A few days ago, two Delta aircraft had an incredibly close call over the runways of Hartsfield-Jackson Atlanta International Airport (ATL), in what can only be described as a massive situational awareness failure on the part of the air traffic controller…
In this post:
ATC sends Delta plane directly into path of another plane
VASAviation has the air traffic control audio and a visualization of an incident that happened a few days ago at Atlanta Airport. For some background on the two planes involved:
- Delta flight DL2472 was an Airbus A321 departing for Rochester (ROC) that was cleared for takeoff on runway 26L
- Delta flight DL1568 was a Boeing 737 arriving from Pittsburgh (PIT) that was cleared to land on runway 26R around the same time
You might not think much of this, except for the fact that the pilots of DL1568 opted to perform a go around at the last minute, as the Frontier Airlines jet landing before them didn’t clear the runway with enough buffer. Fair enough.
So at this point you had a Delta 737 going around above one runway, and a Delta A321 taking off on a parallel runway. That’s not ideal, but also not a huge issue, since they had an altitude difference, and also were both on the same headings, but parallel.
It’s not like the air traffic controller would turn the two planes on a collision course… right? Well, actually…
Somehow the air traffic controller didn’t seem to be processing how close the planes were to one another, but this is where the situation went from bad to worse. The air traffic controller then gave the plane on the left (taking off from runway 26L) the instruction to make a sharp right turn, even though the plane going around (from runway 26R) was to its right.
That’s right, she sent the plane on the left directly into the course of the plane on the right. This sets off a TCAS alert, which the pilots in turn responded to. It would appear that the planes were basically on top of one another, with a vertical distance of around 500 feet.
For those not familiar with TCAS, this stands for Traffic Collision Avoidance System, and it issues resolution advisories to pilots, giving them instructions about what maneuvers to make, based on traffic becoming dangerously close.
At this point the air traffic controller seemed to go into a bit of a shock, and struggled to manage the situation. It would appear that after about a minute she was replaced by someone else, to give her a chance to catch her breath, or something.
Fortunately the planes quickly got sufficient distance, and went on their respective ways without further incident…
There’s so much that was wrong with this situation
On the most basic level, this was human error. My guess is that the air traffic controller’s instructions for a right turn were actually for the plane performing a go around, and not for the plane that had just taken off. As you can hear, the pilots of DL2472 seemed to be caught off guard by their altitude and heading instructions, which is why they followed up.
That’s a pretty major mistake to make when two planes are already so close to one another. She clearly didn’t have situational awareness here, since she was giving another plane taxi instructions while two planes were getting closer and closer to one another.
Another pilot just jumping in and saying “damn boys” on a frequency didn’t exactly help either. Then there’s the whole United States culture around non-standard phraseology, giving landing instructions even before the runway is clear, etc. We just don’t seem to have the safety margins that we once had…
Bottom line
We saw a dangerous close call over the skies of Atlanta Airport the other day, when one Delta plane was taking off, and another was performing a go around, on parallel runways. The air traffic controller seemed distracted, and didn’t initially notice how close the planes were getting to one another. But she made the situation worse when she instructed the plane on the left to turn right, directly into the path of the plane on the right. As usual, TCAS saved the day…
What do you make of this ATL close call?
Did these incidents always occur but were not reported on the media? Or are you on to some new type of surveillance information to get this information?
It is pretty frequent now.
A contributing factor to this may be that US ATC seems to want to actively control aircraft going missed rather than letting them fly the published missed approach. If you have a published missed procedure, it should be safe to fly regardless of what other aircraft are doing. This would let aircraft declare they're going missed and fly the procedure, then Tower can transfer them to approach/departure when they have a chance (rather than having...
A contributing factor to this may be that US ATC seems to want to actively control aircraft going missed rather than letting them fly the published missed approach. If you have a published missed procedure, it should be safe to fly regardless of what other aircraft are doing. This would let aircraft declare they're going missed and fly the procedure, then Tower can transfer them to approach/departure when they have a chance (rather than having to direct them while juggling others) and then the new controller will know what they're doing and can direct them accordingly to get back into the approach.
