0% found this document useful (0 votes)
85 views10 pages

The 2D Tyre Modelling Conundrum and How To Fix It Using Setup Referencing

The document discusses the limitations of 2D tire modeling in accurately representing sensitivity to setup changes. A 2D tire model captures the relationship between tire force and vertical load but lacks fidelity when it comes to small setup changes. The author proposes "setup referencing" which accounts for how tire temperature and pressure change with setup, affecting contact patch length and internal temperature. This better represents real-world tire behavior and sensitivity to setup parameters.

Uploaded by

banabana
Copyright
© © All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PDF, TXT or read online on Scribd
0% found this document useful (0 votes)
85 views10 pages

The 2D Tyre Modelling Conundrum and How To Fix It Using Setup Referencing

The document discusses the limitations of 2D tire modeling in accurately representing sensitivity to setup changes. A 2D tire model captures the relationship between tire force and vertical load but lacks fidelity when it comes to small setup changes. The author proposes "setup referencing" which accounts for how tire temperature and pressure change with setup, affecting contact patch length and internal temperature. This better represents real-world tire behavior and sensitivity to setup parameters.

Uploaded by

banabana
Copyright
© © All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PDF, TXT or read online on Scribd
You are on page 1/ 10

The 2D Tyre Modelling conundrum and how to fix it using setup referencing

Over the last couple of months I have been involved in a number of different car modelling
projects. These have ranged as categories as diverse as stock cars, sports cars to open
wheelers. Yet despite the diversity I have been seeing the same themes emerging. Principally
when using a 2D tyre model (Traction circle radius as a function of load only) you have great
correlation but the sensitivity of the model to very fine changes is extremely small. The
trends still point you in the right direction so it’s still useful but the changes are much smaller
than what you would see on an actual car. The purpose of this article is to explore why this is,
and propose a simple fix for it that I will term setup referencing.

Let me add from the outset that this is an article that is designed to get you thinking. What
I’m proposing here is not the final picture of how to model a tyre. Rather consider this a
discussion paper to get you all thinking about the mechanisms that drive what we see when
we make a setup change on a car.

To kick things off let’s explore what your getting when you refine a 2D tyre model. The crux
of the 2D tyre model is effectively to determining the traction circle radius (or Pacjeka D
term) as a function of load only. This can be done either from tyre test rig data or using the
ChassisSim tyre force modelling toolbox to get the tyre model. When you have done your job
right you should be looking at correlation that looks like this,

Fig-1 – Typical 2D tyre modelling correlation using ChassisSim.


On a number of different articles I have commented that when you create a 2D tyre model
you are taking a snapshot of the tyre at a particular pressure and internal temperature
condition. This is illustrated in Fig-2

Fig-2 – An illustration of what you get from a 2D tyre model.

The setup working well line is the bit of the tyre you have typically taken the snapshot off.
That being said most people would look at this and say great we are ready to start simulating.

However where 2D models struggle is sensitivity. Let’s illustrate this with some hard
numbers that I’m going to take from a V8 Supercar. In reality I could Illustrate this with any
car, but I’m using a Supercar because the change we are about to discuss is well known and is
a very good case in point.

To frame this discussion let’s illustrate some typical V8 Supercar numbers. Typical setup
values in Table – 1,

Table-1 – Typical V8 Supercar setup

Parameter Value
Front roll centre 100 mm
Rear roll centre 240 mm
Front Wheel rate 60 N/mm
Rear Wheel rate 60 N/mm
Front bar wheel rate 40 N/mm
Rear bar wheel rate 10 N/mm
Front track/Rear track 1.6m/1.6m
Front tyre spring rate 305 N/mm
Rear tyre spring rate 305 N/mm
c.g height 0.45m

A representative 2D tyre model for a V8 supercar is shown in Table – 2.

Table – 2 – Representative 2D tyre model for a V8 Supercar.

