data
FedEx.. what am I missing?
GA lurker here, hoping to learn something.
1. I was trained aviate, navigate, communicate and seems to me FedEx nailed it.
2. And although it's unusual to talk cross cockpit, it has saved lives before.. if I remember correctly Pan am tried to stop KML in Tenerife but I may not remember that correctly.
3. If Southwest had understood and rejected, if I understand correctly, it's unlikely that the aircraft would have been in jeopardy of hitting each other... Of course I'm not underplaying RTO consequences.
4. And if I understand correctly, TCAS would not have provided directional instructions.
If everything I have said is true, in my mind, it sounds like FedEx did everything possible... They couldn't go off center line because they didn't know if Southwest was going off center line.
So what am I missing? How could FedEx have any responsibility whatsoever? Under the circumstances what could they have done differently?
Thanks and remember once again, I don't know everything you guys know :-)
1. I was trained aviate, navigate, communicate and seems to me FedEx nailed it.
2. And although it's unusual to talk cross cockpit, it has saved lives before.. if I remember correctly Pan am tried to stop KML in Tenerife but I may not remember that correctly.
3. If Southwest had understood and rejected, if I understand correctly, it's unlikely that the aircraft would have been in jeopardy of hitting each other... Of course I'm not underplaying RTO consequences.
4. And if I understand correctly, TCAS would not have provided directional instructions.
If everything I have said is true, in my mind, it sounds like FedEx did everything possible... They couldn't go off center line because they didn't know if Southwest was going off center line.
So what am I missing? How could FedEx have any responsibility whatsoever? Under the circumstances what could they have done differently?
Thanks and remember once again, I don't know everything you guys know :-)
FWIW, the lowest altitude in my interpolation for the reconstruction video is 540' AMSL. Subtracting out field height of 486' at that point gives 54' AGL. Our answers agree within the margin of error, but I don't think they are consistent with ground contact. For ground contact, the PA would have to have been 25' (like it is when the aircraft actually lands 12 minutes later), and I don't see it could go that low on the GA.
I’m interested to see what part the enhanced vision system (EVS) played as it related to FedEx’s actions. On one hand I tend to believe the FedEx crew, through the use of their enhanced vision system had good SA and his “Southwest Abort” call may have been triggered by seeing the Southwest jet on his takeoff roll, via his EVS. On the other hand, if he had good SA from this EVS system, why didn’t he jink right, or offset from Southwest’s flight path which was runway heading? So perhaps FedEx didn’t see the whole picture with this EVS. This will be a very interesting case study, and I look forward to the findings.
But according to this document from one of the makers of EV (Teledyne FLIR), in CAT III conditions, "there is virtually no difference between how far you can see with a thermal imaging camera and with the naked eye, since the atmosphere is the limiting factor. Radiation does not penetrate through this dense type of fog in all (visible, MWIR and LWIR) spectral bands."
https://www.flir.com/discover/rd-sci...-fog-and-rain/
Comments on whether the specific EV system used by FedEx does any better - and why?
Join Date: Jul 2009
Location: France
Age: 62
Posts: 98
Likes: 0
Received 0 Likes
on
0 Posts
The above ADSBXchange allows for conveniently exporting the data for both planes. Horiz. and vert. separation plotted vs. time. Since both planes are received by the same receivers we can assume that the time stamps are reasonably in sync. The rwy is 17/35 thus latitude can be directly used for horizontal separation.
The lines are the RWY thresholds. The leading craft is to the left in this plot (arriving at same lat earlier). FDX overtakes SWA in the first third, then SWA catches up and overtakes FDX in the last third of the RWY probably around rotation.
The alt profile of SWA indicates at least some situational awareness.
The lines are the RWY thresholds. The leading craft is to the left in this plot (arriving at same lat earlier). FDX overtakes SWA in the first third, then SWA catches up and overtakes FDX in the last third of the RWY probably around rotation.
The alt profile of SWA indicates at least some situational awareness.
Last edited by spornrad; 19th Feb 2023 at 07:23.