I've asked this question to my flight instructor, who believes flight simulators do a very poor job of modelling flight physics for very light airplanes, which makes them useless for preventing accidents.
Considering most accidents are caused by loss of control this explanation does make sense. On the other hand, flight sims might help in dealing with engine failures.
I intend to fly in a plane derived from Zenair CH-701.
Perhaps yes, but it's not easy to qualify objectively.
Simulating a light airplane may require a different approach in some areas, but it's not inherently difficult compared to 'heavy' ones.
However, the truth is, nobody cares that much, so there is much less investment in such simulators. A decent FTD (flight training device, something a step lower than what is legally called a 'simulator' (FFS)) will cost easily several times more than an ultralight, while the risk benefit is arguably lower (compared to real flight).
Secondly, type-rated simulators are rarity even on the 'proper' GA market; typically FTDs are 'generics' that aim to simulate a representative 'class' of aircraft. This makes fair comparison difficult, especially by real pilots who have most experience on a single/few real types.
Then, there are technical features that are even more important for light (and GA) airplanes than for big ones. One is a good visual system: in real visual flying, you obviously use it more, and having a single monitor stuck in front of you will hamper you more than you might think.
Second is control loading: on light airplanes you get airspeed and trim feedback right at your fingertips. There are no cheap solutions to this. You may argue this is not 'physics', but it actually is, this is all part of a feedback between the airplane and you being the part of control system. Lack or poor quality of these things (which is more than common on the simulators used for light planes) will substantially affect subjective and objective qualification of the flying qualities/handling of the simulator.
There are areas where even expensive FFS suffer. A typical example is ground handling. Or, speaking about 'loss of control', many regimes that result from it, notably stall and spin, are actually difficult to simulate well, and are often not required for certification.
That all said, it's an overstatement to say that simulators are 'useless for preventing accidents'. Flying is more about thinking ahead, planning and situation awareness than any specific piloting skills, and even a game simulator can teach you some of that.
They do a good job modeling the majority of the flight physics for airplanes with a few exceptions such as edge of envelope flight characteristics and some power plant characteristics. Eg FSX does not model stall/spin characteristics well or turboprop characteristics well or autorotation characteristics for helicopters well. X-Plane 11 is purported to have good flight physics models in them and they can be more custom tailored to specific aircraft than FSX or Prepar3d can. That being said, some of the aircraft handling actions are not quite the same as the real thing.
In the end desktop sims are not airplanes; they are just mathematical algorithms that closely match how an airplane flies. It’s similar to saying that a synthesized violin sounds pretty good - but it’s never gonna be a hand made 1727 Stradivarius. That being said those same desktop sims are good enough that the same mistakes I make in the sims, I will make in the real aircraft. One thing anything short of a dedicated Level D full motion sim just can’t do is simulate the kinesthetic feel of an airplane. That’s something you just don’t get sitting in an office chair with a joystick and staring at a 20” 2-D screen. While VR sets can simulate what you see out of an airplane, it’s still not quite the same thing.
in the limit of aircraft so light that their weight is of order ~ the weight of the air they displace, the usual and customary rules of aerodynamics i.e., vehicle responses to control inputs goes out the window and the dynamical analysis gets a lot more complicated.
For example, in the case of macready's gossamer condor, deflecting the right aileron down did not roll the vehicle to the left; instead it skidded the vehicle to the right.
So it does not surprise me that consumer-grade flight simulation packages might do a poor job with predicting the response of ultra-ultralights.
I am in solid agreement with two statements made above:
Flying is more about thinking ahead, planning and situation awareness than any specific piloting skills,
One thing anything short of a dedicated Level D full motion sim just can’t do is simulate the kinesthetic feel of an airplane.
Sims are good for practicing the patience part of flying, especially setting up for the pattern and landings, and instrument approaches and flying the procedure. Without the feedback of the plane moving with you, and just having the screen view tilt in front of you, the visual part quickly becomes unrealistic as you make maneuvers. Especially simulating the wind - the screen jumps and bounces around, but sitting in your chair it just becomes annoying.
And depending on the plane you are flying, the joystick/mouse on your computer will allow full range of unlimited movement, while depending on the phase of flight the plane may require increasing amounts of force to actually move the stick in full range, and if not properly trimmed may require a lot of pressure. For example, I was trying out a simulator at a training school. Took off okay in a Beech Baron or something like that, but the trim was set wrong, so I had to really fight the pitch pressure while I searched for the elevator trim to keep from climbing up into a stall.
