AI failure in IMC

[Rojer Wilco]

Has anyone ever had a bona-fide attitiude indicator failure in IMC?

I'm practising randomly timed gyro instrument failures in a synthetic trainer simulating a PA31 (which I fly in the real world), and was rattled by a couple of my performances.

Does anyone know of a drill to identify/verify a suspected gyro or AI failure?

[multi_engined]

During my IFR training I was always taught with any instrument failure, especially that of the AI is to cover it up as it will be hard to ignore the indications if you don't.

Hence, when I detected AI failure... covered it with a post it note. Continued on using the TB for the rest of the NDB approach, first attempt was a little rusty but after a few attempts it became much easier. First attempt was in the sim when no STBY AI was avail. either.

[Arnold E]

RW how did you identify the failure??

[The Green Goblin]

RW how did you identify the failure??

Generally when you're upside down you figure out something aint right pretty quick :}

[Fly-by-Desire]

Some signs to identify failure could be flags, low/no vacuum, turn and bank and backup AI would disagree with failed AI, and coffee starting to pool on the roof. :E I imagine a slow loss of vacuum would be a real nasty scenario.
Try flying a NDB with a fixed card ADF, ASI, Altimeter and compass only, after some how making it to the minima the instructor threw in a engine failure, and lets just say we landed quite soon after that... inverted :bored:

[Arnold E]

GG Is that a fact? please tell me how you would identify you were upside down?

And Please dont tell me you would be hanging in the straps.

[The Green Goblin]

GG Is that a fact? please tell me how you would identify you were upside down?

And Please dont tell me you would be hanging in the straps.

Obviously you don't have a sense of humour however to answer your next question I'm sure Captain Garmin will give you an answer :ok:

[ForkTailedDrKiller]

RW, most AI failures are due to vac pump failure and can be quite insidious (if you don't notice the loss of suction/pressure) or have a light to draw your attention to the problem.

I have had several AI failures in IMC. The insidious ones were picked up because of the slight disagreement between instrument indications as the gyros wind down. A good scan will pick this up quickly.

Once you indentify that the AI is a dud - you need to get it covered quickly - but make sure you have the aircraft S&L on partial panel first. It is surprising how off-putting a wonky AI is to your scan and control.

Even in the FTDK with an electric standby AI close to the real one, it takes concentration until you get the wonky one covered.

Then there was the time that I just happened to be looking at the AI when it rolled upside down and stayed that way - not too hard to pick that one up.

Dr :8

[Arnold E]

Thanks Doc, I have never had an A/H failure in IMC, so was wondering how you would pick it up. It is something that I have considered though. I have had a vac pump failure ( interestingly, it was in a boner ) but was VFR and the first thing I picked up was the DG wandering.
It is also interesting that you have had several A/H failures, Maybe you need to send them to us for repair and save yourself some agro and money :E:E

[VH-XXX]

I once had an AI failure in VMC, boy that was nasty :-)

I knew it was dodgey because I could hear the bearings grinding away over my ANR DC's. Had I not heard that I would have been hard as it started to faulter visually not long after. Was annoyed as it was an RC Allen TSO, cost a fortune.

[ForkTailedDrKiller]

Arnold

I have only had one AI failure in the FTDK - actually a vac pump failure. ForkAir SOPs now dictate that the vac pump is replaced every 500 hrs, and I have made a mental note to NOT ignore low vacuum (pressure in the Bo) at idle. Fortunately the FTDK has an electric secondary AI, so not a big deal.

The AI that rolled over in front of me was in an aeroplane that had a mag compass failure (the oil drained out), the AI failure and a total electrical failure all on the same trip. The owner got real upset when I wrote up the maintainence release - grounding his aeroplane.

That experience was the main motivation for acquiring the Bonanza. I figured I needed an aeroplane that I knew was well maintained if I was going to keep flying SE IFR.

Dr :8

[Old Akro]

Had a vac pump failure on descent a couple of hundered feet before exiting cloud. First noticed that the autopilot seemed to be off, then identified the AI winding down. The thing about real life failures is that it tends to be a bit gradual rather than at all at once. Underscores the importance of a good instrument scan. In my experience the AI gives a lot more warning of failure. It will start to not spin up correctly and will develop a bit of wobble. This whole topic is a good argument for flying your own aeroplane for IFR work. Or at least one you are familiar with and know the maintenance. 2 engines, 2 vac pumps & 2 AI's is a good feeling.