Parameter Value
Initial co-efficient of friction 2.2
Peak Load 850 kg

Now that we have this information to hand let’s explore some setup sensitivity parameters.
One big change in a V8 Supercar is a rear roll centre change. Typically a 10mm will produce
a measurable change. To keep things simple let’s apply a lateral acceleration of 1.4g and use
this acceleration to estimate tyre loads and approximate cornering force. The results are
shown in Table – 3,

Table – 3 – Predictive numbers for a Rear roll centre change.

Setup Load FL Load FR Load RL Load RR FyR V_pred


RRC 240 674.96 133.82 694.58 54.47 9993.5 81.47
RRC 250 671.61 137.16 697.93 51.1 9938 81.27

All Loads are shown in kg, the lateral forces are shown in N and V_pred is in km/h.

What has been presented in Table – 3 is a very simplified analysis. We are simply taking a
static weight and applying a load transfer to it for a typical low speed corner. However what
is revealing in Table-3 is that while the speeds are representative the magnitude of the change
is very small. With the 10mm rear roll centre change, the tyre loads have changed by a mere
3 kg, and the speed has changed by 0.2 km/h. Yet in practice on the car you’ll typically see a
change of 0.4 – 1km/h. In reality the truth is somewhere in the middle but this illustrates that
a 2D model gives you a very good broad brush stroke but does lack fidelity.

The reason the change is so small is that the delta forces produced by the 2D model are
miniscule. This is primarily due to the value of the peak load of 850 kg. When we crunch the
numbers the magnitudes of the changes are so small because the gradients of the tyre force
curve are tiny. This is due to the outside tyre loads approaching the optimum tyre load of the
model. Don’t take my word for it, do the numbers yourself. Remember a simple 2D tyre
model is given by,
Fy = k a (1 − k b ⋅ Fz ) ⋅ Fz
1 (1)
Lp =
2 ⋅ kb

where
ka = initial coefficient of friction
kb = drop off of coefficient with load
Fz = load on the tyre

There are a number of things you can do to mitigate this but ultimately it compromises your
model in other areas. For example to go for setup sensitivity we can drop the peak load. It
gives you sensitivity, but it compromises your accuracy for high speed corners and adding
more downforce. Clearly there is something else at play here.

To further ram home this point let’s consider the nature of load transfer for an independent
suspension. I spoke this at length in my article on V8 Supercars which I entitled a Brave New
World. Remember the highlights for an independent suspension. Due to the nature of force
application points, the suspension geometry will effectively pin the outside tyre, magnifying
the load transfer on this tyre. This load transfer is not equal and opposite like it was on the
live axle. I can tell you right now this is where the 2D tyre model will run out of steam. Let’s
crunch the numbers to illustrate this. To keep our discussion simple we’ll use the tyre model
and I’ll add 15 kg to the outside front tyre. The results are illustrated in Table – 5

Table-5 – Load Transfer numbers for a front suspension

Setup Load FL Load FR Fyf V_pred


FRC std 674.96 133.82 11432 84.02
FRC proj 690 130.0 11426 84.0

This would be the case of a front rear roll centre change of 20mm. As we can see rather than
losing cornering speeds it remains level. This is quite different from what you would expect,
but what this illustrates is the limitations of using a 2D tyre model.

At this point it would be wise for me to not some historical observations about using a 2D
tyre model. It has been my experience that while a 2D tyre model does lack sensitivity
ultimately though it will point you in the right direction for most mechanical setup changes.
However bare in mind the changes you are looking for are extremely small. This is due to the
nature of the tyre model and the way the simulator dives the car. If you are content with this a
2D model is a fantastic tool. However if you want to quantify everything we need to use
something else.

What we can divine from all this is to be very careful when someone claims a 2D model can
give you great model sensitivity. All the numbers we have crunched in Table- 1 -5 using
equation 1 will tell you, that you will be very hard pressed to achieve model sensitivity. If
you are getting model sensitivity there are a number of factors at play here,
• They are using a non linear table for Traction circle radius vs load.
• Camber effects are coming into play.
• Or something isn’t quite right.

