Re: [Jsbsim-devel] C-172 Performance --- More comments

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

----- Original Message ----- 
From: "David Megginson" <dav...@gm...>
To: <jsb...@li...>
Sent: Wednesday, October 13, 2004 9:47 AM
Subject: Re: [Jsbsim-devel] C-172 Performance


    Some more comments, based on my experience with MSFS.  Most should apply
to any simulator that is based on user supplied parameters and tables:

    The C172/182 airframe is a classic.  A good AC to model well -- other
high wing AC could then be extrapolated from it.

    Dr. Lowry,  author of the AIAA "Flight Performance of Light Aircraft"
owns or owned a C172.  He made a lot of detailed measurements on the
airframe, prop, etc. to illustrate his approach to basic performance
modeling.   He doesn't cover dynamics, that's another ball of wax.

    Dr. Lowry hangs out in the sci.aeronautics newsgroup.  I've exchanged
comments with him.

    A few years ago I helped an MSFS guy model the C172 he was flying in a
club.  He also had other 172 pilots critique the FS model.

    Back then no one doing MSFS hacks had ever heard of 'Stability
Derivatives'.  However, we did have the scaled integer parameters fairly
well worked out -- at least qualitatively.  Further, an older MSFS 182
flight model had pretty good values for the SD's.

    I hadn't flown a C172 since 1966.  But, had some idea of roll damping,
etc.  We ended up with an experimental Clp that wasn't all that much higher
than the 'official value' (Clp = -0.47).

    Lowery came up with his 'BootStrap' approach to finding Cdo, Cdi, and
prop coefficents from flight tests.  However, I've never grasped what is
special about his technique.  I do know it is very difficult to extract Cdi,
Cdo, etc. from glide tests; even in a simulator that has a perfect
atmosphere model and no wind.

    I've mentioned Smetana's  SD values for the C172.  I suspect most
published values are based on his computer analysis.

    I've never seen what Datcom would give.  Hopefully quite similar.

    I ended up putting Smetana's SD's in my MSFS AIR file.  Results appeared
good.  I did have to adjust Cl_beta (Dihedral Effect) to a different value
to get realistic wing leveling.  MSFS includes some effect that is based on
the distance from the Aerodynamic Center to the physical CoG.  Independently
of 'Cl_beta' itself.

    Flight testing is very tedious.  It takes some time for TAS to
stabilize.  Weight drops in transport AC during flight testing, thus
specific range changes as one continues testing.  I can set the 'fuel flow
scalar' very low to keep flight weight nearly constant.  And scale any gauge
report of fuel economy back to the unschooled value.

    While we have some logging apps that dump CVS format data for later
analysis, most of my testing and adjustments are done in real time.
Fortunately, one can change the flight/powerplant model in MSFS in the
middle of a flight.  Reload the AC and one can see what effect a change has
made.

    As I've mentioned, it's virtually impossible to get a good overall
flight model unless  most of the applicable parameters and curves are well
set to start.  There are far too many interactions.

    Fortunately, I had a general idea of how torque varies in ICE's as RPM
changes.  We can read SHP in MSFS, so I set the AIR file table that relates
relative "Indicated Mean Effective Pressure" to get rated SHP at rated RPM,
29.92" Hg, 15 C.  "Brake Mean Effective Pressure" BMEP, is lower.  And, easy
to calculated from rated HP and RPM.

    Friction Torque subtracts from IMEP and one ends up with BMEP.  In fact,
one can speak of Friction MEP, etc.

    Above 1200 RPM Friction Torque various approximately with N^2.  I have
one curve of Friction HP vs RPM for an IO-550 and figure the general shape
should be about the same for all AC engines.

    Further, various sources indicate that friction amounts to about 12% of
rated, WOT HP.

    I found that Friction  is composed of Pumping Friction and Rubbing
Friction.  Rubbing friction depends on oil viscosity, clearances, etc.
After warm up it is mainly a function of RPM.

    However, pumping HP varies with induction flow.  It appears the MSFS
model lumps them in one curve.  Thus, if one cuts the ignition, windmilling
drag (negative SHP) doesn't change with throttle setting.  It should.

    IMEP also drops with induction flow.  Due to carburetor, intake
manifold, and intake valve pressure drop.  MSFS has (had) a two point table
that lets one set the ends.  Their default value of 0.98 at the high end
appears reasonable.  When the barometer reads 29.92", the MP will be about
0.7" Hg lower. However, if one drops the RPM with a variable pitch prop, the
air flow drops.  MP reading increases.  As it should.

    Compression Ratio has an effect on IMEP.  This doesn't appear to come
out correctly in MSFS.  Not much of a problem, since I adjust the IMEP table
to get rated SHP.  I found a higher CR also increases Rubbing Friction in
real engines.  Just the same, the normal CR range is 6.0 to 9.0, so the
effect isn't that significant.

    A good engineering model would include such details.  And, allow
students to actually see the effect of changing these components.

    MSFS only models simple turbochargers.  Many WWII fighters used two
speed geared superchargers.  Rather than exhaust driven turbines.  I'm not
that much into such things, but some of the CFS people are.

