F1 TV tyre graphic

[peaty]

Hi,

I've got a question regarding the on-screen tyre graphics that were used for the first time during the Japanese Grand Prix. Initially the graph was a disaster and, in Mexico, Pirelli met with F1 TV representatives to try to help them out (they weren't involved until that point).
So, ahead of the race in Mexico City, F1 announced that viewers will see an improved version of a new on-screen graphic, titled 'Tyre performance'. It looks like the model calculate single tyre energies. I would love to learn how to calculate tyre energy in a corner/lap. Can someone explain me how to do so or recommend me a good read on the topic!?

Cheers

P.S: when I think about tyre energy, the first thing that comes to mind is T8 in Turkey, arguably the most demanding corner in terms of tyre energy

[Tim.Wright]

Tyre energy cannot be directly measured and is not a simple thing to calculate. There are a few different sources of it and for each one you need to know some of the tyre characteristics. For example, energy due to the cornering or braking forces are the forces multiplied by the contact patch slip speed.

But getting the tyre forces and slip from the realtime data is pretty tricky. They have sensors on the car for the lateral slip so this then must be transformed to the contact patch slip. Then with wheelspeed sensors you can (try to) get the longitudinal slip. Force might be measured via strain guages otherwise it needs to be estimated using a handful of vehicle sensors and a load of vehicle information. Once you have the total contact patch force and slip speed you multiply the two together to get the energy in Joules.

That doesn't tell you the full picture though, there are some other sources of energy loss such as rolling resistence and hysteretic losses from carcass deflection but I'd guess the major percentage comes from the tyres forces and slip.

[izzy]

I don't know why they can't simply show us the tyre temperatures. well i do, they couldn't get sponsorship

[peaty]

Tyre energy cannot be directly measured and is not a simple thing to calculate. There are a few different sources of it and for each one you need to know some of the tyre characteristics. For example, energy due to the cornering or braking forces are the forces multiplied by the contact patch slip speed.

But getting the tyre forces and slip from the realtime data is pretty tricky. They have sensors on the car for the lateral slip so this then must be transformed to the contact patch slip. Then with wheelspeed sensors you can (try to) get the longitudinal slip. Force might be measured via strain guages otherwise it needs to be estimated using a handful of vehicle sensors and a load of vehicle information. Once you have the total contact patch force and slip speed you multiply the two together to get the energy in Joules.

That doesn't tell you the full picture though, there are some other sources of energy loss such as rolling resistence and hysteretic losses from carcass deflection but I'd guess the major percentage comes from the tyres forces and slip.

Thank you very much Tim.
So…energy due to the cornering or braking forces are the forces at the contact patch multiplied by the contact patch slip speed…
I’m trying to work backwards, let's see if I'm getting it right…
The equation for the lateral and longitudinal slip velocity are:

Vlat= V*sin alpha
Vlong= V* cos alpha –Vb

Where Vb is the circunferencial velocity of the tire under traction or braking.

I can potentially measure slip angle and steering angle and with that calculate tyre slip angle (considering you could measure/calculate V and Vb).
I think you could use:
1- 2 correvit type sensors (vehicle’s co-ordinate frame). The system measures the forward and lateral velocities directly and compute slip angle using the formula.

2-You could use dual-antenna GPS and/or inertial navigation systems (both earth-based co-ordinate frame).

Am I right!? Am I missing any other way!? Accuracy will change with speed, right!? these instruments measure velocity with a relatively fixed accuracy, so…the smaller the speed the more significant the error.
Is this approach correct so far!?
*Would it be possible to calculate slip angle base on accelerometers!?
So…you won’t be getting a real value but just an approximation, right!?

With regards to the forces…as you said, strain gauges/load cells are the way to go. My question is how do you calculate forces in the contact patch considering your strain gauge is in the pushrod!? Am I missing something or is it just a trigonometry problem!? The resolution will depends upon link geometry, which generally changes with wheel position…So, the push rod load by itself can be resolved into vertical load, but…the result will be "affected" by loads reacted by other suspension arms and wheel position…
In any case, again, you are just getting an approximation of what’s happening at the contact patch...right!?

*Just as side note, you could also use linear pots to measure suspension travel (though you'll be neglecting shock loads, which may or may not be a big deal) to calculate force, which could be good enough for some scenarios…

You multiply both results and that’s just one part of the total energy…
You could calculate/measure rolling resistance and I have no idea on how to calculate/measure hysteretic losses…

P.S: I think I'm missing something though...power is equal to force times velocity, and energy is power multiplied by time

[Tim.Wright]

Pretty much everything you said is correct. The most important thing you said was:

In any case, again, you are just getting an approximation of what’s happening at the contact patch...right!?

