Technical discussion about new Downforcesettings

[Roger]

Fred gave us a formula for D/F some while back. As I read it the down-force is related to vehicle speed (measured in M/s) and it's proximity to the ground - giving a result measured in kg's. I assume that this is because the RT physics engine does not calculate the vehicle shape moving through the air (very complex). So, the downforce number does not relate to the plan area of the car or it's mass.

Where the calculation is odd is that it directlt multiplies downforce by speed instead of squaring it (which would happen in real life).

So, a car with a DFC = 1 travelling at 100kph with 0.150M ground clearance gives 23.713kg of D/F - at 200kph it gives 47.426kg. Raise the car to 0.2M and the d/f drops to 38.183kg.

I must admit (despite it's impeccable source) I am not too sure about the calculation - although I can see that it would simple to execute in game giving some relationship between d/f & speed. I did start set up some tests with markers on the wheels so that I could see how far the suspension was compressed at various speeds - as yet these tests are incomplete.

The implication of having a DFC that produces a number in kg's is that the DFC does not relate well if translated across different car sizes. Let's say you have two cars of exactly the same shape, but one has double the surface area (in simple terms ;) ) the bigger car will produce a greater lift force measured in kg's (negative or positive) than the smaller car. BUT in RT the same DFC number will produce exactly the same lift force.

[Ralph_B]

Thx, Roger, that confirms my thought, that the DFC-calculation in RT is very simplified and not very accurate, same as many other things in RT physics. So at least we have to find a solution, that is also simple and gives a more the less satisfying, for different cars reproduceable result. I am very curios about your tests and looking for the final result. What I have done so far for the TT-cars, gives a relative good result in driving-experience, but unfortunately isn't based on any reproduceable calculations. Hope, we find a acceptable solution, that perhaps could be integrated in the physics-calculator.

[Roger]

The DFC calculation is already in the physics calculator and can be used (if chosen) to calculate the effect of the extra load on the car's springs at speed; allowing the suspension frequency to be adjusted to suit.

A thought...perhaps the phys-calc should work the other way around. Instead of predicting the extra load created by the DFC, maybe it should reproduce a DFC value based on "real" values for lift/downforce?

[Pero_Grozni]

It is doable Roger ;)

But-we have a problem or two. The ridehight influance is murdering us. ATM the influance is too big and therefore it is hard to do something properly.

[Ralph_B]

Have seen the DFC-option in our calculator, but it only affects suspension to compensate possible DF, but has nothing to do with DF itself. So your suggestion could make more sense, Roger.
Obviosly we can't affect the way, DFC works. When I readed your explanations right, it is simply a linear function, only depending from ridehight and speed and not from any other carspecific parameters like mass, size, shape, right? I've assumed this when doing it for the realmod, but wasn't quite shure. Of course the ridehight has a to big influence, but seems to be something we can't change. Was afraid of this, so we have to stay as it is. We only have to find a basic value, that is depending from the cars real DF/lift. This way I see only the real avarage DF as a starting-point. This is the only value making sense and it is a individual property of each car. So my first idea about the produced DF/lift inserting as a direct value like in one of my posts above may work(more the less :rolleyes: ). Perhaps we have to introduce a multiplicator - something like DFC -0.050 or even more seems a little to extreme perhaps, but should be tested. When BB uses a 3.0 to create unreal DF, a bigger value for lift than we used since now should work as well.
Any idea, where I can find DF/lift-values for our cars. Did a google search, but only some sporadic results for cars, we don't have in RT.

[Pero_Grozni]

Technical details-our section ;)
I was just stating that there is no problem with devising a formula to generate DFC(and even z position of Z) if we have info of lift at speed.
Most "Sport Auto" supertests have these two parameter mesured and I have a buncch of them at home so can get the values for many cars, but they are all new cars.
BTW: Downforce is rare in road cars and as far as I know even the mesuring for the 360 modena showed lift(despite having venturi underbody).
the 911 GT3 does have a bit of downforce though.

[Roger]

Yes, it is a linear function (or so it appears from Fred's calcs).

To add weight to the argument for very low DFC numbers (your original case stated above) - the lower the DFC, the less actual effect of having a low ride height (i.e. if you have 1kg of d/f and double it by dropping the ride height you get 2kg - hardly significant to the speed of the car).

BTW has anyone tried to make a car fly? A DFC of about -44.00 should get a 1000kg car off the ground and 100kph :D :twisted: How about "Glider Trophy"?

EDIT: Of course due to the ground relationship - the higher you go, the less lift you have - I reckon that as soon as you start to lift-off you may well land again. OR..... :twisted: if the car has severely restricted droop travel, you MIGHT be able to get it to hover on a downhill section.

