SQ AKL tailstrike report released

[InitRef]

From NZ TAIC

"Boeing 747-412 9V-SMT, flight SQ286, tail strike during take-off, Auckland International Airport, 12 March 2003

On Wednesday 12 March 2003, at 1547, flight SQ286, a Boeing 747-412 registered 9V-SMT, started its take-off at Auckland International Airport for a direct 9-hour flight to Singapore. On board were 369 passengers, 17 cabin crew and 3 pilots.

When the captain rotated the aeroplane for lift-off the tail struck the runway and scraped for some 490 metres until the aeroplane became airborne. The tail strike occurred because the rotation speed was 33 knots less than the 163 knots required for the aeroplane weight. The rotation speed had been mistakenly calculated for an aeroplane weighing 100 tonnes less than the actual weight of 9V-SMT. A take-off weight transcription error, which remained undetected, led to the miscalculation of the take-off data, which in turn resulted in a low thrust setting and excessively slow take-off reference speeds. The system defences did not ensure the errors were detected, and the aeroplane flight management system itself did not provide a final defence against mismatched information being programmed into it. During the take-off the aeroplane moved close to the runway edge and the pilots did not respond correctly to a stall warning. Had the aeroplane moved off the runway or stalled a more serious accident could have occurred.

The aeroplane take-off performance was degraded by the inappropriately low thrust and reference speed settings, which compromised the ability of the aeroplane to cope with an engine failure and hence compromised the safety of the aeroplane and its occupants.

Safety recommendations addressing operating procedures and training were made to the operator, and a recommendation concerning the flight management system was made to the aeroplane manufacturer"

http://www.taic.org.nz/aviation/03-003.pdf

[TTail]

Culture, Error and CRM. From a Longhorn paper.

"An essential for effective CRM is a credible organizational policy that recognizes the inevitability of human error and elucidates a credible commitment to error management.

This policy must be built on trust and a non-punitive stance towards error. Rather than seeking to blame those who err, management needs to understand the roots of error in the organization and to develop an array of defences against future recurrences."

Remind me, what happened to this crew?

[akerosid]

The captain was demoted and subsequently left the airline. The (operating) FO was given a severe reprimand and the relief FO "reminded of his responsibilities".

(Or was that a rhetorical question? Personally, I think most airlines would have sacked a pilot involved in something like this. BA/CX certainly would have.)

[Tony_EM]

I have seen more than a few errors in transferring figures from the loadsheet into the FMS. After the first time, I made it a habit to 'hang around' until I could see the FMS calculate the TOW and that it confirmed what my loadsheet showed.

Working for an LHR handling agent, I spent 3 years on the SQ contract. While the SQ crews showed some of the most professional attitude and abilities of all the airlines I handled, they were not infallible. Yet I am still a bit surprised at the group loss of SA here.

In my expereince, errors occurred most often when using manual load and trim sheets as opposed to system generated ones, since the former were infrequently used and the layout would sometimes cause the pilots to use the maximum figures for ZFW and TOW rather than the actuals. While the usual cross checks would catch 'some' typical errors, the pilots would just have to rely on the ability of the load controller to get it right regarding many others, such as ZFW and CG calcs. Whether manual or system generated, they are only as accurate as the figures that are being inserted. The fact that many pilots didn't know their way around manual load and trim sheets too well always concerned me, but not as much as the declining standards of ground ops and load controllers. The relavent lack of experience and lowering training standards of the latter combined with occasional flight crew errors is just an accident waiting to happen IMO.

The error that should have been spotted first by the FO and subsequently by the Captain seems to have occurred due to bad SA and lack of familiarity with the aircraft type and its typical weights. I would urge that pre-departure procedures are changed to allow the load controller to confirm that 'his' figures are faithfully transferred into the FMS and that the loadsheet is only signed when the ATOW on the loadsheet is cross-checked and confirmed by the FMS.

Centralised load planning and loadsheet generation is another aspect that IMO increases the possibility of such errors going undetected before they reach an unsuspecting flight crew.

We had the ultimate blame culture at BMHS and Aviance, which used disciplinary measures to deal with such errors, even when lack of training and too many flights in a given period were obviously the cause. This just forced people to cover up errors and worst of all, to let them go in the hope that nobody would notice. I agree with the sentiments implied abocve, that humans are prone to human errors. I would add that improving procedures that ensure errors such as these have more chance of being spotted by increasing the number of layers that they must pass through.

