Friction coefficient and temperature/pressure

[ubrben]

Interesting information there, thanks a lot ! I'm not advanced on the calculation point of view, but at least I can explain why !

My question was a bit naive, since you cannot know the EXACT rubber mixture used by the manufacturer. If all of them would provide the composition that if a company (like TüV for example) would test all of them using the same method, I am sure we could get a formula defining the friction coefficient depending on the temperature.

What would be interesting is to evaluate which components in the mixture has the most influence on the grip parameter. Durability would also have to be considered, but I can imagine the curve is not exponential so in the end it is always a compromise.

A shame this discussion is going to a dead end but that micro-waves method is interesting nevertheless ! Who knows what's next !

Sorry to be blunt, but you are being naive.

Even if you know the "exact" composition of the rubber, you still don't know how it was cured, what thickness it is on the tyre, how the tyre carcass influences the amount of energy input into the compound. In turn the race track layout will also affect the energy going into the tyre; long corners or short corners? Then you've got the track surface; what type of aggregate, how worn is it? How much more wear has it got since the last time we measured it. Is it contaminated with dust, how much rubber's down?

On top of that what suspension setup are the team running this weekend, what if we gain 10C tyre temp because they changed the dampers.

In the nicest possible way I'm sick of simulation engineers and people who play iRacer a lot thinking there's a closed form equation for everything... /rant

Ben

[Jersey Tom]

In the nicest possible way I'm sick of simulation engineers and people who play iRacer a lot thinking there's a closed form equation for everything... /rant

I would have given this +2 if I could.

Sorry to be blunt

Better be careful, you'll start sounding like me.

[MadMatt]

Interesting information there, thanks a lot ! I'm not advanced on the calculation point of view, but at least I can explain why !

My question was a bit naive, since you cannot know the EXACT rubber mixture used by the manufacturer. If all of them would provide the composition that if a company (like TüV for example) would test all of them using the same method, I am sure we could get a formula defining the friction coefficient depending on the temperature.

What would be interesting is to evaluate which components in the mixture has the most influence on the grip parameter. Durability would also have to be considered, but I can imagine the curve is not exponential so in the end it is always a compromise.

A shame this discussion is going to a dead end but that micro-waves method is interesting nevertheless ! Who knows what's next !

Sorry to be blunt, but you are being naive.

Even if you know the "exact" composition of the rubber, you still don't know how it was cured, what thickness it is on the tyre, how the tyre carcass influences the amount of energy input into the compound. In turn the race track layout will also affect the energy going into the tyre; long corners or short corners? Then you've got the track surface; what type of aggregate, how worn is it? How much more wear has it got since the last time we measured it. Is it contaminated with dust, how much rubber's down?

On top of that what suspension setup are the team running this weekend, what if we gain 10C tyre temp because they changed the dampers.

In the nicest possible way I'm sick of simulation engineers and people who play iRacer a lot thinking there's a closed form equation for everything... /rant

Ben

There is always a way to make things simple which will be good for most of people. If you take a car with a neutral geometry (no camber, 2-3° of positive caster, no toe-in/out), new rubber, etc, and you test the grip of the tire depending on the slip angle, you can get a nice curve. That doesn't sound too complicated, isn't it ?

With such curve you could map the tire grip decently, as a start. I didn't say all you need is tire mixture, but that would be a good start to have this variable for many tires and make tests on them. You're maybe sick of sim engineers, but in the end, there is always a way to putting all this into equations and make models to get the better out of what's available. Isn't this what racing is all about ? Getting the best out of what is available ?

I understand your feeling, but personally I like to understand how things work, so I can understand how to alter them in the way I want ! Sometimes it isn't possible, but sometimes there is a bloke who spent 20 years studying on this and published his results which are still the reference in those days, so you never know !

[marcush.]

the formula1 people do not have that option ..start on better tyres ...I had the pleasure to work with Michelin on a "works" supply and sure e´ven for me it was obvious these were better than anything i had worked with by a very long shot.

[Jersey Tom]

If you take a car with a neutral geometry (no camber, 2-3° of positive caster, no toe-in/out), new rubber, etc, and you test the grip of the tire depending on the slip angle, you can get a nice curve. That doesn't sound too complicated, isn't it ?

What you described is actually very difficult to do (which supports Ben's point!)

[Lycoming]

for starters, measuring slip angle on that kind of setup isn't exactly trivial.

presumably if you have some significant slip angle, you would also have different normal loads on the inside and outside tyres.

[ubrben]

There is always a way to make things simple which will be good for most of people. If you take a car with a neutral geometry (no camber, 2-3° of positive caster, no toe-in/out), new rubber, etc, and you test the grip of the tire depending on the slip angle, you can get a nice curve. That doesn't sound too complicated, isn't it ?

The fact that you think that isn't too complicated is part of the problem.

You're proposing doing it on a car. Well as I said before the dampers will have a massive impact on the grip itself as will the track surface type, condition and temperature. The driver will have a choice as to how he warms the tyres; weaving, riding the brakes, both? Something else??

So even if you do this test the tyre isn't a constant.

We haven't even talked about how you measure the slip angle and the tyre forces. Even the lightest Kistler hub is probably going to be heavier than the race wheel so your unsprung load transfer will have changed giving you different tyre loads...

Then it rains and the track changes...