[Ralph_B]

:lol:
I've tested this just for fun - all your assumptions are wrong, hahaha! You can't even reach the takeoff-speed, because the car doesn't get the power on the road at about 80kmh and begins to spin. Maybe I'll test with a FWD, that is more stable on straights.

Edit: works theoretical, but can't break this speed-barriere with the Fulvia either, because the drivewheels loose contact to the road and spin. Have adjusted location of DC for this test (with the current TT-settings it's orientated to the back) to avoid a early spin and mounted some megagrippy tires. But interesting, the car actually starts to behave like a hoovercraft coming near the calculated speed.
Repeated with a beetle and set DFC-location to front. a At about 80kmh the frontwheels lifted about 20cm (if it was a realcar) into the air and stayed constantly on this altitude, even when inrceasing speed. The dependance from ridehight is unoverseeable! Something like "getting underair" sometimes seen in Nascar-races and doing a backwards-roll is not possible in RT.

Some kind of fun, but interesting result!

[Roger]

:twisted:

I've had that type of fun too! I have had a car doing a wheel-stand down the straight :D

BTW - loss of grip is why I suggested downhill! :P I'm not THAT stupid :o :OMG: :smash:

Interesting that you experienced the problem at 80kph, this is probably where there is no longer enough tractive force to overcome aero drag.

[Pero_Grozni]

Suggestion-negative Aero Drag ;) might help you after you are airborn :D

[Ralph_B]

As said, tested with the beetle with my physics and set downforce far to the front - so I had traction for accellerating all the time. The beetle weights much less than 1000kg, so set the df to 40.00. Maybe to much for the mass of the car - that may explain, why the effect set in earlier. Didn't care for exact values, only was curios what happens in general.
In the very first test with a Beetle didn't change DFC-location. As it is orientated to the back, the rearwheels lost contact to the road and so the car became undriveable and brakes out uncontrollable.
My statement about breaking the speedbarriere was some kind of missleading - I reached the calculated liftoffspeed with the Lancia, but of course coldn't accellerate more with wheels in the air. But the Beetle-experiment with DF to front shows, what happens, when you increase speed.

What was annother interesting experience: the first time you can feel some effect is at about 40-50kmh when the first noticable loss of grip sets in. That's damn early, but understandable when looking at the linear function, that discribes the increase of DF.

[Pero_Grozni]

The function is not linear as I remember. It is linear in relation to speed, but not in relation to ridehight. The function has two variables.

[MudSnow]

So what is the function from which those numbers come?

[Pero_Grozni]

Look at Physic calculator:
Suspension compression Column AH.

Also some experimental data included. This was made well over a year ago so I dont remember it anymore so well. But you will see what the relation is.

Roger-can you get the formula original again? I would like to have a look at it again :wave:

[MudSnow]

So, a car with a DFC = 1 travelling at 100kph with 0.150M ground clearance gives 23.713kg of D/F - at 200kph it gives 47.426kg. Raise the car to 0.2M and the d/f drops to 38.183kg.

This is what results from calculator 1.04. V2.23 gives 8.7kg. Why the difference? Which is correct?

So according to v1.04, DFC of -0.211 will yield 10kg of lift at 200kph and .15m height. So why did you recommend 0.01 DFC which creates only -0.474 kg of lift at 200kph? And I assume this is for entire car, not only one axle.

[CDK]

