The Great Ongoing Aviation Thread (general and other)

Miscellaneous and Personal Stuff I Must Share
3683

In other updates on existing events (sorry no cites to non-paywalled sources)

  1. In the JFK taxi across wrong runway incident the AA pilots are going to testify / be interviewed with a stenographer, but not audio recorded. It turns out audio recordings are very rarely used in NTSB investigations of air incidents / accidents but are commonly used in rail, maritime, etc., accidents.

    At least one other (as yet unnamed) pilot union is lining up behind APA that audio recordings of these interviews are simply too salacious not to be leaked or simply released to the public while, unlike CVR recordings, there are no statutory / regulatory protections to prevent that. The unions’ perspective apparently being that once such protections are in place, audio recording of interviews would become unobjectionable and commonplace.

    I’ve seen no new insights into what may have been going on rightly or wrongly in the AA cockpit to lead to their mistake.



  2. In the AUS go-around over takeoff event, the weather conditions required the 737 pilots to perform an engine run-up to ensure no icing in the engines before launching. Most jets have this requirement, but the exact weather conditions where it’s required or not, how often to do it, for how long, whether it must be done on the runway, etc., vary greatly. With the result that ATC can’t exactly predict whether a run-up will occur.

    If one intends to do a run-up on the runway it’s polite / expected, although not strictly required, to warn ATC of that before being cleared for takeoff, and certainly before entering the runway.

    According to some leaks it appears at this early stage that a) the controller was assuming Southwest’s usual clear-weather very much hurry-up departure, and b) Without saying so, Southwest was taking extra time for either a run-up or for just getting squarely aligned with the runway before adding power. In contrast to all jet’s habit of starting the power up towards full thrust while still pulling onto the runway.

    In any case the sum of these two things used up all the slack time and then some. Oops.
3684

According to the Port Authority, “an electrical panel failure, which also caused a small isolated fire overnight that was immediately extinguished,” triggered the power problem.

You wonder what catastrophe would happen if they had a little bigger fire.

3685

The good news here is the design of JFK is a series of separate terminal buildings. Each carrier with a large presence at JFK has their own separate building for their own use and share that building only with their own code-share / alliance partners. And not even always all of that partner’s flights.

The affected terminal is the one that all the remaining carriers share. That includes some big names in the industry who’re just not big at JFK.

3686

Referring back to the ATR accident in Nepal and @Richard_Pearse’s cite to the preliminary accident report …

Here’s a very nice pic of an ATR cockpit:

Near the bottom of the pic at center we see the throttle quadrant. From left to right there’s the more or less vertically-oriented speed brake handle leaning to the left, the two large cylindrical throttles, the two smaller sorta-lumpy rectangular condition levers, and the light tan colored sorta-airfoil shaped flap lever.

In the pic the flaps knob is fully forward meaning “flaps up”, and you can see the two other positions of that lever marked “15” and “30” from forward to aft.

In the pic the condition levers are fully aft in the “engines off” position. You can see the other positions marked “FTR” for feather and “AUTO” from aft to front, plus one more position further forward we can’t quite read the label of that doesn’t matter for the story.

Now imagine the flaps are set to 15, and the condition levers are set to “auto”. Notice the two sets of knobs will be almost side-by-side. Notice also that the “flaps 30” and the “FTR” detent positions are not exactly aligned, but they’re also pretty close. And even though there’s some misalignment between the two positions of the two knobs, the total angular range of motion between “15” and “30” vs. “AUTO” and “FTR” are pretty similar. I bet you can see where this is going.

Now you in the right seat want to extend the flaps to 30 at the request of the pilot in the left seat. Reach over with your left hand, grab the flap handle, lift straight up ~1/4" to clear an internal detent, then pull it aft in an arcing motion to the “30” slot then wiggle the lever back down into the “30” detent assisted by a spring.

Now imagine doing that while leaning over to your far left craning your neck to look outside to keep sight of the airport as your co-worker is maneuvering around the traffic pattern inside this narrow valley.

It’d be easy, real easy, to grab the adjacent black knobs and pull them aft one notch by mistake. Apparently too easy. And now your engines are both feathering / feathered while you both are still looking out the windows in this complicated visual maneuver. And by happenstance your co-worker is a bit high in this maneuver, so you don’t need more power; in fact the power declining unnoticed from idle to zero is helpful, not hurtful. At least at the moment. But that will become a problem later, and “later” will arrive pretty soon. Oops.


For sure all this stuff is designed to have different shapes to help minimize these mistakes. The two sets of knobs feel different in the hand. But at the same time, both condition levers and flap levers are moved from every notch to every other notch on every flight. Both “feels” are totally familiar. The hope is not that they’ll feel wrong, but wrong in context. Which really requires your mental context to be “moving flaps”, not “looking outside and teaching / evaluating a maneuver.”

