This article was downloaded by: [University of California, San Francisco]

On: 03 December 2014, At: 10:33

Publisher: Taylor & Francis
Informa Ltd Registered in England and Wales Registered Number: 1072954 Registered
office: Mortimer House, 37-41 Mortimer Street, London W1T 3JH, UK

Vehicle System Dynamics: International

Journal of Vehicle Mechanics and
Mobility
Publication details, including instructions for authors and
subscription information:
http://www.tandfonline.com/loi/nvsd20

An improved Magic Formula/Swift

tyre model that can handle inflation
pressure changes
a b c
I. J.M. Besselink , A. J.C. Schmeitz & H. B. Pacejka
a
Department of Mechanical Engineering , Eindhoven University of
Technology , P.O. Box 513, 5600 MB, Eindhoven, The Netherlands
b
TNO Automotive , Helmond, The Netherlands
c
TU Delft , Delft, The Netherlands
Published online: 26 Nov 2010.

To cite this article: I. J.M. Besselink , A. J.C. Schmeitz & H. B. Pacejka (2010) An improved
Magic Formula/Swift tyre model that can handle inflation pressure changes, Vehicle System
Dynamics: International Journal of Vehicle Mechanics and Mobility, 48:S1, 337-352, DOI:
10.1080/00423111003748088

To link to this article: http://dx.doi.org/10.1080/00423111003748088

PLEASE SCROLL DOWN FOR ARTICLE

Taylor & Francis makes every effort to ensure the accuracy of all the information (the
“Content”) contained in the publications on our platform. However, Taylor & Francis,
our agents, and our licensors make no representations or warranties whatsoever as to
the accuracy, completeness, or suitability for any purpose of the Content. Any opinions
and views expressed in this publication are the opinions and views of the authors,
and are not the views of or endorsed by Taylor & Francis. The accuracy of the Content
should not be relied upon and should be independently verified with primary sources
of information. Taylor and Francis shall not be liable for any losses, actions, claims,
proceedings, demands, costs, expenses, damages, and other liabilities whatsoever or
howsoever caused arising directly or indirectly in connection with, in relation to or arising
out of the use of the Content.

This article may be used for research, teaching, and private study purposes. Any
substantial or systematic reproduction, redistribution, reselling, loan, sub-licensing,
systematic supply, or distribution in any form to anyone is expressly forbidden. Terms &
 Conditions of access and use can be found at http://www.tandfonline.com/page/terms-
and-conditions
Downloaded by [University of California, San Francisco] at 10:33 03 December 2014
 Vehicle System Dynamics
Vol. 48, Supplement, 2010, 337–352

An improved Magic Formula/Swift tyre model that

can handle inflation pressure changes
Downloaded by [University of California, San Francisco] at 10:33 03 December 2014

I.J.M. Besselinka *, A.J.C. Schmeitzb and H.B. Pacejkac

a Department of Mechanical Engineering, Eindhoven University of Technology, P.O. Box 513, 5600 MB
Eindhoven, The Netherlands; b TNO Automotive, Helmond, The Netherlands; c TU Delft, Delft,
The Netherlands

(Received 19 October 2009; final version received 3 March 2010 )

This paper describes extensions to the widely used TNO MF-Tyre 5.2 Magic Formula tyre model.
The Magic Formula itself has been adapted to cope with large camber angles and inflation pressure
changes. In addition, the description of the rolling resistance has been improved. Modelling of the tyre
dynamics has been changed to allow a seamless and consistent switch from simple first-order relaxation
behaviour to rigid ring dynamics. Finally, the effect of inflation pressure on the loaded radius and the
tyre enveloping properties is discussed and some results are given to illustrate the capabilities of the
model.

Keywords: simulation; tyre dynamics models; tyre dynamics measurement; tyre dynamics; Magic
Formula

1. Introduction

Since its conception over 20 years ago, the Magic Formula has fairly quickly been adopted
as the industry standard tyre model for vehicle handling simulations. Over the years various
developments have been made to improve the accuracy and to extend the capabilities of
the model, for example, the method to describe combined slip has been improved and a
special Magic Formula has been developed to handle the large camber angles occurring on
motorcycles. In parallel, research has been done to increase the frequency range by introducing
rigid ring dynamics, contact patch transients and an obstacle enveloping model, also known
as the SWIFT or MF-Swift model. An overview and description of these developments can
be found in Pacejka [1].
The fact that from the start the tyre model equations were published in the open literature
has certainly contributed to the popularity of the Magic Formula. On the other hand, over the
years it has also resulted in a number of different, at times incompatible, implementations.
Different versions of the Magic Formula may be used, equations are sometimes partially
implemented, in-house extensions are added, different axis system or units are used, etc.

*Corresponding author. Email: [email protected]

ISSN 0042-3114 print/ISSN 1744-5159 online

© 2010 Taylor & Francis
DOI: 10.1080/00423111003748088
http://www.informaworld.com
 338 I.J.M. Besselink et al.