It's not perfect but such a process could have helped in this scenario (or am I missing something?)
mostly valid point.
The missed approach procedure for 26R is to climb to 1800 (ATL elevation is 2026 feet) and then a climbing right turn to 4000 on the Rome Georgia (RMG) 132 radial which is less than a 90 degree turn and then to hold.
I suspect the reason they don't want aircraft to do that on a regular basis is because they don't really want planes that go missed because of traffic...
mostly valid point.
The missed approach procedure for 26R is to climb to 1800 (ATL elevation is 2026 feet) and then a climbing right turn to 4000 on the Rome Georgia (RMG) 132 radial which is less than a 90 degree turn and then to hold.
I suspect the reason they don't want aircraft to do that on a regular basis is because they don't really want planes that go missed because of traffic to go into holds - which takes a lot of ATC resources - but rather to quickly resquence them back into the arrival pattern.
If anything, a 90 degree turnout to 360 reduces the time to get back into the arrival sequence
but the command to fly the 360 heading was to the departing aircraft which, as Ben notes, is to the left of the arrival corridor so it never makes sense for ATC to give that kind of command esp. right over the airport.
ATC pushes metal hard and deals w/ professional pilots - the majority of traffic at a primary class B airport - that can quickly process instructions.
ATC screwed up here but TCAS and the Resolution Advisory and the quick acting pilots - 2 crews - eliminated the danger. Making mistakes isn't just an ATL problem; in fact adequately separated parallel runways are common at many large US airports and around the world.
The FAA will figure out why this bad instruction was given, there will likely be retraining, and she will likely return to duty unless there is some evidence that she has made mistakes at a higher than average rate - in which case she might be moved to another ATC facility where there is either less traffic or the traffic isn't so close together so there is more time to react
...ATL elevation is 1016 feet....
This is exactly why we have safety nets like TCAS (which, as a sidenote to Ben stands for Traffic ALERT and Collision Avoidance System - it's a two tier system that provides Traffic Information and/or Resolution Advisories, depending on the situation). It is also a perfect example for a situation in which everything and everyone has to work perfectly in order for these system to work efficiently.
As to your guess "the air traffic...
This is exactly why we have safety nets like TCAS (which, as a sidenote to Ben stands for Traffic ALERT and Collision Avoidance System - it's a two tier system that provides Traffic Information and/or Resolution Advisories, depending on the situation). It is also a perfect example for a situation in which everything and everyone has to work perfectly in order for these system to work efficiently.
As to your guess "the air traffic controller’s instructions for a right turn were actually for the plane performing a go around, and not for the plane that had just taken off": I don't think so. What happened in my opinion is that the Controller missed the notification from DL1568 that they are going around. This means that she was convinced that DL1568 was on the ground, rolling out and that therefore the right turn for DL2472 was safe. Also, have you noticed that DL2472 got the read-back wrong? They acknowlegded heading 340 while the instruction was for heading 360. But that also wasn't caught by the Controller.
Also, the reaction from the pilots to the TCAS Alert needs to be evaluated. At 1:20 radar shows an alert to DL1568. At that point the vertical distance between the two aircraft was 600 feet (with DL1568 being below). That's when DL1568 reminds the Controller that they are in a Go-Around, rather than reacting to the TCAS Alert (my assumption is that the Alert was "level off" as DL1568 was in a climb and DL2472 was above). Two seconds later, the TCAS Alert lights up for DL2472. At 1:29 DL1568 receives a clearance to turn right heading 360 and to climb to 4000 feet. The pilot was hesitating in the read-back (possibly due to the TCAS Alert on the flight deck). At the same time the climb-rate reduced which may have been a result of the TCAS Alert. The first mention of the TCAS Alert was from DL2472 at 1:35 (I assume they received a "climb climb" alert to increase the vertical distance to DL1568). It seems that only then the Controller became aware of the close proximity of these two aircraft.
