737-800 has a different wing and longer fuselage at a minimum. To conclude that a simulation on a 737-300 replicates the 737-800 crash is premature, but it is worth investigating. And if found true on the 737-300, or the 737-800, the certification process needs to be looked at.

So far my favorite - "maybe they got distracted doing the radar altimeter failure emergency checklist". :ugh:

Very interesting. When practising stall recovery in the simulator we are usually in manual flight and only trim nose up a limited amount during the deliberate deceleration prior to recovery. I suspect very few B737 pilots have ever practised a stall recovery where the airplane was in-trim at stick shaker activation.

........and a couple of in-trim 733 SIM experiments today?

We have a direct parallel here with the AB crash at PGF on the other thread. As said before, it required the reduction of power once the stall had been corrected before the nose went too high, or a wing down to stop the climb. Neither necessarily practised although both part of the Boeing 'Upset' recovery brief. This sort of stall is NOT something we practice. There is insufficient elevator authority to control the power/pitch couple which is why a limit on minimum trimmed speed is placed on air test stalls. Had these sim guys followed the Boeing training they would not have crashed either.

However, my (and the main) the focus should be on how it got to the stall, plus a salutory reminder to ALL crew to revise upset recovery.

Sorry guys maybe I missed something here but looking at the plot above do we know why the a/c initially went high (at 5 - 6 miles) although on glideslope at just over 6 miles?

Nice and interesting this "stall-recovery-procedure", but...

I was told and teached how to recognise and avoid such an event. (and yes I did practice stall-recovery-procedures in many types)
Sorry, but 3 pairs of eyes could and should have seen/scanned/observed the (negative) speed-trend!
And as an instructor you know this kind of flights and (for me) you are more "on-the-ball".....just my few cents

testpanel, spot on!

Difference between the signs of an approaching stall versus symptoms of the full stall.

First sign of an approaching stall is decreasing airspeed, reduced control effectiveness and (often but not exclusively) high nose attitude.

This is all basic but perhaps we should also recall that whilst on approach in a swept wing jet transport the speed is below minimum drag speed. So if speed decreases, drag increases ergo more speed decrease! So even if approach power is selected you can't ignore a speed trend. And if the speed gets very low then full power will not overcome the high drag so you have to lower the nose to get some speed on but you need some height to so this.

justice
 

The Main objective of an air crash investigationis to find out what happened.
We owe that to the people involved in the crash and future airtravelers and people who are involved in any way.

Therefore the procedures followed and rules to execute the investigation are of vital importance.

There are however special interests involved which might reduce the investigation to a blame game for the sake of money, reputation and politics.

In order to safeguard the finding of truth and justice we need independent investigation authority.

The only way to reach truth and justice is found in the cooperation of investigative bodies and information sharing to the advantage of the other.

This will require coordination between the respective authorities and it will take time.

For the sake of safety for people in the air and on the ground.

I have read a lot of smart comments in the postings on this forum.
However I doubt that any one actually has full access to the data recovered from the crash site, Air traffic control data, ILS integrity monitoring data, Weather data, Crew data, maintenance records etc.

We have to insist that the investigation is carried out in order to find truth and justice for all involved, then, now and in the future.

Obstruction in any way to this process should be considered obstuction to justice and truth.

Lymbo

Absolutely agree with this. I am very concerned about the level of automation complacency that has crept up on some professional pilots over the years. Many airlines now mandate 'maximum use of automation' rather than 'appropriate use of automation' and this doesn't help with the above problem.

It's just interesting that at stick shaker activation on the accident flight, recovery was always going to be a challenge, to put it mildly. I say 'interesting' but the whole thing is a bloody tragedy. Let's hope some very important lessons are learnt by the industry.

Onset of Stall
 

Excuse the for sounding silly...before I get the ridicule of everyone...but in the 737/300/400/500 level D sims I used to do the controls software on, the pilots recognised the onset of stall by the buffet quite a bit before...so the stick shaker was never an issue. Is it different for a 'dirty' stall or are the sims incorrect?...or the the pilots more aware in the sim?

thanx
 

I would also like to thank Safta and his team for the dual sim run and the results you have shared here. As one poster noted it went quiet in here for a few minutes. :eek:

As stated this needs to run on a 738 sim....and I would be very surprised if it has not with the rough data available from OpenATC as far as location, time, height, and speed. Further, part of any aircraft incident investigation is to load the FDR data into a comprable sim and run it...I would suggest Boeing has either done it or is about to.