[None]

How does SQ generate weight & balance and takeoff performance data?

In a previous job we calculated all of the performance and weight & balance data. Currently it is done by computer software, offering many options for takeoff thrust, flap settings, runways, and intersection takeoffs. Both pilots individually verify that the numbers were transferred correctly from the printout.

It is still required to accomplish a manual takeoff data worksheet every year in training. Through experience, however, it has been found that the computer generated performance data are far more accurate, and require much less effort in making configuration decisions.

[Rockhound]

Akerosid maintains that BA and CX would certainly have fired a pilot for committing such an error. Is this really so? I'm no expert but I have grave doubts. Can we hear from a veteran or two of one or other of those airlines on this subject? As an inerested outside observer of the industry, my gut feeling is that, generally, firings result only from incidences of pilot error where fatalities were involved.
Rockhound

[Tony_EM]

I can only describe what happens(ed) at LHR and up until the data is recieved by the pilots.

Projected pax loads are 'in the system' about 24hrs prior to departure, this gave us the available cargo volume (hold space). About 4 hours prior to dep', we were given the approximate cargo/mail/cou weight and we could then calculate an estimated ZFW. The flight plans were generated by QF ops LHR based on the projected load data and we recieved the nominal fuel uplift about -2hrs at the latest. The operating crew could then increase that amount if required, but usually, every flight was full. I'd say more than half the flights ex-LHR were within 1000kg of MTOW, while about 5-10% were 'redispatch', were loads were so high as to prohibit the ideal fuel uplift. In these cases, the crew would have to make a decision for a fuel stop at some point. Most redispatched flights made it with sufficient reserves, but I've known a few that had to make an extra stop.

We were required to 'practice' a manual load and trimsheet once a month to stay proficient, but always using the system as a back-up to ensure accuracy. However, as I mentioned before, most wgt/bal mistakes are made while inputting data, so it makes little difference whether the loadsheet is system or manually generated. There is of course greater chance of error in the subsequent calculation on the manual, but any cargo/baggage container that is given an incorrect weight, will produce an incorrect CG and therefore stab setting. Really not such a big deal with modern aircraft, considering that notional/actual weight differences in baggage and especially passengers can conspire to make big differences, and the fact that CG limits should (and do AFAIK) take such anomalies ino account. Being seriously out of CG limits will only bite you when other events conspire to expose it, such as loss of an engine during climb, fuel system mis-management, etc.

Back to the procedure;

About 1.5 - 2 hrs prior to dep, we would get the actual and final cargo/mail/cou figures where we would then be able to finalise our hold loadplan. Using the checked in pax loads and location at that point, we would be able to say with some confidence where the CG would be, but until checkin closed, there was always the chance of a last minute hold change.

The interesting part is that with manual wgt/bal, the trim sheets use 5 pax cabin bays and 5 cargo hold areas as the basic locations (or moment arms) to calculate CG location (For the 747). The system does the same until the flight closes at check in. At this point, each seat row is given its own CG effect, so if you had a lot of pax sat in the front of any given bay, the CG would jump forward when the check in closed and the system went from 'estimates' to 'actuals'. I believe there are a number of wgt/bal systems that do this, but certainly not all. If the movement of CG was bad enough, we would have to relocate stuff in the holds, or even chop some stuff to get the CG within limits. Again, since most of our SQs were full, there would be little change when the wgt/bal system went active at check in closure (usually 30 mins before dep.

Another point with the manual load and trim sheet procedure, is that the MTOW is reduced by 2000kgs to accomadate the relative innacuracy of the calcualations and higher probability of error. This was really to widen the trim limits at these weights, which become very tight in the last 5000kgs. So MTOW using system would be 396,625kg while using the manual would be 394,625kg.

Based on the report, it would be safe to assume that their loadsheet and trim was system calculated.

We often had FOB figures of over 160,000kgs for these flights so we would rarely reach the MZFW of 244,939kg, which usually meant fuel in the stab which further reduced the trim limits (a specifically defined area on the trim chart). It is my understanding that the 744 will transfer the stab fuel into the centre tank once that equivelant amount of fuel has been burned from the centre. This momentary and severe CG change is accomadated by the TO trim being no further forward than 21% MAC. Since this aircraft had only a small quantity in the centre tank, I seriously doubt that it carried any stab fuel, nor was it close to MTOW. I would therefore estimate the CG limits as being between 11.5%mac and 31.0% mac. At MTOW, these limits can tighten up to 21.5%mac and 23.2% mac.

