There was an uncomfortably close call at San Francisco International Airport (SFO) a couple of weeks ago, as two jets departing on parallel runways ended up dangerously close to one another on their climb out.
In this post:
United A320 & SkyWest CRJ have dangerous SFO departure
VASAviation has the scoop on what happened at SFO on the afternoon of May 13, 2025. Interestingly, this incident was seemingly reported to VASAviation by a controller at the airport, who stated that this was one of the two or three closest calls he has seen at the airport in 15+ years. That obviously puts into perspective just how serious this was.
Specifically, this incident involves two aircraft that were supposed to perform parallel takeoffs on runway 1L and runway 1R:
- A United Airbus A320 with the registration code N4901U was operating flight UA1152 to Dallas (DFW)
- A SkyWest CRJ-200 with the registration code N246PS was operating flight 5273 to Fresno (FAT)
Both planes had specific routes they were supposed to fly after takeoff, known as SIDs (standard instrument departures):
- The United jet was first given clearance for takeoff on runway 1R, and was supposed to fly the TRUKN SID, which involves a right turn after takeoff
- The SkyWest jet was cleared for takeoff on runway 1L a short while later, and was supposed to fly the SSTIK SID, which involves a left turn after takeoff
As you can tell, this should be a routine procedure, since the plane taking off from the left was supposed to turn left right after takeoff, while the plane taking off from the right was supposed to turn right after takeoff.
However, a moment after takeoff, the United jet made a sharp turn to the left, rather than flying right, as it was supposed to. This meant that the United A320 flew directly into the path of the SkyWest CRJ-200. At their closest points, the planes were just 0.4 nautical miles apart horizontally, and 280 feet apart vertically. This is a beyond uncomfortably close call.
Fortunately the SkyWest pilots received a traffic alert, and they ended up turning to avoid a collision. You can hear the audio for yourself below, along with a recreation of what happened, as VASAviation always does a great job with this.
Why did the United jet fly in the wrong direction?
Given that the United jet was responsible for this close call, pilots were given a number to call after their flight, to discuss the pilot deviation. According to the controller providing details, the United captain called, and shared the following details:
- It was the first officer’s first time departing SFO, and the first officer was the pilot flying
- Around 500 feet after takeoff, the flight management system (FMS) indicated that they should commence a left turn, which they did
- The captain questioned what was going on and was confused, as he knew they should be turning right
- So the captain then punched in the new heading, but at that point the planes were already so close to one another
It sure sounds to me like the FMS for the United plane might have had the TRUKN departure programmed for a runway 28 departure, rather than a runway 1 departure. At least that would explain the left turn after takeoff, rather than the expected right turn.
Everything about this situation was a mess
The United jet flying the wrong heading after takeoff was just the tip of the iceberg here, as there were several other issues as well.
For example, while this situation was going on, the air traffic controller twice issued the United pilots the instructions to “fly runway heading,” which is clearly not what he was actually intending to say. At this point the plane was way off the runway heading, and the air traffic controller intended to say “fly present heading,” but instead, he gave completely different instructions, even after the pilot sought clarification.
This also seems like a situation where at least as an observer, you’d think there are opportunities to improve communication. Admittedly hindsight is 20/20, but when you’re issuing two planes on parallel runways takeoff clearance, it seems like it would make sense to remind them of the other traffic, and of their departure route.
That might sound obvious, but one would think that if the United pilots were specifically reminded that another plane was taking off to the left of them, they wouldn’t try to turn left right after takeoff.
Also, that poor Turkish Airlines pilot! My gosh, how many times did he ask for landing clearance? Obviously there was a lot going on, but it seems like it would’ve been more efficient to just tell the pilot “standby,” based on how much airtime Turkish took up. Like, was the Turkish jet at the gate before landing clearance was actually issued?
Bottom line
A United A320 and SkyWest CRJ-200 had a close call while performing parallel takeoffs from SFO. The problem arose because the United jet made a wrong turn after takeoff, seemingly due to an incorrectly programmed computer.
This situation should’ve never happened, and what makes it even worse is that it wasn’t the extent of the issues. While the air traffic control was obviously overworked, he repeatedly issued wrong instructions, different than what he intended to.
What do you make of this SFO takeoff incident?
@Mallthus - Reagan was 40 years ago. He fired people for not coming to work. What about Obama, much more recent, who not only decided to lower standards, but also also choose based on DEi instead of merit? You don’t don’t think choosing things other than merit is not the problem?
