Tyre Modeling

[LuisN]

I'm responsible for the dynamics of a FSAE team and I'm having problem modeling tire using the tire data from Avon web site

http://www.avonmotorsport.com/resource-centre/downloads

The Formula Ford Slick ones.

How can I do simple manipulation with the data so it be useful to me? There is any paper or material to study?

Thanks

[DougMilliken]

...can I do simple manipulation with the data so it be useful to me?

No.

My short answer assumes that you are: a) starting with little or no background on tires/tire_data and b) the definition of "simple manipulation" is to obtain useful quantitative results in a few days. There is a huge initial learning curve. After several months or years, the answer *might* change to "Yes'.


There is any paper or material to study?

Yes - many, but none of the useful references are short or easy.

This SAE paper, 2006-01-3606 "The Formula SAE Tire Test Consortium—Tire Testing and Data Handling" describes the process involved in lab testing, http://www.millikenresearch.com/TTC_SAE_paper.pdf The link is from the TTC public page,
http://www.millikenresearch.com/fsaettc.html As you read it, consider the possible differences between the TTC tests on a flat roadway vs. the Avon tests on a ~67 inch / 1.7 meter diameter drum (note, there are many other differences between these two lab tests).

Our book RCVD, SAE R-146, has two chapters on tires that introduce the subject, and various simple tire models are used throughout the vehicle analysis chapters.

Pacejka's book, "Tyre and Vehicle Dynamics" has become a standard reference.

The "tire bible", published by the US Department of Transportation includes numerous papers on different aspects of tires, "The Pneumatic Tire" DOT HS 810 561, February 2006. This 3rd edition, ~700 pages, is available online as a pdf. The first (red) and second (blue) editions, are (as far as I know) only available in print. As a broad characterization, these first two editions focus on bias ply and radial tires, respectively.

[LuisN]

Doug Milliken,

Compared to suspension geometry I'm starting to learn tire modelling. The problem is, I was doing some research and calculation with R25B with some graphics that I already know what means, just the basics. It's our and my first year in FSAE competition.

So our tires didn't arrived, more precisely the rims. Now all my work on R25B Hoosier is kind of useless because I have to deal with the Avon Formula Ford 6/21-13. Furthermore I have no idea how to use Avon data, assuming that I have little knowledge in this area.

There is anything that can be done in a week? We are 142 days from the competition and my tire changed.

My dead line is Saturday.

Thanks

[M.logan]

LuisN
Carroll Smith tune to win for basics on tire.


Regards
M.logan

[Jay Lawrence]

Luis,

What kind of work have you done on the R25B's? If you haven't set up any models then you are at square 1 and might want to take Doug's advice. Personally, at this stage, I'd put the tyres on the car and tune it.

[Luniz]

So, if you're in charge of the tires, then let me ask one question: Why did they "get changed"? Who changed them? I thought you were in charge?

What I'd suggest: Try to come up with some very simple load transfers for steady state cornering, then look at the tire curves in this excel from avon and try to "eyeball" the operating parameters (Camber/Slip...) that fit best. Adjust your suspension and steering geometry accordingly and do everything else on the test track. Building and validating a tire model, integrating it in a multibody code and running it in a week is impossible.

my 0.02$...

[LuisN]

Thanks for all,

I'll do an simple geometry, put the tires on the car and tune it with ARB and camber adjustments. I'd like to do some load transfer that and camber gain that helps the tire work but it will not be possible.

How can I take Pacejka's data from avon and plot ideal slip angle versus load?

Thanks again, Luis Nigro

[Jay Lawrence]

Pacejka's data is a set of coefficients for fitting curves. If you don't have curves (tire data from Avon) then you can't really fit it to anything.

What would you be able to do with ideal slip angle information?

[ChassisSim]

LuisN,

My friend you might find the following a really good resource,

Creating tyre models from nothing,

http://www.chassissim.com/blog/chassissim-news/creating-tyre-models-from-nothing

Also a lot of the ChassisSim community have found the following really helpful,

http://www.chassissim.com/blog/chassissim-news/modelling-tyres-using-chassissim-filling-in-the-blanks

I trust this will be of assistance to you.