Either way you don’t have to be a rocket scientist to figure out we are going to need
something else to help us in aiding model sensitivity.

To figure out a way out of this mess let’s consider what the tyre is actually doing. We’ll kick
off this discussion by illustrating a tyre schematic in Fig-3,

Fig-3 – Tyre schematic

What happens is as we apply different setup conditions to the tyre the internal temperature
and hence the tyre pressure changes. At this point you might be thinking so what? There are a
number of factors that come into play here. The first is the effect this has on the contact patch
variation.

FZ
Ac. p =
PT
(2)
Ac. p
l c. p =
wt
where,
Ac.p = contact patch area (m2)
Fz = vertical load on the tyre (N).
wt = width of the tread (m).
lcp = length of the contact patch (m)
As we can see as the Pressure goes up the contact patch length goes down. Consequently
there is less area in the contact patch. So consequently there is less area for the forces to be
applied and the tyre force will go down. In the converse situation when the temperature and
pressure is not optimal, there is more contact patch length and hence greater area. But the
core temperature of the tyre isn’t up to temperature yet. This is where the second effect comes
into play. From the research that Michelin did in constructing the Michelin TaMe Tire model
the overall friction of the tyre is given by,

μ = fn(PL ,Vs , TINT ) (3)


Where,

u = Co-efficient of friction of the tyre.


PL = Local tyre pressure applied at the tyre
Vs = Sliding velocity of the tyre.
TINT = Core temperature of the tyre

This combined with the contact pressure will have a significant effect on the tyre forces. This
is what we are approximating with setup referencing.

So reviewing this let me propose a modification to the traction circle radius equation. The
modification I’m going to propose is the following,

TC RAD = fn(FZ ) fn(Pt ) (4)


What we have here is the 2D model that we discussed in equation (1) multiplied by a
multiplication factor that is a function of hot tyre pressure. This function will look something
like this,
Fig-4 – FFact multiplier of Traction Circle radius vs Tyre Pressure.

At this point some of you might be thinking what drugs am I on and I think a fair few of you
might be a tad annoyed but let’s take a moment to review a things. From Fig-1 we know we
can achieve very good correlation using a 2D model. However when we investigate the
nature of the numbers, the sensitivity needs to be improved. Furthermore if we are not careful
it can lead us up the garden path. Also any body that has spent more than 5 minutes with a
real race car knows how critical it is to get the tyre pressures right. Consequently there must
be something else coming into play and all the results from the Michelin TaMe Tire model
would indicate very strongly that tyre pressure plays a critical role. You also can confirm this
from running the car.

To quantify this we are going to have to run a series of tests on the car. The test matrix we
will run and what to record is recorded in Table – 1

Table – 6: Tests to run for Setup referencing

Test No Setup What to record


1 Base Setup Tyre Pressures + data
2 Base Setup + Fnt TP -2p.s.i Tyre Pressures + data
3 Base Setup + Fnt TP -1p.s.i Tyre Pressures + data
4 Base Setup + Fnt TP +1p.s.i Tyre Pressures + data
5 Base Setup + Fnt TP +2p.s.i Tyre Pressures + data
6 Base Setup + Rear TP -2p.s.i Tyre Pressures + data
7 Base Setup + Rear TP -1p.s.i Tyre Pressures + data
8 Base Setup + Rear TP +1p.s.i Tyre Pressures + data
9 Base Setup + Rear TP +2p.s.i Tyre Pressures + data

It is critical for every run you are recording the hot tyre pressures. You’ll also note I’ve
indicated a sweep of +/- 1 p.s.i. In reality this is going to be dictated by your running
experience and what the car wants.

