bearfoil

At 185 KIAS, they also have a vertical component of 70 knots, (7000fpm)? So, Ground Speed is affected?

Litebulbs

That would be at 2:11:06 for 325kts, from roughly 400kts between 15-50 seconds before?

Machinbird

Probably obvious to the aviators, but this is only 0.4 incremental g above level flight. Add 0.1 g of cobblestone turbulence into the mix, and it is only 0.3 incremental g from the stick to achieve stall warning.
Depending on the pitch rate, this could be the cause of the initial stall warnings.

FE Hoppy

Litebulbs:

2:10:05 FL350 "sharp fall from 275 kt" 275 at FL350 @ 465TAS
FL375 "speed increased sharply to 215 kt" 215 at FL375 @ 375TAS
2:11:06 ? "15 seconds later" isis 185 kt FL380 @325TAS

Not sure of the significance of TAS but these are my rough calculations.

PJ2

BOAC;
I realize you're being a bit facetious and that you don't like automation per se, or perhaps the way it has been handled. There is plenty of room I think, to grow with technology, while ensuring a robust critique based upon the experience and knowledge we, as airline pilots have gained. In the early days when we did so, Airbus was not the least bit interested in our input and critique, and actively so. But much has changed since the 90s. Today, there is no percentage in dissing for dissing's sake - one must get on with it. In fact, one risks becoming irrelevant in such dogged pursuits but I hasten to add that there remain wonderful airliners out there which are completely conventional and a delight to fly. I've flown them and truly enjoyed them and trusted them. I would have been quite happy to have retired off the DC8, but we're all along for the ride whatever it is.

It's complicated. For a greater foundation regarding the notions of automation one perhaps has to return first to the military where FBW was not about "protection", it was about capability, clearly a very different motivation than airline flying!

An airline managements' greatest mistake as automation developed, is buying into the (manufacturer's) notion that aircraft automation was all about saving money in training, reducing crew complement costs and 'easing' hiring practises, (not paying for 'expensive' experience because "experience was in the software" kind of idea).

The L1011, being a design from a primarily military designer, incorporated some wonderful notions of early automation, (while it's contemporary competitor's design remained a bread-and-butter, pedantic conventional design, very successful commercially and 'nice-to-fly' I believe but ordinary). The L1011 was a dream to learn and to fly. It came out of the chute CATIIIc-ready; the -500 series came with a brilliant FMC system.

Boeing advanced the notion of "automation" through integration of the FMC and autopilot system. While it was Airbus that took automation into new territory with FBW, Boeing's triple-seven incorporated a deeper level of automation than the B767, (but not nearly as much as the new A320), but Boeing has fully embraced automation in the B787, and they have pioneered CFRP as a primary structural approach. I wished there was a B787 Ops Manual floating around so we could get a detailed look at what level of automation exists and how the standard problems of fbw flight controls have been solved.

The problems are not of technology but of psychology, perception, expectation and the strong tendency to "normalization" as a way of viewing the world. "Normal" is a design feature today whereas it was anything but in the early days of aviation and automation. Error-trapping behaviours..."recursiveness" are some of the characteristic behaviours intended to come to terms with inevitable human "error", for nobody makes Mistakes. Mistakes are the result of accepting one's actions as being "in-concert" with perceived circumstances and events. Otherwise, such actions are "intentional", (rule-breaking, etc) and we already know that pilots never set out on a flight to intentionally make a mistake or have an accident.

Automation does not interfere with this error/threat management process but the "veil of automation", which has been greatly lifted over the past twenty years or so, must continue to be lifted, and it is not solely the pilot who must do such lifting. The question, as we all know, is..."What made sense to the person such that they acted in a way that, in our wonderful way of invoking hindsight bias, "caused" the accident. A corollary to that understanding might be, - To what extent can design, by itself, break causal pathways and still not render such serious threats transparent to the crew? Much has already been done by virtue of continuous changes in software as a result of exchanges with the industry. But it almost seems as if we are at the point of needing a paradigm-shift to next steps.