This will be the case no matter how many sensors you put on the car since there is no sensor that will measure contact patch energy directly.

The equation for the lateral and longitudinal slip velocity are:
Vlat= V*sin alpha
Vlong= V* cos alpha –Vb

Yup

I can potentially measure slip angle and steering angle and with that calculate tyre slip angle (considering you could measure/calculate V and Vb).
I think you could use:
1- 2 correvit type sensors (vehicle’s co-ordinate frame). The system measures the forward and lateral velocities directly and compute slip angle using the formula.

2-You could use dual-antenna GPS and/or inertial navigation systems (both earth-based co-ordinate frame).

You only need one slip angle measurement be it an optical sensor (Correvit) or inertial+GPS. To transfer this to the contact patches you need to know the complete K&C characteristics of the axle to transform the body slip angle to the contact patch. This is already a massive job. Not super difficult but very long if you are starting from a blank sheet of paper (or m-file).

*Would it be possible to calculate slip angle base on accelerometers!?

In my opinion, not to any reasonably accurate degree.

There are slip angle estimators which use machine learning to transform signals from basic sensors like accelerometers, wheel speed and steering data into slip angle - but this is a discussion by itself.

With regards to the forces…as you said, strain gauges/load cells are the way to go. My question is how do you calculate forces in the contact patch considering your strain gauge is in the pushrod!? Am I missing something or is it just a trigonometry problem!? The resolution will depends upon link geometry, which generally changes with wheel position…So, the push rod load by itself can be resolved into vertical load, but…the result will be "affected" by loads reacted by other suspension arms and wheel position…
In any case, again, you are just getting an approximation of what’s happening at the contact patch...right!?

I see four main ways to do it.

1. Install dynamometric wheels which will give you the wheel forces directly
2. Strain gauge the complete suspension and calibrate it on a K&C rig to get all 6 d.o.f.
3. Use an estimator to calculate the wheel forces based on simple on-board sensors. The estimators have some critical limitations but on the other hand strain gauging a complete suspension or using dynamometric wheels is a nightmare so it's often a reasonable cost/complexity compromise.
4. Machine learning algorithms trained with a good multibody simulation model

The last one, honestly, I've never tried or seen it tried but I'm sure it could be possible.

My guess is that in F1 they strain-guage the complete suspension. Given that indoor tyre testing is banned it practically the only way to get F&M data which is accurate enough for tyre identification.

[Tim.Wright]

P.S: I think I'm missing something though...power is equal to force times velocity, and energy is power multiplied by time

Yup, I fkd that sorry - force x slip will get you the power in Watts, then you need to integrate it to get energy.

[Tim.Wright]

Speak of the devil, Renault were running dynamometric wheels in testing today.

[peaty]

Speak of the devil, Renault were running dynamometric wheels in testing today.
https://pbs.twimg.com/media/EK1sgiEWsAANdEu.jpg

Interesting picture! thanks for sharing!
Are dynamometric wheels the same as the so called wheel force sensor (WFS)/wheel force transducers!?
In the case of dynamometric wheels, are teams using their own sensors or do they use third party ones from suppliers like Kistler or others?

So dynamometric wheels give you the wheel forces directly, right!? how would you go from there to the contact patch forces!? I think my problem is I don't know that much (pretty much nothing tbh) about dynamometric wheels. I'll try to find a good read about them!

[Tim.Wright]

Dynamometric wheels are also known as wheel force tranducers - it's the same thing.

If you know your wheel centre forces and moments you can transform them to the contact patch as long as you know the delta coordinate from the wheel centre to the contact patch.

[peaty]

Dynamometric wheels are also known as wheel force tranducers - it's the same thing.

If you know your wheel centre forces and moments you can transform them to the contact patch as long as you know the delta coordinate from the wheel centre to the contact patch.

so just taking moments and summing forces. I was expecting some other intermediate step. Once again it's clear why achieving "reliable" numbers is so difficult! But I guess a good "estimation" it's better than nothing…
Thanks Tim!

[strad]

As regards the onscreen graphic I view it as just an educated guess and not a real/solid number.

[peaty]

As regards the onscreen graphic I view it as just an educated guess and not a real/solid number.

Hopefully that's what they are doing now (an educated guess)!but the first time...it was pretty much a finger to the wind! According to the graphic, Hamilton could have done until 2021 on those tyres (including the drives from and to Brackley)!
Anyway, I was interested in the real calculation and the graphic was the perfect excuse to finally have a look at it!