Originally Posted by lefthanddownabit
The ICAO document doesn't replace existing NAA documents such as FAA Part 60 or EASA JAR-FSTD. I'm sure the goal of harmonised simulator design standards is a fine one, but the industry has come close to this before but diverged again. It also throws a few curve balls into the process.
While this is true, at least for now, it is likely that the ICAO member nations will adopt the new standards. However, what some may not know is that adopting the new standards will have no effect on the simulators that currently exist … a process that has come to be known as “grandfathering” … and the new standards would only affect the new simulators coming into the inventory.
Originally Posted by lefthanddownabit
Confusingly it introduces a new set of device types without clearly identifying what they are (Types I-VII). From what I can see Type VII is a full flight simulator, so the document ignores the existence of Levels A thru D devices. It also sets some criteria which are in excess of existing (for example minimum lateral visual FOV is put at 200 degrees when many operators are struggling to meet the existing 180 degree standard set out in JAR-FSTD). That is 50 degrees more than most existing simulators. How much training in simulators requires such a FOV anyway? So the simulator world isn't going to change overnight.
True, the simulator world isn’t going to change overnight – but it
IS going to change. It is possible that you haven’t had the opportunity to read the entire set of documents and see the specific requirements including how the Summary Matrix was developed. In short, there were slightly more than 200 piloting tasks evaluated against 12 simulation features, where each feature has 4 levels of fidelity, and each of these combinations were distributed over 15 different kinds of training or license levels, and separated into two categories – training introduction and training to proficiency. This results in a rather massive matrix of just less than 300,000 data points. Obviously, this wouldn’t be manageable by many persons, so once this work was completed, there was a concerted effort to group together those training and license levels that had similar results in the matrix – and where it didn’t make sense to group types of training or licensing requirements together – they were left as a “group of 1.” It was this grouping that generated the 7 types of flight simulation training devices:
Type I includes PPL, CPL, and MPL-1 Core training;
Type II includes training for Instrument Rating;
Type III includes training for a Class Rating;
Type IV includes training (and training to proficiency) for MPL-2 Basic;
Type V includes training for Initial Operator, Recurrent Operator, Recurrent License, and Type Rating;
Type VI includes training for MPL-3 Intermediate; and
Type VII includes training to proficiency for MPL-4 Advanced, Type Ratings, ATPL, Initial Operator, Recurrent Operator, Continuing Qualification, Recurrent License, and training for Recency of Takeoffs/Landings.
The traditional Level A, B, C, and D simulators will continue under the “grandfather” authorizations.
The field-of-view (FOV) requirements were not blindly selected. If you fly an airplane at the designed traffic pattern altitude on the downwind, the location of where you would normally check the runway threshold to begin your turn to base would require you to turn your head just very slightly less than 100 degrees toward the runway. Therefore, for both sides of the cockpit, the required FOV for the visual system would be 200 degrees to fly a VFR traffic pattern. This 200 degree requirement is also just under the logical horizontal FOV for an installation using just 4 visual projectors (going to something like 210 degrees would require a 5th projector). An FOV of just 180 degrees would put whatever 90-degree reference point you may desire right on the very edge of the visual display and any movement might place that point out of view –
and the 180-degree system would also require 4 projectors.
Simulators are currently being delivered with 200 degree FOV visual systems rather regularly.
Originally Posted by lefthanddownabit
As has been stated above, unusual attitude and jet upset training requires data which doesn't exist at present. Incomplete data could well result in negative training. Even stall behaviour is not adequately presented in current airliner aero models, nor does it have to be.
While it is true that a sufficient quantity of data has not been routinely collected that will allow an accurate development of a mathematical model of the aerodynamics of a given aircraft beyond the boundaries of a given flight envelope, it is also true that taking the simulator outside of that envelope is not like it used to be. Aerodynamic math models used to be terribly limited – to the extent that rolling the simulated airplane to a bank angle of something like 60 degrees would confuse the programming and the simulator would often freeze in position. Those kinds of circumstances are not seen in simulators built over the past couple of decades. These “newer” aerodynamic models are fully functional through 360 degrees of roll and 360 degrees of pitch. But – like I said – these models have to be modified with specific airplane data to allow the simulator to perform and handle like a given airplane. It is
this data that is missing for a completely accurate math model to be developed.
Also, while I said there was not a sufficient quantity of data beyond certain limits, that does not mean that data has not been gathered beyond those limits. Data does exist well beyond the supposed “flight envelope,” but it’s not sufficient to be able to relied upon 100%. So going beyond the “flight test validated envelope” is possible – and while not 100% accurate, it is quite accurate just beyond the limit and gets less so the farther away the simulated airplane is taken from those limits. However, there are some things that are still quite accurate regardless of the attitude of the simulated airplane … the visual display will accurately display what the pilot would see in that attitude; the flight instruments will accurately display what the pilot would read on those instruments with the airplane in that attitude; the performance indications of the change of those instruments (rolling, accelerating, climbing, descending, etc.) would be accurate with regard to what the airplane would be doing in that attitude. Where this lack of data becomes problematic is the aircraft response to given control inputs. The simulator will respond to control inputs to the simulator controls – but we don’t accurately know how quickly the airplane will respond and we don’t accurately know the magnitude of that response. Therefore, we cannot use the simulator to teach or test a specific control application, with the expectation that a pilot experiencing that same circumstance can expect his/her airplane to respond the way the simulator did in those areas.
Another piece of information is that we do have substantial data on aircraft performance and handling qualities all the way down to the stall – that means to the stall break or the controls reaching displacement limits. It would be incorrect to say that going just barely beyond that point puts the simulator into “no man’s land.” But, it is also true that the performance and handling qualities at that point cannot be absolutely guaranteed – it will be close, but not exact.
Originally Posted by lefthanddownabit
UA and jet upset scenarios have been programmed into simulators for decades, but risk negative training due to lack of data and the unrealistic way such upsets must be introduced. An early set of UA scenarios I recall simply put the simulator in a particular condition, frozen, then required the crew to recover when freeze was released. Some ramp in the attitude and the crew are requested to be hands off during this process. You can subtly induce a UA by failing attitude indications, but if the failure is identified the UA is never reached and so no recovery is taught.
Because of these limitations and because of the value of training in a simulator rather than risking exposure in an aircraft, there are teams currently working on how to use what is available in existing simulators, recognizing the limitations that exist, and structuring training programs around the areas that could be problematic yet still providing the crew members with very valuable training on what to expect and what courses of action might be considered in an array of circumstances. Teams are also working on what kind of data might be able to be gathered and what kind of compromises might be available for the structure of an accurate math model where we’ve never had math model accuracies previously.