rec.autos.simulators

In the real world the guys involved with car setup are real experts.
They probably have phd's on the subject if there is such a qualification
and could tell us, pretty much on the button, what effect any change
will have.   Their facilities and years of schooling in racing setup
have allowed them to find out this information and setting up a car has
gone from what was a black art in the past to more of a science.

We don't have this kind of grounding.  We are mainly just people who
want to have fun racing, but at the same time, would like to know that
our ability to race isn't being impeded by a dodgy setup.

At the moment that knowledge is very difficult to come by.  We either go
out doing lap after lap, changing one component at a time to see if we
can feel any difference in the physics engine, and try to notice any
difference in the way the car handles.  Meticulously making little notes
of tyre temperatures and speeds through corners or at the end of
straights to try and determine what has been modelled in the game and
what hasn't.  Or we wait 6 months whilst someone else does the work and
then try and set the car up based on what they have discovered about the
game engine, or simply use someone else's setup to see if it is better
than our own.

Take, for instance, the affect of toe-in.  My first assumption would be
that more toe-in is going to prevent the car from accelerating as well.
In the real world surely this is the case.   However, we can't know for
sure whether this is the case in the sim's model and the only way to
find out is to go out and test... and test... and test...   Every new
sim bought means the same tests are required to determine these kind of
things.

Adding more downforce to a car in the real world, would I expect for
tyre-wear to be increased, as I say I'm not an expert.   Once again
though, how are we to determine this in the sim's model without running
many laps to see if this actually takes place.   We cannot take past,
real-world experience with us when we get a new sim.  Basically we are
starting from scratch every time.

So what I would like to see is as much info as possible about what is
actually modelled, so that I don't waste my time trying to make
adjustments with the preconception that they will have a specific effect
on the car which, when in actual fact they do not.

But, if all the info is there will this make setting up a car too easy?
I don't think so.  Take tyre temperatures for instance.  Let's say these
are critical for the way the car grips the track.  Well, here are the
changes that I could think of that could potentially have an effect on
them:-

Toe-in, track abrasiveness, track temperature, tyre pressure, downforce,
camber, driving style.

There are probably others that also have some kind of influence on this
in real life, but like I say, I'm not an expert, though I would want to
know about them if the game had included them.  However, the main thrust
of listing the potential variables is to show that, with enough of them
thrown into the mix, there is still enough variety in the way one driver
will have the car setup compared to another, especially because of
driving style.  Also hopefully, if the model is detailed enough, those
same changes will have effects elsewhere making all setup changes a
matter of compromise in some way or another.

So what I would like to see are race sim developers including more
detail about what they have included in their setup model, where changes
will affect the driving model and also as importantly how much influence
each setup change has in each area of the driving model.   This would
surely save us a lot of wasted time, mental effort and energy for the
majority of us.

Thanks for your time. :)
--
Peter Ives - (AKA Ivington)

No person's opinions can be said to be
more correct than another's, because each is
the sole judge of his or her own experience.

It surely does, but how noticible is it?
Noticible in tire temperature, surely. But in top speed? Not so noticible.

It's not too complex to estimate the longitudinal force from say
2 deg of front toe in (tire with a peak coefficient of say 1.6, slip angle
for max at 7 deg and assume linear, 50% front weight, etc., etc.
=> about 25 pounds.)

For something like a Trans-Am car at say 180 mph that's 12HP,
or about 2.2% of say 650BHP * 0.85 for drive train losses.
With max velocity cubed down by 2.2% => max velocity is down 1.3 mph

Someone who does a (very? ;) careful test could see the reduced HP for
acceleration or the lower max speed, but getting to a max speed
that is around that fast on anything other than a test track might
be difficult... ;)

Shameless plug:   I also ran through that calc because the subject
came up at lunch yesterday and Carroll Smith opined that the
top speed effect from toe-in wasn't as big as I seemed to think ;)

Well, another reason I ran through that calc was to avoid having
to answer (at least one ;) "Why can't I see its effect in my
Very Careful Test?" questions at a later date! :)

Along with most old time WarBirds players I'm still somewhat
traumatized ;) by the "My P-51 is seven mph too slow based
on a 20 mile max speed run measured with a stop watch and
detecting the exact second the pixel of the longitude line on the
RADAR map was crossed" threads of bygone days. ;)

I agree ;)

Some I'm familiar with ;)  that can affect tire temperature readings are:
    front anti-roll bar specs and adjustment,
    rear anti-roll bar specs and adjustment,
    track layout (how long since previous corners, etc.),
    steering geometry,
    how many laps (did the tires get to temperature?),
    how hard did the driver brake to a stop in the pits before measuring,
    brake bias as adjusted,
    front and rear ride heights (affecting downforce and camber)
    ambient atmosperic conditions,
    tire compound,
    tire wear,
        ...