Due to personal time constrainents I will have to generalize here, so plese excuse me for non documented references.
First off I think you guys have done a great job with the TTMod. I have done a lot of physics tuning over the last year and relise just how much research and calculating can go into it. My work has yet to make it to RT for others to try and give feed back on as unfortunetly my co car builder and I are not very fast and get side tracked alot. I had finished what I thought was avery nice version of our car when the TTMod came out in finished version. This sent me back to the drawing board. Interestingly I changed one parameter at a time as I examined the effects of your settings over what had been used in origional versions. Obviousely some things work together and in some cases work against each other when not doing changes all at once. I have worked tuning suspenions both in virtual and real life. I have also had the oportunity to work with stock vehicals (basicaly limited adjustments) and modified race vehicals were you could have a lot of room to improvise. So now to some observations.
I do not think you will find a lot of data on aero numbers as it was not looked at in depth untill well into the seventies. People knew there were effects and performance to be had but a lot of it was trial and error with very little chance to effectively measure it. The access to wind tunnels let alone full size cars in tunnels was just not available on a regular basis.
There are so many veriables to car shape and surface area that it is very hard to calculate what the various numbers would be with out tunnel time. I think this might be a decent example. Take the same body and put it on a frame and then on a unibody would you have the same DF numbers? If you take a VW Beettle and block off between the front bumper and the ground I think it would have tremendouse down force (this is theory but give it a thought). But take that same Beettle and get the front off the ground a little like cresting a hill top and the numbers can go the other way very fast.
The angle, shape and location of the hood (bonnett), windshield, roof and trunk (boot) also all come into play. The underside of the car plays an important part in all of this also maybe more then is often understood.
So it looks like you have your work cut out for you working to get proper figures. With what you have learned about how the RT formulas work you also have built in limitations. From taking a look at the numbers you are using it looks like zero is the middle ground with relitivly small numbers to either side in comparison to the origional numbers.You have done testing that has shown that there are limitations to how far you can go. I know the enginer in us wants to put a formula to this that would work out for each car but I think you may find that you will have to look at the vehical dimensions and shape and try to put it somewhere on the scale between the pre accepted low number and high based on agreement between tuners. I know this is a bad suggestion but I think you will chase your tale with any other approach. Also try to limit how many decimal places you want to go out as this will also simplify tuning. I do advanced control system tuning and it is very easy to get carried away with "fine" tuning. One of the things that I like about RT is the realism so I am all for it.
By the way one of the last things I ended up changing was the DF number and this was before this conversation came up.There was a wonderful change in the way the car responded. In fact I had crutched my first setup with extra down force as I struggeled with some handeling issues.
This was just some observations and ideas and it was nice to get back in on a post. As PG knows this is one of my favorite subjects to talk about. And PG, as I write this I am watching little cars with big motors and wings race where downforce is being used to the max.

Best to all,
CDK

[Roger]

Mudsnow - I don't have the latest version of physcalc with me ATM; if the formula has changed it MIGHT be because P_G was working on ride height/spring relationships - IF any of the cells that automatically refer to others got changed, then the calc will be wrong. OR, we may have made some other changes - I just don't remember - the physcalc is work in progress.

CDK - Good to hear you again. Still dreaming of be-winged V8 monsters over this side of the pond!

You are right in your general observations. I think for our purposes finding a very approximate range for various road vehicle shapes & sizes will do the job. I have been reading "Aerodynamics of Road Vehicles: From Fluid Mechanics to Vehicle Engineering " by Wolf-Heinrich Hucho. This is a little out of date, but carries lots of manufacturer's (mostly VW/Porsche/Audi) data from the late 1970's (ish). We have lift/drag numbers. I think from this, and other research it will not be hard to find a range of DFC values that are appropriate (maybe even consider a formula to alow for vehicle plan area?).

P_G - here is what Fred posted:

"DownForceConstant only affects the groundeffect. And that is only practically effective when the car body is less than 40 cm from the ground. F=v*DFC*e^(-3*((2.5*h)^3)). F is the generated force. v is the car speed. DFC the downforce constant, e is the standard Euler number 2.718... and h is the height above the ground(average for the 4 corners).

Don't ask me where this formula come from, as I find no physical explanation for it. But I recommend that you plot out the e^(-3*((2.5*h)^3)) part sepperately to see how the downforce varies with the car height. The expression is 1 near the ground, and near zero when the height is over 0.5 meters. "

Hopefully, you might be able to put it to better use than I did (watch out for the classic roger - Newtons/Kg cock-up :o )

[CDK]

Roger, Hi. I check in with N-G almost everyday but you guys drew me out with this topic. I was writing from memory late at night in my first post. As I read your last post a key term came up "ground effects". My VW Beettle example was actually closer to ground effects. The Beettle does also just happen to have an interesting body shape for down force. I would think any ride hight numbers would have more bearing on ground effect then down force. Since there is only the DF value to tune with, a compromise would have to reached to allow for the shape and weight of the car.The underside design of the car could then be factored in based on unibody (flat bottomed, plus number) or frame (open areas, negitive number).
Sedan racing here in the states has focused on aerodynamics very extensivly the last few years. The three area that get looked at are in order down force, aero drag and more recently ground effects. An interesting point here is that the cars were getting so well designed in the wind tunnel that they had great numbers and ran very well as along as no other car was around them. Once they got into traffic having a car in front of them would cause what was called aero push (understeer). The lead car would take the air off the front end of the next car and it would now lift instead of having the air push it down. The same could happen from behind as it became common to get right up on the back of your competitor and "take the air off him" making the cars rear end loose traction. Most of the articals I have seen feel that down force as you have stated does not come into play untill speeds start to increase and we may be talking well over 60 mph \100kmph. As I write this I feel your reduction of the DF number to closer to zero was deffinitly a move in the right direction.

