An air traffic controller accidentally cleared two planes to take off on parallel runways around the same time, putting them on a course dangerously close to one another. Perhaps what’s most noteworthy here is how the air traffic controller then acted confused about how this happened…
Continue reading: SFO ATC Clears Two United Planes To Take Off At Same Time, Then Denies It
Share your questions, experiences, and thoughts below.
Was gonna say… pretty sure SFO has the parallel runways… glad everyone’s alright.
(And, if you get the right room, you could watch it all at the Grand Hyatt at SFO!)
The whole reason why tower controllers are in a position well above all of the action is so that they can visually confirm when planes actually take off. He simply should not have given back to back takeoff clearances on parallel runways without visually confirming their location.
The controller's confusion wasn't just "memory loss"; it was likely a Tag Swap or a ghost target on the surface radar display caused by multipath reflections from the heavy metal of two widebodies maneuvering in parallel. When he said, "I don't know how you got cleared," he was likely staring at a scope where the data block for UA2621 hadn't properly snapped to the target on runway 1R, leading him to believe the runway was still sterile.
This is exactly why we need to stop relying on legacy voice loops and mandate CPDLC (Controller Pilot Data Link Communications) for all critical surface phases, not just initial flight plan delivery. If the takeoff clearance had been digital, the system logic would have hard-locked the 1L departure against the 1R occupancy.
Voice communication is an analog relic in a digital airspace. Until we remove the "human in the loop" for routine clearances, we are just waiting for the next collision. The pilots and controllers are just meat servos operating a system that should be automated.
I didn't think parallel departures simultaneously on 1L and 1R at SFO were that uncommon. I've heard it referred to as a drag race, and I've seen them on some YouTube live spotter channels. 1L departing flight instructed to turn left upon departure, and 1R departing flight instructed to turn right upon departure. Perhaps this is only permitted during VFR conditions?
Your AI-generated drama reads like a bad LinkedIn entry.
"It isn't BLAH, it's BLAH BLAH." Irony! Drama! BS! In this case, "The controller's confusion wasn't just 'memory loss'; it was likely a Tag Swap..."
Unfortunately your AI tool doesn't know the difference between a narrowbody and widebody aircraft; or you don't. "...the heavy metal of two widebodies maneuvering in parallel." They weren't two widebodies.
You are posting AI slop.
SFO based pilot for over two decades. Like all aviation errors, this was a chain of events and not a single cause incident. The "holes in the Swiss Cheese" lined up, and safety margins were reduced, but the system caught the errors.
Normal operations in SFO: two departures off the "1's", with 1L having a slight left turn right after getting airborne. 1R has a right turn. Look at any videos - you see that's how airborne separation is achieved. 1L is 1,000' shorter than 1R, but most planes can use 1L when they are using a departure with the left turn.
What likely contributed to this incident here:
- The 777 needed runway 1R for departure for the extra length (ATC now has an aircraft using the right side, but will have a left turn)
- If the animation is accurate, there was some time between the takeoff clearance and the start of the takeoff roll for the 777 (not what ATC sees 99% of the time with northbound departures - most "heavies" need 28L/28R for departures, so they may not be accustomed to the length of time used by a 777 using 1R to turn the plane on both relatively narrow taxiways, line up, advance power/confirm stable engines, set takeoff power, and start the takeoff roll. A smaller plane like a 737 can do all that faster).
- It was night, making things harder to visually confirm. There would also be possible confirmation bias on the part of the ATC controller if he did visually confirm the 737 was in position on RWY 1L, ready to go, when he issued the second takeoff clearance. The start of both runways are slightly displaced (with 1L a little bit further forward) and looking from the tower, a line drawn from a plane on 1R through a plane on 1L almost lines up with the tower. Any beacon / anti-collision lights from the 777 on the right could line up with the 737's. A quick glance and the controller (working at midnight) might think it was just one plane, if the takeoff clearance for the 737 came right when the 777 started moving on the runway. Again, confirmation bias.
- Once airborne, it was clear the controller didn't think he had given two airplanes on the same departure procedure quick, sequential departures, one right after the other. So he either thought the 777 was already well out in front (but I doubt that, or the 737 would have been held a little while for wake turbulence separation) or more likely, he was so accustomed to "1R's turn right" that he forgot the 777 was making a left turn from the right-side runway.
There were still several factors that prevented this from being more serious.
- It is REALLY easy for the trailing aircraft's pilots (737) to see the proceeding one (777) turn in front of them when departing the "1's", and the 737 crew could turn themselves away. Usually, with parallel departures in SFO, one pilot is watching the other departure anyhow just to be sure they do what you'd expect (plus, it's really cool to watch an airborne plane that close, and see them bank away from you!)
- The 777 and the 737 would have dramatically different climb rates, reducing the chance that a turn would result in a collision anyhow if the 777 was not seen by the 737.
- TCAS would come into play to provide an alert.
This was primarily a human error by ATC, of course. There could be improved procedures to help prevent a repeat. At some US airports, when issued a departure clearance, ATC confirms the first departure fix "United 2621, RNAV SSTIK, Runway 1R, Cleared for Takeoff". If the clearance was issued with this extra instruction ("R-Nav SSTIK") the crew in the 737 would have a chance to hear that, and think, "Wait! That's OUR first fix too - they are on the same route - they are going to turn left!" and when issued their own takeoff clearance, challenge or reject it. The 737 crew probably also expected the 777 to make a right turn based on past experience.
Bottom line - it was yet another example of why commercial aviation professionals (Pilots, ATC) need to be rested, alert, and ready for the unexpected at all times - especially more so when anyone first notices a single thing that will be outside the scope of what is "normal operations".
Who would fly from SFO to San Jose ? That's similar to flying from JFK to Newark . Or from Miami to Miami Beach .
It looks like someone didn't pass their city codes test and doesn't know the difference between SJO and SJC.
Okay, I’m an idiot - parallel means not meeting, tight? What’s the problem with takeoffs on parallel runways? Were they set to make turns they would put them on a collision course after takeoff? If anyone decides to answer this I’m sure it’s going to seem obvious once I hear the answer but I’m so lost…
(I was also confused about SFO-SJC till I realized it was SJO so maybe it’s just not a good day for me…)
It seems to be a source of pride that people who post to this site use city codes to describe flights they took, but speaking in this way does not expand the audience base for this blog as some of us do not know that the code for Almaty Kazakhstan is ALA.
You simply need to not be an iliterate boomer and use your brain from time to time. It really isn’t that hard.
They're not uncommon. What doesn't work is providing parallel departures with the same heading. They likely would have collided or at least violated minimum separation.
It's not hard to look up airport codes. There's even a whole website dedicated to it (airportcodes.aero).
you are dealing w/ low IQ people that can't even grasp the concept of two people coming to similar conclusions and really being two different people.
It is advanced physics for some of these people to grasp that SFO has a large number of widebody aircraft and some controller wasn't using all of his tools - including his eyes - each w/ their own limitations - to determine where every airplane was before he issued a new command.
this would not have happened if they went with the plan for new runways which was put on hold in 2021 or so. Then again this is california we are talking about
Thank you for that post, NickW. You should weigh in way more often.
That said, as an ancient former controller, likely mitigating factors could well include (i) combined control positions at that late hour resulting in workload management issues, and (ii) too much reliance on a surface radar screen and not enough oldfashioned looking out the window. Just a wild guess there…you’re in a high tower with lots of glass, so use that tool.