Thus, while the Controller clearly lost situational awareness and bears the main responsibility for the incident, also the flight crews reaction to the TCAS Alerts needs to be evaluated. There is a reason why TCAS Alerts override Controller instructions. But the "weak point" in this system is pilots reaction. And this is, in my opinion, also something that wasn't ideal in this situation.
your assumptions are contrary to some of the facts that are actually known or are not stated, so you are making assumptions based on guessing on the facts.
1. there is no basis for turning any aircraft 90 degrees to their operation; it would have been much more advisable for a 40 degree turn which would have avoided anything but a plane directly in front of the aircraft - but that is not supported...
your assumptions are contrary to some of the facts that are actually known or are not stated, so you are making assumptions based on guessing on the facts.
1. there is no basis for turning any aircraft 90 degrees to their operation; it would have been much more advisable for a 40 degree turn which would have avoided anything but a plane directly in front of the aircraft - but that is not supported by radar.
2. the controller gave wrong directions to the flight number she stated. That is still controller error.
3. a bank of that amount on the arriving aircraft might have put the plane too close to buildings north of the airport. it takes longer for engines to spool up on a go around than for a plane that is already at takeoff power
4. The first level of TCAS is an alert and does not override ATC; the arriving pilots correctly reminded her that they were going around. The second level is a resolution advisory and the departing aircraft stated it at least on the radio but both aircraft might have gotten advisories; it is speculative to suggest what the arriving aircraft got w/o hearing cockpit audio which should have been recorded. Since the aircraft was arriving, DL should have pulled the CVR.
ATC failed at any number of levels and that is rare in ATL just because of the parallel runways.
There really is no proof that the pilots did anything other than what they were told to do - either by the controllers or TCAS.
No, Tim - just, no.
I could dissect what's incorrect about every part of your post. How you turn your assumptions, and a little bit of info, into "absolutes" and write falsehoods with extreme confidence. Again.
Today, I'll just focus on just this one about the instructions for the go around. You said:
"it takes longer for engines to spool up on a go around than for a plane that is already at takeoff...
No, Tim - just, no.
I could dissect what's incorrect about every part of your post. How you turn your assumptions, and a little bit of info, into "absolutes" and write falsehoods with extreme confidence. Again.
Today, I'll just focus on just this one about the instructions for the go around. You said:
"it takes longer for engines to spool up on a go around than for a plane that is already at takeoff power"
That's all I'd need to read from you to know that all of your time in a transport category aircraft has been in the passenger cabin - zero hours on the flight deck.
Time to "spool up" has nothing to do with it. It's about energy management. Jets have WAY more energy on a go around. Engine are stabilized at 50% - 60% N1 when configured for landing, so it's maybe 2-3 seconds to get to full power. Your speed needs to increase by only 30-40 kts. from approach to a go around.
It's very easy to accelerate from having a "good" amount of energy (on approach) to having a "high" amount of energy (on the go around). A plane on a go around, executed at the same time as another rotating for takeoff, will easily outperform the other.
To show how well planes can manage them, and to show how simple and unrushed you can be as a pilot in a go around, I've even done fully hand flown, single engine approaches to a single engine TOUCH AND GO in the sim. It's a great demonstration on how much energy and momentum we carry, compared to takeoffs.
Now compare that to the takeoff - starting at 0 kts (instead of+/- 150kts), and lower (on the runway instead of several hundred feet above it). Takeoffs are a lower energy state - that's why we practice engine failures on takeoffs all the time, and almost never on go arounds.
Your ability to be confidently wrong never ceases to amaze me.
thank you for proving my point
"so it's maybe 2-3 seconds"
which is longer than a jet engine that is already at takeoff thrust.
The departing aircraft was in the air just a few hundred above the runway.
If you quit trying so hard to prove me wrong, you might actually contribute something but all you do is indict yourself.
Seriously, Tim - you don't have the credentials you think you have. Fine. Let's take your points, one at a time.
1) You said: "there is no basis for turning any aircraft 90 degrees to their operation; it would have been much more advisable for a 40 degree turn."