CDK

[MudSnow]

Do we know if these numbers do anything?:

# Trunk/Hood
#--------------------------------------------------------------------
# Angles [deg], relative to car Y-AXIS... used for aerodynamics
TrunkAngle = 0.0;
HoodAngle = 0.0;



I am assuming that size is automatically included in DFC, such that two vehicles of same basic shape but different size will already not have same DFC, so no extra calculations need be made.

[Ralph_B]

Not shure, but I don't think, anybody has ever played arround with those two values seriosly, but maybe worth to be discovered.

[richard]

l think..............

TrunkAngle = 0.0;
HoodAngle = 0.0;

are undamaged, i.e. what ever angle they are in the undamged model ( even if its at a 10 degree slope ), it will still be " 0.0; "

This is then followed by the max damaged angle that the trunk or hood could open to.
As to the aerodynamics effects it causes in RT l do not know.

[MudSnow]

These angles seem to control only the behavior of hood and trunk after damage. So if hood has large negative angle, or trunk has large positive, it will stay down and not blow upward. No effect on downforce of car.

So if we have a chart which says a car has 20kg of lift at 100kph, then the calculator says that I should use a DFC = -0.844 which is very much more than -0.015 that many FTC cars have approximately. So is this the sort of number that we want to use?

So let's see that VW/Audi chart please!

http://www.ascsracing.com/gallery/shouse66073104.jpg

[CDK]

MudSnow has given us a nice shot of little cars with big motors that use max down force.
I think every body is correct on the hood \ trunk angles. It looks like 0 is the standard starting point. I have noticed that the car seems to slow after contact with non movable objects and I have thought it was due to the angle change. I have had a lot of experience with the crashing effect.
Even though we are seeing zero as a middle ground number on DFC meaning it will have little to no effect we also have CDF (center of downforce) Again I think this will have to be estimated based on the general shape of the car as it will have an impact on how the DFC effects it. I feel like I just opened the can of worms here with CDF. So before I go any further what do you guys think about the CDF numbers? I need some time to think about this myself.

CDK

[Ralph_B]

So if we have a chart which says a car has 20kg of lift at 100kph, then the calculator says that I should use a DFC = -0.844 which is very much more than -0.015 that many FTC cars have approximately. So is this the sort of number that we want to use?

Mathematical seems to be right - have experimented a little with so hight values. Maybe, a matter of usage, but I have the feeling, that we go to far with this and loss of grip is overdone. Additionally the influence of ridehight would increase and we get, what we try to avoid: again cars would become fastest always with the same setting, no matter on which surface you go: this time inverted and soft, because with the highest ridehight the influence of negative DFc would be minor! Not good either!
Have tested something in range of -0.025 to -0.080 and feels better for me. But try yourself, guys!

Of course the CGF has a lot of influence on the car-behavior, the higher the DFC the more. I tried to give some sense to the settings I used for the TT-cars. With the relative low upforce I've used so far, the influence isn't to big, but when we increase the negative DF ist will be from big importance. At the moment the CDF is still inaccurate. I've done by feel and depending a bit from visual carshape, but for maximum of realism some actual values wont hurt. But I am afraid, it will get difficult, to find some useful resources for the old cars.

[MudSnow]

Do we have an agreement on the value range that we want to use for the present time?

We need a scale to decide where a car is in the .025-.08 range.

[Ralph_B]

From my sight of view we could have an agreement, but for I am not the god of physics, I would be interested in Rogers and Pero's ideas as well (and of course of some other car-/physics-makers as well. ;)
And yes, still the open question, how to define, which car uses what value in the decided range? Maybe I've a good feeling for reality, but have to confess, physical calculations aren't my strongness and I better leave the scientific part of this to other, more talented guys. I am not quite shure, if it is possible to get some detailed information about the relevant forces for all those classical cars we prefer, so assume it will be more the less guesswork and we have to rely on our feel in most case, unless some of you have a better idea.

[MudSnow]

I have been reading "Aerodynamics of Road Vehicles: From Fluid Mechanics to Vehicle Engineering " by Wolf-Heinrich Hucho. This is a little out of date, but carries lots of manufacturer's (mostly VW/Porsche/Audi) data from the late 1970's (ish). We have lift/drag numbers.

Will you share this with us please? It should cover a large variety of car shapes which we can compare to other cars.

[Buttonz]

All cars produce lift as soon as they start moving. (Apart from formula 1)
So I set all my RT cars to dowforce=0.0 and the replays look far more real. Before it was like watching some scene from a Disney cartoon.
So even setting downforce to zero is cheating really. It should be slightly negative for for true realism (ie some lift).