Many posts are searching for ‘cause’, which unfortunately can lead quickly to blame.
The alternative is to consider the many possible contributing factors to this accident, irrespective of cause. Each contributor is a risk, similar to those encountered in daily operations; these risks have to be managed to assure safety.
Thus the lessons which might be learnt from this accident will probably come from the risk assessments associated with the decision to land and the handing of the flight thereafter. We should focus the discussion on the components of the risk assessments, the possible contributing factors, and how these are managed; the what if’s – constructive speculation without seeking blame.

Risk assessment has dependencies in situation awareness and guidance from SOPs and company safety assessments; the classic individual and systematic views of accidents.
If an operator has overlooked pertinent issues such as wet / tailwind / non grooved / rubber contaminated runway surface, there should still be a later line of defence via the crew – people can create safety.
What this accident might be telling us, like similar recent overruns, is that both the system (organisation) and individual risk management processes are weak; we, the industry and individuals, appear to be taking too many risks.

just read that a jamaican authority has said the plane landed 4000' down the runway with a tailwind of 17mph, quoting a tailwind limit of 16 mph...and that the plane bounced after first touchdown.

also that the pilot was using the HUD.

it will be interesting to find out if the HUD lead him astray...like those automobile GPS units that lead you down a dirt road to hell.

I love this quote from the above pasted article..

"The moral of the story is you don't land an aircraft in a thunderstorm with a dynamic weather environment like that present over the runway," he said.

Question posed in slot #289, from PJ2:Suspect the initial "float" estimate was returned from the manufacturer's Runway Track Analysis Program:

_TRAJECTORY RECONSTRUCTION TOOL FOR INVESTIGATION OF RUNWAY OVERRUNS_
www.cta-dlr2009.ita.br/Proceedings/PDF/59024.pdf

protectthehornet,

Just a small point (1mph) but you transposed the figures from the article. Wind on touchdown reported as 16mph, limit 17mph (or 15kts to those of us that can speak aeronautical units) not vice-versa as you stated.

IGh;

Many thanks, - Though retired off the A340/A330, I still do some flight data analysis so am familiar with the distance calculations as we use them all the time to analyze our own trends. I was more thinking of how this information made it into the public realm without a coordinated news release. I consider it important only because a newspaper or an agency such as the Jamaican Authority can say anything it wishes and I was looking for the original sources of the information. I frankly have no reason to doubt the information, (but also have no reason to believe it), but until we know that this is from the recorders, we're only speculating. I know you're aware of all this...just responding with clarifications.

PJ2

forgive my transposition, thanks for correcting it.

earlier we were talking about a 10 KNOT tailwind componenet limit....and 16 mph is 13.6 knots

wondering if american had a 15 knot limit at this airport.?

and the Wrights were right...even if you have to go to North Carolina...takeoff and land into the wind.

Here is the official report from the JCAA..

Instrument Meteorological Conditions prevailed in the area and heavy rain was reported at the airport at the time of the accident. The aircraft was flying on an Instrument Flight Rules Flight Plan.

The crew reported that after descending through the cloud cover, they made visual contact with the runway at between 1000 feet and 700 feet above ground level. According to the Flight Data Recorder (FDR), the aircraft was traveling at the Vref (landing) airspeed of 148 knots, with a groundspeed of 162 knots, i.e. with a tailwind component of 14 knots, when the wheels made initial contact at about 4,000 feet down the 8,900-foot runway. The FDR further indicated that the aircraft bounced once, then settled onto the runway; the autobrakes then engaged, and reverse thrust and the spoilers were deployed.

American Airlines flight 331 crash: statement from investigators - JamaicaObserver.com

waves-dubai, thanks for the link to the story.

Given this new information, (in which there are no surprises, only confirmations), it will be interesting to see what has been showing up in American Airlines' FOQA data for this aircraft type and their other fleet types in terms of long landings. More importantly, I should think, is, what they did in response to any such trend if such events were in their data.

From AvHerald
 

Accident: American B738 at Kingston on Dec 22nd 2009, overran runway on landing

By Simon Hradecky, created Wednesday, Jan 6th 2010 19:36Z, last updated Wednesday, Jan 6th 2010 19:39ZJamaica's Civil Aviation Authority reported first results of the investigation in a press conference on Jan 6th 2010.

Air Traffic Control offered a circling approach to runway 30 in order to land in a headwind, the crew however declined and elected to land on runway 12 with a tailwind of 14 knots.

The crew gained visual contact with the runway between 1000 and 700 feet AGL and were not concerned at all with the approach.

The airplane touched down about 4100 feet (1250 meters) down the 8900 feet (2713 meters) long runway 12 at 161 knots ground speed in heavy rain and gusting winds and bounced before settling on the runway. The airplane overran the end of the runway at a ground speed of 62 knots.