These sorts of traps exist in every airplane despite being designed out as much as possible. What’s left is just a residual risk of the business.


My real point here overall is not that “anyone could make this mistake; it’s defective design”. Nor is it that the instructor pilot in the right seat was an irretrievable goofball; he was not.

It’s just how mundane and trivial of a physical mistake / mental oversight can snowball into how big of a disaster how quickly. Statistics tell us that given enough opportunities every possible mistake will be made somewhere on some day by someone. That was the place, day, and person where that particular mistake occurred and also wasn’t caught in time to remediate it before the situation became irrecoverable.

For want of a nail …

3687

It looks to me like it says 100% GYRO. Does that make sense?

3688

I had no idea about the details. That set off quite a research project just now. I was able to obtain an ATR-72 pilot manual from an online dedicated aviation subscription library I belong to. But as usual, it was a crappy scan of a crappy set of photocopies. And no table of contents. Cue flipping through a lot of pages. Anyhow, I got the answer.

It really says “100% OVRD” as in “override”.

In typical flight operation, turboprops of all types generally try to turn the prop at a constant moderate RPM and as you add fuel to get more power, the blades adjust to take a bigger bite from the air and hold the RPM down while more torque is extracted and applied to the air. Likewise when you pull the throttles to idle, the prop attempts to maintain a constant RPM by going to a very fine/flat pitch. The prop acts as the equivalent of an automobile continuously variable transmission, trading torque for RPM across a wide range. That typical governing is what the “AUTO” position does on the ATR.

The “100% OVRD” position puts the governor in a simpler mode where it ignores a bunch of inputs and simply tries to hold the prop at the maximum rated RPM. I did not dig into what sorts of unusual situations, malfunctions, or emergencies would trigger the pilots to use this feature.

IIRC @Richard_Pearse has flown turboprops, albeit not ATRs. He’ll be able to give us a more nuanced explanation of the practicalities of engine condition levers and prop governors on turboprops.

3689

Thanks for the details! That makes sense. And I can see why I’d have mistaken GYRO for OVRD. At least I got the 100% right.

Getting back to this:

As well as this example:

I think this is an extremely important point about how human cognition works. It’s really quite incredible that we operate this way, where under the right conditions our brains can lie to us about even what someone said to us–but it’s utterly commonplace. You had an expectation for the callout that was so strong that it overrode everything else.

One can imagine a different situation where the callout was garbled due to noise or whatever, and you had to make a subconscious judgment call about what was said based on experience. You can imagine that the strength of the judgment would depend on how strong your expectation was. It might not take much of an override if it was garbled to a 51/49 probability. But the same effect, when the expectation is strong enough, might even override a 99/1 probability. So you end up “hearing” completely the wrong thing even when it was unambiguously otherwise.

So many bad outcomes from this effect. Unrelated to aviation, but undoubtedly the same thing: children accidentally being left in cars. Always happens after there’s some shift in context. There’s some scenario where normally they wouldn’t have their kid in the back seat, except today it’s different. And tragedy strikes.

Probably a completely inevitable side effect from how human cognition works. We have to operate from ambiguous inputs, and our limited brain horsepower means we have to make approximations. Systems designers, engineering and otherwise, should keep these effects in mind.

3690

'Zactly.

One can try to mitigate these issues as much as possible with a questioning mindset: “How sure am I that I heard XYZ?” then if less than very highly sure, seek clarification rather than pressing ahead with your belief / assumption. But the problem still remains at least somewhat; the meta-perception of your perception is not 100% reliable either. And you may well fail to ask the question of yourself just when you’re busiest and therefore operating at your least reliable.

For darn sure it’s not something one can turn on only when needed. It’s either part of your work- and life-habits, or it’s not.

I encounter people away from work who think the very idea that their own perception and cognition is not 100% certain and reliable is somewhere between inconceivable, anathema, and a grave insult. I can only shake my head in bemused wonderment.

3691

Yes, I flew Dash 8s (200 and 300 series, not the 400 which is quite different) which have a similar engine to the ATR 72. I also have access to ATR manuals as my employer operates them. From what I can see the ATR has a “poor man’s” auto thrust system that provides power control via an engine management selector provided the power levers and condition levers are in their “auto” positions.

In the Dash 8 you would take off with condition levers set to “MAX” which would give 1200 rpm and the power levers had to be manually adjusted to achieve the required torque setting. After take off the condition levers would be brought back to “MIN” which would give 900 rpm. The torque gauges had to be monitored and the power levers adjusted to keep everything within limits. On final approach the condition levers would go back to MAX to give 1200 rpm in case of a go-around. They stay at MAX until engine shutdown where they are pulled back past FEATHER to the fuel off position.