In this day of automation and annunciation, it is hard to understand why when they were lined up on 1R with the TRUKN SID from 28R programmed into the FMC, a major alert wasn't flashed on the ECAM.
Unmentioned issues and questions:
1. On parallel departures (or arrivals) why are controllers not required to stagger them - 20 seconds more separation can cover a multitude of sins.
2. Since there was a runway switch, from 28L/R to 1L/R, and the new runway may not have been briefed (and new runway not set in Flight Directors) is the industry working on software warnings to catch errors like this?
It’s not uncommon for 1L departures to turn right and 1R to turn left at SFO. As counterintuitive as it seems I had always assumed it was to avoid a tight turn to avoid OAK traffic
@ Kiwi -- Presumably not while there are parallel takeoffs, though, for obvious reasons.
@Malthus, the PATCO debacle was almost 45 years ago, and you're still blaming Reagan for what happened this month?! While you're at it, please feel free to blame Vice President Aaron Burr's bad example for inciting today's gun violence epidemic.
Reagan would have wished he had all these technologies of today back then.
PATCO wouldn't be a debacle, it would've been revolutionary first step. Just like how Reagan allowed civilian access to GPS.
We need better IT behind ATC/Piloting. ATC is an eminently programmable task. The data is all there, it just needs feeding into software. It's not that ATC do a poor job, but it's that computers could do it better. The _moment_ a deviation is detected around an airport you could trigger an alert instead of relying on pilots/ATC to luck out and notice it. The whole system needs modernizing and bringing up to speed with the tech which is available today.
That's what I'm trying to say all this time.
But face it Joe, to fully eliminate "unpredictable" deviations, you have to remove the "unpredictable" from the equation, aka humans.
We thank your kind for trying to keep the skies safe. Most of you are doing great so far. But we're now at a point that is approaching human limitations. Time to pass on the responsibility to computers and never have human errors in aviation ever again.
An accident is bound to happen and when it does, there will be a lot of "I told you so". Things don't change until people die. Not only that, but things don't change until a lot of people die. Sad but true.
We're still having humans, the main leading cause of fatal accidents, behind the controls?
I told you so.
An accident did happen at DCA a couple of months ago. There hasn’t been enough “told you so” since then.
To make people feel better... This was a MAJOR screw up by the UA crew. However, it was NOT a near disaster.
The Skywest jet would have to have sped up dramatically and climbed at the same time to hit the UA jet. Virtually impossible even if they wanted to.
It was a big safety issue, but in terms of an actual accident, it was not likely to happen. (And in IFR conditions...
To make people feel better... This was a MAJOR screw up by the UA crew. However, it was NOT a near disaster.
The Skywest jet would have to have sped up dramatically and climbed at the same time to hit the UA jet. Virtually impossible even if they wanted to.
It was a big safety issue, but in terms of an actual accident, it was not likely to happen. (And in IFR conditions they would not be doing simultaneous departures.)
Of course this needs to be investigated further. And this is an opportunity to learn and add safety measures.
“ The Skywest jet would have to have sped up dramatically and climbed at the same time to hit the UA jet. Virtually impossible even if they wanted to.”
You mean like the Blackhawk did @ DCA in January that killed over 70 people?
Why was a computer programmed to fly the plane after takeoff? If the pilots were flying and properly trained, and had paid attention to what runway they were taking off from, they would have turned right according to procedure.
As Mr. Scott used to say, "The more they overthink the plumbing, the easier it is to stop up the drain!" Too much reliance on technology. And they have got to start training air traffic controllers, so we have enough!
The autopilot hadn't been engaged. What the Pilot flying (F/O) did was follow the Flight Director which dutifully commanded a Left Turn according to the (incorrect) instrument departure programmed into the flight management computer
Wondering if the investigation will reveal the 320 was originally anticipating a 1L assignment, programmed the flight computer accordingly, then got switched to 1R and then neither 320 pilot made the necessary computer correction.
It appears more likely that their mistake was even worse in that they might have punched in the TRUKN SID from 28R into the FMC. Starting at 50' AGL, the vertical FD bar would've (correctly) commanded a left turn to follow 284 (28R Runway Mag Heading) due to the incorrect input to the FMC. Note that the TRUNK SID from 28R calls for following the Runway Heading (284) until 520' and then Direct TRUKN.
