Re: [Flightgear-flightmodel] JSBSim and DATCOM: Forces, Moments and AeroRP

[Ralf G. <ral...@cu...>]

Hello!

I am crossposting this to jsbsim-devel and flightgear-flightmodel on
Jon's recommendation. Find my original mail to flightgear-flightmodel
cited below.

I would like to add that just this morning I have verified some of the
DATCOM output. I have done an estimate of best glide ratio and best
glide speed. The results coincide very well with the data from the POH.
So it seems that the DATCOM estimate is at least valid.

For example the POH says that best glide ratio is about 1:11 with 2,5m/s
descend rate @ 90-100km/h. The data output by DATCOM gives me a glide
ratio of 1:12 and a best glide speed of 90-100km/h, of course depending
on the mass of the aircraft.

Maybe I can even get some more accurate data when I do a more detailed
analysis of the AoA-range where the AoA for best glide resides.
Currently the 20 alpha-values that DATCOM allows are distributed so that
the area around the stall AoA is investigated more in detail.

Unfortunately, more detailed information in the POH is scarce and the
manufacturer of the aircraft has been known to be uncooperative in
giving away detailed data (a behaviour which I do not want to judge here).

Cheers,
Ralf

Ralf Gerlich wrote:
> Hello!
> 
> I am currently working on an FDM for an Ikarus C42 microlight and I'm
> getting a bit confused about the buildup of forces and moments with
> JSBSim. Unfortunately I'm very far from being an educated aero-engineer
> so I probably don't understand all the fine details.
> 
> My simulated C42B created from the DATCOM output exhibits very low
> stability in the yaw axis and goes into an nearly undamped yaw swinging
> motion after any deflection and recentering of the rudder. I haven't
> noticed a similar behaviour on the real aircraft. I also see that my
> simulated C42B rotates by itself at 80km/h (which is the actual Vr)
> whereas the real thing needs to be clearly pulled at that speed to get
> the nose up.
> 
> I have taken actual measurements on the real aircraft and plugged them
> into DATCOM+. Thanks for Bill Galbraith for helping me over some
> harder-to-decipher intricacies with this fine software. Working with the
> software alone taught me a lot of things about applied aerodynamics. ;-)
> 
> DATCOM+ already outputs a JSBSim-compatible set of buildup functions for
> lift, drag, sideforce, roll-, yaw- and pitch-moments.
> 
> Some data needs to be filled in additionally, as there are the moments
> of inertia, which I estimated using a small self-written program based
> on some of the weights and geometry I know, the empty CoG or the
> aerodynamic reference point. Specifically the latter is giving me some
> headaches.
> 
> At first I thought that this specific piece of information should not be
> necessary anymore as the moments' coefficients already should properly
> cover that. Then I remembered that the center of gravity of course moves
> in different load&balance situations, resulting in different lever arms
> and thus different moments.
> 
> So I studied the code calculating the forces and moments and found that
> in addition to the moments determined from the YAW, ROLL and PITCH axes,
> JSBSim calculates moments from the force resulting from the sum of the
> LIFT, DRAG and SIDE axes, using the vector from actual CoG to the
> aerodynamic reference point as lever arm (see current FlightGear-CVS,
> src/FDM/JSBSim/models/FGAerodynamics.cpp, line 185 or the respective
> sourcefile in JSBSim-CVS)
> 
> As a first note this looks to me as if the moments are calculated twice
> as in my interpretation - which of course may be wrong - the moment
> coefficients already account for the moments using the CoG as given in
> the DATCOM input data as moment reference center. However, you guys are
> working with this all the time and it seems to work quite well, so there
> must be something that I missed.
> 
> So obviously the location aerodynamic reference point is a really
> important information and if selecting the wrong location might very
> well be the cause of the "misbehaviour" of my simulated aircraft.
> 
> So now I have the problem of how to properly determine the location of
> the aerodynamical reference point. DATCOM provides a coefficient named
> XCP per given angle of attack. This is described as "[t]he distance
> between the vehicle moment reference center and the center of pressure
> divided by the longitudinal reference length."
> 
> Would XCG-XCP*cref give me the longitudinal location of the aerodynamic
> reference point? If so, to get going I could probably just set the
> AERORP value to some point above the empty CoG, which I have given as
> CoG in the DATCOM file, and use the AeroRP-shifting feature to move it
> along the longitudinal axis depending on the angle of attack according
> to the data provided by DATCOM for XCP. Am I right or am I talking nonsense?
> 
> Thank you in advance for any feedback.
> 
> Cheers,
> Ralf