The only reasonable way to incorporate information from a second RA into this process is for miscompare monitoring and if that trips, then it should warn the pilots that autoland is not available.

Well, that's what I would expect. It seems logical to have the AT disconnect when the RA's don't agree (and warn the pilots off course). For autolandings it is required that both AP's are engaged so I thought that is was logical for the AT to require both RA's to agree.It's a pity there's no such thing as a "Cockpit Video Recording" to show exactly what the pilots were doing in the period before the crash. It may answer the question if for example the pilots were too busy briefing the trainee to notice the problem that was developing.

I didn't know that the A/P went into flare mode in Single Ch. approach. I thought that this was only possible with both CMD's engaged. Also, common sense would predict the left RA coupled with CMD A and the right RA with CMD B. Apparentely not though.The aircraft was in CMD B when the incident occurred, and the left RA is reported to be faulty. Maybe an aural callout, something like "Radio Altimeter" will be implemented...I'm sure the investigators could have gathered some information on what the crew was doing at that time, whether briefing the trainee or involved in some other landing procedure. One thing is certain though, they certainly didn't notice the speed bleeding way.As said before, further investigation should reveal more on this tragic accident.

.

.

Well, that's what I would expect. It seems logical to have the AT disconnect when the RA's don't agree (and warn the pilots off course). For autolandings it is required that both AP's are engaged so I thought that is was logical for the AT to require both RA's to agree.It's a pity there's no such thing as a "Cockpit Video Recording" to show exactly what the pilots were doing in the period before the crash. It may answer the question if for example the pilots were too busy briefing the trainee to notice the problem that was developing.

I wanted to add a few things to this discussion based on my understanding of the discussions made in a real pilots forum also. By the way, I AM Turkish-born but have lived in Canada for 35 years now so I can read and understand Turkish.**I want to start with one question first - to the AVSIm forum administrator: Am I allowed to give the link to this forum or is it against general internet forum rules? I am curious - if I can provide the link to the thread, then would the forum administrator here in AVSIM advise me? Thanks...)Apparently if the pilots of this Turkish Airlines flight had DUAL-CHANNEL configuration selected for the auto-land instead of single-channel or flying on "manual" auto-pilot with manual input of air speed, auto-throttle etc..., then the 738 is designed to DISENGAGE auto-pilot if it detects a discrepancy between the left and right radio altimeter (RA) readings (which was the case here). Otherwise, autothrottle will follow ONLY the LEFT (captain's side) RA. This seems to be what happened in this flight. While I am also anxiously awaiting news about why the three people on board were unable to save this plane from stalling and did not seem to notice seriously declining airspeed, I am also concerned with the logic of the auto-pilot design in the first place. ON ANY mode, a soundly designed auto-pilot should always look at ALL the information available and make decisions. This is just plain logic. One does not need to be a professional pilot to figure this part out. Why was not this the case with the 737NG? One possible reason given is cost (again from the pro pilots forum which I read). Apparently the design of such a "logic" or computer programming in the auto-pilot is quite complex and costly. One poster suggested that the airlines lobbied against this and convinced Boeing not to go ahead with such a design which would have made the auto-pilot, essentially "smarter" and much more logical and hence more reliable. Of course, no automation system is perfect and none of my above comments excuse the fact that the drop in air speed was not monitored (or so it seems from the preliminary data) by the pilots. We shall again have to wait for more definite results from the investigation to see what really happened.But I am curious about this auto-pilot design and what members of this forum think about it. It seems like each aviation accident exposes major flaws in the "system". The system is complex and involves many factors: aircraft design, pilot training, weather, pilot error, airlines maintenance practices. I wonder if, after this crash, and the possible legal action being taken by the families of the victims onboard, that there is a possibility that Boeing may be required or at least "recommended" to significantly increase the "fault tolerance" or logic of its auto-pilot systems? Any comments would be much appreciated, especially from aviation professionals. Please note that I am not blaming neither Boeing nor the pilots for this. I am simply pointing out that there are, in fact, TWO SIDES, to this story - one is "possible" pilot error and negligence, and the other is "possible" faulty aircraft automatic systems design, and would like to carry on this discussion from the point of view of the faulty design side of the story. John

Hello,Interesting post.....

Why was not this the case with the 737NG?