Tire parameters and the ones that affect load and load transfer
are probably more important, (and what you are looking for anyway)
but the test condition ones can also be big ***s and as you say,
will make testing a bit confusing. As always, the more careful and
meticulous you are, the more useful the results will be.

;)

Good points, and I've wanted the same things in the past from sims.
I'll keep in mind these points as our group goes forward. With luck, we
can find useful and interesting ways of relating this kind of thing.

The only concerns I have are about our reserves of mental effort and
energy! :)    (Where's that graduate student I was promised? ;)
--
Matthew V. Jessick         Motorsims

Vehicle Dynamics Engineer  (972)910-8866 Ext.125, Fax: (972)910-8216

Hi Peter, Matthew, et al,

I've got a foot in the real world (we work with race teams) and at least a
toe in the sim world (starting back with Hard Drivin' and now with Silicon
Motor Speedway)...so let me take a shot at summarizing parts of this
_problem_ (I don't think that I have an answer):

a.  Adjusting a real race car is a never-ending process, and, many of the
governing bits of physics are not fully understood (and that may be the
understatement of the week).  Even the bits of physics that are reasonably
understood are often nonlinear and linked in very complex ways -- that is
why our introduction to the subject, "Race Car Vehicle Dynamics" runs 900
pages -- shameless plug <http://www.racesimcentral.net/;.
Race teams have smart and dedicated people working on these problems full
time.

b.  Math models of cars _must_ be simplifications of the real thing,
especially in places where the complete physics are not known.  The trick
to using math models in conjunction with real race cars is to learn the
strengths and weaknesses of the model and then apply "corrections" as
needed, to predict the best setup for the real car.

c.  If a game designer divulged the details of how the various bits of
physics were modeled in a game, then the setup problem would be
considerably easier than the problem in the real world.
  For example if the algorithms governing tire temperature, grip and wear
were published, then the tires would be considerably better defined than
the problem in the real world.  With this limited abstraction of the real
world, it might be possible to write another program that optomizes the
setup of the car for the game -- and where is the fun after that.  The
game designer would also be giving out quite a bit of competitive info
about their tire model.

Comments most welcome!

-- Doug

                Milliken Research Associates Inc.

Sent via Deja.com http://www.deja.com/
Before you buy.

I think one of the main points of the plea was for something like:
  "Even though Marketing says our tire wear will tend to
  increase with more down force, it really doesn't work that way
  in the sim so you can stop testing for that."

We can surely do some of that kind of thing, ... maybe. ;)
(Although a developer saying something exactly like that
probably wouldn't say it more than once anyway.
We'd have to learn subtlety.  ;)

Also, just like real world tire companies, we don't want to make
racing too easy either  ;););). And similar to them, we don't
have any interest in publishing design documents for our
competitors use.

One of the main points of the original author's plea I believe
was to be given some indication of what areas of study
would be rewarded with (realistic) improvements in the sim,
and which areas of study wouldn't, or could even be
counterproductive.

A gross racing example would be spending a weekend studying
ones huge collection of various driver training tomes and
autobiographies to learn the real world driving techniques
to exit a corner without falling victim to fatal exit oversteer.
Only to find Sunday night that no matter how one stomps
on the gas or chops the throttle exiting corners in the sim,
the car just sails serenely on.

I once played a spaceship driving game doing a
tail sitting landing straight down onto a planet.
(think Moon Lander type situation)
As a real world journeyman orbit mechanic, I _KNEW_ in my guts
that if I didn't have the thrust to reduce downward acceleration
at altitude I'd never be able to do it near the ground to land safely
and so kept aborting the landing attempt.
Imagine my annoyance to discover from a friend that the designers had
modeled gravity as some altitude under which the thrust magically
increased in order to allow a safe landing. ;(    Yuck.

The opportunities to play fast and loose with physics
come very easily in games, and it annoys users with
domain experience to have that real world knowledge
become a _disadvantage_ in the game because of
physics hacks.  If you know about it ahead of time,
people will generally be willing to accept it and go on.
But if you don't, discovering it yourself annoys, and is
one cause of UseNet flames of the game ;)

Very good point. We'll be including a steering geometry setup option
(i.e., setup a level of Ackermann or reverse Ackermann)
in our SCCA Trans-Am game that an advanced user would really
need some idea of the tire slip angle for peak lateral force
characteristics under different loads to optimize.
(See Race Car Vehicle Dynamics section 5.3 for a picture ;)

In the spirit of the original post in this thread, we have several options:
1. say nothing about the subject,
2. release complete information for how the lateral force coefficients
vary with slip angle and load,
3. option 2 above and also actually show in detail how to use the data, ;)
4. a compromise where we merely swear on a stack of Bibles that the
slip angle for peak force varies with load in a reasonable way, and that the
balance of the car is changed in a reasonable way by the front to rear
distribution of lateral load transfer and it might be well worth a hard core
users time to study the issue of steering geometry optimization beyond
the usual static toe angle settings.