Nope - wrong. There are PLENTY of missed approaches that include more than 90 degree turns. Published missed approaches are designed to separate IFR traffic during...
Seriously, Tim - you don't have the credentials you think you have. Fine. Let's take your points, one at a time.
1) You said: "there is no basis for turning any aircraft 90 degrees to their operation; it would have been much more advisable for a 40 degree turn."
Nope - wrong. There are PLENTY of missed approaches that include more than 90 degree turns. Published missed approaches are designed to separate IFR traffic during multiple missed approaches in IMC - but are rarely flown "as published" when conducted in actual operations. And the 90 degree turn - to DL1568 - was to get them away from the departures on the parallel runway as rapidly as possible. Perfectly reasonable.
2) You said: "The controller gave wrong directions to the flight number she stated."
Nope - wrong. What happened was the controller MISSED the first time DL1568 reported the Go Around, and she never responded to it. So turning the departure to 340 was fine in her mind (and would have been, without a Go Around). When DL1568 REPEATED they were Going Around (which in her mind was probably the FIRST call of one), she turned them to 360 to (in her mind) avoid a conflict and increase separation with anyone taking off on the parallel.
3) You said: "a bank of that amount on the arriving aircraft might have put the plane too close to buildings north of the airport"
I don't know where to begin with a statement that ignorant. First, it's not a 90 degree "bank" - it's a 90 degree turn. Second, there are FAA regs on obstacles within the operational zones of an airport - so the buildings are not an issue. Third, there are minimum climb gradients that all current transport category aircraft have to meet - which every one in Delta's fleet do. There was ZERO wrong with giving DL1568 a 360 degree heading. It's the fastest way to get away from any conflict. The error was that she missed the Go Around call the first time it was made - so she didn't give the turn sooner. And she wouldn't have turned DL2472 at all if she knew DL1568 was "on the go".
Also from you: " it takes longer for engines to spool up on a go around than for a plane that is already at takeoff power"
If you are trying to make a brilliant observation that an engine at a mid-power setting (on approach) will need more time to reach full power than one already AT full power (on takeoff) - wow! Thanks, Captain Obvious! Nobody could have figured that out. As it also turns out, it takes a car going 30 mph longer to reach 60 mph than one already going 60 mph!
But you clearly know NOTHING about critical phases of flight, energy management, plane performance, etc. when you think the plane on the Go Around (already airborne, already at close to 150 kts) has more risk regarding performance than the one taking off. The plane ALREADY FLYING has more energy (both kinetic and potential) and can more quickly climb and maneuver than the one just taking off. It's not because of "spool up" time like you said. Even in the old turbojets (like the JT8-Ds that DID have 6-8 seconds delay from idle thrust settings to full power available), a plane doing a Go Around could FAR more easily respond, climb, and maneuver than one just taking off.
4) You said: "The second level is a resolution advisory and the departing aircraft stated it at least on the radio but both aircraft might have gotten advisories"
No "might". BOTH aircraft get RA's - that's the point of them. They coordinate diverging resolutions, which relies on BOTH of them getting one (one "Climb!" and one "Reduce Climb!", for example). Just further evidence you've never been in a flight deck with TCAS.
5) You said: "ATC failed at any number of levels and that is rare in ATL just because of the parallel runways."
I'd love to see your data on how "rare" ATC issues are particularly at Atlanta, compared to other airports with similar, parallel runway / widely spaced, multiple arrival / "same direction departure" airports (ORD, LAX, PHX, DFW, or MCO come to mind). But as usual, I'm betting you're just shooting your mouth off - with nothing concrete but your own biases to back it up.
Stay in your lane. Continue to pretend to be a financial analyst, and debate PRASM and Pretax Profit Margins and whether to use GAAP numbers or non-GAAP numbers. Because your "I've flown a general aviation aircraft into ATL once" is not giving you the knowledge you think you have to understand, debate, or provide insights on airline operations. It's clearly not working for you.
Still waiting to hear how many hours you have in your logbook, and how many are on anything bigger than a C182.
Another DEI hire gone wrong in aviation...
