34R SYD Review ATSB
 

https://cimg1.ibsrv.net/gimg/pprune....cdf90079af.png


The design of standard instrument approaches and departures, air traffic control and flight crew actions and procedures, and the coding of aircraft flight management system navigation databases are among a number of the factors the ATSB is focusing on as part of the on-going investigation into a loss of separation event near Sydney Airport.

That investigation’s preliminary report, released on 16 January, details that separation between two Qantas aircraft, an Airbus A330-300 and a Boeing 737-800, was reduced to about 0.43 nautical miles (796 metres) laterally and about 500 feet (152 metres) vertically during the incident, which occurred at around 6:30pm on 5 August 2019.

The A330 had been cleared by air traffic control to take-off from Sydney Airport’s runway 34 Right, at the same time that the 737 was on final approach to land on the same runway.

While the A330 was commencing its takeoff run, the air traffic controller with responsibility for managing runway 34 Right, an otherwise experienced controller who was a trainee under supervision for the Aerodrome Controller – East (ADC-E) position, assessed that if the 737 continued to land, there would be insufficient runway spacing between the two aircraft, and so instructed the 737 to conduct a go around.

The preliminary report details that the loss of separation occurred as both aircraft turned to the right.

The preliminary report details that the loss of separation occurred as both aircraft turned to the right, with the A330 turning to the right following a standard instrument departure (SID) from runway 34R (the MARUB 6 SID) and the 737 turning to the right following the missed approach procedure for a GLS (a global navigation satellite system landing system) approach for a landing on runway 34R.

The ADC-E controller, who reported that he had both aircraft in sight, attempted to increase their separation by instructing the 737 to turn further right. As both aircraft converged, the A330 flight crew received a traffic advisory (TA) alert from their aircraft’s airborne collision avoidance system (ACAS).

The A330 first officer, who was pilot flying, then saw the 737 in close proximity and, in response, reduced the aircraft’s angle of bank to reduce the turn towards the 737.

The captain of the A330 radioed to advise the ADC-E controller that their proximity to the 737 was “very close”. The controller then issued an instruction to the A330 flight crew to turn left.

The A330 climbed to 5,000 feet and continued to Melbourne without further incident. The 737 climbed to 3,000 feet and was issued radar vectors for a second approach to runway 34R. It landed without further incident a short time later.

“Preliminary reports outline basic factual information established in the early phase of an investigation. They do not contain findings, identify contributing factors or outline safety issues and actions, which will be detailed in an investigation’s final report,” explained ATSB Director Transport Safety Dr Stuart Godley.

“The ATSB’s on-going investigation into this occurrence will focus on a range of factors including the design and risk assessment of MARUB standard instrument departures and missed approaches from Sydney Airport’s runway 34 Right; air traffic control procedures, controller training and controller actions; flight crew actions and the operator’s procedures for the 737 and A330; coding of flight management system navigation databases; and further analysis of flight data recordings and ATC recordings.”

ATSB Incident status page: AO-2019-041 Loss of separation involving Airbus A330, VH-EBJ, and Boeing 737, VH-VZO, near Sydney Airport, New South Wales, on 5 August 2019

Holy Snap, a few arse cheeks clenching and balls tingling.

edited for unnecessary criticism.

I never paid much attention to Course Intercept 335 or direct to ENDEV. I do now .....

How does one track to the Missed Approach point, presumably climbing, then "mandatory at 600ft turn right track 070"?

It’s an interesting question, however the AIP does state:

A published missed approach procedure must not be flown unless commenced at the MAPT. If a missed approach climb is initiated before the MAPT, the aircraft must track to the MAPT before commencing the missed approach procedure.

So in this case the crew had no alternative but to continue straight ahead to the MAPT before commencing the turn (from the diagrams though it does look like the crew essentially continued to the far runway end before commencing the turn - and given the initial GA instruction was at 400’, the turn may possibly have been a little late). It’s also worth mentioning that the controllers efforts to vector the 737 resulted in them being vectored below radar lowest safe.

Perhaps with the benefit of hindsight the safest course of action would have been accepting that separation standards on the runway to be compromised and allow the 73 to land.

This is a bit like the LAHSO incident in ML several years ago, where the decision to make aircraft go around creates a greater safety risk.

Damned if you do, dammed if you don’t really.

As I understand, the 737 crew should have maintained RWY track to 0.5Nm before the threshold (coincident with the 220’ DA) and then turned right (provided they were above 600’). According to the report the turn was recorded as starting at 0.8Nm beyond the threshold (1.5km).

Now I don’t know how long it would take between initiating a roll control input and a turn displaying on a radar trace but in this case the aircraft was turning 1.3 Nm beyond the MAP while moving at about 2.5 Nm per minute.