In what are differents the others autopilot mounted on the others 737 family ?Are they better than the one of the 737Ng ?The autopilot is obviously a help for the pilots .. but .. the last word for piloting a plane come from (must) the pilot(s) in charge ..Autopilot is a part instrument of the plane .. and as all instrumentations .. he must be know how functionning and monitored ... IMHOAll is perfectible .. and all will fail one day ... and in this case .. the last automation available is the pilot itselfl .. when all is gone .. it's stay a stick .. pedals and power levers ... and a pilot for manage ....Regards.Gus.

Gus, of course I agree with you completely that the pilots must "fly the plane". Once we find out why these pilots were unable to to fly this plane (by disengaging autopilot or at least disengaging the auto-throttle altogether perhaps?), then we'll be able to better understand what happened and why.I was only suggesting that this kind of behaviour by the auto-pilot is "weak". While the pilots should not have ignored the warnings which they apparently received and taken over the plane manually somehow, the first reaction I had (as a mere flightsimmer of course, and not a real-life pilot) was "how on Earth can the auto-pilot be so limited?". The auto-pilot seemed to be working "as designed" but it is the very "design" that I'm questioning here, and not what the pilots should have done etc... If commercial pilots are to rely on auto-pilot (AP), it should at least have features which enable it to use "backup" or "redundant" systems from where it should get its information before it (the AP) makes a final decision. The decision AP made here was very limited and basically wrong, because it received its data from (what seems to be) ONE source - the LEFT RA! There did not seem to be any other source of information that the aircraft's automated flight management systems relied on at all, other than this one left side altimeter. This is what appalls me. Also, do not forget that the visibility that day was also very very poor - something like below 1 mile I believe so the pilots had very poor visual confirmation of the runway, let alone how close the ground was. Of course, I am aware that the pilots seem negligent because of their apparent failure to address a serious drop in air speed. If the pilots were properly monitoring the air speed indicators (which, as far as we know, were NOT malfunctioning), then this accident may not have happened. The final report, when it does come out, will hopefully clarify this situation so we should not speculate too much here before knowing more facts.Of course, there is also the issue of pilot training and the arguments that no AP is fool-proof and again, I agree with these. The AP is not to be relied on to make critical decisions, especially since today's experienced pilots already know that AP has many limitations. It is the pilots who must decide quickly in such situations and stop believing that the AP will do everything correctly. And if it IS proven that the pilots were just distracted by training, or chatting or simply were not paying enough attention to flying the plane and ignoring the warnings they were receiving, then certainly, the majority of the fault will lie with the pilots. But this does not mean that the basic underlying logic design of the automatic flight control systems were not at fault does it?If pilots are to truly trust the AP, then - maybe - more care has to be taken in designing more fault-tolerant APs by aircraft manufacturers. When there is a will, there is a way

This accident sort of reminded me of the old pilot joke:Q: What is the ideal cockpit crew?A: A pilot and a dog

So we can see that simple crew errors and misconceptions can lead to disaster, but although changes are often made to systems and procedures when such things are discovered, it is still something which the crew, and indeed their training and understanding of things, bear a large responsibility for.Al

