Raw data flying
 

Most of the aviation community believes automation has made flying safer, but there is a fear that pilots fail to see that automation is a double-edged sword. Pilots need to understand the advantages of automation while being aware of its limitations. Experience has shown that automated systems can make some errors more evident while sometimes hiding other errors or making them less obvious. In 2005, the British Airline Pilots Association (BALPA) raised concerns about the way airline pilots are trained to depend upon automation. BALPA felt the current training leads to a lack of basic flying skills and inability to cope with an inflight emergency, especially mechanical failures. The union believes passenger safety could be at risk.

My MCC students, flying an fnpt2 737 sim yesterday were flying a DME arc approach with the autopilot, followed by an Raw Data VOR/DME leading to a circling approach. Not bad for a lad who has just completed his IR with less than 200 hrs

It's called "The Children of the Magenta." American Airlines has BALPA beat by a good 15 years because they identified this way back in the late 1990's.

Many operators, mine included, are starting to move back to basics with our training - a lot more non-autopilot work, basic IFR tracking, real-world emergencies that don't end at the conclusion of the memory items, etc.

I'll stop there because I can feel the "250 hour" rant in me starting. ;)

To be honest most of the 80 hour wonders start out quite proficient in manual raw data flying as that is what they did during their training, both on the real plane and in the simulator. However those old 20.000 hour longhaul captains with three landings in two months, all done on automatics down to 200ft are much less proficient if confronted in the simulator with an exercise in raw data flying.

And yup, quite a few carriers encouraged raw data manual flying for the last 20 years in their SOPs already, so it is nothing new for them.

Its my belief that only by spending significant time flying raw data do you build an effective instrument scan which will enable you to monitor the approach effectively when the FD is coupled. So many of the mistakes I see day-to day in simulator training are down to ineffective monitoring.

Non PC, I agree that raw data flying helps build the skills required monitoring.
Re sim; an interesting observation, but what appears to be ineffective monitoring could also indicate a weakness in the principle of monitoring or in understanding of how it should be applied.

No technical system would use a lower integrity system to monitor a higher one – you would not use a Cat 1 AP to monitor a Cat 3 system. Whilst comparing pilots to machines might be incorrect, why then do we use the less experienced crew member as a monitor?
We need to self-monitor – another source of weakness, but this is more likely constrained by workload, but so too for the PM in todays SOP dominant operations. More time is spent comparing the perceived situation with that expected in an SOP, with little thought as to actual (real) situation – our minds are becoming synthetic.

Children of the Magenta Line targeted a specific operator and certain situations; the click-click slogan is good, but where is the explanation how you determine when a click-click situation arises, and how is this identified – awareness, raw data; plane, path, people.

I have always thought the term 'PM' is wrong, and I prefer 'PNF'. 'PNF's' job is manyfold apart from 'monitoring' but he/she should remember that it may be a matter of seconds before he/she needs to become 'PF'. There is no real need for 2 crew to 'monitor' the A/P, 'PF' should be doing that as long as input and mode selection are checked, but when manual flight is happening, then the 'M' really begins.

Up to that point experience level is not that significant, and providing he/she is trained to a basic standard and imbued with company SOPs/procedures, 'PNF' should be more than capable for the task. With a new/inexperienced PNF it is a judgement call whether or not - and when - to 'inflict' A/P disconnect on them.

I prefer PNF as well.

I've often challenged crew in the sim as to why they are not actively monitoring the other pilot - that is, an obvious speed deviation or other such thing. Quite a number of times I've had the student(s) turn around and tell me they would do it on the line, but they didn't want their sim partner to get "caught out." Especially so on the rides, where I've only twice heard some one call deviations. Often it's more of an "under the breath" comment that they think only they can hear. ;)

I'm also training a wide range of experience - from 250 hours all the way up past 20,000 (where I'm not quite too sure who is training who some days). There is often just as much reluctance on the part of the experienced pilots when doing sim with someone with much less experience for fear they destroy confidence.

From this standpoint, I think we do need to re-evaluate (as trainers) how we are teaching what the PNF's job actually is and how best to communicate that a speed deviation call is much more important to catch than, say, making sure all the paperwork is complete. Unfortunately, way too much emphasis (of which I'm guilty myself) is put on the "learn all the SOP's for your ride" during an initial, which naturally carries through into the rest of their career.

It is generally agreed that automation addiction or automation dependency is the prime cause of a serious degradation in today's airline pilots manual raw data instrument flying skills. I think we are stuck with that forever as the manufacturer's and operators all are dead against anything other than full use of automatics.

