Thursday, September 23, 2010

Not All Malfunctions in a Simulator are Simulated

I'm on vacation at a lake in British Columbia, just as I threatened. The lake is lovely, as is the town, the accommodations and the company, But I'm not saying who that was nor blogging about my vacation. I'm blogging about the fact that I have to leave this lovely lake and drive an hour or so to the airport, and renew my IFR. I find the school and check in with dispatch, where I do some paperwork. They tell me my instructor's name. I've never met anyone with this name before. It's the same as the name of a plant, and it doesn't tell me anything about them, except that perhaps my instructor is the child of late-blooming hippies.

Oak (not the real name, but I'll use that to avoid excessive circumlocution) arrives and he is a he. He doesn't look like a hippy, is familiar with the information I gave the school on my experience and needs, and seems to know what he's doing. He proposes we sim today, then fly tomorrow. He's concerned that the fact that I've never flown this airplane before might be a problem. I'm concerned that I have to take a flight test, and AAAAAHHHH! FLIGHT TEST! I'm not worried about flying a new airplane compared to that. Every time I have flown a new airplane, it's turned out to work like all the other ones I have flown. The only tricky bit is usually landing nicely. On an instrument ride if you get to the flare without screwing up too badly, you've passed. I know someone who had to have the instructor land the airplane for him on his initial IFR, because the instructor had arrived late and postponed the flight test later into the dark, for a candidate who did not yet possess a night rating or any night landing experience. (And if he's reading this, please get in touch: both paper and e-mail is bouncing).

We go into the simulator room and start one up. It's a typical flying school simulator with a yoke and pedals and movable controls on a dashboard, into which is set a simulated instrument panel on a computer monitor, connected to a console running some flavour of flight simulator, possibly even the Microsoft one. Once upon a time school simulators had real gauges all the way across with little motors running everything, but those are more expensive to construct and maintain. Real airplanes are going towards glass now, so the simulated ones do too. The pilot seat is on rails bolted to the floor, and after it has been adjusted to the comfort of the pilot, it does not move in any way related to the simulated movement of the airplane. The whole things has been certified by Transport Canada as conforming sufficiently to the characteristics of whatever it simulates to be used for certain types of flight training. As this training is just to acclimatize me to the local navaids and procedures and to assure the poor flight instructor it's okay to fly with me, its certification doesn't currently matter anyway.

I start running through the instrument checks. This is partly a simulation of how I will test all the instruments on the flight test to show that I am a good little pilot, partly a test to make sure the simulator is working properly, and partly a chance for me to familiarize myself with the location and operation of all the instruments. You know how in your car you can put your finger on any control pretty much without looking, but in a rental you have to wait for a stoplight and hunt around for it? Well same in an airplane. In a simulator of this type some of the switches tend to be rinky-dink little things and not located in logical places, either. "Where are the cowl flaps again?" All the installed instruments are familiar ones to me, and I know how to use them, but figuring out how to adjust the rightmost digit of the ADF frequency can be like trying to turn off the windshield wipers on the rental car. Just when you think you have killed them, they turn out to be set on intermittent.

Above is a literal museum example of the oldest kind of ADF I have used. I think anything older would require separate manual tuning of the sense antenna.

I think I may have ranted thusly about ADFs before. Those of you who know or don't care about the various ways one might set the frequency on an ADF receiver can safely skip the following two paragraphs with no loss to the narrative. If you want to know what it's actually for and what it does, not just how to tune it, go here.

