Aerobatics in Flight Simulator

By Andrew Herd (3 May 2004)

Our Rallye having shed a cam as it was being taxied in for its annual and the RV9 having only just reached the stage at which we can sit and make airplane noises in it, I took advantage of the unexpected spare time to do an aerobatics course. After about the fourth lesson, as I was trying to get my head around the finer points of the barrel roll, it occurred to me to do some practice in Flight Simulator - after all, I have used the sim to do virtually everything else that can be done in a real plane, so why not aerobatics? The default installation even includes a plane designed for unlimited aerobatics, the Extra 300S, and if FS2004 didn't work, well, I could just use one of the combat flightsims instead. So I sat down to give it a trial - and this piece is the story of what I found.

It might be worth backtracking a little before I get into the meat of this tale. Like most private pilots, I have experimented with limited aerobatics - wingovers and chandelles and aileron rolls in a Yak a friend of mine part-owns (the shot on the right shows an aileron roll in the Yak and while it is has well coordinated controls which make it a delight to fly, I just wish the designer hadn't chosen to exhaust the engine through the cockpit. The Yak must have the smelliest interior of any plane I have ever been in and even half an hour in it is enough for me, though I am told you get used to it). All of these figures can be practised without any trouble in the sim, given an appropriate aircraft, because one thing they all have in common is that they can be performed with the pilot's attention on the view up front. The one exception is the wingover, which really should have the entry and exit parallel to a convenient line feature, but when you are up there, any maneuver that feels good must be good, right?

Wrong. If you ever indulge in competition aeros, you will find yourself being judged by anal retentives with giant protractors and pernickety minds, who will happily penalise you for being a couple of degrees off line. Besides, what point is there doing a thing if you aren't going to do it well?

Light aeros kept us happy for a long while. There were some hilarious moments: the time my friend Roger inverted us at the top of a wingover (that's where the calculator went, said I, as it whizzed past, going down, except to us, it was up); and another when I inadvertently hit the PTT button and gave ATC the benefit of my impression of the famous Robert Duvall scene from Apocalypse Now, while Rog loudly hummed the theme from Die Valkure - who knows what the guys made of me growling, 'I just lurrrrv the smell of napalm in the morning,' but fortunately no-one has ever said anything about it.

We also discovered that the one really big problem with digital cameras is that a three second shutter lag isn't ideal for getting shots of maneuvers that are over in four seconds, and that you need a really firm grip when you are pulling more than a couple of G, but half the fun of life is finding these things out for yourself, even if everyone else knows the answer.

But though our plane is certified to do a remarkably wide range of aerobatic figures, I have never been that keen to push it anywhere near the limit, chiefly because it is 30 years old. Just because the manual says you can do a thing doesn't mean to say that you should do it. So when the opportunity arose, I signed up to learn how to do it properly, in a 160 horse Slingsby T67 Mk II. As you can see, this is a mean, lean, yellow machine and though it has barely half the grunt the Extra puts out, it can still perform the vast majority of aerobatics without any height loss.

Now while the Slingsby is a pleasure to fly, with an exhaust-free cockpit, it has a slightly evil reputation, earned because it has two types of spin. The first is a perfectly normal, nose down spin; the second is a flat, very high rotational rate descent spin that is extremely disorientating and which has claimed several lives. The flat spin is rare, but controversial, with some maintaining that the condition only occurs because of faulty control inputs, while others claim that it is caused by a fault inherent in the airframe. The one problem I have noticed is that the stick is located unecessarily far forward, which could make it difficult to do a smooth push to the stop after stopping the rotation. But whatever, by anyone's standards, the T67 is a nice plane - it feels and handles like a thoroughbred.

This makes a good moment to talk about spins. A while ago, spinning was a prime topic in the forums and I am sure that many simmers went away with the impression that most pilots do spins all the time - whereas the reverse is the case. Many GA pilots under a certain age have never done a spin because most jurisdictions removed the requirement to learn how to recover a spin from their syllabi years ago. This sounds whacky, but it was done in order to emphasise spin prevention, which has the force of a certain amount of logic. To hold a C of A, a plane has to be designed to be as safe as possible, unless it is specifically intended for aerobatic use - which means it should give lots of warning before a gentle stall, with the least possible tendency to drop a wing into a spin. The reason you rarely hear of GA planes spinning is because they are designed not to and while the vast majority can be persuaded to spin, you usually have to work on it and some even have to be held there with full pro-spin controls - so that if you centralise everything, they will recover themselves. I find this a most comforting thought. Even the T67 gives plenty of warning before it spins, enough time to pick up the wing with opposite rudder and release the back pressure on the stick, which is all it takes, mostly. An accelerated stall is another thing and many aerobatic planes will flick out of high G maneuvers into a spin if you don't listen to what they are trying to tell you, but that is way beyond anything the average 172, or PA28 pilot is likely to get up to. Unintentional spinning is exceedingly rare, though that doesn't mean to say you can afford to forget about it.