All the Best

Danny Nowlan
Director
ChassisSim Technologies

[LuisN]

Jay,

With ideal slip angle vs normal load combined with load transfer I'd be able to determinate the Ackermann geometry.

-o-O-o-

Danny Nowlan,

Thanks, that is the magic that I need.

[ChassisSim]

Luis,

You are welcome my friend.

All the Best

Danny Nowlan
Director
ChassisSim Technologies

[Kevin Hayward]

Jay,

With ideal slip angle vs normal load combined with load transfer I'd be able to determinate the Ackermann geometry.
.

Luis,

Not really. you will be far short of being able to determine your Ackermann geometry. Note that when a front tyre is turned the direction of force is not perpendicular to the car. The component of force towards the rear of the car of the inside tyre can be used to improve initial turn-in. Some of the information that is peddled where you set Ackermann purely on how the peak lateral slip angle changes vs load is an incomplete analysis of the problem.

The message that you should be getting from the previous posts is that learning from an experimental framework is where you and your team should be at the moment. You can learn a lot about the effects of temperature, load, camber, ackermann etc on your tyres with a few test sessions. Danny Nowlan's approach to gaining tyre information from testing is exactly the sort of thing that will benefit your team and you should be working towards that.

Kev

[Jay Lawrence]

^What he said :)

[LuisN]

Kev,

I agree with you that is a simple approx to the problem, but due some problems in our team I have no chance to do some empirical tests.
We have only one set of tires, we don't have time to put the car on the road to test. It just killed my work.

But you agree that if I work with the peak, or ideal, slip angle vs load I can determine if the geometry should be Pro-Ackermann or Anti-Ackermann?
(SteeringGeometry - By OptimumG (http://www.optimumg.com/docs/SteeringGeometry.pdf))

For our first year on FSAE competition is one great analysis, due the fact we have no car to collect the data to design the next or make some adjustments.
Like Danny said, I'm trying to keep "simple", kind of complex but simple compared with other teams or models, to get my car tuned before it gets out the manufacturer, and so our garage to the race or our short week of tests.

Thanks all

Luis Nigro, FEB Racing - UNESP Bauru, Brazil

[Claude Rouelle]

Luis,

There are 12 causes for the yaw moment; 4 Fy, 4 Fx and 4 Mz of the tires (actually there is 13th one which is the aerodynamic Mz of the car, not negligeable of a FSAE / FS with big wings and big wing endplates).

Kevin Hayward is correct to say that the Ackermann geometry should take into account the FX of each tire, not only the rolling resistance but even more the Fx under braking (and acceleration for a FWD or 4WD).

After all, even in steady state you will need the "some" tire Fx at one or more wheel to compensate for the aero drag and the 4 tires rolling resistance. That is where your diff model could be useful too.

The tech tip we publish is a good introduction to suspension design but it is not the complete thing. For more info on that topic (and on many others) you should consider attending our seminar, there will be one in Brazil in November.

That being said, I advise you to first perform an analysis with only the tire Fy (in the chassis coordinates, Kevin is, again, correct so take the initial toe and steering angle into account) before you use the influence of the FX and Mz. At least you will know the influence of each part of the sum.

Kevin,

The seminar never stops improving. You should attend the 4 day seminar again, very different than the lat one you and/or your students attended; 900 pages. More info, quality and quantity gained from now 17 years of OptimumG activities in passenger Vehicle Dynamics and Race Car Engineering which has pushed us to always think out of the box. And if you come to Denver stop by to try our new simulator.

[LuisN]

Claude,

I agree with Kevin that Ackermann is dependent of a variety of forces, all 12 forces in the car. But as I mentioned I do not have these data to make an more complete study.
As I said, I didn't say that Kevin was wrong, but that I have no time or data to do some complex analysis and I'm trying to do some simpler thing that still be valid.

I applied to your seminar in Brazil.

Thanks.

I don't know if I made clear my problem. My problem is, lack o tire data and don't have any car to test.
With these condition, how I design one suspension and steering geometry?

This is what I have, the car weight, Avon tire data, engine approximate torque.

Luis Nigro, FEB Racing - UNESP Bauru, Brazil

[Claude Rouelle]

Luis,

Then 3 solutions (even if there are no perfect solution)

1. Design a suspension with adjustable Ackermann and go figure out what is best by testing when you will have the car ready. At least your front upright and steering rack installation / design will take that into account and will be ready for that test.