An ADF can be set to either a three or four digit frequency, that is between approximately 190 and 1750 kHz: the AM radio band. I believe all the NDBs that the ADF has to receive are in the three digit range, from 190 to 600 kHz, but I speculate that ADF was designed to also receive higher AM radio frequencies, so as to use commercial radio stations for navigation. Radio towers are published with their frequencies on old charts and still better than nothing in the middle of nowhere. The oldest sort of ADF I have used has one big knob that moves an indicator along a scale which you carefully examine to see which frequency you have selected, and then you listen really carefully to the Morse identifier, because you could easily be a few off. That sort is a museum piece now, and pretty much was then, but I doubt I was the last person in the world to use one of those in IFR conditions. The next oldest kind has three frequency adjustment knobs, the way an analogue transponder has four knobs to change the squawk code. The leftmost knob changes its corresponding "digit" through the range one to seventeen (or maybe fourteen, I don't remember) and the other two change their digits through the zero to nine range. The more modern kind (although "modern" and "ADF" don't go together well) has a digital display and the knobs are more compact. Typically there is one big knob with a smaller knob in the centre. They correspond to the three knobs on the older unit like this: the big outside knob does the job of the leftmost knob; the smaller inside knob does the job of the middle one, and then you can pull out the middle one and then when you turn it, it does the job of the rightmost knob on the old unit. You could be stuck for a long time if you didn't know the pulling out trick.

In simulators like this one, for some weird reason, you don't get to pull out the inside knob to set the frequency. I guess someone was shown how it was supposed to work, didn't realize that the person showing him had pulled out the knob, so made up his own way, and pilots are so used to ADFs working different ways that they shrugged and considered it a limitation of the device. So in the sim you turn the outside or first knob to set the first one or two digits, the 1-17 thing, then you turn the second knob to set the second two digits, from 00 to 99. Like that's not a pain. You have to do this sometimes during a missed approach, where one beacon is the missed approach point and the missed approach required you to track to a subsequent NDB, so you have to scroll through up to 50 frequencies while cleaning up the airplane in a simulated go around.

I dial in the frequency of the NDB nearest my simulated airport so I can test the device. It identifies and points, so I then try to tune the first NDB I'll need during the flight. It's at this point that I notice that as I adjust the third digit, I'm only getting even numbers. "How do I get odd numbers on this?" I ask Oak, who is sufficiently confused by the question to clue us into the fact that it's broken. We switch to the other simulator and go through testing everything again before flying.

The simulated ATIS unexpectedly (to me) works, and I copy it, then address Oak as "ground" for my clearance. He gives me the Abbotsford Seven departure off runway 07, The first seven means that there have been six revisions to the procedure since it was first published and the second seven means that the runway is oriented such that the airplane will be on a heading of 070 degrees magnetic as I take off. The departure procedure is a number of instructions, such as the fact that I'm supposed to climb on runway heading to six hundred feet then hang a right to 202 degrees and wait for ATC to give me vectors on course.

The procedures are not a problem, but I do get caught up in getting every detail right, and my altitude control is abysmal. That's not abnormal for the sim. The trim is in the wrong place and doesn't have the same feedback as in an airplane. I make a few dumb mistakes but demonstrate a basic knowledge of how to fly an airplane, so tomorrow we are scheduled to fly one.

I don't want to spend too much time in the simulator because it doesn't even try to simulate the airplane I will be flying, and you do gain habits to appease the simulator. This article suggests that because much better simulators that are used for airline pilot training do not do a good job of modelling aircraft behaviour as the airplane goes out of control, simulator training may be to blame for pilots responding incorrectly to emergency situations, causing some crashes, such as the Continental runway excursion in Denver and the Colgan Dash-8 stall in Buffalo.


Aluwings said...

Re: "simulator training may be to blame for pilots responding incorrectly"

Negative flight training in simulators is a pet bug of mine. Not only due to the technical limitations but also from bad habits that are ingrained by simulator training practices in general.

Case in point: The training to deal with Electrical Smoke in the Cockpit: for years crews were taught how to laboriously follow out a very long and complex checklist - pulling many CBs to isolate the cause of electrical smoke etc., And if this emergency happened in the middle of the Alantic Ocean, maybe that was the only choice.

But realistically, most of the time the best option involves landing the aircraft asap, regardless of landing weight, etc.. After the Swissair crash near Halifax, in which valuable time was lost working through a long checklist, the training emphasis in Canada was significantly changed.