I am writing this piece as when I tried practicing aerobatics in FS2004, I found there are some significant gotchas - a couple of which can't easily be overcome. Microsoft's view on the subject, as I have it, is that on the basis of the reasoning above, FS is not designed to simulate departures from controlled flight and there is no intention to change that. This is fair enough, though when I asked the question of the development team, I didn't think to ask why the Extra had been included among the default planes, given that the whole purpose of the Extra is to fly the kind of maneuvers that FS2004 does not support - like knife edges and avalanches - and leaving it there is a bit like permitting simmers to smell a glass of vintage wine, but not to drink it.

Just so we can all play in the same ballpark, the obstacles to using Flight Simulator for aerobatics are:

1. the limitations of the simulator at the extremes of flight

2. the 'tunnel vision' imposed by conventional monitor setups

3. the lack of control forces

4. and the lack of G forces

The interesting thing about this list is that the items on it apply to all simulators, even combat flight sims. When next-door's ten year old is shooting me down - again - in Forgotten Battles, I console myself with the thought that if the game really was as real as it got, then Walter Nowotny's 250 kills wouldn't stand out much. With persistence, even an average combat sim pilot can rack up hundreds of victories and the games would sell far fewer copies if they were so real that it was necessary to fly twenty missions before you so much as clapped eyes on the opposition. Real air to air combat was more a case of hours spent freezing your butt on patrol or escort duties, followed by enounters that were over before some pilots knew they had started - the guys in question being the unfortunate ones who made up the kill statistics. Home simulators dispense with the boredom and they also free you from having to cope with a whole clutch of things that gravity and human physiology enforce on pilots, which is one of the reasons why combat simulation is more popular than doing it for real - you can't get hurt and the enemy tow around little tags so you can see what they are up to at all times. Personally, I wouldn't have it any other way, but the reality is that flying combat-type aerobatic maneuvers is a whole magnitude more difficult than the simulations would have you believe, or we would all be flying fast jets.

Let's take the items on the list in turn. Flight Simulator has reasonably realistic stall behavior, but push the limits any harder than that and the bets are off. I have yet to come across a plane which will do a convincing spiral dive; aggressive flight model tweaking is necessary if developers want their planes to spin; and flick rolls do not appear to be possible, which isn't entirely unexpected, when you consider that a flick is actually a class of a stall. Many Il2 and Forgotten Battles planes also exhibit odd behavior when they depart from normal flight, though the planes in the Aces Expansion Pack are better in this respect - apart from the way the Ta-152 develops such a stable spin that if you don't rescue it after the first turn the chute is the only way out. The one lesson you can take away from this is that simulating departures from normal flight is not a priority for game developers - though it is easy to argue that they should be if simulations are to become less arcade like. Some maneuvers actually depend on the plane making a departure from normal flight, the flick being a prime example.

Monitor formats also put significant constraints on aerobatic simulation, the reason being that when when you are doing aerobatics, you don't spend much time looking straight ahead. Taking the loop as an example, in a T67, you dive to pick up speed to 120 knots, then begin a pull up, looking quickly left and right to check the wings are level and that the aircraft isn't yawing as it is pulled up into the vertical, after which you tip your head back (it would be up, were you in level flight) to pick up the horizon behind, align the wings with it as the plane goes over the top and then keep your head held like that until you spot whatever line feature the loop is based on as the aircraft slides into the vertical, before pulling through. Just about the only time you look at the panel is to check the airspeed indicator before executing the pull up. A loop is not so difficult to do in a simulator, beyond that quick look from one wing to the other early on and even that can be done with efficient use of the hat control.

Now try a barrel roll. Here the plane is continuously rolling and pitching around an imaginary line drawn several thousand feet up in the air - a barrel roll being a loop combined with a slow roll. The best pilots can do a barrel roll without spilling a pint of beer balanced on the seat; and so can I, as long as the seat is in the bar. I find the best way of doing a barrel roll is to find a small cloud level with my horizon and to roll the plane around that, which means looking at a point somewhere left and above my normal eye level to do a left hand barrel. Again, this is possible in FS, as long as you are slick with the VC, can handle the moment when your attention moves from the airspeed indicator to the cloud, have a key set up to centralise the view forward and nothing major goes wrong. The trouble is that all the problems with barrel rolls tend to set in somewhere around the point at which the plane becomes inverted, where instead of pulling back while you roll the plane, you have to push forward, in order to float the aircraft over the top. The movements needed to do this are hardly intuitive and in the simulator, if you lose sight of your cloud, it is virtually impossible to handle all the control inputs at the same time as adjusting your viewpoint, a problem that clearly doesn't arise in a real plane. To give an idea of how difficult a good barrel roll is to fly with a simulator type forward view, I have included an avi of one taken from the Slingsby - as you can see there is no visual reference for half the maneuver and notice how far behind the AI lags. Do not ask how many times I had to fly this figure to get the shot, but I assure you it was fun trying.

OK, you say, why don't we use TrackIR? Great piece of kit, I reply, but how does it keep tabs on the spot on your forehead when you are looking directly up? Of course you can't look directly up, because otherwise you would lose sight of the screen, so TrackIR might help doing aeros in the sim, as long as you accept the need to restrict your head movements to a degree that won't feel right if you also do aerobatics in a real plane. But if the opportunity ever arises, I will have to give TrackIR a try, because it is the only answer I can think of for this particular problem.

