Separation issue over Bight, 20/9/13

[Rowan McKeever]

Just saw this on ABC News... http://www.abc.net.au/news/2013-09-20/aircraft-in-near-miss-over-great-australian-bight/4971478

[Kazuya H]

If the report is correct, it looks like it was possibly QF642 (VH-EBI) and QF581 (VH-EBO) on the Sydney/Perth route. Will see if the LiveATC archive has anything on Melbourne Center and post here if I find anything.

[Alex T]

QF576 (VH-EBS) Per-Syd - http://blogs.crikey.com.au/planetalking/2013/09/20/two-qantas-jets-involved-in-serious-mid-air-near-miss/

[A McLaughlin]

Have just done a playback on Planefinder (12.23pm AEST - look over the Gulf of St Vincent) and the jets appear to be 1000ft vertically separated, and several miles apart laterally...not sure what the fuss is about.

[Greg Hyde]

Just on Ch10 Sydney News

A/c closed to 700 ft.

ATC stood down pending investigation.

[Robert S]

FlightAware/Google Earth plot shows 900ft vertical at 3.24km distance. Clearly there are significant limitations in this data... I'm presenting it "as is" only.

I spent some of the evening yelling at journalists at Twitter who were sensationalising about near misses, which was about as useful as expected (ie. not very). Most of them were freaking out that 1000ft was too close. :rolleyes:

[Robert S]

ATSB page is here: http://www.atsb.gov.au/publications/investigation_reports/2013/aair/ao-2013-161.aspx

12:13 AEST, approx 10NM West of Adelaide

[Robert S]

Ben Sandilands updated his article at the Crikey link above. He is now claiming this:

"The jets came within 700 feet of each other in the incident at a very high combined velocity, which occurred as they were flying mirror image flight paths in opposite directions between Perth and Sydney. ... Information says that as the aircraft followed the RA directions the amount of space between them diminished to about 700 feet. This may be the nearest two airliners have come to each other while at cruise altitudes and speeds in Australian airspace."

The funny thing is that he's even provided a PlaneFinder image in the same article, yet failed to correct himself. Is it really so difficult for journalists to understand three dimensions... especially someone who regularly writes about the aviation industry? Sigh.

Clearly we don't know the actual distance for sure yet, but the case for "within 700 feet" is hardly convincing to say the least.

Part of me is wondering if the 700ft rumour is because FlightAware shows QF581 hit 38300 feet, without realising that it also shows QF576 at 39200 feet at the same time and some distance to the north.

[Robert S]

Ten late news is now claiming they came within 200m and 5 seconds!

Errrr.... reaaaally doesn't look like that.

In fact this whole news report looks like it's been made with the Lazy Journalist's Story Generator.

[Jarrad Phillips]

If you playback FlightRadar 24, starting at 0210 QF581 gets to within 750ft of QF576 just west of Adelaide.

[Robert S]

That's not what I see. At the time of this screenshot (UTC 02:17), QF581 is shown at 38250 ft (and descending) and QF576 is shown at 39200 ft (and climbing). They are laterally separated... I went to measure it in Google Earth and discovered that the two co-ordinates shown are identical to those in the planefinder screenshot I posted earlier (which measured up as 3.24 km apart).

This said I can't explain why the position of the aircraft on FlightRadar24 doesn't match up with its own co-ordinates... you can see in the two screenshots that they don't make sense. It has QF576 at longitude 138.3094 E and QF581 at longitude 138.2958 E, yet QF581 is shown further east than QF576. Something wrong there.

[Jarrad Phillips]

Well given the ATSB will have access to the actual radar data one would hope it will offer a little bit better fidelity then the internet data we were looking at.

[Ash W]

What kind of seperation loss would make TCAS kick in? Would have thought well before 700ft.

[Geoff Br]

The following link may assist with the rules of separation..

http://www.skybrary.aero/index.php/Airborne_Collision_Avoidance_System_(ACAS)

From all accounts it looks like one a/c was cleared to climb in front of another on an apposing route. Pretty normal stuff, however, it seems like either the altitude change clearance was either given with not enough thought about apposing speeds, or the climbing a/c did not commence or climb as quick as the controller thought they would.

I notice in the ATSB investigation report http://www.atsb.gov.au/publications/investigation_reports/2013/aair/ao-2013-161.aspx

(12 months for final report) that they say the controller cancelled the flight level change clearance and the a/c had descended back to FL380. This being the case, why the RA?? Maybe both a/c received the RA and the controller then cancelled the flight level change when both pilots reported clear of conflict. Anyway, many possible causes for the LOS, and great that the TCAS technology worked as designed and, apparently within 12 months we shall all find out what happened. Though....Can we wait 12 months to determine if there 'was' a problem and what if any resolutions were recommended?