As Peter pointed out, option 1 can be disheartening.
As Doug says, option 2 and 3 might be going too far for several reasons,
but the original poster's desires might be very well satisfied by at least
option 4.

Hehe. This also touches on something else I've been thinking about. ;)
As I have a background working with aerospace codes to do
that kind of optimization, (and company responsibility do
exactly what you mentioned ;) I'm a little afraid no one would let me
into races of my own game.   I may have to stick with default
setups to be fair. ;(
(Except that my lack of great sim driving ability will protect me
from most accusations of "inside information" cheating anyway! ;)

--
Matthew V. Jessick         Motorsims

Vehicle Dynamics Engineer  (972)910-8866 Ext.125, Fax: (972)910-8216

One interesting point made was that it may be possible, if too much
information was given, to create a program that is able to calculate the
optimum setup for a car based upon that information.  I wouldn't want to
disagree with that, but feel that any particular program would only work
well under very specific circumstances.   I could see a program created
that is able to calculate the optimum setup to allow for a vehicle to
accelerate to its maximum in the shortest possible time given that, the
driver is able to get the best out of the car, the car is driven in a
straight line and the track is completely flat and even.  Once you start
having to deal with cornering angles and track camber etc then the
definitive setup would surely be much harder to come by as too many
'what ifs' get added to the equation.

A crew chief of some sort was mentioned as a possibility and, although
the idea has potential, it would need to be very sophisticated to work
well and would therefore require a lot of development time to get right.
I say this because I have a game that already tried this -  F1 Racing
Simulation by Ubisoft - which in theory was meant to take all the hard
work out of setting up the car.  In practice though it didn't seem
detailed enough to be able to contribute much in the way of improvements
to the balance of the car.  It worked by asking questions where the
driver could select one from a multiple choice of up to 5 or so answers
and then, from the answer given, the next question to ask would be
chosen.  Unfortunately the questions asked soon became too generic and
it felt like the same old ground was being raked over every time the
chief engineer was consulted.  Also, whatever changes were made, they
didn't appear to have much affect on the car's handling.

So, if developers don't want to give away too much about there
algorithms used to define the setup/physics/handling of the car, then
that's ok as long as we are pointed in the right direction.  It would
still be advantageous to know, in more general terms, how the setup
changes relate both to the physics model and almost as importantly, to
each other.  I say that because, I assume, that with a good physics
model a change in one component may likely affect how well another
component of the setup works - or am I talking out the top of my head?
:)

So like I say, help is what I'm after, not necessarily the company
secrets. :)

To quantify the kind of problems I have at the moment with regard to
setting up a car I think a little anecdote is required.

Recently the track Elkhart Lake/Road America became available for GPL
and I downloaded it on saturday.   On that day I was only able to try it
out for about 10 laps with the Lotus on defaults with 5 laps of fuel.  I
managed a 2:16.4, but noticed that the Lotus had a tendency for the back
end to want to come round at the slightest opportunity, but mainly under
braking.

Yesterday I decided to see what I could do to improve matters and put in
25 laps whilst at the same time, with what limited setup knowledge I
have, tweaking things here and there to see if I could rid the car of
its problem and by the end of the 25 laps I had managed a 2:12.28 and
also seemed to have a handle on preventing the backend from wanting to
come round - so I thought.

By today though I had given it some more thought and I wondered whether
what had really happened wasn't that my setup changes had made the
desired improvements but that my ability to deal with the car's inherent
problem had improved over those 25 laps.  So it wasn't that the car
driving model had improved, but that my awareness of its weakness had
improved and so I was able to get more out of the car that way.