It appears in the ATR you have the power levers in a “notch” and the condition levers in AUTO and then power is changed by selecting TO, MCT, CLB, or CRZ for take-off, max continuous, climb, and cruise power respectively.

The ATR FCOM says to put the condition levers to 100% OVRD during approach in turbulent conditions to better control the aircraft speed, but otherwise they stay in AUTO.

The condition levers have a lever that must be lifted to allow it to be moved out of the AUTO position towards feather. There’s a good video of that in action below:

It looks like it would be difficult to accidentally move both condition levers out of the AUTO detent but muscle memory is a funny thing. Your brain can be saying “Flaps 30” while your hands are confidently carrying out a different task.

In a similar vein to @LSLGuy’s anecdote, the BAe146 had its flap lever near the first officer’s left leg and the gear lever further forward sticking out of the instrument panel, a typical arrangement in two crew aircraft.

I never had a problem with it when I was a full time FO but when I was a captain I occasionally had to fill the FO’s position and fly with another captain. A couple of times I caught my left hand automatically reaching for the flap lever instead of the gear lever immediately after take-off, as in the captain calls for gear up and I sense my hand wanting to grab the flap handle and have to consciously direct it to the gear handle.

Although a simple mistake to make, it can have disastrous consequences as putting the flaps up at low speed and altitude can lead to a stall. I had one instance where my FO did it to me. We took off, I called for gear up, and he put the flaps up to zero. Oops. To his credit he recognised his mistake right away and put the flap handle back to the take-off position. In that case the flaps will either lock in their current position or move back to where the lever has been set. In our case the flaps moved back to 18° and we continued the flight uneventfully.

The brain and body are a curious system.

3692

Alaskan Air had a brief system standown after two tail strikes on Hawaii bound planes. Weight figures were off by 20,000 to 30,000 lbs. A backup software solution was loaded and I presume a safetygram to pilots. Also a runway incursion in Honolulu. A jumbo blew through a stop in front of a landing plane. The landing plane had a fair amount of room to exit the runway but that didn’t excuse the incursion. Hawaii has been in aviation news lately. I liked things duller as a resident.

3693

Here’s a decent intro article on “expectation bias” as it applies to aviation:

Backing up a couple of layers in their taxonomy of articles to “Human Performance” or “Human Behavior” is a pretty good round-robin tour of state-of-the-art thinking on these topics.



After parking the airplane and plugging in external electricity I bet one of the standard FO actions is to raise both condition lever interlocks and pull both condition levers from AUTO to OFF. So there’s plenty of muscle memory leaning towards the mistaken move.



A thing in common with the ATR mishap and the first part of your anecdote is the subsequent dual mismatch between both pilots’ mental models and reality.

In the ATR case, both pilots thought a) the flaps were at 30 when they were really at 15, AND b) the engines were producing thrust when they were not. In your case, had you and / or the FO not immediately recognized the glitch, the discrepancies would have been a) gear down when you thought it was up and b) flaps up when you thought they were down.

One of those two discrepancies will come to light before the other. And then while the crew is distracted resolving the one discrepancy, the other may mature at the worst time. Discovering you have one problem is unusual and unwelcome. Discovering you really have two is much more likely to lead to a bad outcome.

In other “news”:
I’m riding the commuter train to work this morning as I type. My FO for the next 4 days is brand new. Like graduated from his final QA flight 3 days ago. I don’t yet know what he did before; probably flew RJs for at least a couple years, maybe even five. But it’ll still be a fun learning experience for both of us. It’s a fairly civilized trip, early mornings to late afternoon and few long-ish legs per day, so lots of time to yak & learn / teach.

The number and variety of human performance issues that arise with newbies and their flaky “muscle memory” both physical and mental will be interesting. Lots of opportunities for what we call “negative transference”, doing something in response to some cue that was appropriate for the airplane you used to fly, not the one you are flying.



Ouch!

3694

More details (the first two are the same article, the third is Hacker News discussion):

https://news.ycombinator.com/item?id=34847550

3695

Very interesting. The back and forth on ycombinator was all guesswork, and mostly by IT types, not pilot types. Still interesting given its limitations. Frankly I started skimming that pretty quickly once the blamesmanship reared its head. So there might be a nugget or two in there I missed.

I wonder how / when / whether the FAA / NTSB will put out any detailed post-mortem.

Most carriers use if not the same software, at least the same approach to the bureaucracy and the software that supports load planning.

