Sokay, just a nav database error.

[ktcanuck]

SFO-Bound Pilot: 'Waiting For The Airplane To Hit'

Reporting Anna Werner

SAN FRANCISCO (CBS 5) ―

It's a flight that the pilot Mark Taylor said still gives him nightmares.

January 15, 2008. United Flight 901 is coming in for a landing at San Francisco International Airport. Behind the controls is Taylor, an experienced pilot who's flown thousands of hours in 747-400's for United Airlines.

He said the crew was making the last leg of an international trip, an 11-hour return from Frankfurt, with more than 300 people on board. "It was basically uneventful," he said.

But as the plane descended through the fog under the guidance of its navigation system, Captain Al Langelaar, sitting in the cockpit next to Taylor, said: "Something's not right, go around." Taylor said he pulled up on the controls right away, but was puzzled, until he looked out the window.

"I saw nothing but water," he said. Taylor believed the plane was going to crash into San Francisco Bay.

"At that instant, that instant I was waiting for the airplane to hit," he said.

Fortunately, the 747-400 began to climb.

It's not supposed to happen in those airplanes; sophisticated onboard computers are designed to allow them to practically fly themselves.

"Today's modern jetliner will fly you to your destination, land you on your runway, and roll to a stop," said Jim Curtis, a 747 trainer. Curtis said pilots are taught to rely on those computers to take them to their destinations. "It will fly to the correct approach," he said.

And to be certain the programming is accurate, information for every airport in the world is updated every 28 days. So Taylor said on that January afternoon, the information they'd entered looked correct. But while they thought they were headed for the runway, they were actually headed for the Bay.

Taylor said the crew attempted a second time to land at SFO but had the same problem with the navigation system. At that point, he said, the plane had only minimum fuel so the crew was forced to divert to Oakland, where the weather was clear.

They landed safely. But Taylor still didn't know why it happened. As he describes his experience at the time, "I'm in a state of shock going, how did we get here? And it was, it was kind of disorienting. Go from this to that point? How did we get here? This shouldn't have happened."

But United documents obtained by CBS 5 Investigates suggest the answer. They indicate that Honeywell, United's contractor, made an error in its database for the navigation system.

Passengers like Scott Siera never knew. "Nothing really felt out of the ordinary until maybe less than a mile from the runway," he said. "I was looking out the window and feeling that we were awfully close to the water."

At the time, he wrote a post about his experience on a frequent flyer site. And after CBS 5 Investigates told him what Taylor had said, he told us: "I would definitely commend the pilot for gunning the engines and getting us out of that situation."

But Taylor never got a commendation from United. The company declined an interview with CBS 5 but in a statement, confirmed there had been an "anomaly in the navigation system information", a problem the company said has been corrected. A United spokesperson disputes Taylor's claim of a near-crash. (United's full statement is posted below.)

But in the company's documents, a United lead engineer wrote that the database problem had "serious flight safety implications."

Contractor Honeywell International also declined an on-camera interview but said in a statement that, "Honeywell stands behind its products and firmly believes they are safe". (Honeywell's full statement is posted below.)

However, in an email from January 18, 2008, which was part of the United documents obtained by CBS 5 Investigates, a Honeywell manager admitted to United that an "error" in Honeywell's database caused the problem and said, "we apologize for any inconvenience."

Mark Taylor never received that email but said the experience haunts him.

"I have a recurring nightmare that is just a simple continuous running loop, of just looking out the window," said Taylor, who said he's been diagnosed with post-traumatic stress disorder as a result of his near-death experience. "That image of looking out the window and seeing nothing but waves and water. I see it when I'm awake, I see it when I'm asleep. I carry that with me all the time."

As a result of his PTSD diagnosis, Taylor can no longer work as a pilot. He has filed a lawsuit against Honeywell, which the company responded to in its statement below.

Honeywell statement to CBS 5:

"Honeywell stands behind its products and firmly believes they are safe and did not cause Mr. Taylor's claimed emotional distress. Honeywell will continue to defend itself aggressively and looks forward to vindication in court."

United statement to CBS 5:

"United did a complete investigation of this flight. The crew determined that there was an anomaly in the navigation system information, after following the standard procedure of verifying their location. The crew was in Air Traffic Control communication and radar contact throughout their entire approach to SFO, and ultimately determined the safest action was to divert to Oakland. The flight landed safely and without incident. We have since worked with the manufacturer and with our pilots to ensure we corrected the issue."