I don’t think it even remotely fair to label them ‘muppets’ or accuse them of not following the missed approach procedure even if the turn was delayed by 20-30 seconds. Let’s not forget the ridiculous 737 autopilot disconnects as soon as you push TOGA so it’s a pretty busy period.

Not saying it was textbook but the situation was bound to occur despite the variation in reaction time. To blame the crew overlooks all the procedural issues which made this an entirely likely outcome.

I thought about this option too - but then what happens if the A330 aborts their takeoff? That would probably be a lot worse?

GLS Rwy 34R (12-45) 14 JUN 19
The missed approach point is variously indicated on the chart as
1. RW34R (Fly Over Fix)
2. MAP at DA (Vertical Profile Strip)
3. MANDATORY 600' RT. (MISSED APCH: strip + Vertical Profile Strip)

The approach coding in the FMC is from the RW34R fly over fix (34R THLD)
According to the report graphic the 737 was turning right by less than half way along the runway.
ie. Turn commenced less than 12 seconds after coded MAP.
Less than 20 seconds from being instructed to GA from low level with configuration changes etc.

Muppets Paddleboat? You must be good!

Maybe I am being overly critical. But from the report old mate flew 2,5 km past the MAP without conducting the turn, a turn which is written in big bold, capital letters with MANDATORY scribbled on there..

Put another way, if they had followed the procedure as published, would there have been a breakdown in separation? What if they tried this kind of thing into ZQN?

Anyway, not saying they're solely to blame, ill retract my muppet comment.

In my question I should have said "presumably climbing, already above 600ft". What are you expected to do if you execute a MA above 600ft? Turn immediately anyway, or continue on to the MAPt as per AIP?

As it turns out, the actual report says the 737 commenced it's GA at 400ft.

The other more interesting nuanced sleeper trap lurking here is even if you keep the "To" waypoint updated with either a direct-to or a direct-to course intercept, for some approach codings, the LNAV armed (white below) VOR/LOC doesn't become visible until you cross the FAF. In other words, you can't tell if you will have TOGA to LNAV functionality until late in the approach (1600' in this case OLSOG). Most ILS approaches you get LNAV armed with or just after VOR/LOC capture.

If the To waypoint hasn't been updated you won't get it at all, It is difficult to notice something by its absence (LNAV). Even ff you have done everything correctly, you have to actively search for it to confirm it is available after crossing the FAF. That to me is one of the the big HF lessons I have taken away from this event.

This raises the old chestnut of what to set in the MCP for Boeings, the QDM or missed approach heading (070) in this case? We set the altitude, but we are divided on the HDG.

When the 737’s got a missed approach track of 070, and the 330 ahead of it is turning onto about 105 to pick up the 075 radial - surely it’s a moot point whether the 737 turned at the MAPT or 12 seconds later. Things were going to get untidy regardless.

They should be vertically separated at that point, the 737 at 2000 at the 330 at 3000.

If they had started the turn at the MAP, this would have looked rather different. Ie

https://cimg4.ibsrv.net/gimg/pprune....5a9fb0b230.jpg

You might be assuming a bit much about the 330’s climb performance.

Sure, anythings possible.

I put it to you however that the 737 was at 920ft when the A330 rotated. Yet by the time the flightpaths converged, the A330 was higher. If it was struggling for climb performance, how'd that happen?

Paddleboat.
possibly consider that after 800ft. AGL, the 737 was above acceleration altitude and accelerating, retracting flap etc. while the 330 was doing the noise abatement climb with an acceleration alt somewhat higher than the 737.

Indeed! Certainly a possibility.

It was put to me that the reason a breakdown in separation was inevitable regardless of the 737's lateral track was due to the A330's lack of climb performance, at least relative to the 737. Your reasoning I believe supports my position. If the 737 had turned at the correct time and followed the path in my quick and dirty diagram, then there would have been more than sufficient time, given the A330 is back at 180kts until established on the radial, and the 737 is accelerating, for their to be vertical separation as provided by the respective procedures by the time their tracks intersected, if they would at all.

Except that is not what the chart says. By your logic if you were told to go-around at 1000' you would fly all the way to the runway end then start the right turn. By that time you would be level at 2000'.

By the looks of the ground track the crew flew the Missed Approach correctly as they started the go-around just above the minima. So they had to climb to 600' first before commencing the turn which took away some of their separation from the 330. They were a little late in the turn but as mentioned above the 737 is a handful in a unexpected go-around.

The turn is at 600' not at the MAP. If they started at turn at 220' then they would have turned straight into the cranes at the port. So they flew the MAP correctly as they were pretty close to the minima

The AIP is quite clear - you climb to the missed approach point and then follow the missed approach instructions. If I were issued a go around at 1000’, I would be continuing to the MAP (roughly .5nm from departure end threshold) before turning, and if that means I’m at 2000’ before starting the turn, then so be it.