Then you go to another track...

None of this is impossible, but I hope you might be starting to appreciate that given the resources required it's rarely the right way of spending a racing budget.

Ben

[MadMatt]

I understand there are lot of parameters involved and that it is very difficult (or nearly impossible) to know them all. That wasn't the point of my message, although I can understand you want to explain why it is difficult (or impossible) to solve this. It also depends on the degree of precision you want to reach, compared to the difference you can see in the results. I just thought there was some kind of general model of the friction coefficient depending on the temperature of the tire, which seems to be wrong.

I am maybe not that well into vehicle dynamics or aero, but I'm a bit surprised to see people trying to complicate things that much (ubrben). I''m not trying to get values 99.99% close to the reality otherwise I wouldn't be on that forum but behind a computer in a R&D team, but for some kind of "amateur" simulation, I thought there was a model available for this which would be accurate withouth being too complicated. Of course the more you want to get close the reality, the more you need to take in consideration (as you mentioned).

The fact that you are pissed off by people trying to build model of "every" physic phenomena is up to you, but I find interesting the fact to try to understand these principles and put them into equations. If some genius had not done this during the past centuries, you wouldn't have been able to type on your keyboard to write your message. There still phenomena that cannot be described (I cannot find the English word for what I'm trying to say), and solving them can get you the Nobel price, so nothing is impossible. It is just more or less difficult.

[marcush.]

The biggest issue is the lack of testing preventing you to build up a solid experience base from which you could guestimate your approach.
with all the variables involved you still would most likely have some black spots in your armoury and if you accidently tapped into those areas chances are -you got a bad day in the office.

On the other hand with simulation you sure can build a model of the tyre and give it some response over the parameters you think are those who drive performance.If you happen to get this close to reality you will be able to predict tyre response to almost every situation .
With every race and race condition you can hone the model you are building by deleting parameters that will have only minor influence on performance or readress parameters you thought were not that important .
So in effect you would build a map of combinations that work and others that did not work and that will give ýou a working range.

Unfortunatelly you are constantly changing the test -by developping the car in other areas as well...so sure this is a Hercules task ...and i wonder if it were not better to follow your instincts and simply look for a steady balance optimising tyre contact patch over time instead of chasing a pink elephant..

[ubrben]

I understand there are lot of parameters involved and that it is very difficult (or nearly impossible) to know them all. That wasn't the point of my message, although I can understand you want to explain why it is difficult (or impossible) to solve this. It also depends on the degree of precision you want to reach, compared to the difference you can see in the results. I just thought there was some kind of general model of the friction coefficient depending on the temperature of the tire, which seems to be wrong.

I am maybe not that well into vehicle dynamics or aero, but I'm a bit surprised to see people trying to complicate things that much (ubrben). I''m not trying to get values 99.99% close to the reality otherwise I wouldn't be on that forum but behind a computer in a R&D team, but for some kind of "amateur" simulation, I thought there was a model available for this which would be accurate withouth being too complicated. Of course the more you want to get close the reality, the more you need to take in consideration (as you mentioned).

The fact that you are pissed off by people trying to build model of "every" physic phenomena is up to you, but I find interesting the fact to try to understand these principles and put them into equations. If some genius had not done this during the past centuries, you wouldn't have been able to type on your keyboard to write your message. There still phenomena that cannot be described (I cannot find the English word for what I'm trying to say), and solving them can get you the Nobel price, so nothing is impossible. It is just more or less difficult.

I think you have too much faith in science...

There are lots of models of rubber friction vs. temperature the Persson model that JT posted some links to is one of the most widely used in the tyre industry, but the number of assumptions that need to be made to do a calculation mean that it's not a usable predictive tool for racing in any way.

The TaMeTyre model form Michelin (actually IPG did it for them...) is good in a generic sense, but still uses flat-track force and moment data to populate it. Flat track tests of different compounds rarely correlate with track tests, so the only way of adequately characterising the compounds is to track test them, which is the very thing you want the model to do for you in the first place!

I'm not pissed off, I just have some very direct practical experience of this field, which I think is relevant to the discussion rather than fantasising that science can solve any problem.

Ben

[Jersey Tom]

I thought there was a model available for this which would be accurate withouth being too complicated.

You can make whatever model you want. Here's a model for "friction" as a function of temperature: mu = x1 + x2 * Temp. Accurate centered around some point and within a small range. Or on the other end of the spectrum you have the WLF equation set.

Either way, the model almost doesn't matter. That's the easy part. Make it whatever you want. Getting appropriate data to define the model parameters is the challenge. Just like a spring model is F = -k * x... that doesn't tell you anything until you know 'k' for any given spring.

[ubrben]

I thought there was a model available for this which would be accurate withouth being too complicated.

You can make whatever model you want. Here's a model for "friction" as a function of temperature: mu = x1 + x2 * Temp. Accurate centered around some point and within a small range. Or on the other end of the spectrum you have the WLF equation set.

Either way, the model almost doesn't matter. That's the easy part. Make it whatever you want. Getting appropriate data to define the model parameters is the challenge. Just like a spring model is F = -k * x... that doesn't tell you anything until you know 'k' for any given spring.

+1

That's generally the difference between science (and academic engineering for that matter) and the real world where a result (however imperfect) is needed within a defined timescale and budget constraint.

Ben