More AFCS Mayhem....

[212man]

https://assets.publishing.service.gov.uk/media/65cb8411103de2000eb8f3a6/AW189_G-MCGT_03-24.pdf

Synopsis
The Search and Rescue helicopter was on its third approach, in poor visibility, to collect a casualty from a site adjacent to high ground. The Pilot Flying (PF) selected a mode of the Automatic Flight Control System (AFCS) which would bring the helicopter to a hover. As he did so, the helicopter unexpectedly yawed towards the high ground. When a further selection was made on the AFCS to effect a go-around, the helicopter accelerated towards the terrain while maintaining height. The Helicopter Terrain Awareness Warning System (HTAWS) triggered a visual and aural caution terrain alert. The crew immediately made a climbing turn onto their planned escape heading during which a warning terrain alert triggered. The helicopter recovered to a safe height and returned to its home base. The unexpected yaw was caused by a mismatch between the previously selected AFCS heading reference and the heading flown by the PF. While the helicopter and the flight control system were found to be serviceable and performed as designed, the crew did not have a complete understanding of the functionality of all the AFCS modes.

Other factors included:
● Overriding the engaged modes by manually flying the helicopter.
● A lack of clarity between the role of PF and Pilot Monitoring (PM).
● Ineffective communication and co-ordination between the pilots.
● Imprecise application of Standard Operating Procedures (SOPs).

he operator took a number of safety actions to raise awareness of the event, improve knowledge of the autopilot modes and include the event as part of their initial and recurrent training.

[HeliComparator]

Looks like poor training and a crew who have never become comfortable with the automation, thinking that “proper pilots fly it manually” perhaps? A bit sad in this day and age.

[212man]

Looks like poor training and a crew who have never become comfortable with the automation, thinking that “proper pilots fly it manually” perhaps? A bit sad in this day and age.
It's a curious report. From my understanding of the description of the system in the report, and AFCSs in general, my assumption is that the manually flown right turn was done without his feet on the pedals, hence the FTR micro-switches were not pressed and the HDG bug did not track the heading changes.

[Droopy]

Like HC I'm retired now, but most recently was on the 169 with a very similar AFCS. The latest phase upgrade got quite a few people excited, "we can do this, that, offshore landings will become much better managed, etc etc..." But the weakness of all this automation capacity is the amount of training and more importantly currency so we mandated no use of all those esoteric modes as they all required investment which the company/client wouldn't pay for.

[HeliComparator]

Like HC I'm retired now, but most recently was on the 169 with a very similar AFCS. The latest phase upgrade got quite a few people excited, "we can do this, that, offshore landings will become much better managed, etc etc..." But the weakness of all this automation capacity is the amount of training and more importantly currency so we mandated no use of all those esoteric modes as they all required investment which the company/client wouldn't pay for.

This is definitely a valid point. With a highly automated aircraft you need to have the same training and skill in manual flight, but also you need to have a lot of training in the use of the automation. So the overall training “burden” is definitely greater. But then if all that training is done properly, the overall level of safety is a lot better.

I've been out of the game for 10 years now but I wonder if the CAA policy for recurrent training and testing still mostly revolves around manual flying skills, with skills in the use automation being an afterthought? Or has CAA extricated itself from the Stone Age yet?

[Hot_LZ]

The training of AFCS modes and SAR modes is quite thorough and is exercised on an almost daily basis while on shift. The caveat being that the some of the older generation who were ninja at manual flying still haven’t got to grip with trusting the automation fully.

LZ

[Cornish Jack]

Can somebody tell me whether the AFCS is primarily electronic, hydraulic, or a combination ? If the latter, what are the proportions and relative distributions in the system hierarchy ?

[[email protected]]

CJ, sensors (gyro, accelerometers etc), AFCS computers, linear and parallel actuators - all electronic. The flying controls - rods, levers, wires etc to hydraulic servo jacks.

The AFCS computers can either input directly to the servo jacks (usually SAS) or use open loop actuators directly onto the flying control runs.

That is a simplistic and generic overview and not aircraft specific as there are many variations.

The main problem with modern AFCS is that they are made complex by engineers and need to be treated like any other computer - ie garbage in equals garbage out.

If you tell it to do the wrong thing, either through an error, ignorance or lack of practice - it will do what you have asked it to - even if that isn't what you wanted or expected.

[212man]

This is definitely a valid point. With a highly automated aircraft you need to have the same training and skill in manual flight, but also you need to have a lot of training in the use of the automation. So the overall training “burden” is definitely greater. But then if all that training is done properly, the overall level of safety is a lot better.

