Come now, john, the authors put that paper on the internet, and we all know everything on the internet is true!

j s, it’s a little harsh to state that the authors are ‘wrong’; they have an opinion which should be respected.
However, I agree that the article is of little value because it appears to be biased by hindsight and tailored towards a particular agenda.
A meaningful discussion requires an understanding of the design principles and objectives of many years ago, the assumptions made about piloting skills, education, training, and procedures. These should be compared with today’s standards and expectations.
Many of the ‘problems’ stated in recent times stem for this gap in understanding; - different points of view, different operational environment, training standards, social expectations, etc.

j s, vilas, :ok:

I disagree, not every opinion necessarily merits respect.

While john_smith makes an (IMO quite correct) one-word response, it might be worth (yet again) trying to parse what's being said. For one thing the article posted (which is a blog post - not a peer-reviewed paper) is very hit and miss with its use of sources (including an article from 1987 - a year before the A320 even went into service). For another, it makes assumptions about the reasoning behind design decisions which are at best speculative.

Furthermore it makes this statement:
which is fundamentally incorrect. In an aircraft with direct cable control, the aircraft will respond based on whichever of the two inputs is physically dominant (which is not necessarily the right thing to do), and in more complex modern aircraft with apparently "traditional" controls, the response may not be immediately obvious. An example would be the all-hydraulic B767 controls of EgyptAir 990. As it turned out, with one crew member pushing forward on the yoke and the other pulling back, the result was a split elevator condition which exacerbated the aircraft's departure from controlled flight.

In fact, a lot of the potential objections made by a subset of the piloting community around the launch of the A320 were based on the (as it turns out, incorrect) assumption on the part of those pilots that they knew what the response of their current aircraft types would be when presented with control-based edge cases.

The Boeing FBW system in the B777 retained (or to be more precise, reimplemented) the tactile feedback, but this came at an engineering cost - namely that the hardware and software required to provide this feedback was as complex again as the entirety of the Airbus EFCS. The sole reason the Boeing system requires a physical bypass control and the Airbus system does not relates solely to the potential scenario whereby the feedback system itself develops a problem.

The fundamental principle of the Airbus design involved confronting a basic truth that had been around for some time - namely that with hydraulically-assisted or all-hydraulic controls, the tactile feedback the pilots were getting was itself an electro-mechanical simulation, and as such open to potential failures and complexities that weren't really necessary.

As a postscript, it's worth noting that the naysayers the blog post refers to were predicting a slew of accidents caused by the change in control system philosophy - and to this date there have been precisely zero accidents attributed to the change.

flight path stable fbw is inherently incompatible with any force feedback. (the case of airbus fbw)

speed stable fbw is another story (the case of boeing fbw, and gulfstream it seems)

Pilots in the loop
 

Ehem,

I'm the author (there's just one of me) of the work quoted by GoldenRivett above, and I'll clarify that the quoted conclusion relates strictly to the mechanisms that are used to resolve dual input scenarios. On that function there is IMO sufficient evidence in the form of ATSB and NTSB reports to indicate that it is a less than effective implementation, for the reasons stated. I'd note that at the time I wrote it I hadn't seen the AF447 final report, which is why it's not referenced, again in that incident we see the same behaviour.

By resolve 'control contention' DozyWannnabe I mean the PF knows the PNF has his hands on the controls and can either tell him to get his hands off or relinquish control. Dozy you're reading more into it that statement than there is, but that's probably because of how I worded it (thanks).

Regarding FBW and displacement rather than force feedback, the reason that displacement is viable is that the protection laws can prevent you over-stressing the airframe. Hence there's no need for force feedback to tell the pilot to ease off. As Airbus point out you can get rid of a whole lot of complication and reliability problems by not providing force feedback, I agree and that too is in the quoted post. There are other reasons why it's essential to have protection laws on an Airbus, but I won't bore you with a discussion of C* based feedback flight control laws and FAA regulations.