    Note that a geared supercharger adds a torque load to the basic engine.
A turbocharger isn't free, either.  It increases exhaust port pressure.
Which means the low temperature reservoir temperature isn't as low as it
would be otherwise.  Thus, lower thermodynamic efficiency.   As far as I
know, MSFS doesn't model such effects.  However, the powerplant model is
quite adequate for turbocharger or normally aspired engines.

    I haven't looked at the JSBSim powerplant model for some time.  Perhaps
it could benefit by adding some of the elements mentioned above.  I'd expect
the prop model is equivalent to that in MSFS.  Only that a different pair of
prop coefficients are used.

    Dr. Lowry's text has quite a bit of info on the C172 prop.  And, an
approach to approximating prop coefficients that covers the normal operating
range of J and Beta; which also applies to variable pitch props.  Ideally,
one should be able to set only a couple of scalars to define a basic
propeller operation over a wide range.  A computer program (or SS) could
fill out the required prop tables once a few operating points were known.

    For the PA-28 one can refer to McCormick's 1995 text.  McCormick
owns/owned a PA-28, so many of the examples in his text related to this AC.
I put a couple of hundred hours on a 150 HP Cherokee.  Again, long ago. At
least I remembered a few things about its performance.

    As far as 'accuracy goes', I'd disagree that ones flight model doesn't
have to be that accurate, 'since  real AC vary all over the place'.  Seems
to me that a good flight model should represent the ideal AC.  One fresh out
of the factory.  Not one misrigged, with inaccurate instruments, etc.
Ideally, a pilot could compare an accurate PC AC with what he flies.  And
see how much has real AC has deteriorated from a factory fresh model.

    I know Flight Gear lets one log flight data.  And, write scripts to
generate standard test flight profiles.    This can be done in MSFS, however
almost all the adjusting and measuring I do is with test instruments that
can be displayed on the panel.  With digital displays so I can see exactly
what current MP, SHP, TAS, Cdi, Cdo, etc. etc. are.   Rates are also
significant.  My current 'XML' test gauges display 'V dot', 'VS dot', 'p',
'q', beta dot', 'a dot', etc.  One day I'll add Specific Potential Energy
(simply h) and Specific Kinetic Energy '1/2 V^2) and also display the sum.
The rate of change of the total specific energy is a function of current
(normalized) drag.  Important to military fighter AC.

    Numerically competent individuals like to see the 'numbers'.  I feel one
can learn a lot by watching values that go beyond the basic panel
instruments.  At least for me, it gives me something to do on long test
flights. ;)

    With an accurate and verified 'flight model', one can see the effect of
changing from a 'climb prop' to a 'cruise prop' without fooling with a real
AC.  He can watch Rs and see how it changes as RPM is changed in a CS prop
AC.  He can see what BMEP is doing.  All this should help him set up a real
AC for efficient operation and minimal stress on the powerplant.

    This  is all really old stuff.  It was well understood by 1935.  But,
perhaps not so well understood by actual pilots.  Who have typically learned
rules of thumb that may or many not help them  set the most appropriate
operating conditions.

    With accurate and verified flight models, real pilots, engineers, FBO's,
etc. can lean many things right on his PC.  Safely, and at low cost.

   By hook and crook we have been able to calculate a large number of
engineering parameters from available MSFS variables.  Finding the room to
display them all ends up limiting me more than figuring out how to calculate
them. ;)

    No doubt, similar test gauges could be created for JSBSim.

    Ron




> On Wed, 13 Oct 2004 09:11:32 -0400, Bill Galbraith <bi...@ao...> wrote:
>
> > This is of particular interest to me. Detailed testing of the C-172
would ...
> >
> > So, what is the solution here?  Well, I've know guys that will do "bar
stool
> > and yard stick" flight tests (yeah, really using a bar stool and a yard
> > stick for their instrumentation). You spend 10-20 hours in the air, with
> > some simple measuring tools (stop watch, hand-held force guage) and a
> > clipboard and record data this way. Probably not going to happen here.
>
> Not for most makes and models, but we do have a lot of pilots on the
> list and people can and do contribute in-flight data for the simpler
> planes.  I remember that Alex Perry did some tests a few years back
> for the 172, and I did some for the PA-28-161.
>
> > Therefore, you are stuck digging through POHs, Jane's All The World's
> > Aircraft, and anything else that mentions the words "Cessna" and
"C-172".
> > You have to decide if the information provided is valid for your
aircraft,
> > and if there is enough information to be useable (set-up, conditions,
> > procedure, etc). You compare your results with generalize curves in the
POH,
> > keeping in mind that you might perform better than the POH predicts.
> > Phugoid, short period, steady heading sideslip, flap response... Yeah,
good
> > luck. You won't find any useful data.
>
> Just so -- for phugoid, dutch roll, yaw-roll coupling, roll rate, etc.
> we have to rely on general impressions -- does it feel about right to
> pilots who have actually flown the plane (difficult to measure, since
> computer controls move so differently), and does its handling match
> published pilot narratives (i.e. "tends to spin", "docile in stall",
> "needs a lot of rudder in a turn", etc.).
>
> For some models, we do have information available from published
> sources.  For example, Roskam publishes linear coefficients for quite
> a few aircraft, though they come from a computer model rather than
> actual flight tests.  NASA also publishes some data.
>
>
> David