Hello Al (chock),Of course there is strong evidence so far that the pilots somehow did not check the airspeed and may have been able to avoid this accident. However, my point goes far deeper than just pilot training and this accident alone. Perhaps because I am not in the airline industry that I question like this, but sometimes, it is good to be an "outsider looking in" rather than an "insider looking out". I am the "outsider", looking into a situation where there was a known issue with the altimeter on the pilot's side and this was ignored by the airline. Also, there were 8 previous flights with this defective altimeter where obviously, no fatal accident occurred and we don't know why yet (probably the pilots took care of the fault and handled the situation somehow, if there was an altimeter failure at all in the previous flights).But - and this is my point - WHY would an aircraft manufacturer WAIT for such an accident to occur to re-design their autopilot and auto-throttle so that it IS fault-tolerant? I realize that certain training and aircraft design issues are overlooked since an manufacturer and training staff cannot possibly think of EVERYTHING in advance. But why THIS? Why do aircraft manufacturers design a crucial system like automatic pilot which is highly complex, and NOT design it to be more logical? If the auto-throttle could have been designed so that it disengaged after seeing a major discrepancy in the altimeters of the left and right side, AND NOT put the thrust levels to idle but simply disconnect and give audible warnings, then things may have been much happier for this crew. After reading posts in other forums by 737NG type-rated pilots and other professional pilots, I can't help but think that BECAUSE aircraft systems in todays jetliners ARE so complex, the very fact that they are so complex CAN lead to pilot confusion when under stress and hence to pilot error. So if manufacturers are to make such a complex system, they need to ensure that -a- pilots must be trained on them properly, taking into account all scenarious as much as is humanly possible BUT ALSO ... -b- the AP and AT must be designed to react in a SAFE and logical manner in case of problems like this. Based on what I have learned (and please correct me if I'm wrong) right now, Boeing's design seems to not even "flag" or consider the LEFT RA to have failed at all when it went to -8ft. Hence none of the usual warnings would have been seen by the pilots. That's the findings so far in any case. But the AP makes NO comparison of the readings of the left and right RA and also, apparently, there is altitude indicated by other systems in the aircraft (I think this is the GPS or TCAS, I don't recall now). Since AP was (most likely) NOT in dual-channel configuration, it did not disconnect and blindly obeyed the left RA. This should never have been designed like this when the aircraft was first put into service in the first place, at least in my opinion as an "outsider looking in".Professional pilots are, of course, much more able to comment on the intricacies of this accident. They are the "insiders" who have worked with these systems for years. This is why pilots tend to see the causes of this accident very differently than a mere flight-simmer like myself. But sometimes, people who work IN a system (pilots and other pros) may not admit that there are faults with the system itself. This is not to say that they don't know what these faults are, but realize that there many real-world problems which cause the system not to function as well as it should. This type of problem is more (or seems more) urgent to crazy flightsimmers like myself, who are just "observing" from the comfort of my chair in front of my PC and in my comfy room, with limited knowledge. But, logic is logic and this cannot be denied. I am not "nit-picking" here about a minor detail in the auto-pilot which would not have jeopardized anyone's lives. I am talking about a seriously major issue with AP or AT design which resulted in an airplane completely falsely assuming that it had landed!! I hope you can see the difference here. And please, I am not saying I know more than any pilot neither nor am I blaming any aviation professional in any way. I may sound "naive" in my ideas here but that "naivete" may be what is needed to try to question the very foundations of the system. The "system" being the whole idea of measuring the importance of anything and everything in this world with money and "cost effectiveness" first, and human lives SECOND. So human life, if it is lost, is less important than cost. This is a well-known fact, even admitted by many pilots in the forums I have seen. I have seen numerous postings from pilots that airlines are increasingly being run by people with "business" knowledge and knowledge of "administration", whose priorities are profits and the bottom line first, safety and aviation second. These people do not know or do not really consider as important, the technical side of aviation, aircraft design, pilot training, ATC factors etc... They calculate every decision with (mostly) the financial cost-effectiveness in mind and safety takes a back seat to finances. Deferring the repair of this defective altimeter by Turkish Airlines was a possibly serious error, probably based on time and money and availability of repair crews etc... Boeing not designing the AT to get information from both RAs was also a possibly (note I say possibly, since I am not a pro and do not know all the facts about flying nor about this accident) serious omission. Whether or not there was pressure from lobby groups to not design this fault tolerance I cannot say. But suffice it to say that it was not done.I hope I made some sense here, since my post is quite long. I would not expect all airplanes to fly themselves in full automation but SOME design decisions seem to be questionable and I feel that logically, the AT should not have been designed like this in the 737NG. Again, I don't want to offend anyone, because it is finally the pilots who must take control of a potentially "imperfect" or "illogical" design issue with airplanes and fly the plane (which did not seem to occur here in this case), but some design imperfections are, let's say, "more imperfect" or more urgent, than others.All this said, I just want opinions from people, particularly from who are in the aviation field and are (ideally) 737NG type-rated, hoping that I can learn somethings from them, since this is the purpose of these forums: Just to have a discussion and to learn something in the process.Cheers!JohnJohn