A few hours ahead of departure estimates are made of passenger, cargo, and fuel weight. These estimates come from actual passenger & freight bookings for that flight for that day, tweaked by historical averages for that flight on that day of the week. A couple hours before departure the “final” fuel load is calculated based on the planned total weight, winds, altitudes, alternate airport requirements, etc. With some room versus any limitations for what if the passenger/cargo weight runs a bit heavy vs. the estimates. If there is not such room, then either the cargo or passenger weight is capped to prevent an overweight situation. About an hour before departure the pilots capture a set of that planning data on paper or on their tablets. That’s all the weights, and the resulting takeoff performance info versus the runways and weather du jour.

As passenger’s check in or are boarded, and as cargo and baggage are loaded below, the estimates are replaced with actual numbers. Once the fuel is loaded, the final actual fuel load is also relayed back to HQ. Of course, all this is computers talking to computers, with all the potentials for glitches you might expect. Not that people making phone calls as it used to be done would be more reliable; they’d probably be less so.

Once we lock up and depart the gate, all that weight & performance stuff is re-computed as final then sent as both a text message to our printer and as a data blob to our on-board performance computer.

Finally, the pilots go through their “printout” of the plan from an hour ago, the final text message on the printer, and the final data as uploaded into the airplane’s performance computer and displayed there to make sure all the figures match within normal tolerances. And to resolve any discrepancies that exceed tolerances.

What there is not, except by informal experiential awareness, is a formal reasonableness check that this all makes sense. A glitch between these various systems should come to light easily. A glitch farther upstream in the bowels of the computer / bureaucracy that misstates some number that’s then fed into all the subsequent calculations and that are then taken for granted by all the matching cross-checks is a more subtle problem and may not be noticed.

Weight has always been one of those topics where it’s simply not possible for the pilots to personally audit the cargo weight numbers. Or really the passenger weight numbers. Although one could do the mental math on headcount * average weight → total weight. But that just moves the mistake gatekeeper back to “What is the real headcount?” At some point you need to believe that the people and computers whose job it is to do that stuff right every time every day did in fact do it close enough to right this time today.

Regardless of the source of the error I must say that I’d hope most pilots would notice a huge discrepancy on what’s normally a maximum effort flight such as a full plane flying to a destination near the airplane’s max range.

OTOH, a partly full flight going a much shorter distance has a lot greater potential for legitimate variation, where weighing e.g. 130,000# or 150,000# are both perfectly plausible numbers from one day to the next. The greater variety of flight durations and fullness of flights you personally operate, the wider range of weights sound “normal” to you.

3696

Emergency services were alerted to the crash at 12:02 p.m., the Little Rock Police Department said in a Facebook post.. There was a line of storms moving through the area around the same time, according to CNN Weather. There were no warnings alerted for the storms, but the National Weather Service issued a special statement calling for gusts up to 50 mph. The highest gust at the airport was 46 mph, recorded at 12:02 p.m.

https://www.cnn.com/2023/02/23/us/arkansas-plane-crash-five-killed/index.html

3697

More on the CSB:

3698

Another 5 fatality crash in only two days:

No information on possible cause.

3699

That’s some really bad reporting. “This just in, something happened somewhere”.

Why even put it out.

3700

The New York Times Feb. 25, 2023

Five Killed in Medical Plane Crash in Nevada, Officials Say

  • The Lyon County Sheriff’s Office said its dispatch center received multiple calls around 9:15 p.m. local time about a possible aircraft crash in Stagecoach, Nev., which is about 45 miles southeast of Reno. It was not immediately clear where the flight had originated and where it was headed.

  • Sheriff deputies, along with firefighters and search and rescue teams, responded to the calls and searched for about two hours before finding the airplane around 11:15 p.m.

It must have been remote to take 2 hrs to find it. I’d hate to think there were survivors that died in the cold while waiting on rescue. A sad situaiton all around.

3701

Just as an incidental note, that sounds like a fun and satisfying experience. Passing on the learnings of years of experience to the next generation, as it were, is intrinsically rewarding.

OTOH, keep a close eye on the young whippersnapper! :smiley: It’s like the old adage that for every surgeon, there is always a “first time”! We do our best to mitigate the risks, but “first time” rarely equates to “best performance”!

3702

Turns out he had been a training / QC captain for a regional airline flying RJs for the last few years.

There were lots of nuance points about 737s and about our employer’s arcane procedures that I shared with him. But he didn’t need me at all. You wouldn’t know he hadn’t been on this jet for a year already. And his overall awareness and aviatorness was superb.

This isn’t an entry-level job and he was far from an entry-level aviator.

But yes, I had a great time. Mentoring the newbies is by far the most satisfying feature of the job from my aged POV. Others in my role think otherwise & just want to sit there. Shame on them.