IL2 The Flight Model (FM)
Without attempting to be morosely obvious, we can all agree that the
most important aspect of any flight simulation is the Flight Model
(FM). The desire to pilot-- in some sense-- aircraft which we otherwise
could not is the reason why we part company with our hard earned and
spend, frankly, absurd amounts of our free time faffing about with this
software. It is the fidelity of this virtual "flying"-- may we
say, the immersion of the simulation-- which is the key component. And, in many respects, this immersion is the offspring of the FM.
As mentioned elsewhere, the central components of the IL2 sim-- the FM
and the physics-- cannot be changed. So, our VPB mods have instead
focussed on manipulating the existing FM to suit our purposes. As one
can well imagine, this has been a long and hard slog, full of pitfalls,
errors, confusion and failures. Even now, at Phase 8,
we do not yet grasp all aspects of the original FM; but we do understand a
bit, and enough to make some very different flight behaviours.
Problems With The FMD and EMD Files
Each aircraft in IL2 is assigned an FMD file (flight and structural
properties), which in turn refers to an EMD file (engine). We will not
try to present a tutorial on FMD/EMDs
here; that kind of information can be found elsewhere. Therefore, if
you find these terms and file structures incomprehensible, let us
recommend some reading on these files before proceeding.
Once opened, the FMD reveals some of the main problems with, and
limitations of, the sim's FM. The first realisation is that the FM is
in fact a bi-polar fluid dynamics simulation. Now, indeed it is true
that aerodynamics is a form of fluid dynamics, and surely there is
considerable cross-over between the two. But they are not the
same thing; aerodynamics is a sub-discipline of fluid dynamics with a
great deal of speciality. This observation is very well demonstrated by
the orientation of the polar forces in the FM, where aircraft are moved
via a forward and aft lift/drag
polar. While this method cannot be said to be invalid, we believe that
a vastly better approach would be to have a lift/drag polar for each
wing, and to treat the empennage as a deflecting body.
This kind of fore/aft polar model, to be sure, makes for some very odd
aircraft behaviour. In the first case, it causes aircraft to feel as if
they are mounted on a rail; neutral directional static stability is
probably not even possible using this method. It also makes for very
strange stalling behaviour, in that a stall is caused by insufficient
lift over the wings, not the empennage. Further, it is virtually always
the case that wings stall asymmetrically, something which is impossible
in IL2 as there is only one lift/drag polar representing the entire
wing surface (or surfaces). It is this very behaviour that causes a
real aircraft to drop a wing during a stall (especially under g),
sometimes so violently that the machine is inverted. It is thus that
stalling behaviours such as lateral rotation and spinning are scripted in the IL2 FM; rather like setting off a penalty time for entering this condition.
The next major problem-- and likely the worst of them all-- is that specific aerofoils do not exist in the IL2 FM. Yes, you did read that correctly.
Opening the FMD for the first time, it is a natural reaction to search
for the aerofoil section listed for any aircraft to be sure that this
is given accurately (misconceptions on this matter are commonplace, so
one does tend to check). No such thing can be found. In fact, the FM
applies a single, generic sort-of aerofoil to every aircraft, one which is so rudimentary that it appears to work by means of a coefficient of pressure distribution. Such a model is not suitable for use in a flight sim, but it is convenient computationally
if one is employing a single polar to represent all of the wing
surfaces. In other words, a short-cut. This observation explains why
the aircraft in IL2 are so pathologically identical in their flight
behaviour-- they have to be. Every design of the time, according to
1C's FM, had the same (and highly generalised) wing profile! This
situation, we must add, rather smacks of yet another conceptual
failure-- "What's so important about these aerofoil shape thingies;
what's the big deal?"... Oh dear...
So, what to do? In the first instance, we made sure that all of the physical data
in the FMD was correct: span, wing area, weights, Vne, DM values,
various control sensitivities, and so on. We have experimented
extensively with values in the Polares section, and by often by sheer
trial-and-error managed to find ways to tweak these values to produce
what we regarded as better (more convincing) behaviour. No doubt some
further discoveries await us in this section of the FM.