So, I went out on the track again today, but this time back with a 5 lap
default setup and on my 7th flying lap I did a 2:12.27 followed a lap
later by a 2:12.26 (no lie).  You can't get much closer in lap times
than that. :)

So it seemed that the setup changes achieved little, though to confirm
this I should have then gone out in my own setup to see what sort of
time I could have managed today.  Unfortunately, I was not able to try
this out.  Me old computer is playing up again.  Maybe tomorrow.  :)

I meant to make a few notes of my changes to post here, just to show
what a complete numbnuts I am when it comes to setting up a car and to
show how easily it is for the likes of one like me to make a complete
mess of the whole thing, but didn't get the opportunity what with the
game suddenly wanting to lock up on me. In fact as I type this I have
had to restart twice due to computer failings. :(

One final thing, it's excellent to hear from those at the pointy end of
game development, but it would also be nice to hear from those in the
same position as myself, who wouldn't know a slip differential from a
slipped disk.  I can't be the only one who gets frustrated like this or
are most games players out there happy to stick with whatever setup
they're given, ne'er to twist a nut or turn a***in anger.   Maybe
most sim racers are too scared that, because they don't know what
they're doing, that they will make a complete pigs ear of things.  

Now who knows, if more info was available to the player, then s/he would
be more likely to get involved with that side of things.  Surely it must
be a good feeling to change something on screen and then go out on track
and see/feel that your change has had a benefit.
--
Peter Ives - (AKA Ivington)

No person's opinions can be said to be
more correct than another's, because each is
the sole judge of his or her own experience.

I understand and to some extend subscribe to your viewpoint.

In practise however, I think a little more information accompanying
simracing games would go a long way.

Consider what we're trying to do when driving GPL (I find it hard to use the
word "playing" in these matters, sorry). We're controlling a reprensentation
of a car in a representation of a world, deprived of most of our sensory
input. Although GPL's force feedback implementation is arguably the best
we've ever seen, there's still ample opportunity to misread the car.

Continuing the GPL example, I would have liked torque plots for the engines
and tyre diameters to take the guesswork out of gearing and shift points.
Also, since the developer has to reproduce the tracks he uses in the game,
decent track maps, with corner radii, track camber and gradients can't be
too hard to produce, IMHO (notice I'm neatly bypassing licencing issues here
;-)). Off hand, I can't remember what else is needed to calculate a maximum
theorethical, constant radius speed for a turn, but if some maths can show
your car can only take a certain corner at, say, 65 mph you can stop trying
to make it at 80mph. Basically enough info to calculate some reassurance to
show we're on the right track.

Don't get me wrong, I see your point about a calculated best setup. However,
I have read your book (Liked it, a very interesting read for anyone with an
engineering background) and don't see the kind of math you use in there
implemented in a hurry to produce such a setup program. Besides, I've had so
many revelations reading about real life vehicle dynamics issues, I doubt
I'd use it if there were one. One has to make a clear distinction between
people who treat simracing games as just that: a way to "live" the
experience for themselves and those who simply treat it as a vehicle to

those who are trying to be"-brigade). I think the former category would
appreciate some more user documentation to come with their simulated race
car.

Regards,

Jan.
----

To comment on another point that someone made, I agree that it would be
nice if games came with more information on things like the track geometry.
Put very broadly, game designers could include the same information that is
available (or for GPL that _was_ available) to the teams that are racing.
Note that the section in RCVD referenced above starts with a description
(from Sterling Moss) about the race analysis department at Mercedes Benz in
the 1950's.

I've already written a draft of a "tuning guide" for 'Cup cars, which
should be released if (when?) car tuning is turned on for Silicon Motor
Speedway.  There has been a lot of good info in this thread and I will
probably incorporate some of these thoughts into the SMS tuning guide --
thanks to all.

-- Doug

                Milliken Research Associates Inc.

--
Don Scurlock
Vancouver,B.C.

In my case however, I think that it is probably because I am simply not fast
enough at the moment for the setups
to make any big difference, I am still searching for that sub 1:30 at Monza
:O((( I would hope that once you reach
a certain level the setups then come into play.

Peter Hooper

This however was not what I was trying to say.

I will assume that you SELL this software, for a profit to an audience
limited to bonafide racing teams. Even if we were able to buy it (both in
terms of cost and as an acceptable customer), we probably wouldn't have the
required data to gain any usefull info from it.

What I meant was: it's one thing for an amateur coder to produce a
simulation setup printout program as freeware. It's another to produce a
similar program which solves a complex set of inter-related differential
equations. I'd applaud someone who does, but I don't see it happening any
time soon. Should this opinion plant a seed in some fertile mind, all the
better. <G>

Anyway, I would like to express how much I appreciate an esteemed RL racing
personality like yourself coming here and showing an interest in the world
of simracing. This is the kind of recognition I strive for each time I don
my virtual helmet. Thank you.

Jan.
=---

<Snip>

Eldred
--
Tiger Stadium R.I.P. 1912-1999
Own Grand Prix Legends?  Goto  http://www.racesimcentral.net/

Never argue with an idiot.  He brings you down to his level, then beats you
with experience...
Remove SPAM-OFF to reply.