HazelNuts39

Machinbird;

Sorry for not expressing myself more clearly. The point I was making is that according to BEA "the plane started to climb" sometime after 2:10:16, whereas 10 seconds of "0.3 incremental g" would give the plane a ROC of 5800 fpm at that point. This is just to illustrate the difficulty I had with BEA's indications of timing, when constructing my 'tentative' time-history.

gums

Salute!

Well put, PJ.

Seems to this old pioneer with the first completely FBW system that we need to separate the functions that protect the airframe and those that make the plane easy to fly in its intended mission. Our design did both, as with the Airbus, but not to the same degree.

- Gee command increase related to bank angle for a level turn. Same for pitch attitude, as a constant one gee command results in ever-increasing pitch attitude and resulting AoA if speed slows. This makes thing easy for the crew, but I wonder if it reduces "airmanship" and basic skills.

- reduced requirements upon the crew to use trim and train for "unusual" conditions

- control laws that seem to be related to specific portions of the mission, as in "TOGA"

- confusing alarms and cautions when things begin to fail

And the beat goes on.

As all know here by now, I am a big proponent of FBW systems in terms of performance, but not necessarily for making it real easy to get from point A to point B without having basic piloting skills.

Thanks for your views, PJ.

Litebulbs

Just trying to picture at what point it stopped being the thing it was designed to be and turned into an object under the affect of gravity as 34 seconds after the last speed reading, it was passing FL350 at -10000fpm and possibly had no forward momentum at all (10000fpm being 107kts), just dropping vertically, wings generally level at reasonably constant pitch attitude, with engines still at 55% and no normal airflow over the stab.

It just does not make sense, but I am an engineer, not a pilot.

BOAC

PJ - it was not really 'facetious' but a genuine caution against DW's suggestion of putting the a/c straight into Direct when things go awry - it just would not fit with AB OR the airlines psychology, although to a non-AB pilot it seems like a damn fine idea right now.. It is the last phrase of your first sentence that is correct - the 'way'. I LOVE automation - having 'grown up' from an alt hold /wings level A/Plt, work out the drift with Doppler and find the Canary Islands with DR, I progressed to the 737NG. Life became so easy. I had to force myself to throw in a visual arrival here and there, however, rather than let the LNAV/AP take me round an ILS. Loving it does not mean I wish to hand over FLYING to it. How would an AB operator take to crews flying a detail in Direct (however that could be done?) just for 'practice'?

"paradigm-shift" - indeed. My opinions on the training for these systems and the views of management, insurers and accountants have been expressed elsewhere and do not need repeating. If this turns out to be 228 pax dead because the 'automation' confused the hell out of the crew, that would be a crime, whether it be due to inadequate training or 'unexpected' software loops.

DozyWannabe

Really? It's a bit at odds with what my late prof discovered when visiting Toulouse:

True, but Prof Mellor's visit was in 1993, and the implication is that pilots have been in the requirements-gathering loop since the start of the A320 project.

Now, if you're saying that they weren't interested in input outside the pilots that defined their requirements-gathering group, I don't know. But I do remember the hyperbole that was flying around back then - not helped by the Habsheim clusterfunk, so it's understandable that AI (as they were then) were probably on the defensive. Sorting out the wheat from the chaff when it came to line pilot feedback must have been an unforgiving task.

As I said earlier, I'm not sure. I really think it was an honest look at where the state of the art was in aerospace technology and seeing if that couldn't be turned towards making line pilots' lives easier. "Protections" and safety enhancements were just a way of selling the idea to the public - remember that the A320's development passed through the year 1985, which was the worst year on record for jet-era civil aviation in terms of fatal accidents. Even the quadruple-failsafe 747, arguably the safest design flying at the time, was discovered to have a single point of failure in the case of rear pressure bulkhead damage blowing out the tailcone and ripping out all four hydraulic systems. If the mass of cables and hydraulics could be reduced to a failsafe electronically-controlled system and save a significant amount of all-up mass at the same time, the design decision is a no-brainer.