2. That one will take time and won't be ready tomorrow; use the tire curves from Avon and try to make it a tire model in Matlab, Excel or OptimumTire to understand the camber, vertical load, pressure, camber sensitivities and use this model as an input of a simulation.

3. Use another ire with tire data more available. The TTC is the best solution I can imagine.

[LuisN]

Claude,

I choose the second one.
Using the approximation of Danny so I can get some useful data from Avon.

Thanks

[Z]

I recently received a PM from Luis asking the following question. The following answer is, of course, just my opinion.


... Furthermore, I have lack of tire data, how can I continue to do my work on suspension design? ...

Luis,

Here are two "TRUTHS" about tyre-data.

TRUTH 1. Regardless of how accurate your "tyre-data" is, it (almost) NEVER EXACTLY matches the behaviour of those very same tyres on your car.

This is because any manufacturer supplied tyre-data is only describing the behaviour of those particular tyres in very specific, and very limited, conditions. Namely, they might be very new tyres (or maybe quite old?), being tested indoors in a cool climate (or maybe quite hot?), on a very artificial sand-paper-like steel-belt (or a drum?), with an unknown number of previous heat cycles (maybe none?), and probably a whole lot more unknowns than I can think of right now. Best ask someone from the TTC how many other unknowns there are.

Bottom line here is that the tyre itself is only a small part of the "equation" when it comes to overall tyre performance. The other parts, often the largest parts, are the road surface, climate conditions, history of heat-cycling, etc. I would not trust ANY of the specific, manufacturer supplied, tyre-data to be within 10 % accuracy when you are trying to match your real testing on that old parking-lot. The data could easily be 50% off, depending on conditions.
~o0o~

TRUTH 2. If you randomly draw some "freehand tyre-curves" that are within the usual range for such things, then that "made-up" tyre-data will be a VERY ACCURATE representation of some tyre, sometime, on some road, in some weather conditions, somewhere in the world.

Unfortunately, it is probably NOT an accurate reflection of your particular tyres, right now. But that doesn't matter, because the freehand tyre-curves are FREE! So you can afford to draw lots of them, so they cover the full realistic range of all real tyre behaviours. You can then carry out lots of VD simulations using all these different sorts of "tyre-data", and see how your simulated car behaves.

You then look for the freehand tyre-curves that give your simulated car the "best" performance, however you choose to define "best". Your last step is simply to go looking for real tyres that have curves ROUGHLY like the freehand ones that gave the best simulated performance. You will never find EXACTLY the right tyres. You are only after the right general direction to head in.

Bottom line here, you can use 5 significant digits (or 8?, or 12?) for your "freehand-tyre-curves" in the your VD simulations. But don't expect the real tyre-data to be anywhere near as accurate. Maybe the first digit, at most.
~o0o~

To sum up, a practical approach to choosing tyres for a first year team like Luis's.

1. Start by looking at what tyres other teams are using. Look more closely at the more successful teams' tyres, and at teams with similar cars to yours (ie. total mass, power, etc.). Most importantly, check carefully what tyres you can actually buy, and can be delivered within a reasonable timeframe. (Hint: You are looking for "hi-grip", racing-rubber, not hard-wearing, econo-tyres.)

2. Meanwhile, do lots of "theoretical" simulations of your car with your own "made-up" tyre-curves that ROUGHLY cover the range of the tyres you expect to use (eg. start with "Mu" values ranging from 1/2 to 2, but concentrate more on the higher end). This should give you a good general idea of how to set-up your car, for any given tyres in that range.

3. When your car is finished, do lots of testing on that "parking-lot out back". Compare your simulations with the real performance. Based on these results, and time permitting, perhaps try some new tyres that have tyre-curves that now seem "better" for your car. Or else recommend to next year's team that they try tyres that have more "whatever". (Hint: Generally you are after higher Mu, but driveability, longevity, etc., also count...)

Z

(PS. For Ackermann I suggest you aim for as much "Pro-" as you can get from a reasonable steering-linkage. Ie., aim for outside-wheel = ~30 degrees, inside-wheel = ~45 degrees, at full-lock. Your VD simulations will tell you why, even with roughly "made-up" tyre-data!)