Lastly regarding the CVR I thought that it has been established recently that public release of the full CVR tapes does not happen now like it did in the past and instead transcripts are made available...or is that a country specific ruling?.

Glueball wrote:
Who said that there was a flag indicating the RA1 failed? The RA1 was simply showing -8 feet with NO FLAG! On the other hand what you say it's true. Who is checking whom with what reference..
Would be nice to have a simple miscompare feature build in RA system..

Posted by lymbo:

I agree with your post but relatives should and deserve to know that as it looks right now, this a/c had: 2 wings, a fuselage, 2 WORKING engines, no problems with primary nor secondary flight-controls, 3 pair of eyes in the flight-deck, etc etc, they "only" had a R.A. problem.....and they crashed (a for the rest complete perfect a/c)

Several people have made reference to Professional pilots ...

Commercial Pilot does NOT EQUAL Professional Pilot (unfortunately)

This thread is positively overflowing with helpful and well-intentioned comments of "adding a few lines of code" here and there, "improved system logic" etc...

The results of such efforts would be:

- More complicated systems. Already, pilots are just barely able to keep track of all details of every system; it has already gone so far that systems descriptions are removed from the pilots' manuals, on the basis that they neither need, nor indeed can, understand how the plane works.

- More design errors. Not only the pilots, but indeed the engineers, can no longer keep track of how the systems work. I'm not tallking about "simple bugs", but system design failures and unpredicted behavior.

- More bizarre failure modes. This is the result of the previous points; when the systems get increasingly complex, they will also fail in more complicated ways. Cascading faults, and redundant, error-monitoring systems switching each other off, resulting in systems shutting down or acting strange that would appear to be unrelated to the original error.

- Increased cockpit confusion. Systems that nobody fully understands, failing in ways that nobody can quite predict, probably even providing "helpful" error messages explaining that they just went offline... Or, equally bad, failing to a degraded mode without telling the pilots, meaning the pilots have no way of knowing what state the aircraft really is in, and therefore how it would react to any further system degradation. Aircraft reacting differently to the same event on different days, depending on whether some particular computer is in "Degraded state 2" or "Degraded state 4" while another computer is in "Secondary Alternate", or not.

- Increased revenue for software design companies.

Thanks, but no thanks! :ok:

RadAlts - post 1596
 

Please read my post 1212. As many have said a radalt is a luxury to help the pilot not replace a function of him/her.
However a system that feeds verifiable erronious information to the system, (by system I mean the pilots and automation that together fly the plane), is ridiculous. if the radalts disagree they should both be discounted and the "system" told to disregard their inputs. if they disagree due to flying over mountains or whatever, then fine, they drop out: they are only a luxury. There are only of use during the landing phase in any event.
I don't have an axe to grind here, I've never flown anything with any form of automation not even engine control - well ok a governor - but what I am sure about is that airliner automation will become more and more sophisticated, the pilots job will be further de-skilled during the 99.99% of the flight, due to the continuous financial pressures on the industry the quality of the intake into the flying schools will gradually drop, the expectations and acceptance of "perfect automation" will increase and the new generation pilots won't know what the hell is going wrong when it does. it's therefore vital that the automation is constantly improved to fail safer, and communicate it's failure to the crew in a simpler way.

I'm not saying it's Boeing's fault; it's just another item that can be improved through experience to make the likelyhood of the cheese holes lining up less likely.

James

Rainboe wrote:

and he misses the point. This was not a random event, not a 'stray bit in a stream of good data'.

The circumstances here were predictable based on the design of the systems. A proper evaluation of the system - as designed - would have revealed the opportunity for a malfunctioning radio altimeter to cause a reduction to idle thrust in flight.

After the final specification stage, bad fortune was all that was necessary to place this link in the chain of causes.

So, yes, those developers (and again, he does not recognise that the tasks he mentions go far beyond mere 'software development') did get it wrong. Until that 737 began its approach at AMS, they were lucky, not expert.

Lymbo,

Another interesting first post. Investigator or wannabe?

You mentioned. If you wish to write so expertly about your chosen topic, you would do well to learn that while 'truth' is at the heart of the topic, 'justice' is for the courts. You, in the Netherlands, probably suffer as much as we do here in France when people confuse justice and safety.