I have often noted in Flight Radar24 that some times an a/c "pauses" for a period of time and re-adjusts their position, both in the flight direction and sometimes laterally. On many occasions I have observed a/c which appear in LOS conflict, only to merge and carry on. Presumably, the FlightRadar positioning and the actual positioning are not accurate. I live on final for 01 YBBN. I can see on FlightRadar24 an a/c which should be directly overhead my house, is some where else. Usually close though to indicated positioning.
I can hear the aircraft crew saying on my a/c scanner they are at X altitude, when I immediately look on Flight Radar24, it indicates another altitude. I have found FlightRadar24 may not accurately indicate the a/c exact altitudes and positioning, especially when in transition between levels. But, I consider the positioning, altitude, speed etc it does indicate is fantastic and we should not expect it to be as accurate as the ATC facilities, which cost $zillions to produce and operate. I would imagine the Flight Radar24 people would not say their system is accurate enough for air traffic control.

[Robert S]

Pretty normal stuff, however, it seems like either the altitude change clearance was either given with not enough thought about apposing speeds, or the climbing a/c did not commence or climb as quick as the controller thought they would.

I notice in the ATSB investigation report http://www.atsb.gov.au/publications/investigation_reports/2013/aair/ao-2013-161.aspx

(12 months for final report) that they say the controller cancelled the flight level change clearance and the a/c had descended back to FL380. This being the case, why the RA?? Maybe both a/c received the RA and the controller then cancelled the flight level change when both pilots reported clear of conflict. Anyway, many possible causes for the LOS, and great that the TCAS technology worked as designed and, apparently within 12 months we shall all find out what happened. Though....Can we wait 12 months to determine if there 'was' a problem and what if any resolutions were recommended?

I have often noted in Flight Radar24 that some times an a/c "pauses" for a period of time and re-adjusts their position, both in the flight direction and sometimes laterally. On many occasions I have observed a/c which appear in LOS conflict, only to merge and carry on. Presumably, the FlightRadar positioning and the actual positioning are not accurate.

It's really too early to be commenting in this level of detail, but the impression I'm getting is that the Perth bound aircraft started its climb after it passed the Adelaide VOR, which was the point of intersection of the two paths. The problem is that it was still passing or was to be passing the Sydney bound aircraft within a range of less than 5nm, so a loss of separation occured. They would have needed to have passed each other and had the gap open up to 5nm before it could have climbed without a LOS. I cannot imagine that it was intended to climb ahead of the Sydney bound aircraft.

The wording of the ATSB initial brief summary appears to be that the climbing aircraft had the clearance cancelled and returned to the lower altitude as directed, while the Sydney bound aircraft is said to have received the TCAS RA and responded to it.

I get the impression that these events were happening at much the same time. Also the climbing aircraft would not have been able to instantaneously switch from a climb to a descent... it would take a few moments at least.

The ATSB will give us the exact chain of events and timings in due course so there's probably not really much point speculating about the detail now.

With respect to the live flight trackers sometimes being off, keep in mind that they are slightly better with old data than live data. The old data basically comes up as a series of data points. FlightAware is better for this than FlightRadar24. The data points are probably not too far off a lot of the time. Everything in between is just joining the dots. Critically in this case the timing information may not be spot on and is only shown to the nearest minute.

When you're trying to watch live, the sites show the aircraft continously moving but they don't have continuous data. The aircraft are animated based on the last known information, then when a new point of data comes in, the aircraft is shifted to the new known position, then the animation continues from there based on the previous information until the next data point comes in and so on. This becomes worse when aircraft are turning - as they often are in approach and departure patterns - because the tracking sites have no concept of the aircraft banking and don't include that in the estimated path. Therefore the aircraft will often appear to have significant lateral movements, but it's just the site catching up on the delta between the estimate and the reality.

[Geoff Br]

.....
With respect to the live flight trackers sometimes being off, keep in mind that they are slightly better with old data than live data. The old data basically comes up as a series of data points. FlightAware is better for this than FlightRadar24. The data points are probably not too far off a lot of the time. Everything in between is just joining the dots. Critically in this case the timing information may not be spot on and is only shown to the nearest minute.

When you're trying to watch live, the sites show the aircraft continously moving but they don't have continuous data. The aircraft are animated based on the last known information, then when a new point of data comes in, the aircraft is shifted to the new known position, then the animation continues from there based on the previous information until the next data point comes in and so on. This becomes worse when aircraft are turning - as they often are in approach and departure patterns - because the tracking sites have no concept of the aircraft banking and don't include that in the estimated path. Therefore the aircraft will often appear to have significant lateral movements, but it's just the site catching up on the delta between the estimate and the reality.

Agree Robert. I often notice the trackers have a very blocky path on turns, but when you return to it a few minutes later, the data points and mapping catch up and show a more accurate representation of the flight path.
FlightRadar24 and LiveATC are a fantastic combination.

[Robert S]

Aviation Herald has another churn over what's probably the same data source as the others:

http://avherald.com/h?article=468b8432&opt=0

I'm not sure that their conclusion that "Radar data suggest minimum separation reduced to 700 feet vertical and 5.6nm lateral at 02:13:55Z" is right... their Google Earth shot seems to show much the same data which shows they did pass closer than that. In fact if their conclusion that the minimal lateral distance was 5.6nm is correct, then that would mean no LOS occured in the first place.

