Ted.P wrote:
I think the answer to the OP is “almost anything”.
I agree. It’s Murphy’s Law, plain and simple. If anything has the slightest chance of going wrong, it will. However, the common denominator is the pilot, keeping the pilot out of the loop as much as possible will prevent more accidents than any other thing you can do. Complex machines are better operated by automated systems.
For a recreational aircraft, it’s a bit counter productive to automate everything, unless the aim is only to transport people safe from A to B where the humans are reduced to spectators.
Peter wrote:
And this is on certified aircraft… I wonder what this thread would look like in Non Certified?
Indeed as the NTSB report about experimental homebuilt aircraft conclude, it is “non-standard” systems and higher performance (more difficult handling) that is what differentiate experimentals and certified aircraft in the accident rates. But, the problem is the next or later owner, not the original builder. The original builder knows the systems and handling, and the next owners tends to just jump into the seat and fly, with no training and no study of the systems. In short, the next owners tends to treat the aircraft as if it was any other certified aircraft, when there is nothing certified in an experimental aircraft. An experimental aircraft requires more attention, because it can have strange systems and “strange” handling characteristics, and this can be different from seemingly identical aircraft.
Sure… it comes down to the “investment” which the pilot puts into the aircraft.
There is a proverb in the USA: beware of the man who has only one gun, because he probably knows how to use it. (the context being that there are loads of people who have a big collection)
But also modern avionics are not trivial, and it is not automatic that everybody who can get a PPL (in some simple training type) can ever learn the systems in a current “glass” aircraft. I think many people refuse to accept that, and obviously the industry will very happily sell a $1M SEP to anybody with $1M to spend.
I used to know a guy who had unlimited funds and ended up “flying” a Citation (with a permanent RHS FI). He could not drive a manual gearbox car. He spent years trying to do an IR. Eventually he gradually downgraded back to an ancient MEP which he is comfortable with. But at least he was smart enough to know this.
any good examples of things that could kill you
Keen on semantics, as I suppose I must have a quite bad reputation for, I have been annoyed by this style of phrase from the beginning of the thread. Too tabloid-like for me. More proper wording could be
To the first wording, the answer must be “very few, and very unlikely” – a wing or empennage breaking of, or the engine very suddenly vibrating so much that it breaks off its mounts (or separates from the firewall with the mounts and all)
To the second wording, the answer must be “almost everything” – as has already and rightly been pointed out.
Peter wrote:
I wonder if any SR22+Avidyne+DFC90 owner can confirm that it really does this?
I don’t own one but have flown around 50 hours behind one, don’t know if that qualifies? I confirm that if you preselect an altitude and press VS+ALT, it will set +500 fpm on the VS if selected ALT > current ALT and VS <= 0, and -500 fpm if it’s the inverse. I don’t think it’s confusing or dangerous, but rather very convenient.
One habit in IFR flying on such a machine is to always set your bugs where you’re going. Before takeoff, you would set the cleared altitude. Another good habit I learned during my IFR training is to manually fly to a safe altitude (that is CAPS altitude on the Cirrus; 500 ft AAL on the G1 to G3) before touching any button – flaps, autopilot, anything. And in the climb on a DFC90, you would always use the IAS mode, never VS, as Flyer59 already pointed out. There’s really no way I can see how this AP behaviour could become dangerous if you follow a few basic procedures.
Any AP which has an IAS hold or Level Change mode (what ever it’s called) whereas it will adjust pitch to hold IAS vs a power setting is much supieror in terms of stall prevention than using the VS mode, especcially in climb.
The usual procedure in an airplane thus equipped would always be to set IAS to take off safety speed (which usually is Vy + a safety factor) and to engage IAS / FLCHG above the minimum engagement altitude of the AP. It will then maintain this IAS, the Vertical speed will then be a result of the power setting as well as pressure altitude. Vertical speed in a climb is never a fixed value, it decreases progressively. If you want to cruise climb, that is to climb with a faster IAS than Vy, then all that happens is vertical speed will decrease a bit.
Most airplanes which have this feature will come with a nice set of instructions at which speeds to execute the climb. Of hand I recall the MD11: 250kt IAS to 10’000 ft, 290 thereafter until reaching Mach .80 and then Mach .80 until TOC. For a GA airplane it will look simpler: Vy+ 5 to 10 kts from deck to level off, while Vy will vary with altitude. Therefore, it is necessary to select a climb speed relative to the altitude which keeps with the best rate speed and well above stall speed.
In piston airplanes, VS mode is often used in descent, as IAS mode doesn’t work too well in piston engines. In a jet or turboprop, you can reduce power to flight idle and fly again at a prescribed speed regime. In a piston, you can not do that without shock-cooling the engine, hence most people use VS mode in descent. Which does not mean you can’t do IAS but it needs a bit more thinkering to figure out which IAS to program to get a decent ROD as well as an engine preserving power setting.
What does surprise me is what Alexis sais about the Avidyne set: Is it true that protections only work in IAS mode? That would be highly counterproductive. They would be especcially necessary in VS, even more so than in IAS mode.