Stick shaker too late
 

safta's experiment shows a big hole in the cheese. If normal recovery won't work, then I submit the stick shaker is coming on too late. Mind you the software revisions to account for trim state bodes to be complex. More likely is a revision to stick shaker recovery procedure when the trim is way back. We also have the Buffalo crash to consider.

Up to a couple months ago, nobody would believe that a crew would just sit there while the A/P wound the trim back until the stick shaker came on.

As far as minding the store, usually PF minds the instruments and PM is dividing his time between looking outside and and crosschecking the instruments. But here we have a teaching situation. Possibly a critical portion of those 100 seconds while the speed was deteriorating were consumed by PM giving teaching such that all attention by the junior pilots was focused upon a PM who was not monitoring.

I used to fly the B737-200 a very long time ago. I seem to remember that one exercise during flt sim refreshers/base checks involved trimming forward (to counter the effects of rising airspeed and thrust induced pitch-up) whilst accelerating, using G/A thrust, from an in-trim, S&L condition at very low airspeed and that during this manoeuvre the tailplane trim jack would stall due high aerodynamic loads generated by the elevator, thus preventing continued nose-down pitch trim operation (a feature of the actual aircraft). To counter this, forward stick pressure would be eased, thus reducing load on the trim jack and hence allowing nose-down trimming to be reapplied. Forward stick pressure would then be resumed to control pitch-up as the aircraft continued to accelerate and the process would have to be repeated a further two or three times, producing a “porpoising” effect, before normal trimming, in concert with forward stick pressure, could be continuously maintained.

Perhaps current and experienced B737 Classic/NG pilots would advise whether my distant memory of this phenomenon is correct and if so is it also a feature of the B737-800? If correct and applicable to the -800, perhaps this unfortunate crew’s finally attempted stall recovery was hampered and ultimately thwarted by a series of accelerated stalls associated with the “porpoising” manoeuvre? I would be very interested to hear the views of experienced B737 pilots on this matter.

PS. If this has been covered previously, please indulge me as I haven’t time for a detailed trawl through the 1660 posts to date.

safta
 

One thousand, two thousand, three thousand, four thousand, five thousand, six thousand, seven thousand, eight thousand. Still seems a long time to allow for the deer in the headlights aspect and the recovery altitude would tend to suggest you did not have anywhere near the same amount of excess speed to bleed off as the accident crew, so being slower you managed to incur enough nose up trim to guarantee an accelerated stall on recovery a lot quicker and higher on the approach than they did.

I wonder if anyone else attempting this could confirm if it is possible to fly the aircraft out of the stall with something less than full power but sufficient to arrest the descent against the nose up trim yet without using so much power that the nose pitches up beyond the capacity of the trim system to maintain a level attitude?

Oh for goodness sake - when will you realise that the above scenario is a non-event and has probably occured thousands of times, due to a myriad of different causes, and has probably never caused an accident before, due to pilots monitoring their goddam thrust settings! We've all experienced A/P and A/T misbehaviour and drop-outs and its our job to manage them.

Anyhoo -Safta - thanks for that sim info- good information, although mindful of the recent A320 crash and a few other incidences I reakon the same holds true for any aircraft with a moving stabilsor trim (i.e any aircraft with a wide speed range). Once you allow things to get so out of hand that you have a stall warning with a bucketful of nose up trim then you are in a close to irrecoverable situation.
As stated by rainboe and others and ignored by all the armchair experts, the absolute key is to not allow the situation to arise to start with. (or not have moving stabalisor trims - which means either you have to cruise at 150kts or take off and land at mach 0.8)

Saftas information is great - but come on guys, 30-40 secs of uncorrected negative speed trend and 8 seconds of shaker? Do we really blame the aircraft if a situation goes unchecked for so long?
I really hope and suspect that there is a deeper cause to this crash, as I find it hard to belive that the 'human factor' can be that off the ball to ignore such a major developing issue for such a long time.

With ground filling the window, I have no doubt that anyone would apply full power. The Boeing recommended upset recovery (which all EU crews should have completed, I believe), says you may need to reduce power in a nose high low speed situation AND bank if necessary to get the nose down. Read the 'TOM stall' thread, and the PGF AB crash thread to see what the pitch/power couple can do.

Only the FDR will tell us what happened in the attempted recovery. It is pointless guessing.

I suspect there will be a lot more upset training mandated!