And the entire concept of "new airplane, new day" in training, where one serious emergency failure is "cleared up" all of a sudden and a new flight begins with an entirely new scenario, incorporates all sorts of negative and incomplete training.

Aviatrix said...

The electrical smoke issue wasn't so much a failure in simulator training as an erroneous concept in emergency handling, period, although I suppose the lack of choking smoke and melting plastic dripping from the ceiling gives the pilots a false idea of what that might be like.

SR111 was very early in my life as a pilot; I can't remember a time when smoke-in-cockpit did not mean we land NOW and ATC pushes the fire button. I wish I had a photo of the big firefighter in full gear looking at a C152 that he would in no way fit inside. The airport didn't have its own fire department, so he was from the local town and I think coming to the airport was an interesting diversion in his day. I showed him where the fuel lines ran, and how to shut off the fuel and electrical, for future reference.

k1mgy said...

Before this article I'd always assumed, and swallowed the line, that modern simulators are capable of reproducing every flight condition in full fidelity.

Who tests and evaluates a simulator? FAA in the US? If so, how did something obvious as crosswind slip by?

Aviatrix said...

It's not so much a matter of 'slipping by' as the fact that airplanes become less predictable and much harder to model at low speeds. You can't even predict exactly how the same real airplane will stall three times in a row in practice.

Sarah said...

Thanks for the links clearly explaining how an ADF works. It's still unclear to me how it works with a std. AM broadcast station, but the rotating phase signal transmitted by an NDB - cool.

... recovering electronics geek and Tesla fan.

Best technology of the 20th century ( near magic ): electronics & radio. And Aircraft, of course.

Aviatrix said...

Oops, wrong technology. The VOR transmits a signal that varies in phase with the direction from the transmitter. But the NDB is extremely simple, transmitting the same signal in all directions. The only difference between the NDB and an AM radio signal is that the NDB transmits only a Morse code identifier, while the AM radio station transmits traffic reports, oldies music, used car lot commercials and, until last week or so, Doctor Laura.

Sarah said...

Er, well, no. I know why you were thinking of VORs from my comment, but I was interpreting the "Theory of Operation" paragraph in the avweb link you mention.

Given a vertically polarized AM signal, the ADF will find it - so in that sense I was wrong about the signal being special for an NDB.

I need to find a EE. Don't really understand this without a little math. In rough terms, the loop antenna gives a +/- 180 degree bearing and the sense antenna disambiguates that.

word verification: flies

Aviatrix said...

Oh, I didn't read the avweb article, just linked it for the pretty picture. I didn't realize that it went into that kind of detail. Ah, I see it doesn't really, just flies right up to it then swerves away at the last moment.

The article doesn't make clear, but I think from your comment that you're already aware, that all electromagnetic radiation consists of perpendicular oscillating electric and magnetic fields, and both NDB and AM radio antennae are vertical, giving vertical polarization of the resulting electric fields Yes, the loop antenna gives the line of propagation of the while the sense antenna disambiguates the direction of propagation.

This paragraph should be a clear logical explanation of how the sense antenna does that, but I don't remember ever having a firm enough grasp on the topic to convey it to others. I'll figure it out and then do a blog on it later. Anyone who wants to help may e-mail me.

Ed said...

A neat (though expensive) example of the negative effect of simulator training which I was told about ages ago was a training captain who was sent to Boeing to do the acceptance checks on a new aircraft prior to its delivery flight to his airline.

After the test flight went well he decided to exercise the fire handles (just the fuel cut-off stage, not the extinguishers, of course). All four engines ran down as expected. Normally in the simulator that's the end of the exercise and the next job is tidy everything up ready for the next so he did that on the real aircraft. Unfortunately, the effect of putting away the fire handles was to introduce gobs of fuel into the still hot but not turning engine with fairly dire results, including his taking somewhat early retirement.

Don't know how distorted that story might have become in repetition or in my memory but it's a relevant point that in the real world there's not such a neat end to situations.