As far as source data, I conducted what I consider to be a solid literature search, I'd note that the references quoted go from 1987 through to 2007. The use of side-stick controllers was a hot topic back in the 70s and 80s, and the US DoD was funding a lot of research into it that resulted in the USAF's F-16 side-stick controller (as an example). Their conclusions are still pertinent and I see nothing that has rebutted them. I also cautioned about drawing too general conclusions from work in the lab, my experience is that when you get into field trials things can often be very different from a human factors perspective.

I'd also note that the work by Corps (1988) was written by an insider to the Airbus program about the Airbus teams experiences with FBW and side-stick controller, you cannot really get a more contemporaneous and pertinent source than that. I summarise from that report how Airbus developed their dual input resolution functionality, it took several iterations and some incidents in the field before they settled on the final form. The paper by Corps is still good read, I recommend it.

Amadis, yeah everything on the interwebs is so true :) That's why I put my big smiling dial on my blog, tell everyone who I am, what my background is (system and aerospace safety, human factors etc) and give a list of my peer reviewed publications. All posts are open, I invite constructive comment, respect opinions other than my own and if I get a fact or conclusion wrong I correct it.

I hope all that helps in clarifying my position, which is that I am neither an Airbus-Boeing fan boy nor a hater, the real world's a little more nuanced than that.

Here's the link to the infamous post

Pilots in the loop? Airbus and the FBW side stick « Critical Uncertainties

Did you notice that the work you have quoted lists affiliation with Douglas Aircraft Company, builder of competing MD-80 design at the time?

Welcome to the forum.

I'd argue that the implementation is at least as effective as the alternatives for the reasons I mentioned above (particularly the unexpected side effects of dual input in the B767 as evidenced in EgyptAir 990).

And yet there are other incidents involving aircraft with linked controls where the same thing happened (e.g. Birgenair 301, West Caribbean 708).

Indeed, but I'd argue you're quoting him out of context to make your point. If you were to take a look at my earlier posts, you'll see that I'm well aware who Gordon Corps was - in fact he's something of a hero of mine. His experience with the control setup actually goes all the way back to the Concorde "minimanche" experiments of the '70s and early '80s, when he was still at the ARB.

As far as the design of the Airbus FBW control system goes, the overriding ethos of the design was to use technological advancements to build a fleet of aircraft whose degree of similarity in terms of flight deck layout and handling characteristics would be unprecedentedly consistent across all types, from the short-haul narrowbody to the long-haul widebody. Because of the need to support legacy types, this was something that Boeing and MD couldn't hope to match, and in business terms, looking back across 27 years, it seems that Airbus were remarkably prescient.

The tendency for a group of pilots to react negatively to such changes is nothing new. If you go back to D.P. Davies' "Handling The Big Jets", he notes that a number of pilots fought the introduction of stick pushers tooth-and-nail. Hell, you can go all the way back to Eddie Rickenbacker's decision not to have autopilots fitted to Eastern's fleet at the time of their introduction - which ended up being rescinded very quickly!

I dont think there is much disagreement regarding the advantages of the AB FBW/protections and uncoupled sidesticks. Some advantages that have just been mentioned are:

• They are cheaper.
• It is easier to implement commonality over fleet types.
• And of course people can "get used" to the feel of the lack of tactle feed back.

However a key point with the "Pilots in the Loop" article is that tactile feed back requires less cognitive processing than visual. Hence:

• Under heavy work load the pilot is likely to consume less of his cognitive resources on interpreting the machine interface and will have more left over to deal with what is actually going on with the plane.
• And maybe there is a very small chance that an accident like AF477 would be less likely to occur.

I dont think anyone is implying accidents like AF477 would never never never never ocuur in a plane with more orthodoxed FBW/protections and connected yokes/sidesticks.