As with most things, it comes down to money, in the sense that airlines cannot afford to take an aircraft out of service for every little component that goes on the blink. So it is important to determine how critical a failed component is before it means the thing cannot take off from a legal standpoint. If any failed component meant an airliner could not take off, then hardly any of them would ever be able to make a flight, so some components are backed up by others, to avoid minor technical faults grounding an aircraft, and some components are not deemed 'critical' if they have failed. Such non-critical failed components are noted in the aircraft's log and may possibly be flagged with a 'non operational' tag or sticker, to advise crews of that component's temporary status whilst it awaits repair. I daresay that of all the pilots you find frequenting AVSIM, there probably isn't a single one who has never flown an aircraft with at least something not working properly on the panel.This means there are basically two types of components on an aircraft in terms of failures: the 'no go' and the 'go' items. If some items are bust, you can still fly safely as long as you are aware of the fault and make allowances for it, providing they are of the 'go' variety. For example, the DU source selector switch on the overhead in a 737 NG changes which source the flat panel monitor displays get their data from, which would tend to make you think that if one data source is not functioning, you could switch to another and be good to go, but that is not the case, if any source is not functioning, it is a 'no go' system failure, and the thing cannot fly legally until it is sorted out, because it means one of the aircraft's critical back up systems is out of action.Other items may control a less critically important function, for example, if a window heat indicator bulb is blown on the overhead (quite common on the 737), then it's not a big deal, and there are in fact a lot of spares carried for things like that. You might be interested to note that if you look at the overhead on a 737, you'll see a tiny white cross on each panel. The purpose of that is to tell you where to tap the panel in order to sort out any loose connections, as that white cross is placed right over where the electrical connector is underneath the panel, which gives you some idea of how things can occasionally go on the blink in aircraft, but can usually be easily sorted out, and are thus not a major concern.Having said that, if for example, the autothrottle wasn't working properly, since you can manually operate the throttles, it would not legally prevent the aircraft from being flown, even though it is a system most pilots would deem as fairly important, so on occasion, common practicality overrides legal issues, and you might get pilots not wanting to fly something that was at least legally perfectly okay.Again, it all comes down to a crew knowing what will and will not work if a known deferred fault is on the aircraft. You are right of course, it would be great if there were always tons of back up systems on aircraft that meant it would not suffer such outages, but there comes a point when you have to determine which of these will keep an aircraft on the ground and which will not. And when that happens, it should be the crew's responsibility to take account of any possible implications if a component is not operating, but can be worked around.You can equate that to your car: If for some reason your car would not go into reverse gear, you could quite easily drive it a thousand miles, you'd just have to remember not to park it facing a wall when you got to your destination. But imagine if your headlights weren't working; on a thousand mile trip, there's a good chance you'd be driving in the dark at some point, so what is not necessarily a critical fault might therefore become one if you did not plan around it. In the case of that 737 with a radar altimeter fault, if it was in the log, the crew should have briefed how it might possibly affect an autolanding approach, and kept an eye on any clues as to things going awry. If they were not trained on the implications of such a fault, they cannot be easily blamed, but if they were, then it was their responsibility.I think we are all aware that most pilot's wages do not really reflect the level of responsibility placed upon them, but when you agree to do a job, you agree to the responsibility of it too. The same is true when you get in a car, nobody is paying you to drive it, but you are responsible for other people's lives when you drive, and you should accept that responsibility when you get behind the wheel. There certainly is pressure on pilots to not refuse flying an aircraft with occasional faults, but that is the job, and if someone does not like that, then they should not take the job. It's not always fair, or wise, but that's the way it is.Al

Hi Al,Have you seen the final report for the crash? I had not searched for it and had not realized it was out in May of this year. It is basically faulting Boeing's design and recommending that it be changed, as well as faulting the pilots and recommending more training for stall situations during approach.I for one found the wording of the Netherlands investigation committee quite clear and blunt with regards to the need to have adequate warnings and design changes for the altimeter / retard/auto-throttle systems on the 737NG and quite glad that they made this recommendation.I think now, money or not, Boeing will have to at least consider some redundancy and warning systems for the altimeter / auto-pilot coupling logic.I am curious about what you and anyone else think about this also.Here is the link to the final report, which is publicly available: http://www.onderzoeksraad.nl/docs/rapporten/Rapport_TA_ENG_web.pdfJohn