[Arm out the window]

As the doc says, a good scan will help identify when something isn't right, when power plus attitude doesn't quite equal performance any more.

This'll work even in reasonably turbulent conditions as long as you're aiming for a particular power/attitude combo; eg wings level & balanced gives you a continuous slow turn that you have to keep correcting, or your normal cruise power setting and attitude don't give you level flight at the usual speed any more.

Practice is the key, of course, and the more comfortable you get with reacting to a situation where things don't compute the better off you'll be. To that end, have an unusual attitude recovery procedure ready to go, both with and without the AI.

If the AI's working, simply setting a sensible attitude / power combination, balance and trim will settle things down quickly (for when your attention may have wandered away from attitude to read an approach plate or something, you look up and you've let the attitude go somewhere you didn't want).

If the AI's not playing the game, something like this works for me:

Altitude - above / below safety height.

Speed - increasing / decreasing / OK > decide what to do with power.

Set 1 g if you have a g-meter. Balance.

Turn needle (or compass, if no turn needle) - which way am I turning? > aileron input until turn needle off stops, or DG stops turning, then remove input.

Nose high or low? (Altimeter / airspeed) > Pull or push until airspeed reversal or altimeter stops, remove input (controls central).

Balance / assess situation - do I need to climb? If so, do it straight away.

If not, set sensible power, balance, trim, sort yourself out from there.

Limited panel flying (ie no direct attitude reference) is possibly going out of fashion these days, but is a great skill to have in my book - lots of practice and a good sensible technique are the key.

[Baldnfat]

The coasters that the local watering hole put under your squash are great for plugging the clock holes when one goes US. :ok:

Don't be alarmed about being a bit messy on the sim. Practice, Practice, Practice.

I have sat in an old wreck practising plugging the coasters in. Takes me waaaaay back to a better time. (Frasers):O

[By George]

Air India lost a 747 when the Captains ADI failed and he followed into the sea. If my memory is correct so did Korean out of Stanstead. This is why the support role in a multi-crewed aircraft is so important. In both cases the Co-Pilot did nothing, said nothing and its hard to believe considering his side was OK. They where both 'Classics' with an engineer, who should also have picked it up. Single pilot is another matter, scan and more scan I guess, never had such a failure myself so who knows how you would handle it. One would think other clues would become quickly noticeable like air noise, T/B and basic panel etc. You don't get much time once a large aircraft is pointed in the wrong direction. Terrible thing this gravity business.

[Exaviator]

Air India lost a 747 when the Captains ADI failed and he followed into the sea.You are quite correct "By George", It was in the mid 70s and I just happened to be in COLOR=Blue]Bombay at the time and as a result of the subsequent flight delays it caused got stuck there for a week all at Air India's expense.

The was caused by a specific power failure to the A.H. portion of the captain's Flight Director which did not cause "Failure Flags". The F/O did in fact advise that his indications differed from the captain's and that bank was increasing but this information was ignored. (Before the days of CRM).

It also happened just after take-off, at night and in poor vis with the aircraft in a left climbing turn out.

With the benefit of hindsight, it later turned a check of the test button on the suspect ADI, and not receiving the correct indications, (Bank, Tilt and Flags) would have immediately indicated a power failure of this particular kind. An action I later incorporated into my own companies sim training on the 747.

Ces't la Vie :hmm:

[Arm out the window]

The 'Crash Investigation' show's take on that was frightening and frustrating -as the situation developed, there were so many indications to the crew about what was happening that they didn't act on. Discrepancies between the AIs should have been easily resolvable with a bit of a basic att/pwr/performance scan, but they let it turn into a total shambles with disastrous results.

Yes, hindsight is a fine thing, but no airline pilot should be let loose without the skills to resolve something like that safely.

[glekichi]

Had one in a PA31 - the gyro decided to topple straight forward and the autopilot decided to follow it. It was gradual of course and we caught it quickly but I wondered why the ICUS pilot was flying a descent only to look over at his AH and see it centred nicely. Autopilot would have followed it straight into the ground I reckon!!