I will reiterate the point that experience on type and especially familiarity with the typical loads would/should have redflagged the figures being loaded into the FMS. Adherence to proper procedures by all 3 flight crew would almost certainly have identified the error.

Another interesting factor is that by using the Derate 2 procedure, would require an additional stab adjustment which for their weight assumption would have been + 0.66 on the stab. Not much I know, but at the weight and speed they were actually rotating at, there is the possibility that such a stab adjustment could have 'encouraged' the nose to pitch up, though I would say that the pitch tendancies in the respect would be more ascociated with the amount of thrust being generated.

Finally, the reason for such a long post and so much ancillary information is that some errors CAN be spotted by an observant crew that faithfully adhere to procedure, but there are many wgt/bal errors that will be totally invisible to even the most experienced and intuitive of flight crews. While I accept that modern aircraft are much more tolerant of CG and trim errors, my concern lies with the declining standards of training and experience, especially centralised load planning, that seems to spreading throughout the industry. I know in this case the wgt/bal calculation was not an issue, but with examples of crew errors such as this one, and in mind of all the others that I have seen in my time, is anyone else as concerned as I am about the increasing possibility that such 'simple' errors could become more common and harder to spot? I would like to see the industry return to dedicated load controllers that get at least 3 months training with the appropriate authority rather than the poor sods from checkin that get as little as 2 weeks training before they are let loose on big jets without supervision. What I see now are apes that can mimmick the procedure of wgt/bal but they do not have the experience or ability to spot errors before they reach the flightdeck.

[reynoldsno1]

I can understand how small errors may go undetected, but surely you have to be pretty switched off to miss 100 tonnes (100,000 kg or 220,000 lb), a Vr 33kt lower than expected......

[Tony_EM]

I can understand how small errors may go undetected, but surely you have to be pretty switched off to miss 100 tonnes (100,000 kg or 220,000 lb), a Vr 33kt lower than expected......

It is a stinker isn't it?

The point here is that the operating pilots were relatively new to type, yet had 1000's of hours on the A340 which did see such TOW figures regularly. It is a known fact that during times of stress (centre fuel tank error), people revert to their original or predominant training, experience and understanding. It is also concievable that having spotted one mistake, they became slightly oblivious to spotting the other.

transposition errors is a subject that was well covered in 'The Naked Pilot'. Well worth a read for those that find it impossible for highly trained professionals to make seemingly blatant mistakes.

[exeng]

<<Akerosid maintains that BA and CX would certainly have fired a pilot for committing such an error>>

In the early days of BA B.777 ops a similar error was made. (The ZFW was entered into the TOW in the FMC) It did not result in a tailscrape but was close.

The crew were not fired.


Regards
Exeng

[kaikohe76]

I am not in any way pointing the finger at any person involved in this unfortunate incident. However are we not getting to a position of relying far too much on automatic / computer generated data and accepting that data as gospel. In my many years flying, like a lot of us I suppose, I operated electric / glass flight deck jobs as well as the steam driven machines. Yes you accepted the computer / automatic generated data for what it was, but before any figures were calculated and inserted into the systems, you most certainly ran a manual check of the figures, just to be sure.
I do accept that the F/D involved in this incident were recently off the A340 where normal ZFWs / TOWs etc would have been less than the 747.However I would still have expected one of the three guys on the flight deck to have twigged that something was not right with the weights at the time. Easy to be critical and that is not my idea and I suppose we are all human after all.

[TTail]

Would you get fired from your airline?

If you were the captain and this happened to you, what would be the consequences for you personally?

I know incorrect loading of take-off speeds into the FM(G)C has occured on several occasions within my company, however the most severe outcome was a tailscrape of less severity than SQ at AKL.
I also know that the crew involved kept their positions and that the cases were made available to the rest of us as a way to avoid similar encounters in the future. Ofcourse, they were all debriefed by management pilots but for other reasons than to allocate blame.

I was hoping this would be the response from most managements in the "civilized" aviation world...

[Flight Detent]

Hi all,
Well, if I've said it once...................................,
if only they had a Flight Engineer aboard, he does his own check calculations on AUW, AND also checks the TO card!!!