The other main reference in the FMD file is to the aircraft's engine(s)
(EMD file). Here again we found many curiosities, and alas more
demonstration of the fact that the team has no real understanding of
period aircraft engine function. [ Interestingly,
Daedalus Team must have encountered the same trouble. We understand
that in 4.11+ a number of the power features which we will mention
hereafter have been reversed, similar to our own solution. There is no
question that Daedalus' changes to the EMDs have been positive, if
still insufficient in our eyes. ] The use of of the WEP key is a case in point. This device seems to have been informed by watching episodes of Battlestar Galactica--
push the red button and *pfffoom*, off you go! The use of WEP also
interfered with any kind of revised boost settings within the EMD,
these in an attempt to get the engine to perform correctly at height,
by becoming entangled with the daft compressor points system in the FM. What on earth is that?? We still don't know.
In our own work, the WEP button has been thrown out. The only engines of the
period with anything like this kind of button were water-injected
members of the Pratt&Whitney R-2800 family. Therefore, we feel that
it is better to integrate all forms of over-boosting into the throttle
system, which in IL2 allows for a 100% - 110% setting. The same is also true for the Luftwaffe's Kommandogerät system,
which certainly lends itself well to the integrated over-throttle
method (being an automatic control device to handle mixture, boost,
pitch, supercharger gear setting, and so on). With WEP
removed, we were then free to modify the supercharger pressure settings
within the EMD. However, once again these changes resulted in other odd
behaviours. The power system in IL2 applies a kind of boost to engine
performance near to the supercharger engagement height. This oddity
results in a "tent pole" effect, where the power is maximal near to
these values and diminishes as one moves away from them in height (but
at least it affects all engines equally, so as not to confer any
specific advantages). Once again, this is not how such devices actually
worked! Indeed, most USA engines had three supercharger settings (not
two)-- Neutral, Low and High. Trying to implement three supercharger
settings in the sim resulted in an even more ludicrous mess, so we gave
up. Since superchargers (and turbochargers) do not function as they should
in the FM, we have decided to use a generic EMD method which gives
better power performance at height, but uses only one or two
supercharger settings (as appropriate) in a homogenised engine
operation: a sacrifice of historical engine function for better
FM Updating-- A Way of Life
Why is it that VPB issues so many Buttons (the
file in which the FMDs and EMDs are hidden) updates? In order to get
the aircraft to show better performance characteristics, it was
necessary to create custom engines within the EMDs for each aircraft.
It is a direct confirmation of the dire effects of not having proper
aerofoil and drag data in the sim that one cannot simply apply an
engine which has been properly configured to all of the machines which
did in fact use it, and then realise anything at all like
the correct performance for each respective type. This kind of detailed
customisation for nearly 500 types, one can imagine, is a lot of work.
On such a scale, it is always the case that errors, glitches and
unforeseen effects creep into performance. These are discovered in the
course of our flying, and once noted the problem is hunted down and
corrected. These corrections then, naturally, have to be flight-tested,
and then off we go again-- it will be a very long time before this type
procedure is no longer required. And each such correction, no matter
how small, requires a new Buttons file.
Update No. 1 -- Pro-Blue Bias in the FM
Responding to several requests, let's take a detailed look at some of
the more obvious problems with the sim's stock FMs as regards bias and
unrealistic favouritism towards Blue aircraft.
Sadly, finding such cases in IL2 sim is easy. Let's start with the Bf
109 G-2 vs the Yak-1b; two classic contemporaries. These two machines
make for a good FM comparison as they similar in size and take-off
weight. The Bf 109's aerofoil is NACA (root) 2R1 14.2% (tip) 2R1 11%,
while the Yak-1's is (root) Clark YH 15.15 (tip) YH 8.8. Disregarding
any additional lift devices for now (the 109's slats), we can make the
observation that the Yak's aerofoil will give more lift (Cl) and have slightly more drag (Cd);
YH has a greater critical AoA and better boundary flow than 2R1. By all
means, run an analysis on these two sections at your favourite Reynolds
and you'll see these observations confirmed.
Now, it is important to mention at this point that there is no
possibility to "interpret" nor "debate" this data. The behaviour and
relative characteristics of these two aerofoils are an unanswerable aerodynamic fact.