Post 1509 anyone? Might just have saved the day.

OK, so we mod the stall warning system to take altitude into consideration, so that at low alt. it fires earlier.

I know, we'll take the alt. information from one of the Rad Alts - but it's giving false readings because the plug's corroded or the VSWR is up the creek.

Ah, yes.......

Onset of Stall
 

My last post seems to have been unregistered.

Surely there are more signs of "onset of stall" before the stick shaker comes in...Is this true of a 'dirty' stall? I am talking about airframe buffet and in particular whether the sims reflect this correctly...

On 737/300/40/500 Level D sims I worked on for FAA/CAA approval the pilots only ever went to "onset of buffet" before recovery. Do the simulators reflect this "onset of buffet" with gear and flaps configured for landing correctly?

(edited for typos)

I would tend to agree. In that sense, broadly speaking, this is a very similar type of event to the Spanair MD-80 in Madrid last year where (IIRC) pulling a circuit breaker to isolate a non-critical faulty sensor, as a side effect caused the critical take-off configuration alarm system to become inoperational.

Proper systems design and testing seems to become more and more important, as complexity increases.

Interesting reading from a lot of experienced people but it mustn't be forgotten that it was a preliminary report talking about the RA and AT etc. I think the Dutch investigators stook their necks out divulging that type of info so early.

There could be the odd little suprise or two drop out of the full accident report.

BOAC, I bow to your superior knowledge but is this an "upset recovery"? Surely it's an "Approach to Stall Recovery" according to Boeing? I quote:-

"The following is immediately accomplished at the [I]first[I] indication of stall buffet or stick shaker

PM Advance thrust levers to maximum thrust
Smoothly adjusting pitch attitude to avoid ground contact or obstacles
Level the wings (do not change flaps or landing gear configuration)
Retract the speedbrakes
When ground contact is no longer a factor:
Adjust pitch attitude to accelerate while minimising altitude loss
Return to speed appropriate for the configuration

The Upset Recovery in the Boeing QRH also states inter alia:-

"These techniques assume that the aircraft is not stalled. A stalled condition can exist at any attitude and may be recognised by continuous stick shaker activation accompanied by one or more of the following:-

Buffeting which could be heavy at times
Lack of pitch authority and/or roll control
Inability to arrest descent rate

If the airplane is stalled, recovery from the stall must be accomplished first by applying and maintaining nose down elevator until stall recovery is complete and stick shaker activation ceases."

Badly written post, FFB - I was referring to the ensuing fun in the sim. It was the 'elevator' the sim guys did not have due to the trim, just like the PGF 320 - and probably the BA038 had they had significant engine power to recover from the near stall and the TOM 'incident' - hmm!

Ok all is forgiven!

So what do you suggest?
Millions of $ redesigning and retrofitting a new radalt system to the entire 737NG fleet?
Or reminding pilots that when they are not getting the thrust they need from the autothrottle, they may switch it off and apply thrust manually?

8 seconds of stick shaker!!!!

No wonder you couldn't recover. What was your minimum speed during the manoeuvre?

Sounds like you wanted to prove something and did. But it's not very relevent as I don't recall an 8 second gap being mentioned in the press release. And of course you weren't using an NG.

Did you try recovering from the shaker without delay?

Please, read up on that accident. The pilots actually did not make that mistake. They made a much more basic and fatal one. Just as in this 737 case :(

- the really worrying thing is we appear to be having to 'remind' pilots more and more.

Don't forget to throttle back when you land

Do monitor your speed on finals

Do have some idea of min stall speed before you try it

etc etc.

Something is not right.

I confess to being rather bored by all this poring over and implied criticism of the aircraft systems. It seems that one Radio Altimeter failed - so what?

Commercial pilots are all required to hold an Instrument Rating. This means being able to fly the aircraft solely by reference to instruments. To do this, one develops the ability to scan across the instrumentation - attitude, speed, altitude, heading, back to attitude, etc. Especially on any approach - be it manual or automatic - a professional pilot should maintain that scan. This where the first indication of something unusual will be apparent. If it is manual flying, a manual correction is made. If it is autopilot flight, the automatics may need adjusting or, if there is a serious discrepancy, even disconnecting. Either way, a necessary correction to the flight path must be effected. This is what professional piloting is all about.