The investigation still tries to determine why the airplane landed long. Analysis of cockpit voice and flight data recorders is still ongoing. The investigation also focusses on the actual weather at the time of the accident, runway friction tests have been scheduled to start Jan 6th. Tyres, brakes and several other parts are being sent to the US for closer examination.

From the same Jamaica Observer article in post #295:

That about covers it.

Flight Safety;

Re, "That about covers it", yes, I think so.

So the question now is, "Why?" Why 2 similar accidents at the same carrier?

Where are the root pathways to this over-run and are they different than Little Rock, and for that matter, AF at Toronto? Long landings are not unique to AA, so what's different here, if anything, and is there anything new to learn or are there old things to learn that weren't learnt before, or are there old things to re-learn because people and processes have changed? Is this organizational, or crew-related? Under SMS and the FAA audit processes, these are some of the key questions.

PJ2,

You and others keep making references to the AA flight 1420 Little Rock MD82 accident on June 1, 1999, as a possible comparison. This accident had almost nothing in common with the current accident, except that both aircraft ran off the end of the runway. Just read the details of the flight 1420 accident and findings, and you'll see that they have almost nothing in common, from a pilot decision point of view. My point is that the lessons to be learned are entirely different for these 2 events.

I don't dispute this, but I'm always willing to learn.

Perhaps some one can cite the differences in the lessons to be learned.

On the other hand maybe some others might cite the similarities in the same lesson needed to be re-learned.

Blame is not my game, but lessons are.

lomapaseo;
As always, mine too. So here goes an attempt:

Flight Safety;

Out of respect for the process, I don't wish to get into a pointing-to-the-crew discussion here because we don't know enough about the Kingston accident yet.

That said, you are correct regarding some differences. 1420 did not land long and crosswind was a factor. For 1420, the spoilers were not armed and did not deploy, therefore normal deceleration through the available (normal) braking system was not available.

From the comments available on AA331, the braking, spoiler and reverse systems functioned as designed/intended. Crosswind was not a factor. We can speculate but have no statements on the tailwind experienced by AA331.

From what is known thus far, the decisions regarding weather however, may offer some similarities. Statements in the AA1420 NTSB Report indicate that the approach was continued when, collectively, factors were pointing towards a go-around. While visibility was substantially better at Kingston, it wasn't VMC and rain was indicated as "heavy". Examination of the decision to use 12 vice 30 has not been released to the public so we don't know these details nor do we know the reasons for the 4100ft float.

As has been pointed out here, a late go-around is possible and is part of the CATIII certification providing the engines have not been reduced towards idle for the landing. As the document from Transport Canada states, once a "low-energy state" has been entered, any attempt at a go-around is a high risk, undemonstrated manoeuvre.

We do not know yet whether a go-around from the runway, (before the selection of reverse and with N1's not yet at idle) was possible and indeed, this belongs wholly in the area of uninformed speculation.

Some differences, some similiarities. We know that AA1420 got high on the approach; we do not know if AA331 "got high" or whether the distance covered in the air was just above the runway in a long float or still in a slight descent, the threshold having been crossed somewhat above 50ft - none of this we know.

I hope this clarifies a number of points, Flight Safety. If there is further, we should discuss it for the lessons, as you say.

PJ2

Thanks to PJ2 for more perspective.

I take it as some important differences between the earlier event and this one, coupled with the need to continue the investigation to resolve the unkowns.

I'll stand by for the new lessons learned and the possibility of re-inforcing some from the AA Little Rock event.

There's an unresolved question here: How does an aircraft come to touch down 4,000' down a runway at Vref speed? If the approach, which was an ILS, was stable to the flare (and there's no reference that it wasn't) then the aircraft must have floated for ~10 seconds (2800'/273.6 fps). How does it do that and remain at Vref speed until it touched down?

The choices seem to be either, the approach was not vertically stable in the late stages, the approach speed was higher than Vref, power was not initially retarded at the flare, or there were changes in windspeed and direction sufficient to offset the energy that should have been dissipating during the float.

Any one of those explanations requires more information than what has been released so far.

ELAC

Late gust of heavy wind from the rear strike anyone's fancy?

PJ2

It seems most overruns come from tunnel vision on the pilot's part. At some point, the pilots are no longer open to going around, the low energy state means a go-around maybe more dangerous than a late touchdown and trying to stop. And after thousands of landings on wet runways that work out, touching down looks a better bet. Sometimes it isn't.

I think training is sometimes to blame. We encourage landings, not go-arounds, salvaging the approach is too often tolerated by instructors. After a couple of overrun accidents in the USAF, we had to, in multi-motor planes, brief touchdown point, landing distance, runway conditions. Dry day, no issues, it was common to have to tell a pilot, "Go Around" when they passed the distance remaining that they had briefed. Once in the flare, landing becomes a predestined ending to the flight. You do learn a thing or two about low-energy, crosswind go-arounds teaching LTs landings in a Galaxy.

All three of these accidents (2 AA, 1 AF) could have easily gone around, held or circled, but once mentally committed to landing, tunnel vision took over.

GF