I'm not sure that was ever the case, certainly not after the Mt. St. Odile accident - the saving that Airbus were pushing to airlines came not from reducing training in terms of aircraft handling, but from the fact that with the A32/3/40 series, the conversion training was completely minimal - almost a case of simply changing the numbers the pilots had to work with.

2-crew aircraft had been a fact of life on short-haul since the days of the One-Eleven, DC-9 and Jurassic 737, and the A300 made it possible on long-haul operations as well (shortly followed by the mighty 747-400). I think that's very important to remember.

The "contemporary competitor's" tale is in fact a classic case of cutting corners in design to attain a commercial goal, in this case beating a competitor to market. I've heard many tales of pilots waxing lyrical about how good the DC-10 felt to fly - and this also illustrates the point you make about what constitutes "Normal". In the case of the DC-10, the designers were so confident that what they were doing stuck to tried-and-true methodology that they suffered a failure of imagination - specifically regarding what happens in the case of explosive decompression in an aircraft of that size.

Hmm... not sure. What Boeing did with the 777 was very similar to what the "q-feel" engineers did in the '40s and '50s, which was take the new technology and implement it, but then use that technology to provide a facsimile of what came before. It always puzzled me in the days of the early AvB flamewars that those who said they were suspicious of the Airbus technology simply because of the fact that they did not trust computers were perfectly happy with the B777, which had all the requisite feel of the older airliners - provided by computer-driven force feedback.


This is true, and AI certainly managed to make a rod for their own backs in terms of their relationship with the piloting community based on their product evangelism and sales tactics in the early days. I like to think that they recognised that was a mistake and have learned from it.

IMO it should never have been allowed to fall in the first place. The industry as a whole reacted to the advent of second-generation automation in a completely dysfunctional way. Not to blow my own trumpet, but I've had some really kind and supportive PMs over the last few weeks we've been discussing this subject, and the common thread has always been "Thanks for explaining it in a way I can understand". Understanding it is not difficult - I'm a layman when it comes to aviation, and until I got hold of a copy of HTBJ recently, a lot of the AoA discussion went right over my head, but when it comes to modern FBW systems in general and the Airbus model in particular, I firmly believe that all the dry language and engineering talk in the books can be boiled down to simple rules of logic by anyone of average intelligence.

In this case it's simply:

Normal Law -> All systems go, you can rely upon the protections
Alternate Law -> There's something wrong - protections will try to help you but if you need full authority it's there.
Alternate Law 2 -> You're missing data required for the protections to work, but you've got full pitch and trim authority via your sidestick and thrust authority via the levers.
Direct Law -> Exactly what it says on the tin.

I didn't intend to suggest that, so maybe I miscommunicated. What I was saying is that as part of recurrent training, pilots should be taught how to deal with controlling the aircraft in Alternate Law and Direct Law reversions so that if it happens on the line itwon't come as a shock.

Welsh Wingman

One final post at this stage, only in response to your excellent and thought provoking post, before I bow out.

I certainly don't want to get sucked into a philosophical debate about Airbus v Boeing (which admittedly you are trying to "kill off at birth" at this stage by your reference to the automation on board the B777 and also the impending B787) and for which you are far better qualified.

My "gripe" is the present interface between man and machine, which just feels unsatisfactory, and I could write for days on the disaster of manufacturers selling automation on the basis of crew cost savings.

If you are going to start pushing pilot error for AF447 "flying" into a stall and for not "flying" out of a stall, as per the media rush in some quarters upon publication of the 27 May 2011 BEA statement, flight crews simply have to have had a fair amount of (at least simulator) training at "flying" (i.e. direct law for Airbus pilots) at cruise altitude. How many have, to their own entire satisfaction? How many airlines facilitate small aircraft flying for their pilots? The one thing that I would say about many of my former colleagues who flew the Concordes - they were always flying one small plane or another in their own time e.g. Hutchinson and Cook.

