jsbsim-devel

Re: [Jsbsim-devel] DATCOM+ released

[David C. <dav...@co...>]

> I've thought about that issue. The .DCM files allow comments in them, so
> wouldn't one large files with ALL the models in it be useful (okay, up to a
> certain point). There are some parameters in the DATCOM input file which
> are real headscratchers, and being able to look at the values that other
> aircraft used might be nice. No, I'm not talking about wing span, but
> surface roughness, and how to input surface deflections correctly.


A web application, a la aeromatic, might help.  It would be a big project 
though, and I'm very short of time these days :(


Dave
-- 
****************************
David Culp
dav...@co...
****************************

Re: [Jsbsim-devel] DATCOM+ released

[Ron F. <ron...@at...>]

----- Original Message ----- 
From: "Bill Galbraith" <bi...@ao...>
To: <jsb...@li...>
Sent: Wednesday, July 28, 2004 10:04 AM
Subject: RE: [Jsbsim-devel] DATCOM+ released


>
> Looks great!  Maybe we're going to need a repository for the
> input files
> (*.dcm) we create?
>

>I've thought about that issue. The .DCM files allow comments in them, so
wouldn't one large files with ALL the models in it be useful (okay, up to a
certain point). There are some parameters in the DATCOM input file which are
real headscratchers, and being able to look at the values that other
aircraft used might be nice. No, I'm not talking about wing span, but
surface roughness, and how to input surface deflections correctly.

  I assume DCM files have siginficant AC dimensional data.  Smetana's low
speed applications probably use similar input.  Too many for me to tackle.
;)

    However, one of the guys I know is into this, and has used Smetana, etc.
to work out SD's for a few AC.  He also has quite a bit of Roskam, etc. data
from various sources.  Roskam's S&C Part 1 gives SD's for different AC,
depending on the edition.   If nothing else, a good way to check calculated
results.

    Data bases for AC that have been dimensioned. would be useful for most
any of the programs which calculate SD's.


    Smetana has some data files already worked out, such as for the C172
airframe.  It appears the FlightGear models were based on the resultant
SD's.

    ----

    "Theory of Wing Sections" gives wing polars for different Reynolds
numbers and generally one for 'standard roughness'.  It appears the later is
most appropriate for real AC.  Real AC may not even stay as smooth as
'standard'.   Flush rivets are used, so they don't generally have an effect.
However, there are all sorts of interference effects that increase drag past
what a simple calculation would give.  McCormick shows how cooling drag,
antennas, etc. built up a lot of extra drag in a WWII fighter.

    Rather than trying to calculate Cdo, known maximum speeds and HP/Thrust
allow one to set zero lift drag for actual performance.

    However, once one has done some AC, it may be that appropriate values
for 'roughness', etc can be entered in the DATCOM data to get a more
realistic end result.

    Control deflection limits are often given in the FAA TCDS files.
Aileron and Elevator deflections are typically asymmetrical.  Average L-R
aileron deflection determines the roll moment, while 'Adverse Yaw' is a
partly a function of asymmetrical deflections and accounted for separately.

    Up elevator is typically higher than Down Elevator.

    Even the + and - Rudder limits may be a bit different, but not by much.
+/- 30 deg is typical (but may drop to +/- 5 deg in jets at cruise speeds).

>Jon and I have kicked a couple of ideas around about making it easier to
>build a DATCOM model, etc. One possible idea was a fancy Excel spreadsheet,
>with questions and blanks, and examples of other aircraft values. All those
>could be stored on a second sheet of the spreadsheet, one column per
>aircraft.

>Bill

    Further, one can use Excel to calculate most anything.  Such as the
required Cn_dr to cover an engine out. I think  'vpi-nasa.xls' has some
calculations for such matters.

    So, if one doesn't get Cn_dr from DATCOM, he could still try different
values in a SS to find what is required.  Even use Excel to iterate to the
needed value.  Any extra calculations could be added to extra SS pages, and
left out of  a data only SS.

    Ron

Re: [Jsbsim-devel] Aitfoils at high alpha

[Ron F. <ron...@at...>]

----- Original Message ----- 
From: "Jon S Berndt" <js...@ha...>
To: <jsb...@li...>
Sent: Wednesday, July 21, 2004 6:17 PM
Subject: Re: [Jsbsim-devel] Aitfoils at high alpha


> On Wed, 21 Jul 2004 18:03:02 -0500
>   "Ron Freimuth" <ron...@at...> wrote:
> >
> >     I think the shape of the CL curve is important in modeling spins
(also
> >Cm vs alpha).  I suspect CL  gets stuck a bit past the bottom of the dip
at
> >20 degs in a stall/spin.  Something has to keep Alpha relatively constant
> >since a spin tends to stay in a constant state.
>
> Ron:
>
> Given what you know about simulation, let me state these questions to
> you (and anyone else that wants to answer):
>
> 1) What do you think is the value of providing some modeling of high
> alpha lift/drag characteristics - even if the data is sort of generic,
> providing perhaps (at best) only "plausible" dynamics?

    I've been sticking the humped CL vs AoA curve in my FS AC for some time
now.  Even in jet transports, which are never purposely stalled.

    MSFS does not set Cdi to IDK * CL^2.  Rather, it bases it on
{m*Alpha)^2.   Which gives virtually the same Cdi in the Linear region of
the lift polar, but continues to increase Cdi past stall.

    I saw one curve of Cdi in some image that suggested Cdi continued to
increase even faster than alpha^2 past stall.    JSBSim's CL, CD table
should probably be set for an increasing CD past stall, rather than dropping
as CL drops.

    Cm(alpha) is clearly important.

            ---------------
    Side note: I've finally figured that Cdo for Flaps always varies as
sin^2 (deflection).  Essentially deflection (rad) up to 30 deg.

    That works for split, ordinary, and fowler flaps.  Only the magnitude of
the sin^2 (def) need be set.  That's given in a 1930's formula and depends
on flapped width, 'depth', and type.


> 2) Do you have any insight or suggestions about modeling high alpha
> lift/drag characteristics, given the use of the JSBSim coefficient
> buildup method?
 > Jon

    JSBSim and MSFS have similar flight models.  FS has many tables that
multiply the basic SD's by a factor which varies with alpha.  It's 1.00 or
near at alpha =0, but typically drops as alpha increases.  JSBSim uses an
array for a SD that has rows for each alpha.

    Default tables in the FS AIR file are triangular shaped, and drop Cl_p,
Cl_beta, Cn_r, and Cn_beta on each side of AoA = 0.   I've modified them as
I found information on realistic variations.  Cl_p(alpha) typically
increases between alpha = -12 deg to alpha =  +12 deg.  But, drops rapidly
in the stall regions.

    Cl_beta increases over that range, at least if the wing has quite a bit
of sweep.  I've typically set the factor to 0.7 at alpha = 0 and 1.4 at
alpha = 12 deg. The increase in Dihedral Effect is quite apparent at lower
speeds, where AoA is high.

    Similar tables modify Cn_r and Cn_beta vs alpha.

    All four of the above affect the tendency to spin.  However, they are
not enough.

    While some CFS2 AC, and FS AC may develop spins of a sort, I don't think
the flight model has what it takes to model them correctly.

    I need to get my email out.  I'll go more into what I think is required
to model at least some spin characteristics later.

    NACA Report 517 goes into some aspects of spins.

    Ron