Except on a precision approach the missed approach point is the DA/DH, it is not a fixed point on the ground like on a NDB/VOR/RNAV. So you are not going to continue descending to that They want you to turn right away not fly 335 for 3 miles then turn. Same on 16L.

If anything is clear... there is confusion amongst the industry about how these things are meant to be flown. Usually it never matters. But in this instance, it has brought to the surface a potential confusing issue. Which is what the ATSB are checking :)

11 ON THE SENSATIONALISM SCALE
 

So The New Daily has reported on the loss of separation incident thus:

My bolding.

Really? 330 at domestic weights.... 2500' before you roll out of the turn. It's common to accelerate from V2+10 at 2500' on that one due to the speed restriction.
Bear in mind the 737 single click toga gives 1000-2000fpm...

So the bus would outperform massively in this circumstance.

The MAPT in a procedure may be:
a. the point of intersection of an electronic glide path with the applicable DA/RA Height; (ie the corresponding point on the GS with 220’)
b. anavigationalfacility;or
c. afix;or
d. a specified distance from the FAF, or
e. a waypoint.

A published missed approach procedure must not be flown unless commenced at the MAPT. If a missed approach climb is initiated before the MAPT, the aircraft must track to the MAPT before commencing the missed approach procedure.

Of course you won’t continue descending. They very much do want you tracking to straight ahead to that point before the turn, and not turning early. There shouldn’t be any debate about this......

Whatever you pilot fellows read, don't fall for the 'trainee ATC' bull****. The trainee has a rated ATC sitting/standing beside them, watching every move.

There's only so much an ATC can do, can't fly the plane for you. That's been recognised, operationally, procedures changed to reflect this some time ago.

This, to me, looks like an airspace design issue. Everyone let down, pilot, pax. ATC.

What a **** show.

I sent the ABC an email about that and that has now been removed. :ok:

I tend to agree with Porter. Crappy design.
A few rough estimates;
At 400 ft they’d be about 1200m from the MAPT as the crow flies.
They’d be doing maybe 82m/sec.
Time how long it takes you to say
ATC : “Qantas 123 this isn’t going to work, carry out the standard missed approach procedure”
QF: “ Ahhh understood standard missed approach procedure, Qantas 123 going round”
PF “ Righto, we’re gunna go round”
PM “ Yip”
PF “ Going round Flap fifteen”
three second pause for the actions
PM” positive rate”
PF “ gear up”

It takes me between 17 and 20 seconds if I do it at a pace I think is realistic/likely.
So now you’re 400m past the MAPT climbing back through 400ft and doing exactly as the procedure calls for, tracking 335 waiting for 600ft where you’ll begin your turn onto 070. How long does that take? 200ft at 1000fpm is another 12 seconds and 980m. So now, after flying the machine nicely, you’re 1.4km past the MAPT.
So my thoughts are that if everything goes really well, you’re going to be over the runway when you turn but very likely is that you’re going to be well down the runway as things rarely go exactly to plan. All it would take is one drawn out communication and you are turning where this aircraft turned having done the job very nicely.

"Go Around, follow the 330, he's doing 140" would have worked. :E

"Too close, I'm switching to guns!" :}

The rule set is simple; at the end of the day separation gets hand balled to the pilots either VMC or IMC...

Was the spacing that tight? Looks like the 330 may have been getting airborne about when the maggot went round

Outstanding. :ok:

Interesting discussion on the event.
I do this approach often and when PM I usually ask my sidekick if in the event of a GA where they intend to turn on to 070.
I get about as many variations as given above.
It is a Sh!t design that allows the varying interpretations. Only 1 is correct and I'm not sure it's the one coded into the FMC.

However 2 lines in the report Table Summary of key events stand out.
Time 1831:04 The ADC-E instructed the A330 flight crew to line up runway 34R. The 737 was on final approach at about 2.8 NM.

Gutsiest move I ever saw Mav.

Yes but the preliminary says,

Isn't that at least part of the problem, that the 737 appears to have been 600-700 feet late executing the right turn?

2.8 nm = ~ 900’, with a Dash 8 still to vacate. Very gutsy indeed.

They want you to turn at the altitude. I've done it numerous times and even had ATC get up us for not turning soon enough. They do not want you to fly toward the runway. That is why there is a mandatory turn in the MAP. Just do what it says.

Given the confusion out there and this incident some clarification is obviously required.

Neville,
Get the AIP changed.

The obvious intent of the procedure design is not in line with the AIP rules re tracking to a MAPT before turning.

Nev,

I don’t mean to be contrary but for the sake of clarity the AIP is clear re the expectation of tracking:

https://cimg3.ibsrv.net/gimg/pprune....80b5ce6d1.jpeg

if you go around from 2000’, the expectation is that you will track to the Mapt then turn. Not turn straight away.

Track to the mapt. It's clear.
I've always been given a heading for the many GAs I've had in Sydney.