No, sadly there will always be accidents, but we must try to minimise them as much as possible. Not being able to see what the other pilot is doing with the controls is a major lack of important information, in my opinion. If that pilot is following SOPs all well and good, but if they do something totally odd and out of the blue and you can't see or detect that, then that is a problem.

But controls that move don't always help either. What about Asiana 214 that got slow on finals and crashed into SFO?

If you believe what the advocates say, thrust lever movement - or the lack of it - should apparently have been a big cue, but clearly wasn't.

Uplinker, it potentially wasn't a cue because PF was only about 20 approaches into flying a brand B after some years on the Airbus. The fact he was also holding 40 pounds of out of trim force suggests rather more basic issues.

Neither aircraft can save you from a basic inability to fly!

Agreed. He obviously was not looking at his instruments either, which is pretty fundamental.

It could be argued that as a result of his being accustomed to a thrust lever that did not move, he was not alerted when it didn't!

Derfred
He was an experienced 747 captain. Where do you think he got used to non moving throttles? What they both got used to was total reliance on auto throttle and not monitoring speed.On approach you monitor speed and not throttles. On heavy jet with that inertia you don't drop 30KTS on approach unless you are doing stall recovery and what about standard call outs? -5 KTS should have elicited SPEED call, -10KTS should have been a scream. When you check someone you have to be on your toes. This accident is a mirror reflection of an accident in 1990 in Bangalore only the aircraft was Airbus 320.

Of course.

How anyone could fly an approach without keeping a very close eye on their IAS is beyond my comprehension.

How anyone could let the other guy fly an approach without keeping a very close eye on his IAS is also beyond my comprehension.

But having thrust levers that move is a tactile feedback that the autothrottle is doing something. It is something that Boeing pilots appreciate and Airbus pilots have to learn to live without. In changing from an Airbus to a Boeing he had possibly learned to live without it so it could be put down to one of those many contributing factors.

Derfred
Captain under check was an experienced 747-400 captain where are you bringing in the Airbus and Airbus pilots don't sleep on approach they get into the habit of watching the thrust N1/EPR itself. Yet in 1990 Airbus crashed in Bangalore for the same reason. There also not single call out for speed even when speed dropped 27 KTS below VLS. There is only one way to fly an approach aircraft doesn't matter and no excuse for poor flying, period. Is there any tactile speed back for Air speed indicator.

vilas

No, he wasn't. He had flown the 747-400 as an F/O. Prior to this accident he had spent something like 9 years on an A320.

vilas,
Obviously not. But according to the NTSB report, with more handling practice, the feeling for the need for pitch trim should alert the crew to falling airspeed.

"More manual flight for Asiana pilots. Asiana’s automation policy emphasized the full use of all automation and did not encourage manual flight during line operations. If the PF had been provided with more opportunity to manually fly the 777 during training, he would most likely have better used pitch trim, recognized that the airspeed was decaying, and taken the appropriate corrective action of adding power."
Crash of Asiana Flight 214 Accident Report Summary

Goldenrivett, Derfred
I quoted two accidents. 1990 Airbus no feed back in the stick, throttles static fully serviceable 3 year old aircraft unlimited visibility aircraft fatally crashed killing even the pilots. 2013 SFO Boeing 777 conventional yoke, tactile feed back moving thrust levers weather and aircraft same as Bangalore crashed on finals. The only similarity in the two is pilots did not monitor speed. Bangalore it was -27KTS in SFO -30KTS. There are two categories of pilots first category lost their raw data skill and scan over a period of time by over reliance on automation. They forgot that instrument scan is no different on auto pilot or manual flying, only difference on auto is pilot just monitors all parameters. Second category they never acquired the skill in the first place. I see some cases where an experienced captain is unable to fly raw data during type rating on a new aircraft. Why should that happen if your scan is intact? Same classic T is used in PFD of any aircraft. Faulty or insufficient training and non critical refresher and checking are main causes.

As I said, I don't disagree with any of that.