[200hr Wonder]

Our SOPs have us check the lat and long of each fix and or heading/milage to ensure that they are in the correct place as part of verifying an approach. This is of course even more important to do the first time flying after a database update. I have my self found errors and as such did not conduct the RNAV approach and instead opted for a traditional nav aid approach. Always double check, errors are not that uncommon!

[scopiton]

Few years ago I recall listening to AC pilots stating that they discovered FMS had not the proper data when they tried to set an approach to frankfurt and had to compute it manually.

[Cobra64]

If it was an error from the database provider (Honeywell) wouldn't more than one plane have been affected? Surely United flies more than once a day into KSFO with the 747. Also, if they were so close to the ocean, was that a TAWS warning? Was ATC watching them fly into the ocean? Lots of unanswered questions here.

[cdnpilot77]

If it was an error from the database provider (Honeywell) wouldn't more than one plane have been affected? Surely United flies more than once a day into KSFO with the 747. Also, if they were so close to the ocean, was that a TAWS warning? Was ATC watching them fly into the ocean? Lots of unanswered questions here.

I agree, however having never flown an airplane with FMS or TAWS I am not entirely familiar with their limitations. It almost sounds as though Mr. Taylor just got spooked and replays it in his mind remembering it being a lot worse than it was. It's just like when I got in a hockey fight at 16 and hit the other guy once in the side of the helmet...but the story now goes that I knocked him out cold with 3 punches (not a true story).

My point is that stories get embellished to prove a point. It's human nature because we like to believe that events are more significant than they are. Maybe this guy just didn't want to fly anymore and wanted the money. Frivolous lawsuit in the US?? No way?!?!

However, there is a very good chance that I am way off base here and it is totally legit...either way, keep your stick on the ice and try to get the helmet off before you punch, thats smarts!

[1300+]

The first couple cycles of the Universal FMS database this year, had issues. All of the coordinates would check out, but in the missed approach (where there was a track to fix), they wouldn't cycle correctly. Upon further review, although all the coordinates were correct, the distance between the mawp and the first waypoint of the missed approach was 0.0. We missed due to low fog, and were able to maneuver visually, but it makes you wonder how these things happen. As for checking coordinates or track/distance..I check both now.

[ahramin]

Always check distance and bearing. Have the map out. Tune in traditional navaids. Look out the window.

One good thing though, at least this person is no longer flying.

[55+]

Sounds like a Jeppesen AIRINC coding error.............

[G.N. Thompson]

This was a much better outcome than Mt. Erebus long long ago. Double check everything.

[ahramin]

I don't think the Mt. Erebus crew had much to double check with. They had no way of knowing the waypoint was incorrect.

[bcflyer]

I'm kinda curious what kind of approach they were shooting into SFO.. Everytime I've been there in bad weather its been a good old fashioned ILS. We use the FMS for the arrival but after that get vectored to an ILS.. If they were doing an RNAV approach the first time and it obviously didn't work why not just revert to old school ILS... I agree lots of unanswered questions here.

[PilotFlying]

Our SOPs have us check the lat and long of each fix and or heading/milage to ensure that they are in the correct place as part of verifying an approach.

Your SOPs say so because the regs say so. Every pilot flying a Canadian-registered aircraft is required, by the CARs, to perform this cross-check prior to conducting a GNSS/RNAV approach.

Cheers,

[ahramin]

How about posting that CAR PilotFlying.

[PilotFlying]

How about posting that CAR PilotFlying.

I can give you the AIM reference off hand, but will have to dig around the CARs to find it.

AIM 3.16.18 Avionics Databases, 2nd paragraph:

It is good practice to verify that retrieved data is correct, and
it is mandatory to do so for approach data. Verification can
be accomplished either by checking waypoint co-ordinates
or by checking bearings and distances between waypoints
against charts.

Cheers,

[iflyforpie]

I remember this one from a few years back...

http://www.tc.gc.ca/civilaviation/publi ... 04/565.htm

[ahramin]

I remember that one iflyforpie. I could understand someone getting PTSD over that one. It was a total fluke that they didn't just disappear into the Pacific Ocean.