Therefore, the data in the sim's FM must agree with these
observations-- at least in a relative sense (since we know there are no
actual aerofoils in the sim engine)-- or it is wrong. So, what then do
we find? Opening the respective FMs, we see that the Bf 109's raw lift
force (parabCxCoeff) generates 55% more lift
than the Yak-1. It also has slightly less drag, which is in line with
the actual aerofoil data. The 109's critial AoA is also greater (by
~15%) than the Clark YH aerofoil. Worse still, the Bf 109 G-2's
Take-Off wing loading in the FM is given as 175 kg/m2, when in fact it
should be (using the 'usual published' wing area; proper wing area is a topic for another day) 192.9 kg/m2!! Conversely, the Yak-1b's Take-Off wing loading is basically correct (~167 kg/m2).
Well, but what what about these slats? Surely, the Blue delusionists
will cry, these must explain everything!? Errr... no. In the first
case, leading edge wing slats (LEWS) operate at low linear airflow
moments so as to increase the Cl/Cd
of the aerofoil under these conditions. That is to say, basically
increasing lift/drag near to the stall. The total effect of the LEWS on
the Bf 109, for example, would represent something like 8-11% of the
total Cl/Cd, depending upon
various conditions. LEWS have obviously no effect at speeds higher than
their operational range. That said, none of these observations are
relevant to the IL2 sim. The more cynical readers out there will likely
already have guessed that the operation of LEWS has no effect whatsoever within the sim's FM. Nothing. Visually they open in a very nice way and look the part, but this is just eye candy.
This profound departure from reality is all the more suspicious when we
examine the FM data for aircraft which are not on 1C's favourites list.
The relative data for the Fw 190 vs the La-5, for example, while
perhaps not strictly accurate, are at least in relative proportion. The
two nearly identical aerofoils (Fw = NACA 23015/23009; La =
23016/23010) do show the correct relative properties, so that the Fw
190 has slightly less drag and the La-5 slightly more lift. These are
fine within the context of the sim's FM.
When we see such absurd deviations from reality-- as with the Bf 109--
no sensible person can accept that these values are the result of
error. Recall, also, that this so happens to
be the same aircraft with gigantic make-believe sheets of armour plate
installed into them. The pattern of these abuses of reality is clear:
this is deliberate cheating. The Messerschmitt is obviously a
favourite, so they have consciously made it virtually invincible. But,
more than this, the truly preposterous scale of this cheating points again to the offensive old Luft Hero-Worship
drivel that sim software marketers so dearly love. Quite frankly, from
an historical (and moral) point of view, this is rather sickening.
Of course, the Bf 109 is not the only type on 1C's favourites list. Some other aircraft which feature notable cheating include:
Buffalo Model 239 (Finn)-- a preposterous UFO lacking in connection to any reality
Ju 87, especially the D models-- thoroughly absurd
Macchi C.202-- advantageous tweeks in many aspects
MS.406-- hugely armoured, must be filled with helium to achieve such lift values
...And so forth. We could go on and on here, but it's simply not required. Any objective reader by now has got the point.
At the same time, a number of aircraft have been unfairly and ahistorically downgraded as compared to their actual performance. The most obvious, and serious, such case occurs with the entire Yakolvlev family. Pick any example-- they have all been (literally)
sabotaged. We suspect that there is an identifiable reason for this
behaviour, but that topic (as well) must wait for another discussion.
Some other notable victims of downgrading:
Spitfires-- all marks, similar to the Yaks
Pe-2-- all versions, constructed of
paper so that this large robust twin-engined bomber has less DM
strength than virtually any single-seat fighter
LaGG-3-- all models, again farcically weak with less DM robustness than an A6M 'Zero'
...And on and on. Has anyone, perhaps, at this point noted any certain pattern to these problems? Perhaps relating to which side these example flew for? Hmmm... How "surprising", indeed....
Update No. 2 -- Fighter Aircraft Performance: The Guide Book
The recent publication of Fighter Aircraft Performance of WW2 has at long last provided the basis for a coherent and factual analysis of any flight sim's FM.
We have made several claims about the errant nature of IL2's relative
aircraft performance characteristics in these articles, and now one may
read exactly why this is so. We invite everyone who is interested in
these topics to have a read of that volume. Moreover, we invite anyone
who is currently engaged in the creation of flight simulation aircraft
behavioural characteristics to read it in detail. Please! If the data
being used by your development group has not the same level of factual
and scientific propensity as that in FAPWW2, surely you must then ask yourself why and discard the data currently being used.
Here is a link to the book and site. There is a Forum for discussion
for all those who enjoy such things. Go forth and educate yourselves....