To use a B777 example, to avoid Airbus-specific issues, it's easy enough for a flight crew to deal with an ice-blocked RR Trent engine FOHE leading to a single engine rollback at cruise altitude. The test of the interface is when you have a double engine rollback on final approach for the same reason, and whether that flight crew (having mostly been monitoring systems for 10 hours+) reacts quickly and decisively enough to adjust the flaps to at least get their B777 over the airport perimeter obstacles that could prove catastrophic. I am not convinced that there would invariably have been a "happy" outcome irrespective of flight crew (intra-airline, let alone inter-airline - is not part of the manufacturer agenda to keep-up with aviation demand to an extent that would push traditional pilot training resources beyond breaking point?).

My worry with the level of automation, which undoubtedly has prevented "pilot induced" accidents (you only have to look at the statistics), is where it leaves us when things suddenly go wrong and I mean really go awry e.g. such that the flight envelope degrades and the plane is handed over to the pilots in less than ideal circumstances (at night, with conflicting airspeed data, at cruise altitude, in inclement equatorial weather). That is where AF447 should be treated as a wake-up call, because I am not seeing many posts from current commercial pilots to the effect that modern training and SOPs have flight crews all ready for this eventuality.

I have my doubts about how alert and ready Capt Dubois in the LHS would have been, let alone his F/Os, but will keep my own counsel on this (at least until the full CVR transcript is released and we really know what was really happening in that cockpit, particularly as they were stalled and on their way down before Dubois even returned). We will see.

P.s. several old colleagues were devastated when their precious L-1011s were sold to the RAF, and they would vouch for every sentence in your post! Nothing more ever needs saying about the DC-10 being pressed into service (the Paris crash still angers, even after all these years, because of the earlier Windsor Ontario incident with that cargo door) when the Tristar was crippled by the RB211 induced bankruptcy/nationalisation at RR.

HazelNuts39

According to BEA, "the descent lasted 3 min 30, during which the airplane remained stalled". Since the descent actually lasted 3 min 22, that means the airplane stalled before reaching its apogee of FL380. I use the word apogee because that part of its trajectory was partially ballistic at less than 1g. My tentative trajectory confirms that.

Litebulbs

But when did it stop going forward and just start dropping? The wing stopped working, so the the magic that pushes it up there went, but its inertia was also dispersed, so it fell to earth, no longer being a controllable device, just a horrible loss of life.

It apears to have been recording data all the way down, so it was intact and powered by its main generation systems (not been questioned) in a reasonable stable attitude, just not going forward.

3holelover

You meant to say "extraordinary" there, didn't you? ...c'mon... fess up?

grrr.r.r.r.r.r.rrrrr

HazelNuts39

Litebulbs;

It never stopped going forward, the wing did not stop working, it just became very inefficient, producing more drag than lift.

Litebulbs

Maybe when it was at FLxxx, but not down low and it goes back to GS107 equaling 10000fpm ish.

BJ-ENG

PJ2 & Gums - Nicely put..

Just to add some quotes from two publications well worth the read, and particularly relevant.

Nancy Leveson (Safeware - System safety and computers -1995)

Computers and other automated devices are best at trivial, straightforward tasks. An a priori response must be determined for every situation whereby an algorithm provides predetermined rules and procedures to deal only with the set of conditions that have been foreseen. Not all conditions are foreseeable however, especially those that arise from a combination of events, and even those that can be predicted are programmed by error-prone humans.

Sidney Dekker (The field Guide to understanding Human Error -2006 ) in his preface nicely sums up his 'New View' of human error as follows:

What goes wrong:

Human error is a symptom of trouble deeper inside a system. To explain failure, do not try to find where people went wrong. Instead, find how people's assessment and actions made sense at the time, given the circumstances that surrounded them.

How to make it right:

Complex systems are not basically safe. Complex systems are a trade-offs between multiple irreconcilable goals (safety and efficiency). People have to create safety through practice at all levels of an organisation.

Note: N Leveson is Boeing Professor of Computer Sc at the University of Washington. NASA advisor on the Shuttle software development process.

S Dekker is professor of human factors, School of aviation Lund University Sweden. He is also has experience as a pilot, type trained on DC-9 and A340.

PJ2

BOAC;

'k, that clarifies and d'accord with your historical views as well...shoulda known.