The critical thing to be paying attention to here is the lateral acceleration data. Due to the
nature of this test if there is a change it’s very easy to isolate what caused it. We can also
easily quantify the change. For example if we are dealing with a front tyre pressure change
the delta FFact will look like this,

a y _ cur − a y _ ref
ΔFFFACT = (5)
wdf ⋅ a y _ ref
Here we have

ΔFFFACT = Change in Force factor


ay_cur = Measured lateral acceleration for the change.
ay_ref = Measured lateral acceleration for the base setup.
wdf = Weight distribution at the front.

Clearly if you are measuring a change at the rear you simply measure substitute the wdf with
weight distribution at the rear. Once this data is calculated, Fig-4 can be readily populated.

The next step in the process is to match our setup data with the tyre model. The critical thing
to note here is that the more roll centre you apply or the stiffer you make one end of the car or
the more static tyre pressure you apply, the hot core temperature and pressure of the tyre will
increase. These adjustments are a race engineer’s primary tools when they are dialling in core
tyre temps and pressures. What we are going to do here is match this to our baseline setup
and attach multipliers to our baseline setup. This is the key to setup referencing. While this is
not perfect at least it gets us into the ball park and this is our goal.

The most effective way to do this is to use your race car simulator to match your expectations
with what occurred with the car. Some of you think this is fudging. It isn’t. What you are
doing is using the race car simulator to quantify what’s going on with the car. This to me is
the biggest benefit of using race car simulation because it forces you to understand what is
going on with the car. This is the biggest pay off to using race car simulators which is why I
maintain you must do it yourself.

So to achieve this we need to do assign the base setup and the pressure multipliers. As an
example let me show you the sample controls in ChassisSim,
Fig-5 – Setup reference dialog from ChassisSim.

As can be seen the base setup is indicated in the first column. Please note The tyre pressure
you enter here is the warm tyre pressure from the base setup. This is very important. The next
step in the process is to assign the pressure multipliers. Some baseline suggestions for
numbers in p.s.i is illustrated below,

Table-7 – Some rough rules of thumb for the setup sensitivity parameters

Parameter Value
Tyre pressure 1*(TPcur – TPref)/TPref
Roll centre 30-40*(rccur – rcref)/h
Pitch Centre 30-40*(pccur – pcref)/h
Spring/ Third Spring Rate 10*(kcur – kref)/kref_total
Bar rate 10*(kcur – kref)/kref_total

The variables here are,


TPcur = Current tyre pressure
TPref = Tyre pressure of the reference setup.
rccur = Current roll centre
rcref = Roll centre of the reference setup.
pccur = Current pitch centre
pcref = Pitch centre of the reference setup.
h = c.g height.
kcur = Current spring, third spring or bar rate.
kref = Spring, third spring or bar rate of the reference setup.
kref_total = Sum of all the spring and bar rates of the reference setup.

These numbers are subject to car factors such as motion ratios and parameters. However the
key to getting this right is to use these pressure multipliers to dial in some typical changes in
setup and dial this. Remember at this point we are using the race car simulator as the ultimate
motorsport calculator.

Also too bear in mind what we are presenting here is an approximation and don’t forget that.
Before a lot of the readership turn their noses in disgust at this let me just remind you of some
approximations that work pretty well,
• In most CFD analysis for aircraft air viscosity is ignored. However it gets you in the
ball park.
• The calculation of damping ratios ignores the tyre spring rate. It still doesn’t stop it
being a valuable tool.

Remember the key here is we are using setup referencing to quantify what’s going on with
the car. From that we can make the appropriate determination of setup changes.

In closing then setup referencing provides a very interesting perspective about how we can
use race car simulation to classify the sensitivity of setup changes with the car. Also too just
remember this is not a technique etched in stone. Rather this is something to get you thinking.
Consider this as a first step into quantifying just how sensitive a setup change truly is.
Remember the key to this technique is to classify the tyre force sensitivity to hot pressure
changes and matching that to our setup information. Consequently this is a useful
approximation for extending the 2D model of the Traction circle radius vs load characteristic
of the tyre. This knowledge provides valuable insights to determine how to make your race
car go faster.

You might also like