The two flights involved were UAL601 (an Airbus A320, departing Runway 09 for Vancouver) and UAL437 (an Airbus A320, departing Runway 15L for Mexico City). At the time of the incident (0240Z), weather was rainy, low clouds, and winds south to southeast at 8 knots.
|KIAH 090153Z 17008KT 7SM -RA OVC012 24/23 A2988|
|KIAH 090253Z 16007KT 10SM BKN014 OVC030 24/23 A2989|
The airport was in a southeast flow, with arrivals using the north parallel runways, and departures using the south runways (15R, 15L, and 09).
The controller was apparently working a departure push. He had three runways at his disposal, and was consistently loading each runway with a “line up and wait” aircraft, then launching and immediately reloading with the next “line up and wait” aircraft. Other than timing, the controller’s main focus was to issue initial flight headings that would ensure a diverging traffic flow while pointing each departure toward their destination airport. Flights off of Runway 15R were cleared for takeoff with instructions to “turn right” to headings of 2-2-0 or greater. Flights off of Runway 15L were cleared for takeoff with instructions to “turn left” to headings of 1-3-5 or 1-1-0. Flights to northern destinations were launched off of Runway 09, with an apparent plan to issue instructions to “turn left” to various headings, such as 3-4-0.
This is where the error happened. At around 9:40PM, when he launched UAL601, the controller said “turn right heading 3-4-0″ instead of “turn left heading 3-4-0″. The resulting conflict is illustrated below, in the FlightAware route printouts. The dashed red line approximates the flight plan route. The conflict is shown where the two blue departure tracks cross the added orange lines (representing the other flight). Both pilots reported that their onboard TCAS detected a conflict and issued resolution advisories. Also, note that there was some weather painting on the radar presentation, to the south and west of the airport; these aircraft had climbed into the soup, and were thus unable to provide any visual separation.
The audio was posted at LiveATC.net, and is accessible at this link. All times listed below are minutes:seconds into the audio recording. An example of a correct Runway09 departure clearance, with turns to the left to Runway 15L departures, is at time 03:34. Here is a transcript showing the error, and the mad scramble to prevent a possible collision.:
TWR: “UAL437, Runway15L at WV, Line Up and Wait.”
TWR: “UAL601, tower, are you ready to go?” (confirmed)
TWR: “UAL601, Runway09 Line Up and Wait. Let me know when you’re ready to rock.”
TWR: “UAL601, turn right, right turn heading 3-4-0, Runway09 cleared for takeoff.”
UAL601: “..to 3-4-0, cleared for takeoff Runway09, United 601.”
TWR: “UAL437, turn left heading 1-3-5, Runway 15L cleared for takeoff.” (ack)
TWR: “UAL601, what’s your heading there?”
TWR: “UAL601, just stop your heading, stop the turn right there sir.”
UAL601: “OK, we’re heading 2-1-0 now.”
TWR: “UAL437, turn right heading 1-8-0 sir, and maintain 3,000 there.”
UAL437: “Stopped. Turning right, immediately….”
TWR: “UAL601, stop your turn. Copy? Stop your climb and stop your turn for UAL601.”
UAL601: “United 601 is descending for a radar … resolution.”
TWR: “UAL601, roger that sir, you were given a right to three-forty heading and, uh, that aircraft now he’s at 2,600 feet indicated now sir, he’s not a factor for you, you can continue that right turn to three-sixty please.”
UAL601: “OK, we’ll be clear momentarily, and then we’ll continue the turn.”
TWR: “Alright, when you are ready to go give me a three-sixty heading sir.”
UAL601: “3-6-0 heading.”
TWR: “UAL437, maintain .. actually, continue climb via the (?) departure and give me a 1-3-5 .. actually a one-thirty .. one-twenty-five heading for United 437, one-twenty-five.”
UAL437: “Three for four thousand. We did get a TCAS climb out of that.”
TWR: “UAL437, roger that, and let them know your heading when you go to departure. Have a good flight, sir.”
Was On-the-Job Training (OJT) Being Conducted?
Multiple news reports, including a detailed account by Jeff Pegues at CBS, have noted that unnamed sources say OJT was in progress at both the Houston and Hawaii NMAC incidents. Listening to this audio, and from my perspective as a former air traffic controller (and Whistleblower!), I would have to say that OJT is possible, but if so, it appears the OJT instructor never cut in. Sometimes, in ATC OJT, the instructor gives the trainee a lot of latitude and also wants to see the trainee clean up their mess. This may be the case here, or it may be there was simply no OJT in progress. Listening briefly to the controller an hour later, though, I can say his phraseology was a lot better, much more concise.
The audio recording includes other voices. which appear to be for a second tower controller, working the north parallel runways for arrivals. This is common for LiveATC.net, to gang up multiple frequencies onto one combined recording. Some clearance delivery activity was on the audio, suggesting that this ATC function may have been combined up with a Local Control position at the time of this NMAC incident. The NMAC controller appears busy, but at the same time, his work appears to be very routine: timing launches off of three runways, and issuing headings to fan out the departures. Quite possibly, the tower was minimally staffed and multiple positions were combined so that just two or three controllers could handle the entire workload.
If OJT was in progress, this error is astonishing, because it means the OJT instructor failed in his/her critical duty to monitor the trainee. Just as importantly, though, is the question of whether there was a Supervisor on duty in the tower cab, or a tower cab coordinator or local assist position. All of these positions had some responsibility to ‘save’ the controller error. In any event, the Public needs FAA to produce position logs to show how positions were combined and how many controllers were actually on position; FAA also needs to account for all other controllers who were on duty but not present in the tower cab and assigned to an ATC position.
FAA Turns on the Spin Machine
Interestingly, it appears that the first news of this 5/8/14 incident was first reported a full two weeks later, on 5/23/14. The video clip at USAtoday incorrectly reports it was a Friday night incident; in fact, it was Thursday night, but the official aviation time (UTC, in Greenwich) was 0240 on Friday. This is a common error in reporting ATC incidents.
The initial news stories had lots of detail, as evidenced by the USAtoday video above. The few questions (such as, what the staffing was, or whether OJT was in progress) could have been easily answered by FAA. Instead, FAA has taken a tack that aims to obscure this NMAC/error. One day after the USAtoday video was posted, ‘an FAA official’ made statements that led to the following content within an LA Times article. Amazingly, it implies the pilot strayed off course.
This FAA revisionism, with no mention of the controller error that was so prominent when the story first came out, simply does not cut it. FAA is our aviation safety agency, so we depend on FAA to deliver true safety, not phony spin. So, would one accountable person at FAA PLEASE provide us with the position logs, training records, internal memos, and other hard evidence, to clearly explain what happened at KIAH?