Meanwhile Ben Sandilands has put his axe back on the grinder.

http://blogs.crikey.com.au/planetalking/2013/09/21/adelaide-fall-out-airservices-is-dangerously-incompetent/

My view - serious questions to answer are not assisted by abject hysteria that clearly doesn't align with currently available information.

EDIT: Looks like AIPA is weighing in on this too: http://www.theage.com.au/travel/qantas-seconds-away-from-midair-collision-20130921-2u68u.html

"The two aircraft ultimately missed each other by just 700 feet (213 metres), said Mr Woodward, with QF581 flying directly underneath QF576."

One wonders what he is basing that on.

[Mick F]

Perhaps eyewitness accounts from the pilots Robert.

Mick

[Robert S]

Hmmm, think I'll just wait for the accurate data on that one.

It looks like QF581 also crossed JNB-SYD QF64's track a few minutes minutes later while over the Yorke Peninsula. QF64 was maintaining FL370 according to FlightAware and probably didn't pass closer than 3.5nm. I wonder if that's what some of the passengers saw, especially the guy on QF581 who reported seeing another aircraft underneath theirs, even though they were actually the lower of the two aircraft in the LOS encounter with QF576.

[Chris B.]

What kind of seperation loss would make TCAS kick in? Would have thought well before 700ft.

Ash,

TCAS Resolution Advisories (RA) and Traffic Advisories (TA) depend on a closure rate between two aircraft. Thus RAs are generally issued within about 20-30secs of collision and a TA between 25-45secs. That is to say an aircraft approaching another from underneath for example can trigger an RA some 2000ft before impact due to the aircraft's rate of climb.

[David Knudsen]

The ATSB have released an update on their findings here (http://www.atsb.gov.au/publications/investigation_reports/2013/aair/ao-2013-161.aspx)


Summary:

Loss of separation between Airbus A330 VH-EBO and Airbus A330 VH-EBS near Adelaide SA on 20 September 2013

On 20 September 2013, a loss of separation occurred between an Airbus A330 aircraft, registered VH-EBO (EBO) operating a scheduled passenger service from Sydney, New South Wales to Perth, Western Australia, and an Airbus A330 aircraft, registered VH-EBS (EBS), operating a scheduled passenger service from Perth to Sydney. Both aircraft were within radar surveillance coverage at the time of the occurrence.

Airservices Australia advised that EBS was cruising at flight level (FL) 390. The flight crew of EBO were cleared to climb from FL 380 to FL 400 and the aircraft commenced the climb. Soon after, the controller cancelled the clearance and the aircraft descended back to FL 380. The flight crew of EBS received a resolution advisory alert from their aircraft’s traffic collision avoidance system. However, EBO’s flight crew advised that they did not receive any indications on their traffic collision avoidance system of the presence of EBS. A full system test conducted on EBO’s traffic collision avoidance system after the incident confirmed it was not functioning.

Recorded data from the two aircraft showed that the minimum vertical separation was 650 ft when the two aircraft were 4.1 NM (8 km) apart laterally. The minimum lateral separation was 1.6 NM (3 km) when the aircraft were 870 ft apart vertically. At that time both the vertical and lateral separation were increasing as the aircraft were on separate one-way routes. The vertical and radar separation standards were re-established a short time later. (A loss of separation occurs at that flight level when two aircraft are within 1,000 ft [305 m] vertically and 5 NM [9.26 km] horizontally of each other).

The ATSB investigation is continuing and will include:
• further analysis of the ATC radar and audio data and the recorded data from the two aircraft
• analysis of the context in which the controller’s actions occurred
• examination of the TCAS computer and related components from VH-EBO
• review of the reliability and availability rates of TCAS.


The fact that EBO's TCAS was found to be "not functioning" is surprising, I'd have thought it would have some pretty decent self-test functions.

[Kazuya H]

Here is the link to the article on SMH.

http://www.smh.com.au/travel/travel-incidents/qantas-nearmiss-collision-avoidance-system-not-working-20131010-2va5i.html

Second last paragraph:
"Qantas said in a statement that while the traffic collision avoidance system on the Perth-bound plane was working intermittently, this "wasn't a contributing factor to the incident".

Looks like it was working, but only intermittently.

Here is the link to Qantas response:
http://www.qantasnewsroom.com.au/qantas-responds/statement-on-loss-of-separation-incident-a330s

[Robert S]

It was pleasing to see that Fairfax's article today grew up and spelt it out - "The initial report makes clear that the planes, which can carry up to 300 passengers, would not have hit each other even if they had continued on their paths before the alert was made."

Disappointing the industry people pushing agendas were fuelling and validating the nonsense flowing in the media on the day it happened. I still wonder what they think they were achieving by doing so. You're not going to get proper safety outcomes by playing the headless chicken game.

[Rich W]

Bit embarrassing for Alan Joyce. I saw an interview he did on Channel 10 while on the new Jetstar Dreamliner. When asked about the separation issue he said something like, "Well that's why people trust and travel on Qantas because we have experienced pilots and the excellent equipment and safety standards on our aircraft that prevented any collision".

Looks a bit embarrassing if one of your avoidance systems wasn't working correctly in a potentially critical moment.