His experience would have recognized the very low TO EPR, not to mention the early correction of the fueling mistake.

This really is criminal..........He would have also aggresively advanced the thrust levers in response to the stall warning, even without the captains sayso,....and......he would have known that the tail will strike the runway when the attitude exceeds 11.5 degrees with the MLG wheels still on the ground ie. the u/c selector gate not retracted!

All these things are basic, and these companies still just put on an extra pilot and somehow think that that will solve all the forseeable problems.....when will they learn that no pilot can replace the experience, professionalism and dedication to the job that the FE reflected each and every flight.
He saved them millions and they didn't even know it!!! - until now!

Just another example, one of very many!!!!

Go on...razz my words if you will...but you cannot deny the truth!

Cheers

[exraaf]

I think most people in this forum are under the false impresson that the Captain was fired.

No, he was NOT fired but demoted from a Captain on expatriate terms to a Co-pilot on local terms. (Essentially saying please leave) He chose to leave the airline on his own. I understand that he is working as a Copilot now and expected to make 340 command soon.

The tragical part is that he is a very good operator and a very cool headed and decent guy, a kind of person who is least likely to make this kind of mistake. The whole incident is an invaluable lesson to all of us in aviation.

[Buster Hyman]

Tony_EM.

I most definately agree with your sentiments about centralisation of load control. It is another example of the systematic problems encountered when bean counters are in control.

You could probably get by with well trained gate staff checking the FMC, as you say, but I believe the most worrying problem is cultural. I'd say we've been in the same boat in different parts of the world, especially when it comes to "foreign" (to us) carriers. Language barriers can cause confusion & mistakes down a phone line & I used to have faith that any mistakes could & would be picked up by the operating crew. This raises yet another issue, whereby we end up depending on the operating crew too much. I guess the last thing they should have to worry about is what's been done on the ground, out of their control. Surely they should have the confidence to rely on the load controller & the ramp t/l, albeit with the usual checks & balances.

As an aside, I wonder how FD crews would react to loadsheets, via ACARS, being fed directly into the FMC? I don't know if it is possible, or feasible for that matter, but if their only job in this area was thorough verification, would it be worthwhile?

Just a thought.

[411A]

Have always, when operating heavy jets, had a look at the takeoff figures, and wondered...is this reasonable?
Sometimes it is not, on more than one occasion.
Common sense should prevail.

Don't be like the SQ crew who taxied to R27 in MEL ( a long time ago), when the weight required R34....take the time for a reasonable look at the probabilities that the numbers just might be wrong.

Of course in helps if you have a really switched on Flight Engineer, but with two crew types, even more important.

[RRAAMJET]

Good grief, 411, a sensible posting with no vitriol...

Rockhound: in the past decade, I do believe CX would have fired a Capt. for this much damage - especially if he was on their "hate" list, which they most definately had. Justifiably, CX are very proud of their widebody professionalism (although sometimes comments by senior managers used to make me wonder....!), and guard it aggressively as part of their good brand image.

Tony EM makes a very good point about knowing some rough performance numbers for your aircraft and wt. I often think some approx. numbers for a wt spread would make a good memory item. It's the same basic airmanship as knowing approximate EPR's for approach settings, etc - don't just rely on the autothrottle to take care of it for you.

About the FE thing...I remember an incident at CX where an L-1011 was mis-fuelled and the FE (who was a top-notch bloke) missed it on his uplift calculation. The #3 nearly flamed out in tank-to-eng mode with a stuck fuel indicator, and the wing was overstressed by massive imbalance.

So it just goes to show, s**t can strike anyone, anywhere, at any time in this business.

:uhoh:

[320DRIVER]

This incidient is a real eye-opener and by no means an " exceptional" occurence. I know of at least one other nearly identical issue resulting in a tailstrike reported at:

http://www.bst.gc.ca/en/reports/air/2002/A02F0069/A02F0069.asp

There was also a similar incident at my present outfit and luckily the CM1 noted the discrepancy (a too low Vr inserted in the MCDU) during the take-off roll and avoided further problems by delaying the rotation until reaching the ballpark figure he had in mind for the given weight.

With the advent of the Less Paper Cockpit concept, are there any airlines moving towards removing the paper take-off speeds bug card, i.e.?