[PA39]

T&B...DG....Alt.....VSI...compass. It comes as a surprise......but good experience!....so they say! Had a few fail over the time, mainly in VMC but had one fail over YSTW during an approach one dark and stormy night. it may come as a shock but personally i learned to never really rely on the AI/AH.

[FokkerInYour12]

Has anyone had an electric AI failure?

[ForkTailedDrKiller]

Has anyone had an electric AI failure?

Oh dear! Yes, I have had one of those also - but only the secondary AI.

In VMC at the time.

Dr :8

[mcgrath50]

As only a VFR pilot, we are taught to trust our instruments 100% if we stray into cloud. If the failure is gradual how can you be sure it's the instruments mistakenly banking and not your inner ear?

[ForkTailedDrKiller]

As only a VFR pilot, we are taught to trust our instruments 100% if we stray into cloud. If the failure is gradual how can you be sure it's the instruments mistakenly banking and not your inner ear?

You can't!

If you do a proper pre-flight and pre-TO checks and pay a bit of attention to monitoring your intruments in flight, then your chances of "straying" into cloud AND having a concurrent instrument failure are close to zip!

Dr :8

[PA39]

Cross reference my dear boy....scan......all going back to primary instrument AI.....you'll soon pick up that something is amiss. if its a vac pump the DG will go too, so T&B..DG...VSI...Alt returning to AI after each one scanned.

[mcgrath50]

Thanks guys. Makes sense! The question was more aimed at (deliberate) IFR flying, I don't intend to fly into those fluffy things yet, they look scary :ok:

[ForkTailedDrKiller]

The question was more aimed at (deliberate) IFR flying

OK Mac (:ok:), that's what IR training is all about. Instrument scan; interpretation; relax and take in the big picture rather than fixate on the AI; partial panel; recovery from unusual attitudes; etc, etc, etc!

Dr :8

[Exaviator]

Yes, hindsight is a fine thing, but no airline pilot should be let loose without the skills to resolve something like that safely.

Yes, you are quite correct AOFTW, but aviation is a continuing learning curve, and many, if not most of our current procedures and training programes have been developed because of the mistakes or misfortunes of others. :sad: That's why there is no substitute for experience.

[Jabawocky]

....................and how does one gain the experience.... lots of f:mad:ups and toppled A/H's etc :ok:

[Rojer Wilco]

Quote: Arnold E:

RW how did you identify the failure??


Well that's the thing. When I said that I was 'rattled' by my performance in the sim, it's 'cause I identified it way too late on a few occasions. I can set the sim to fail instruments of my own (or it's choice) randomly in a 20 or 30 minute window, while practiciing short IFR hops in prep for my upcomming IR renewal.

Scary stuff.

I can fly quite okay on a partial panel. I was trained for that. It's failing to identify when the bloody thing's gone that has put me upside down a few times.

Speaking of toppled things... I had five pax on, leaving Alice one July dawn last year heading for WA. Pax half asleep until... the AP put us into a steep descending spiral to the left. Whoa... is everyone awake now ladies and gentlemen?

Later the LAME told me the problem was in the AI, and that it was about to topple completely. Chilling. I'd just handflown the thing home two and a bit hours through solid IMC. So now this finds me with a special interest in improving my proficiency in this area.

[glekichi]

Has anyone had an electric AI failure?


Yep. A lot easier to identify when everything just goes black!!

[Jabawocky]

Not always so......I have not had one airborne, but if your "glass panel" is suffering from some electronic gyro issues it can show you some funny results.

Have a look at this picture.....UNDERSTAND this one point, when flown and there is airspeed and other sensory inputs the maths sorts out the error, but its a sign things are not functioning as they should. NOTE also that not all glass panel gear is as smart, I bet many TSO suppliers do not have the same ability, so do not apply my observations across the board.
http://file050b.bebo.com/17/original/2010/06/12/02/4525920200a12572752419o.jpg

[Roger Greendeck]

Have had a failure of an electric one and a few, surprisingly, electronic ones. The electric one was due to water ingress during a missed approach in a particularly heavy shower. Thankfully is was the PNF's side and it came with an off flag so it didn't cause any disorientation.

The others have been in an aircraft that compares both AIs so in each case it has come with a warning. Surprised how often it has happened. I would have thought modern avionics should be a bit more reliable.

[Roxy_Chick_1989]

Great thread guys!

A really interesting read.