One way however to force an improvement in manual flying skills is to require all instrument proficiency checks to be conducted on a 50/50 basis. Half the test session on automation to check that specific skill and the other half with no automatics and that includes the flight director off. Since the instrument proficiency test is usually over two hours then it means one hour of button pressing and one hour of pushing and pulling. Any takers?:eek:

The problem with that is probably 75% of current candidates would be unable to fly the manual raw data part of the test within the Regulator's skills test tolerances. If the pilot cannot pass that half of the test because of incompetence then someone has to make the decision to fail the candidate. That will inevitably cause some heart-ache in crewing. That costs more money to re-train the failed candidate. So it is back to full use of automatics all the time. :{

Offered it yesterday and offer denied...oh well..as long as i enjoy it.:ugh:

When you have people routinely disconnecting AP at 200ft on CAVOK days, you can't expect them to be confident enough to disable both AP and FD at 3000ft and fly without magenta...

I like that idea Centaurus. I already routinely fail the autopilot or give it as an MEL during training, but to force pilots to have to hand fly during a proficiency check would ensure we bring the skill level up, avoiding the very situation FlyingStone describes.

A manually flown ILS and go-around is a mandatory part of the rotary-wing proficiency check. Do you not do this for fixed wing?

For the training I've done on the ride, the go around is normally given off an RNAV or other non-precision circling approach, which is hand flown from the MDA through to the go around (normally followed by some form of engine malfunction). The ILS is then done as a single-engine approach which, for most of the types I've flown, the autopilot is not approved to be used for - so you have to hand fly it. But that's about it - maybe, 10 minutes of an hour and a half ride are hand flown - assuming you don't have to demonstrate stalls or steep turns.

It would be nice to see a little more hand flying though, which is why we've started incorporating autopilot failures into our training. At least that way we know they're getting the exposure.

This is a significant remark and needs attention and training for the individuals prior to checking otherwise failure rate would be too high and it is all the result of automation encouragement or mandate by many airlines.

The airline I do most of my contract sim instructing with has a good training program. Each pilot gets two by two hour handling sessions and two by two hour support sessions each six months. These are training sessions and merely graded 'satisfactory' or 'improvement needed', i.e. non jeopardy.
The third two hour session in the six month cycle follows within a day or two at most after the training and is a proper pass/fail check to the regulator's requirement.
So I tell them we will train hard and check as soft as we can legally get away with. During training they will have strong wind, turbulence, often no autopilot, sometimes no flight director. If someone is struggling with a particular exercise I may reduce the environmental difficulties, by easing up on the turbulence or wind and setting visual conditions for a while. Then we practice it some more until even the most average pilot usually 'gets' it before going in hard again. This may be at the expense of some minor item called up in the syllabus, but that can wait until next time if need be.
On the check ride the weather is set to minimums but turbulence and crosswind wound down to more modest levels.
During the check they can use the automatics except for one raw data approach and go around. Knowing that they trained hard, surprisingly some pilots actually voluntarily disconnect the automatics on approach even when not required to. Such is their confidence. It's a pleasure working with these guys.

Mach E Avelli;
interesting post. May I ask what methods and techniques you use to improve the scanning of the pilots and correcting deviations . I've seen experienced pilot getting embarrassed from their lagging behind in raw data flying and as you said their scan and handling improve after the second or third approach if we have time. during my training briefing I cover the unreliable speed indication pitch and thrust setting as a ref , avoiding large correction and including the symmetrical thrust in the scan.
Regards

+TSRA
Thank you for sharing Children of the Magenta.
Regards

Johnman, assuming that the trainee is beyond basic flying school and already has a good grip of the automatics I do the following:
Obviously brief on the exercise and give approximate pitch attitudes and thrust settings - but make the point that there is such a wide weight variation in a large transport aircraft that these can only be a rough guide. Also temperature, icing and altitude will further the approximation, so they should not fixate on learning exact numbers, simply have a fair idea of target settings.
Let them settle down in a particular configuration at the desired thrust and speed using the automatics. Wait for any pitch trimming to stop. The aircraft I am currently teaching has no auto-throttle, but if it did, I would then have them note the N1 and disconnect it.
Then simply tell them to disconnect the autopilot and make only the very slightest inputs - the idea being that nothing has changed, so no great control inputs should be necessary.
Build each configuration change from there by initially re-engaging the automatics, setting the new configuration, let it stop trimming, disconnect etc. Once they get the idea not to interfere too much or over-control, and use their support pilot to do all the donkey work with systems, switching, checklists etc, their scan seems to come good - usually within an hour or two. Possibly because they are not over loaded. Gradually their scan opens up to the extent that they can also manage PNF's activities.
I am no behavioural scientist so I dunno, but this training method seems to get good results.

Thank you that seems to be one good way to help solving a difficulty . I do encourage pilots in addition to keeping a good eye on monitoring automation to fly manually whenever it is feasible . I was surprised to read the following article.
Pilots play with lives by flying manually for 'ultimate kick' : India, News - India Today.