I think it is essential to keep it as it is a good thing to have in the crews' sight who can quickly check the speeds if something is amiss. It will not guard against performance calc errors but would be heplful in case of entry errors in the FMS as it would provide a visual crosscheck at all times.

I look forward to comments/suggestions on the matter.

[planecrazi]

I have heard of a case where the crew entered the figures in an A340 , back to front, V2 first the Vr and then V1. After take-off, the aircraft lifted then tried to climb out at supposedly (V2) V1 plus ten and proposed a steep pitch nose up shortly after airborne. Crew then decided to fly on basic climb out pitch angle of 12 degrees instead, and realised the mistake later. Mistakes happen and will continue. It's up to us to slow down, pay attention, cross check and know your aeroplane.

[Tony_EM]

With todays LCC schedules demanding ever tighter turnarounds, the fact is that many crew MUST accomplish their tasks in short time. The risk of making a mistake is increased not only because of rushed tasks, but that the many figures they deal with can be transposed from the previous sector. Something I have seen often.

I became increasingly frustrated that I could not go a single week without making at least one stupid mistake, however much I tried not to. What I resolved to do was have my own way of systematically checking my work. At times of high stress (delays, confrontations, lack of sleep, just having had my butt chewed by a pilot/mngr/pap/ramp, etc, I would turn double checks into triple checks and force myself to do them slower and more deliberately concentrating on the big stuff first like getting the loadsheet 100% accurate. Eventually, I realised that the trick wasn't to make no mistakes at all, it was to ensure that every mistake was spotted before it reached the flight deck, since they would have far more to consider and deal with themselves. My job was to make their job easier, since theirs is most definitely the harder, but if I couldn't do that, then at least I should not make their job harder or distract from their tasks.

It could be argued that the fuelling error did contribute in some way to them overlooking the procedures that would have identified the bug-card error. This was another lesson I learnt; when things start to go wrong and you are deflected from your usual procedure, then check everything else again.

I'm very happy that this mistake did not harm any individual, yet can still add to other pilots' wisdom when faced with similar circumstances.

[sycamore]

With all the technology and electronics available in modern civil/military a/c, surely a strain-gauge system on the undercarriage could determine the true weight and CofG of the aircraft instantaneously and feed it to the FMC and any input errors would be readily apparent..
Sycamore.

[Rockhound]

RRAAMJET,
Thank you for your insight. Nevertheless, clearly, even the CX of yesteryear would not have taken lightly the decision to fire one of their captains for allowing a tailstrike of the SQ@AKL variety. Akerosid was exaggerating, I'm sure. The subject of punishment and penalties for pilot error interests me and I once tried to get a debate going on PPRune but was unsuccessful, probably because, being regarded as an interloper from outside the shop, I often get frozen out by you professionals. So, what have I got to lose, here goes...
I understand that the QF skipper who botched the landing at BKK in Sept/99 and virtually wrote off an entire 744 continued on as a captain with the company. Fair?
The CAL clowns who took off from a taxiway at right angles to their assigned runway at ANC in Jan/02 were not demoted, as far as I know. Fair?
I don't think it's fair that the handling crew of SQ6 (TPE Oct/00) were dismissed but I do understand that the company had no viable alternative (because fatalities were involved?).
Do you or anyone else wish to comment?
Rockhound

[IBTheseus]

"The CAL clowns who took off from a taxiway at right angles to their assigned runway at ANC in Jan/02 were not demoted, as far as I know. Fair?"

You can be assured that those involved with this incident were all permenately removed from flying duties after they landed from the flight. The FO is still with the company in a non flying, non operational capacity.

CAL has recently been very quick to deal with safety issues like this.

[Algy]

Sycamore,

...such as this...

http://www.craneaerospace.com/new_dev_airweighs.htm

[JW411]

The old Air Bridge Merchantmen (Vickers Vanguards) had just such a system donkeys years ago. I have watched them on many occasions stopping on the taxiway while they check-weighed themselves.

I have often wondered why the idea didn't catch on but I suppose we would have to ask the bean-counters that question.

[sycamore]

Thanks ALGY, unfortunately I can`t open the details, but at least its good to know someone is working on it, as it`s been around for about 100 yrs.!!!

[Tony_EM]

Interesting stuff. So many of the newbies I trained came up with the idea, but I was told by some engineers that such systems were not terribly accurate since the stiffness of the oleo compression action would cause it to stick until the force was great enough to 'jog it'. Ultimate accuracy required zero friction.