A fbw system is inconsistent with the simplistic notion of the 'big red button'. The B777's reversion has never been discussed here but it does not have such a reversion either, although it is closer to DozyWannabe's notion.

DozyWannabe;

Thanks for engaging the broader (off-topic?) view. I'll respond to the first point and think about the others!

When I initially checked out (and upgraded as captain at the same time) on the A320 in 1992, the airplane still had the "Interim Standard", which did not have VNAV. The only AFS descent regimes were IDLE/OPEN DESCENT, SPD/HDG-VERTICAL SPEED or much more rarely SPD/TRK-FPA, (rarely used, because the descent from cruise was not the place to use a "path-oriented" tool that had not calculated the ToD. One could just hand-fly the descent which I did most of the time including ATs off.

Although the airline wanted its full time use, (from "just after takeoff 'til the end of the landing roll", was the SOP in the first manuals until we simply fought back and got it changed), automation was just a tool, in my back-pocket to use "when-if". That was my (cantankerous?) attitude then, and it didn't change when I retired off the A330/A340. Just to be clear because there's a lot of "I" here, I'm describing what was the case at the time, and not "holding court"...I don't like such behaviours but sometimes one has to speak out of personal experience.

So learning the airplane was a challenge and re-learning it when the "Full Standard" was introduced some time later. We got real FMGECs for the first time and the air was full of "what the hell is it doing now?". But disconnection was the rule because all the guys had flown the Lockheed, Douglas and Boeing equipment and knew, and flew it like a regular airplane and engaged the AP/AT when happy.

During that initial period we saw a lot of "why did they [Airbus] do this?" moments. We received a few visits from AB during the introduction of the airplane into the fleet. At meetings which the entire group of guys (who weren't flying) attended, we provided our feedback from our experience. I don't want to fully describe the engagement and reception but it was dismissive and even arrogant and it was that way over a long period of time. I was so frustrated at not getting answers or an ear that I went to other sources and found someone at the FAA who not only had an ear but participated in the certification work for the airplane in the US, so he knew what he was talking about. He's the guy who landed the airplane in Direct Law on manual THS trim only, just to see. We carried on a rich dialogue for a few years until I lost track but it was our experience as experienced pilots but new on the A320 that Airbus never came to the table to listen.

They may have had their "study group" and I've heard ET talk which I thoroughly enjoyed but it was a long time before AB began to actually listen to the end-users. That changed over the years, for reasons.

The documents posted by PerkyPerkins bear very careful reading, especially the parts about stall recovery and especially the parts regarding simulator work in reproducing states beyond "normal". I highly recommend this reading.

I completely dismissed as more public nonsense and talking without the benefit of actual evidence, the Spiegel article and professor Hutig's comments because, for reasons, I already knew that simulators were GIGO devices but that simulators replicate quite well, normal aircraft and system behaviour. We need to be far, far less credulous when reading what the media has to say about this accident. Shields and Crap Detectors need to be fully deployed!

Got to go.

Lonewolf_50

ventus45, your cross control theory is intriguing.

(Wouldn't the pilots feel the yaw displacement as well as the roll if the rudder was out when the ap went off? It should be in a trimmed position up until the AP goes off, so quite possibly no kick and no roll). What could happen, however, is that as the aircraft changed speeds, the rudder may not be in correct position for the new airspeeds. If you aren't on the pedals, a yaw/roll moment could develop if the aircraft isn't trimming or finding new positions for the rudder by itself.

To that one can factor int a case of mild vertigo for PF, a pilot used to the incredibly smooth balanced flight modern airliners are capable of providing.

The RLU position may not be so hard to explain:

by the time the controls were passed, it is likely that the (originally PNF) now in control pilot used both hands and feet to try to recover, so it's hard to say if the Rudder was displaced many times before impact, or perhaps stuck on high and remained stuck. With the THS remained stuck ... it can lead one to a symmertrical line of inquiry.

PJ2

3holelover;

oops, yes, extra...ordinary, yes, that's what I meant...really.