Ultimately, it's similar to a lot of aircraft crashes, in that it is not one thing alone which goes wrong and causes the accident, but one thing leading to a cascade of failures which then add up to a crash, for example the notorious Eastern 401 L-1011 crash in the 1970s, where the accident was caused not by anything actually wrong with the aircraft in the sense that it would prevent it being flown and landed safely, but by the crew all messing about with a failed light bulb on the panel and neglecting to realise they were in a descent.In both the L-1011 crash and the more recent accident, the crew were not watching the instruments and simply assuming the aircraft was doing what they expected it to do, rather than monitoring stuff to make certain that it was.What I will say here though, is that it is easy to be judgemental about such things, because I have been guilty of it myself in the past too: I once got distracted on take off by a simple buzzing vibration noise in the panel (probably an inconsequential loose screw, one of many which hold the panel in place), but instead of concentrating on flying the aeroplane, I instead concentrated on trying to figure out what the vibration was, and in the process, I let the airspeed drop to just above the stall speed at a fairly low altitude and in a nose up attitude. On that occasion, which was an aerobatics training flight, I had a very good instructor with me, and he was smart enough to know that letting me get right on the danger point before he reminded me of it, would serve as a valuable lesson which I would never forget, and one which would drill smarter piloting behaviour into me, by scaring the bejeezus out of me about how stupid I had been, for if we had stalled and spun, we'd have been dead, and it would have been my fault. Unfortunately for some, they get a similar lesson but without the benefit of the luck that I had on that occasion to be accompanied by someone who knew better than to get distracted.Al

Ultimately, it's similar to a lot of aircraft crashes, in that it is not one thing alone which goes wrong and causes the accident, but one thing leading to a cascade of failures which then add up to a crash, for example the notorious Eastern 401 L-1011 crash in the 1970s, where the accident was caused not by anything actually wrong with the aircraft in the sense that it would prevent it being flown and landed safely, but by the crew all messing about with a failed light bulb on the panel and neglecting to realise they were in a descent.In both the L-1011 crash and the more recent accident, the crew were not watching the instruments and simply assuming the aircraft was doing what they expected it to do, rather than monitoring stuff to make certain that it was.What I will say here though, is that it is easy to be judgemental about such things, because I have been guilty of it myself in the past too: I once got distracted on take off by a simple buzzing vibration noise in the panel (probably an inconsequential loose screw, one of many which hold the panel in place), but instead of concentrating on flying the aeroplane, I instead concentrated on trying to figure out what the vibration was, and in the process, I let the airspeed drop to just above the stall speed at a fairly low altitude and in a nose up attitude. On that occasion, which was an aerobatics training flight, I had a very good instructor with me, and he was smart enough to know that letting me get right on the danger point before he reminded me of it, would serve as a valuable lesson which I would never forget, and one which would drill smarter piloting behaviour into me, by scaring the bejeezus out of me about how stupid I had been, for if we had stalled and spun, we'd have been dead, and it would have been my fault. Unfortunately for some, they get a similar lesson but without the benefit of the luck that I had on that occasion to be accompanied by someone who knew better than to get distracted.Al

This reply is a bit late but I re-read at length the step-by-step descriptive sections of the events in the crash report again and found that it was quite disturbing that none of the three pilots on board (including the training pilot) thought of monitoring the airspeed and then trying to react to it by simply disengaging the auto-throttle. This would have probably done it. This way, the ILS would be managed by the auto-pilot but airspeed would be managed by the pilot or first-officer. The airspeed was the big problem here. The pilots DID PUSH THE THROTTLES FORWARD in response to drop in air-speed, but apparently they did not push it forward enough. The RETARD mode PUSH THEM BACK AGAIN! The document suggests that A/P would have MAINTAINED the airspeed if the throttles were pushed to near or at maximum, but that pilot intervention would still be required with throttles at full since now, the plane would try to climb due to excessive thrust, which could pitch the nose up dangerously steep. So disengaging A/P altogether or just the A/T would have probably saved the plane. The visibility being very low that day (roughly 1SM or possibly less I think), it would have been wise to ONLY disengage A/T and control the thrust levers manually to recover the aircraft.So you are right of course, in saying that the pilots were distracted but sadly, could not learn their lesson. There was also an important point made in the final report about REPORTING : reporting mechanical problems promptly to engineers/mechanics by pilots. The malfunctioning LEFT (pilot's side) altimeter was apparently not dealt with due to negligent reporting of the issue or it was reported and not fixed. This is why, even though I AM Turkish, I will not be caught dead flying with Turkish Airlines and find the commercials we see on TV (in European channels like Deutche Welle or Euronews) quite hypocritical, making Turkish look so much more professional and safe than they really are...John