But for the purpose of red-flagging serious discrepencies, I suppose it has its merits. I always thought big pressure pads strategically located at holding points near the threshold would have worked, but cost and durability would be difficult as would persuading every airport to install such a system.

I still think well trained and experienced load controllers are the way to go, even if such systems are used to back them up and reduce obvious errors.

[Rockhound]

IB Th,
Thanks for the info. That's reassuring.
Compliments of the season.
Rockhound

[alf5071h]

With increasing technology today’s crew are more remote from the workings of systems (the total system: FMS, aircraft, other crew) and thus they are more vulnerable to any safety threat or error potential. Similarly the crew has less contact with reference manuals and paper work; training has evolved to meet the current needs that have reduced the focus on manual / mental calculation.
Therefore, agreeing with RRAAMJET, it is necessary to seek more defenses to the threats cause by these changes; the defenses are many and varied - CRM, airmanship, personal culture or general treat and error management. In the spirit of seeking a safe new year I offer a few operating tips that kept me scrape free:

Avoid the use of ‘cross fill’ between FMS as it removes the opportunity to check for erroneous entry that can be detected by comparison – “mine is bigger than yours” routine!

Additional knowledge (a requirement of good airmanship) can provide rules of thumb about takeoff weight and speed; in my aircraft these were:
VR approx 2 kt / ton above the basic 100 kts at MZFW, applicable for all three takeoff flap settings.
V1 approx VR-10.
V1 ‘wet’ max VR-20, but not below Vmcg (always know Vmcg for the aircraft type; it does not usually vary with weight).
V2 approx VR+10, but within a range of +7 to +15

Calls and communication are always a threat area. Defenses in this area require well thought out procedures and personal discipline. Beware the ‘parrot speak’ error; parrots cannot evaluate and compare, where as humans can; thus they must take part in a dual checking system. Only call the speed / bug that is indicated on the instrument; this means reading the instrument (and understanding the context), only call what you see. One aid in this area is to state the value of a sped bug when setting it i.e. “V1 set 103 kts”. This communication enables another crew member to detect an error either in his or your calculation / setting. Use this concept in other checks, don’t just call “checked” or “set”; these are meaningless as error detecting checks.

Any other good tips out there?

[Dan Winterland]

The 744 does have the option of the Weight and Balance Computer (WBC) which weighs the aircraft from the pressure sensed in the oleos and also computes C of G. I asked in an earlier thread if Singapore had the WBC option and the answer was in the affirmative. The crew must have therefore ignored the EICAS message.

[kaikohe76]

As I suggested earlier on this topic, electronic and automatic generated data is great, but surely this still requires a final manual check. On the SQ at AKL, it appears all three of the guys on the flight deck did not appreciate or `twig on`, that the figures and V speeds they were about to use were incorrect. I would strongly assert that, had a manual check been completed someone would have noticed the mistake. On my last aircraft the F/O normally calculated the T/O data and speeds etc directly from the tables, the card was then passed to the F/E who ran a complete recalculation on the F/Os figures. Only after it was agreed that both sets of figures matched totally, were these used for the take off. When the take off was very limiting indeed and quite often as a matter of course, the Captain also ran a manual check.This sounds a bit drawn out I know, but in fact on the DC10 the whole two calculations normally took no more than about six minutes. I would say this was time very well spent indeed. Electronic and automatic data for performance calculations must be checked manually, otherwise how do know what you are seeing on the card is correct.

[pedds]

The comment made above regading BA or CX management reaction in a similar situation by sacking those involved I think would be highly unlikely. Having worked for BA for 32 years, my impression is that if those involved were prepared to be as constructive as possible in the investigation, they would lose large chunks of seniority and the Captain his command for a period. The culture was very much one of expecting you to admit to your infallibility and assisting in the prevention of a similar situation in the future. BALPA was very much involved in such serious investigations, one of its more useful functions in it's BA directed activities.

Having said that, as someone else pointed out, there had to have been a massive loss of SA by all three pilots for the various glaring warning signs to have gone unnoticed.

[safetypee]

Some related safety articles

http://www.boeing.com/commercial/aeromagazine/aero_11/erroneous_story.html

http://www.boeing.com/commercial/aeromagazine/aero_04/textonly/tr01txt.html