It’s getting old.
would have been the most premium mid-air collision
ATL likely over scheduled. Should reduce takeoffs and landings accordingly. Greedy DL of course.
in your dreams.
ATL has far better operational performance than any other large hub - any NYC airports, CLT, DFW, DEN, etc.
ATC pushes metal and expects traffic to comply. The Frontier pilot -who probably wasn't near as familiar w/ ATL - was moving slow.
The two DL pilot crews were the two legs that kept this from becoming a disaster; in just about every incident where one party messes up, the other two do...
in your dreams.
ATL has far better operational performance than any other large hub - any NYC airports, CLT, DFW, DEN, etc.
ATC pushes metal and expects traffic to comply. The Frontier pilot -who probably wasn't near as familiar w/ ATL - was moving slow.
The two DL pilot crews were the two legs that kept this from becoming a disaster; in just about every incident where one party messes up, the other two do what they need to do to fix the other party's mistake.
you can't stand that DL runs the world's largest and most efficient passenger hub in the world and does it in great partnership with the City of Atlanta, the FAA and all partners.
You cant understand that was a joke post because you also suggest that EWR should get further cut even when an incident has nothing to do with traffic - like the UA plane hitting the light-pole/bus.
Daniel of course, he is known as The Timcel for a reason. Fired, failed, neutered DL “analyst”.
jokers can be shown to be the idiots that they are at any time.
Of course it had no basis in fact...but he opened himself for a reminder of just how well ATL is run
Ben, is it DL 2472 or 2474? You referenced both.
ATL ATC runs a pretty tight ship but they were clearly flustered by the error of the controller that gave a conflict heading for the departing aircraft.
I have flown a GA arrival on ATL 26R and departure on 26L and they always recognize that there are two lines of traffic that cannot mix.
and they want all traffic to keep moving. It should not have been a surprise that the DL arrival had...
ATL ATC runs a pretty tight ship but they were clearly flustered by the error of the controller that gave a conflict heading for the departing aircraft.
I have flown a GA arrival on ATL 26R and departure on 26L and they always recognize that there are two lines of traffic that cannot mix.
and they want all traffic to keep moving. It should not have been a surprise that the DL arrival had to go around.
The real issue is why she gave a change to runway heading before the departing aircraft cleared the end of the runway; that is very rare precisely because of the possibility of a go around. The fact that she couldn't get right and left straight is concerning.
We talkin’ in a Cessna 172, or a little bigger?
bigger
Oooh! Me like-y!
Bigger than a C172? Good for you. If we want to discuss the particulars of how ATL ATC will handle a C182 arrival, we'll send for you.
The adults need to talk now. Go back to coloring in your room.
then you won't be part of the conversation.
16,000+ hours in the logbook, with over 10,000 of them in transport category multi engine turbojet aircraft.
Your move, Cessna Boy.
Retired seems a reasonable next step for ATL ATC.
Hopefully he didn't follow through on removing the bookmark (or can at least read this one)
"Well, actually…" Hold onto your Biscoffs! Phew. Glad everyone's alright.
(Now, for Alert/Eskimo to suggest we shouldn't have humans in cockputs or towers...)
Indeed that is the normal reply from our dear Eskimo.
There have been too many close calls and two major preventable tragedies in our recent past, so I guess I can understand why he's so afraid/upset? Maybe?
The problem is he incorrectly asserts that "AI" is what could have prevented those accidents. AI is not developed sufficiently to replace human ATC, despite his baseless claims. (No corporation/government has announced otherwise.)
He is also talking from...
Indeed that is the normal reply from our dear Eskimo.
There have been too many close calls and two major preventable tragedies in our recent past, so I guess I can understand why he's so afraid/upset? Maybe?
The problem is he incorrectly asserts that "AI" is what could have prevented those accidents. AI is not developed sufficiently to replace human ATC, despite his baseless claims. (No corporation/government has announced otherwise.)
He is also talking from the position of someone who has never worked ATC, and back when I was in that position, I might've thought the same. The job isn't what you think it is when you're on the outside.