Scalars in the Flight Model

I am currently working on an aircraft of similar configuration to the Cessna 208B Grand Caravan. There have been some problems with the flight dynamics, despite having direct access to the manufacturers and their data. With the supplied data entered into the flight model, flight behaviour was (to put it mildly) “a wild ride”! When compared to the default Caravan, which is a relatively stable platform, I was at a loss as to why the aircraft behaved so differently despite having similar figures so I took the step of putting our figures individually onto the Grand Caravan files.

Weight and Balance, aerodynamics, dimensions, all were set up and tested in turn before moving on to the next and nothing caused an issue until I reached the flight tuning section. This is basically a section of scalars, much as was present in the old aircraft.cfg, and the expectation is that for a scalar to have no effect on the things it is scaling, a figure of 1.0 will be used. i.e. the original figure is multiplied by one, so remains constant. Here is where I found the root of our problem, and also some very strange behaviours.

Firstly, the use of XYZ_Maxangle_Scalar. I’ll give an example of this issue - Rudder is set up to the correct size, deflection angles and coefficient. Wing dihedral is set to the correct amount of 3 degrees. Rudder_Maxangle_Scalar=0.5 (left at the Caravan setting). Behaviour is okay, with a yaw-induced roll from the dihedral effect which is perfectly manageable and doesn’t exceed about 20 degrees of bank. Increase the scalar to 1.0 (which logically should be the default - everything scaled by 1, so each setting is actually the figures defined) and it is possible to complete a 360 degree roll on rudder alone, as if it were an aileron roll.

Quite why you would want a scalar on the maximum angle of a control surface, I don’t know – If I want 20 degrees, I will set 20 degrees rather than setting 40 degrees and a scalar of 0.5. However, from the behaviour displayed I suspect there is more to it than just scaling the angle.

Then there’s the Pitch / Roll / Yaw stability scalars – the Caravan has these set to 0.1. I altered them to 1.0 (again, expecting a scalar to multiply by the amount, i.e. x1, giving the stability expected from the coefficients, dimensions, angles which had all been carefully set up from the manufacturer’s data. Instantly, the aircraft became uncontrollable. It appears that increasing the stability scalar actually increases instability / decreases stability. To cross-check myself, I then went back to the original Grand Caravan files and adjusted only the stability scalars (from 0.1 to 1.0) and the aircraft became distinctly unpleasant to fly.

Going back to the SDK tutorial, which I had been advised to read, there is brief mention of stability scalars but nothing about their default settings, which (if anything other than 1.0) should really be very strongly noted. Quite why the default setting should be anything other than 1.0 I do not know. The tutorial then instructs you to adjust the effectiveness scalars, giving an example of 1.25 for aileron effectiveness – if all of the basics are being read correctly by the flight model why do we need another two scalars to affect them and create the correct response?

All in all this appears to suggest either

  1. that inputting real-world data does not work in the flight model so scalars have been introduced to try and artificially create control, or
  2. there is a default setting for the scalars at which they do not affect the defined figures but that default setting has not been documented and is not 1.0.

The feeling I am left with is that you can input pretty much any figures you like for the bulk of the flight_model.cfg, because it’s the scalars which actually control things. That’s not conducive to accurate flight dynamics, particularly when the manufacturer is on board and supplying us with the correct figures! It certainly takes away from the level of accuracy I am happy (and used to) working with.

Slightly off-topic, but on the subject of accurate figures – I wonder if anybody has found out why the fuselage length in the Grand Caravan and in the tutorial is -1 ft, particularly in a section of the tutorial dealing with accurate data input

5 Likes

You’re probably better off asking this in the ‘SDK Discussion’ subforum. More likely to find answers there.

A fair point - rather than double-posting, can this thread be moved across there?

it might depend on your forum level, but if you can click the edit icon, pencil, at the side of the thread name, you might be able to move it yourself, with the drop-down box.

I moved this as requested

1 Like

This topic was automatically closed 30 days after the last reply. New replies are no longer allowed.