I've been out of the game for 10 years now but I wonder if the CAA policy for recurrent training and testing still mostly revolves around manual flying skills, with skills in the use automation being an afterthought? Or has CAA extricated itself from the Stone Age yet?
Hi HC, I believe it has moved on from the heady days of 2004 where the TSLG minutes documented that the CAA would allow the use of ALT hold on a proficiency check, while copying down the ATIS! I think a combination of factors - the FOTIs being exposed to the modern types with their currency flying with operators, and the Head of Training Standards having a 'road to Damascus' moment after he participated in the S92 JOEB, and switched views by 180 degrees.

[hargreaves99]

TSLG? (Training Standards Learning Group?)

FOTI? (Flight Ops Training Inspectors?)

JOEB?

[212man]

TSLG? (Training Standards Learning Group?)

FOTI? (Flight Ops Training Inspectors?)

JOEB?
Joint Operational Evaluation Board - the JAA predecessor to the EASA Operational Suitability Data (OSD). The report I refer to is here : https://www.easa.europa.eu/sites/default/files/dfu/Sikorsky%20S-92A%20JOEB%20Report.pdf

[gipsymagpie]

A heading bug on the wrong heading, a turn towards high terrain, inadequate understanding of the AFCS and manually fighting the AFCS. Hang on, that's the G-SPAU accident in 2002. Whilst in that case Airbus changed the AFCS heading mode logic .

https://assets.publishing.service.gov.uk/media/5422fa2140f0b6134600081b/dft_avsafety_pdf_023427.pdf

[jimf671]

This is definitely a valid point. With a highly automated aircraft you need to have the same training and skill in manual flight, but also you need to have a lot of training in the use of the automation. So the overall training “burden” is definitely greater. But then if all that training is done properly, the overall level of safety is a lot better.

I've been out of the game for 10 years now but I wonder if the CAA policy for recurrent training and testing still mostly revolves around manual flying skills, with skills in the use automation being an afterthought? Or has CAA extricated itself from the Stone Age yet?

And if just one combination of settings is omitted from the training then that is the hole that somebody will fall down.

[SASless]

There is no way to 'train" for every situation or every kind of failure or malfunction.

There are pitfalls, potholes, and other obstacles to operating with modern avionics and devices and one should try to stay familiar with what they are.

The Heading Bug situation is caused by poor design, inadequate certification processes, and manifested by failing to use the system in a proper way.

One would think in this day and age that could be eliminated as we have so many examples of where it has played a role in mishaps down through the years.

How would we want the AFCS system to deal with "Heading" ?

[Cornish Jack]

Thank you Crab - so the direct 'wiggle the output' is hydraulic ? Query triggered by memories of Colorado Springs 737 et al and, the, almost unreported, 744 LY out of Heathrow.

[212man]

manifested by failing to use the system in a proper way.

Personally, I think that is more the issue, based on inadequate knowledge of the system functions. This accident (https://aviation-safety.net/wikibase/72613) in 1985 wasn't even using the upper modes, it was the TRE trying to demonstrate the inherent heading hold function of the ATT mode, by lifting into the hover with feet away from the micro-switches on the pedals. Problem was, they had already been released on a different heading, and the aircraft turned without them being depressed again. So, as it lifted, it turned onto the memorised heading.......as it was designed to.

[HeliComparator]

There is no way to 'train" for every situation or every kind of failure or malfunction.

There are pitfalls, potholes, and other obstacles to operating with modern avionics and devices and one should try to stay familiar with what they are.

The Heading Bug situation is caused by poor design, inadequate certification processes, and manifested by failing to use the system in a proper way.

One would think in this day and age that could be eliminated as we have so many examples of where it has played a role in mishaps down through the years.

How would we want the AFCS system to deal with "Heading" ?

There isn't a way to train for every situation but there are are some basic principles to be reinforced, such as not flying through the Upper Modes other than very briefly, using both pilots to ensure that everyone is aware of what modes are engaged and what modes are not engaged. Plus of course an understanding of how the upper modes behave.

[albatross]

I remember that event. Much Laughter on the part of those not involved! However a lot of folks sure noted the causal circumstances and learned from it.

In this case they mention “As he did so, the helicopter unexpectedly yawed towards the high ground. When a further selection was made on the AFCS to effect a go-around, the helicopter accelerated towards the terrain while maintaining height.” Does this perhaps mean the go-Around button was pushed which would, in some helicopters, cause the aircraft to roll wings level and commence a climb straight ahead?

Personally, I think that is more the issue, based on inadequate knowledge of the system functions. This accident (https://aviation-safety.net/wikibase/72613) in 1985 wasn't even using the upper modes, it was the TRE trying to demonstrate the inherent heading hold function of the ATT mode, by lifting into the hover with feet away from the micro-switches on the pedals. Problem was, they had already been released on a different heading, and the aircraft turned without them being depressed again. So, as it lifted, it turned onto the memorised heading.......as it was designed to.

[Ex Machina]

An earlier event. Same aircraft, same unit: https://www.gov.uk/aaib-reports/aaib-investigation-to-leonardo-aw189-g-mcgt