Stewart platform-based flight simulator

So, that's a professional hexapod- a. k. a. Stewart platform-based flight simulator (image source: Wikipedia - "Stewart platform"). By moving the simulated cockpit, it simulates the movements of a plane during its maneuvers.

But the range of maneuvers it can realistically simulate is certainly limited - it can certainly simulate a straight flight, but I doubt it can give its occupants the realistic impression of a full barrel roll.

Given an existing, recent (i. e., modern and advanced) siulator for commercial aviation:

What are the actual limits on maneuvers and their parameters (i. e. turn and turn radius, banking and bank angle)?

Bonus question: How does this compare to the flight situations encountered in a typical flight?


2 Answers 2

  1. What they can do. Motion systems can produce accelerations that stimulate the acceleration sensors in our inner ear. Higher frequency acceleration can be sensed instantaneously in surge, sway and heave. From this answer:

    Motion systems can do a pretty good job at reproducing direct accelerations, until the actuator stroke runs out of course. So for short term accelerations.

    This direct acceleration sensing is essential for pilot-in-the-loop motion control, that is why we have the sensors in the first place. The velocity information we get from our peripheral vision is an integrated signal and has 90° phase shift, causing a delay in sensing of motion.

  2. What they cannot do. Produce sustained accelerations. Six Degree of Freedom motion systems get by by tilting the sim, so that a component of gravity can be used for the feeling of sustained motion. But with some restraints on tilt velocity and angle, the tilt must happen imperceptibly and is of course also limited by the actuator stroke:

    • Imperceptively, by rotating at a rate below what the inner ear sensors can detect, and by not changing the visual image horizon.
    • Limited by stroke: in a 6-DoF system, using full tilt means that some of the other DoFs cannot be used. Full aft tilt has the front actuators fully extended, so linear surge cannot be actuated anymore.

Motion software contains a lot of fine balancing and frequency filtering code, to make optimal use of the available stroke and of the peculiarities of the inner ear acceleration sensors. Large stroke systems are essential for pilot-in-the-loop tasks in airliner and helicopter pilot training, and make the difference between being in an immersive simulated atmosphere, or being aware of the fact that you're inside some device firmly on the ground.


It's not restricted - the sim will continue to move through any maneuvers - but realism degrades the further away they get from zero.

Full-motion simulators don't map motion 1:1, but rather trick the senses. Maneuvers are performed with velocity and acceleration. What the human body senses is acceleration. And what the brain processes most actively is jerk - the rate of change of acceleration.

This means that, if you want to simulate the bank in a 360-degree barrel roll, you just need to simulate the initial jerk from initiating the roll. Say it's 10 deg/s^3 for 2 seconds to initiate the roll, then -4 deg/s^3 for 5 seconds as it stabilizes. But if you do 8 deg/s^3 for 1.5 seconds to initiate and -3 deg/s^3 for 6 seconds to stabilize, the human inside won't know the difference.

So that's what you do. The sim cabin jerks almost as much as a real plane, but the acceleration ceases sooner than it would in reality, and where the real plane would've been stable, the sim cabin will gradually return to its neutral point.

The effect of motion on a collimated surround display is immersive enough that first-time visitors often mistake a simple static sim for a motion one. Actually adding a bit of motion, then slowly returning it to neutral after every maneuver, completes the illusion.

You don't have to simulate the entire motion, just hint at it enough to corroborate the information the pilot's brain gets from the picture.

  • 1
    $\begingroup$ It's not really "unlimited" though. This type of sim can present a wide range of sensations, but try sustained inverted flight in one. "g" will never point up... $\endgroup$
    – Ron Beyer
    Aug 23, 2019 at 20:38
  • $\begingroup$ @RonBeyer Inverted level flight is perhaps the one thing that can't be simulated at all, because the direction of lift changes. Inversion in a barrel roll, on the other hand, is within type 7 simulators' capabilities. $\endgroup$
    – Therac
    Aug 24, 2019 at 6:04
  • $\begingroup$ Straight climb, straight dive, knife edge, etc.... Those are things this simulator cannot do. Any maneuver that causes "G" to be outside the limits of tilt. It also cannot simulate more than 1g (at least not for more than a second or so), for example in a level turn with 45 degrees of bank. There are a lot of things that this simulator can't do, not just inverted flight. These simulators are good for certain envelopes, but they can break down outside those physically speaking. $\endgroup$
    – Ron Beyer
    Aug 24, 2019 at 12:32
  • 2
    $\begingroup$ @RonBeyer Level turn with bank simulation works acceptably. The platform simulates the initial transition to the bank quite well, and then returns to 1g. Which doesn't matter, because you aren't going to get 7g turns, the kind that make you unable to lift your arms and black out intermittently, and thus would have to be simulated, in a transport aircraft. The purpose of the platform is to replicate the sensory cues that interfere with flying by the instruments, not to produce a perfect reproduction of everything you'd feel during an actual flight. $\endgroup$
    – Therac
    Aug 24, 2019 at 14:02

You must log in to answer this question.

Not the answer you're looking for? Browse other questions tagged .