The idea that the pilot is simply a bystander to the electronic wizardry and only gets involved when the automatics provide a warning is absolute nonsense. There may be need for software criticism in order to improve its performance but do not let us overlook the fact that the pilot is in charge of the aircraft, whatever the automatics are doing - or trying to do.

Piloting comes first - automatics may help the pilot but are subsidiary to his/her primary function. Pilots are responsible for flying the aircraft. They may indeed use the automatics but then it is essential that they should watch the automatics to ensure they deliver the required aircraft flight path and performance. Failing to do this is nothing less than professional negligence.


JD
:)

1677 Jumbo
 

Jumbo

Maybe it was professional negligence, but assuming nothing else comes out in the report regarding the throttles being closed, do you not agree that in this particular accident, the plane would not have crashed and lives lost had the duff radalt not caused the throttles to be backed off?

For some reason, incompetance, distraction, whatever, the crew didn't notice the loss of airspeed. sadly the cheese holes lined up: they wouldn't have done (in this case) had the radalt performed correctly.

Improving safety is surely about reducing the holes in the cheese?

James

Jumbo Driver and Rainboe - I agree. The Wright brothers didn't have a Radio Altimeter. either.

If the wings were undamaged and contained some fuel, and the engines working and still controllable to some degree, that was all the crew needed to land on an airfield that they could see. Anything else was a bonus.

There may be very valid reasons why the crew got themselves into the position that they did, time will tell, but trying to blame electronics and computer systems is like the Supermarket clerk being unable to work out my change because " the computer is down"

Must go, my mobile phone has apparently received a text, and I need to find a passing 5 year old to work it for me.

757 Driver
 

Well this time it did lead to an accident and I fear this sort of explanation in the light of public awareness of what triggered the accident would not go down well with certain groups, say your passengers for instance. Yes the crew would appear to have been incompetent, but having exposed a new hole in the cheese to the scrutiny of investigators and the traveling public alike, it will be incumbent on the powers that be to institute a technical fix in addition to any remedial training requirements both specific to TLH and any more general requirements they may come up with. That's just the way the world works.

In other words incidents like a malfunctioning RA#1 that trips the A/T into an un-commanded mode at an inopportune moment is a benign failure that doesn't concern anyone outside the industry, until such time as it is linked with an accident like this one. Then it's no longer benign.

Here's a couple of things for people to consider regarding rad alts and autopliots.
Rad alts are primarly autopilot sensors not primary flight instruments. There is one rad alt per autopilot which is used in the landing phase.
To meet the design requirements for autoland with multiple autopilots, you must have the autopilots operating independantly. Autopilots will be powered from separate electrical busses, utilize separate hydraulic systems and receive information from separate sensors. Because of this requirement, it is not possible to use rad alt info from another side to replace a bad sensor.
Rad alts monitor the aircraft height above ground. Its operates between 0 and 2500feet. Height is determined by bouncing raido signals off the ground. The frequency is in the 4.3Ghz range. Above 2500 feet, it is not used. Each rad alt has 2 antennae. One transmit and one receive. On most of the modern Boeings, the rad alt R/T and both antennae are monitored for faults and will supply fail flags for a hard fault. Intermittent faults are much harder to detect and may not show as a failure.
I spent most of my 30 years in aircraft maintenance working on B747s. I never saw a rad alt stop one from flying. I don't know of any aircraft that doesn't have an MEL to cover the dispatch of an aircraft with an inop rad alt.

No redesign of the RA. Yet.

So far, this occurrence is unique and most probably fulfils the "acceptable failure risk" criteria.

Air transport is still a very safe form of travel. This nowadays tends to be forgotten whenever someone dents an airplane.

If the number of occurrences becomes excessive, then consider a redesign, or at least a revision of the certification criteria - for example, as actually happened some years ago for birdstrike protection.

In my opinion, the issue that has occurred too many times recently is allowing the aircraft to lose safe flying speed for any reason. This accident and Buffalo are recent fatal examples. I still suggest the solution in post #1234 as a cost effective technical measure, if any such is implemented. It should in any case be slightly more cost effective than redesigning and retrofitting radalts, IMHO.

Very well said stunt pilot as i alluded to some pages ago. I think i said let's not get sidetracked. And still we hear about the intricacies of Rad Alts etc... :ugh:

Put recents incidents together and formulate a picture. See threads A320 Perpignan, Tom Stall, etc... Build your own picture. But please keep an open mind until the full report is out. There is far more to this than first meets the eye...