What we do with the S-Tec 55x is that we use VS in climb with a default value of 500 fpm below 10’000 ft and 300fpm above. Obviously you will have to watch your IAS all the time, but 500 fpm in lower altitudes is a safe ROC with climb power (the airplane will do somewhere between 1000 fpm to 600 fpm between SL and 10k ft) so while you need to watch IAS during climb one can pretty safely assume that 500 fpm will not overtax the capability of the airplane. Clearly, what that does is that it will increase the time/fuel and distance to climb considerably, however it will also reduce engine heat and allow climb at a lower fuel flow (e.g. 25/2500 iso WOT/2600).
In descent we use 500 fpm ROD from TOD to pattern altitude. Again, with 500 fpm you will be able to keep below the Vne speed at a power setting which will not shock cool the engine most of the time.
Having an IAS mode would be the single most important reason for me to change from the S-TEC to the Avidyne if it ever becomes available for my plane. And I fully agree: Climbing in VS mode is one of the gotchas you can encounter.
The envelope protection of the DFC90 works in all modes, of course. In coupled operations it will always prevent a stall, in uncoupled or FD operations it will announce bank angles too high or speeds that are too low or too high.
I just wanted tp express that while a wrong altitude preselect would make the airplane descend into the ground if the pilot used VS not much can happen if you select a speed too low: the envelope protection would lower the nose.
A typical SR22 (95 percent?) will also habe a KGP-560 EGPWS systems (TAWS B), so if you descend inadvertently you will get the whole “sink rate”, “too low” or “pull up!” program.
But IMHO the only right way is to use either AP (pitch and roll hold) or IAS after take-off.
LeSving wrote:
I agree. It’s Murphy’s Law, plain and simple. If anything has the slightest chance of going wrong, it will. However, the common denominator is the pilot, keeping the pilot out of the loop as much as possible will prevent more accidents than any other thing you can do. Complex machines are better operated by automated systems.
Isn’t this how people get killed? The pilot is there to monitor (among other things). You can’t do that properly if you’re behind the plane or out of the loop.
If you look at the presentation that prompted this thread, you’ll note that the author flies G450 and his list has two items: ground spoilers and slow onset hypoxia. Quite a short list for such a complex plane loaded with things to go wrong. And several things have to happen for those spoilers to kill him. Mitigation? Procedures. Disarming the spoilers as the first thing even before retracting gear (not assuming checklist will be completed) and periodically checking cabin altitude. So we are’re looking at things which can kill you if you mismanage them and which can leave too little time to (identify and) rectify the problem (i.e. you can’t afford to make that mistake).
Martin wrote:
Isn’t this how people get killed? The pilot is there to monitor (among other things). You can’t do that properly if you’re behind the plane or out of the loop.
True, but pilots get distracted, tired, reckless, bored, stressed, confused and so on. Something is bound to go wrong at some point because a pilot will not be 100% “on” all the time. Mishaps can be minimized through procedures, training, two pilots instead of one, but can never be removed entirely. By automating complex systems, you remove these kinds of mishaps entirely, leaving only mechanical faults. For instance, a single lever engine control (FADEC) will relieve the pilot for lots of monitoring and micromanaging of the engine. It will remove the chances of the pilot doing something wrong in that area. It can be something as stupid as pulling the wrong lever, because two levers are positioned “too close”, but it can also be less chance of a mechanical faults due to more precise and optimal managing of the engine.
If aviation was invented today, it would never ever be “manual”. Everything would be automated. The would be no pilots, no manual ATC. Everything would be designed from the ground up as fully automated logistics systems. Luckily (for us) it isn’t 
In my opinion, the only thing the pilot is good at is flying the aircraft using stick and rudder. Anything preventing/distracting him/her from doing that is lazy/poor engineering (or money perhaps). A perfect aircraft is one with no or a minimal of checklists, no procedures. These things should never be so complex that they cannot be remembered as back bone reflexes. At a very minimum, checklists should pop on a screen when needed. Monitoring is for machines. The aircraft should monitor the pilot, not the other way around. A simple aircraft, as a Cub, is a perfect flying machine. From there it only goes worse, until ending up in the other end in a modern jet fighter which has systems that is built to augment the pilot, not work against him.
If aviation was invented today
If aviation was invented today it would be prohibited. Absolutely no question. I guess hang gliding would be allowed, as rock climbing and scuba diving is.
Maybe in very “macho” and “can do” countries, e.g. South Africa, aviation would be allowed, but only in the wilderness where there is nothing to crash into.
checklists should pop on a screen when needed.
I wonder why this is not present in light GA (usually) and I guess the answer is that avionics interconnectivity doesn’t support it. After all, what is the point of displaying a checklist if there is no means for the system to verify that the action was followed? GA planes are way too simple for that.
But even if you heavily automate stuff, so many pilots will not understand the systems, but they still need to. Look at AF447 – two 5000hr+ ATPLs, 14 exams, ex gold plated elite French national aviation academy stuff, sitting there totally baffled. There are more subtle examples where the pilots did not know about e.g. the interaction of a failed RADALT and the auto throttle.