Even without TrackIR, why not use the instruments to fly the maneuver? Well, that would be fine, were it not for the fact that in a real aircraft, a single loop is enough to topple the gyros, following which they are inclined to sulk for many minutes, which is why you see a control for 'caging' the instruments in some planes. You cage the gyro, fly the figure, release the gyro, no problem. You will notice that the horizon doesn't topple in the avi, but I flew that after a pause to get the camera set up and a barrel roll is such a gentlemanly figure that the instruments sometimes don't realise they should quit. Fortunately for us, Flight Simulator gyros are made of such tough code that they don't fall over like real ones - so as long as you can tolerate certain lack of reality in this respect, you could use the panel to fly most figures, though bear in mind that even in FS, the artificial horizon gets funny when the nose is pitched more than about 45 degrees away from the horizontal and that flying aeros with your eyes fixed on the panel kind of misses the point, given that your view is more often something like the one in the shot on alongside, which is a right handed barrel roll. As you can tell, I enjoy barrel rolls.

The lack of control forces is a harder nut to crack. I have tried force feedback joysticks, but none of them are sophisticated enough to give an impression of what goes on in a real plane that is being flown to the limit. If we go back to the loop, the entry is flown with something like a 3 G pull up, with the result that a plane like the Slingsby is knocking on the verge of a stall as it passes the vertical, with all the associated buffet that goes with that; and then the buffet goes away and the controls get progressively lighter as the plane nears the top, before everything becomes heavier again and the buffet returns at the 3/4 point, as the nose goes back down through the vertical. At this point, another 3 G pull is necessary to complete the figure, but this time at a much higher airspeed than on the way up and depending on how fast the plane is going, aileron snatching can occur. Even the best force feedback seems to use algorithms that don't duplicate any of this stuff and furthermore, if you pull too hard, many FF setups commit the cardinal sin of 'yielding' momentarily. While this gives a quite good impression of what aileron snatching feels like, it never seems to happen at the right moment. I cannot think of a way around this.

Does it matter that control forces can't be simulated? Well, yeah, I think it does, because 'feel' is an important part of flying many aerobatic figures - for instance, when the plane starts to buffet, you relax the back pressure to prevent a stall, and when the controls become light, it communicates the need for proportionately larger inputs and so forth. This lack of feedback is the Achilles' heel of combat sims, because when you are flying a big hairy fighter right on the edge, the lack of feel can result in unpredictible stalls, which I find quite a problem in Forgotten Battles. Heresy follows, because the very worst joysticks for flying aerobatics or combat type maneuvers are ones with big heavy springs like the Hotas Cougar, which, as supplied, gives the impression that flying fighters requires biceps like an all-in wrestler, whereas most real planes of this type can be flown with one finger. While I am on the subject, the other weakness of sticks like the Cougar is that the springs impose uneven control forces on the diagonals as compared to the fore and aft and side to side axes. While one set of control surfaces might be, or might become disproportionately heavy in some phases of flight in a real plane, the increase in pull required remains even, without any sudden increases in resistance.

The lack of G is the final problem. Though sims make a gesture in this direction with red outs and tunnel vision, I cannot imagine G effects will ever get simulated to any signficant extent, though curiously, you don't notice the phenomenon as much when you are flying a real plane as you do when you are a passenger. But it is there, all right. Sir Isaac is ever with you in aerobatics, because your stalling speed varies with the square root of the G force experienced, so that at 4 G it has doubled. That means that a plane like the T67, which stalls in level flight at 50 knots, needs a hundred to keep flying with 4 Gs on the meter and it is very easy to miscalculate and pull that on the way out of a badly executed vertical maneuver. Fortunately, the plane is usually going fast (sometimes too fast) at this point, which saves you, but a G-loaded stall is a frightening experience the first time it happens, because the aircraft departs abruptly from controlled flight and it can be quite tough working out what is happening for the next few seconds. Interestingly, some instructors advise intentionally putting the plane into a spin if you can't identify an unusual attitude, though in the normal course of events you just level the wings and recover to the nearest horizon. The screenshot here shows the view looking half right as we are about to roll right way up out of a half Cuban - not so difficult in the sim, but if you imagine yourself accelerating upside down at a 45 degree angle towards a windshield full of fields, you will believe me when I say it takes an effort of will to stay in charge of the plane.

There is one other feature of aerobatics that is hard to simulate. This occurred to me yesterday, when I blew a stall turn and left us hanging in space, nose pointing heavenwards, while our already low airspeed declined to nothing. The engine was roaring away at 2600 rpm and I could hear air traffic berating some guy for a late downwind join, yet my main impression was of a long silence, while Sir Isaac made up his mind which way he was going to have us go. In those circumstances it is difficult to ignore the sound of your heartbeat.

Which is why I always leave the beer in the bar.

Andrew Herd
andy@flightsim.com