Not even close to what the sleeping Air Canada pilot came within a few feet of causing in SFO
Yet another reason I am not even flying over U.S. airspace, let alone into it. Sort your acts out people!
just because the majority of internet aviation sites are in English and focus on US aviation doesn't mean there aren't pilot and ATC errors in other countries.
and it is also interesting that this article comes right after the article about someone's perception of cockpit communications failures.
The primary reason for this will likely be incorrect input of the navigation into the FMS probably by the FO that flew the route which he didn't...
just because the majority of internet aviation sites are in English and focus on US aviation doesn't mean there aren't pilot and ATC errors in other countries.
and it is also interesting that this article comes right after the article about someone's perception of cockpit communications failures.
The primary reason for this will likely be incorrect input of the navigation into the FMS probably by the FO that flew the route which he didn't understand correctly.
ATC botched the communication after the incident began to unfold but the captain did realize something was wrong and communicated it to the first officer.
and a question will be asked if UA procedures require both pilots to verify what is input into the FMS. If the first officer incorrectly input the route into the FMS and the captain should have caught it, the captain is on the hook.
Thank you. We don't want you.
Oh brother
It's really difficult for me to understand the complete lack of situational awareness. The takeoff clearances were both on the same channel so the UA pilots should have been aware of the other plane taking off on their left side.
Oy: Humans behind the controls.
Here’s a fun fact. That same SkyWest airplane used to be operated by PSA, and was involved in a (nonfatal) crash back in 2010.
https://avherald.com/h?article=42607d9d&opt=4096
Some aircraft are cursed. The AA CR7 that collided in mid-air with the Blackhawk @ DCA back in January (also a PSA aircraft) was involved in an accident @ CLT in 2017 when it collided with a deer on the runway causing significant damage:
https://asn.flightsafety.org/wikibase/319924
What makes you think the controller didn’t want him to turn to runway heading?
There will be an investigation, it seems premature to speculate on the controllers intent.
Because of their error, the UA plane was to the left of Skywest when they should've been to the right. Having UA fly runway heading would have them on a heading of 010. With Skywest being assigned a heading of 350, this places the planes on a converging course again. Later giving UA a heading of 280 ensures the planes were on diverging courses.
"As you can tell, this should be a routine procedure, since the plane taking off from the left was supposed to turn left right after takeoff, while the plane taking off from the right was supposed to turn right after takeoff."
What an awkward sentence. How about stating "since the plane taking off from the left was supposed to turn left *shortly* after takeoff"?
I' glad I'm not the only one that noticed this. I had to reread the sentence a few times even though I knew the intent.
Well in fairness most of the time when departing off the 01’s 1L turns right and 1R turns left, so it is not surprising that the FMC could be programmed that way.
That happens quite a bit at SFO where two planes take off on parallel runways and they turn based what runway they're on. Wouldn't common sense dictate that if there's a plane on the left runway that plane is going to turn left?
Sorry, did you mean to include the link to Vasa?
https://www.youtube.com/watch?v=lOpRf0IeGAo
Why do we have so many near accidents, or worse? Whether they're due to pilot errors, ATC errors, errors by some other ground crew, or errors by some workers at a plane assembly plant, they're all HUMAN errors. No one wants to admit it, but the reality is that the average quality of people working in various sectors of the economy has significantly deteriorated, especially after the pandemic.
At the end of the day, in some ways, ATC in the US has never recovered from Reagan's mass firing of PATCO controllers.
Like pilots, air traffic control is a highly technical, highly specialized, and highly stressful job. Naturally, we want to have the very best people in these roles and we want them to have the tools and backup needed to do their jobs as close to perfectly as possible.
But we...
At the end of the day, in some ways, ATC in the US has never recovered from Reagan's mass firing of PATCO controllers.
Like pilots, air traffic control is a highly technical, highly specialized, and highly stressful job. Naturally, we want to have the very best people in these roles and we want them to have the tools and backup needed to do their jobs as close to perfectly as possible.
But we have also decided (as a society), that public employees, regardless of their roles, shouldn't make private sector money, so it's become incredibly difficult to recruit and retain the caliber of talent our systems really require. Sadly, I don't think this is going to change until there are multiple large accidents with mass casualties where the blame can't go anywhere but to ATC. Only then will we start to rethink our practices around ATC.
That's a ludicrous statement without evidence. The Reagan firings were ~40 years ago. Most likely all those involved are long gone. Come on, have a more nuanced & intelligent hypothesis.
but was the UA flight properly catered?
It's been a rough couple months for UA.
Interesting UA is proud of calling themself a premium airline without catering.
Well done. I actually chuckled.