Notwithstanding these issues, the TNO MF-Tyre 5.2 tyre model has reached a mature status
and is widely used in the industry for vehicle handling studies. The model equations are
documented in TNO [2].
To move forward from MF-Tyre 5.2, several targets have been defined:
• To improve the description of camber, that is, to have an explicit formulation and control
over the camber stiffness and to extend the capabilities of the model for handling very large
camber angles. This will make the special ‘motorcycle’ Magic Formula superfluous and
improves processing of measurements according to the TIME procedure.
• To include the effect of inflation pressure changes in the Magic Formula. This will eliminate
Downloaded by [University of California, San Francisco] at 10:33 03 December 2014

the need to have separate parameter sets (tyre property files) for different tyre pressures. It
will allow evaluating tyre behaviour for pressures not in the measurement programme and
it also leads to a reduction in the total number of measurements required.
• To make the description of tyre dynamics consistent between MF-Swift and MF-Tyre. For
example, when including the dynamics of the tyre belt, the path-dependent (basic) tyre
relaxation behaviour should remain unchanged.
In addition, various other enhancements have been made, for example, in the description of
rolling resistance and overturning moment. This paper aims to introduce the MF-Tyre/MF-
Swift 6.1 model and accompanying equations. Due to space restrictions and the extent of the
changes that were made in comparison with the MF-Tyre 5.2 model, turn slip extensions [1]
will not be discussed.

2. Contact point, loaded radius and calculation of slip

Traditionally, the tyre force and moment characteristics are defined in the tyre road contact
point. The location of this point is defined by considering the tyre/wheel combination as an
infinitely thin disk through the plane of symmetry of the tyre, as is shown in Figure 1.
It is important to note that in practice the forces and moments are not measured at the tyre
road contact point C, but at the wheel centre. In order to process the measurements and to
calculate the forces and moments at the ground contact point, both the height of the wheel
centre and inclination angle γ are required. Consequently, the distance from wheel centre to
ground contact point, that is, the loaded radius Rl , should be represented accurately in the tyre
simulation model.

Figure 1. Force, moment and kinematic variables of the tyre road contact (ISO sign convention).
 Vehicle System Dynamics 339

First, centrifugal growth of the free tyre radius R is calculated using the following formula:
   
R0 2
R = R0 qre0 + qv1 , (1)
V0

where R0 equals the non-rolling free tyre radius, V0 a reference velocity,  the wheel rotational
velocity and qre0 and qv1 the model parameters. The tyre deflection ρ is the difference between
the free tyre radius R and the loaded tyre radius Rl :
ρ = max(R − Rl , 0). (2)
Downloaded by [University of California, San Francisco] at 10:33 03 December 2014

The vertical tyre force Fz is then calculated using the following formula:
       2 
R0 qFcx Fx 2 qFcy Fy 2 ρ ρ
Fz = 1 + qv2 || − − qF z1 + qF z2
V0 Fz0 Fz0 R0 R0
· (1 + pF z1 dpi )Fz0 . (3)
Various effects are included in this calculation: a stiffness increase with velocity (qv2 ), ver-
tical sinking due to longitudinal and lateral forces (qFcx , qFcy ), a quadratic force deflection
characteristic (qF z1 , qF z2 ) and the influence of the tyre inflation pressure (pF z1 ). Further, Fz0
is the nominal load and dpi the non-dimensional pressure increment, see appendix. For large
camber angles (e.g. motorcycle tyres), a modified approach for calculating the vertical force
is necessary, taking into account the contour of the tyre. Its discussion is outside the scope of
this paper.
The vertical stiffness cz0 at the nominal vertical load, nominal inflation pressure, no
tangential forces and zero forward velocity can be calculated as follows:

Fz0
cz0 = qF2 z1 + 4qF z2 . (4)
R0
In the expressions for the effective rolling radius and contact patch dimensions, the vertical
stiffness adapted for tyre inflation pressure is used:
cz = cz0 (1 + pF z1 dpi ). (5)
The forces (Fx , Fy ) and moments (Mx , My , Mz ) at the ground contact point are functions of
vertical force Fz , various slip properties, inclination angle γ , forward velocity Vx and inflation
pressure pi . These nonlinear relations are captured in the Magic Formula. The longitudinal
slip is defined as the ratio of the longitudinal slip velocity Vsx and forward velocity Vx :
Vsx Vx − Re fr −  
κ=− =− =− = − 1. (6)
Vx Vx fr fr
In this equation, fr is the angular velocity of the freely rolling tyre. When executing mea-
surements the latter part of this equation may be used to determine the amount of longitudinal
slip. In the tyre simulation model, the first part of the equation is used, requiring an explicit
equation for the effective rolling radius Re . The ratio of the forward velocity of the wheel
centre Vx to the angular velocity of the free rolling tyre fr equals the effective rolling radius
Re . The following empirical formula is used:
   
Fz0 Fz Fz
Re = R − Dreff arctan Breff + Freff , (7)
cz Fz0 Fz0
where Dreff , Breff and Freff are model parameters. When no measurements are available, the
suggested values are Dreff = 0.24, Breff = 8 and Freff = 0.01. The effective rolling radius
 340 I.J.M. Besselink et al.

defines the slip point S. Note that the location of S is different from the contact point C as is
shown in Figure 1. Point S is also used to calculate the sideslip angle α using the lateral slip
velocity Vsy :
 
Vsy
α = arctan . (8)
Vx
It is clear that in order to get an accurate representation of the tyre characteristics consis-
tent definitions for slip should be used. Though one can argue on the exact definition of slip
variables, it is clear that in any case consistent definitions should be used for both the mea-
Downloaded by [University of California, San Francisco] at 10:33 03 December 2014

surements and the simulation model. The definitions given in this section have been used since
1996 in the various MF-Tyre simulation models (including MF-Tyre 5.2). When modelling
the tyre-road enveloping and relaxation behaviour the dimensions of the tyre contact patch
are needed. The empirical expressions for half of the contact length a and half of the width b
read as follows:
     
Fz Fz ρ ρ
a = R0 qra2 + qra1 ≈ R0 qra2 + qra1 , (9)
cz R 0 cz R0 R0 R0
      1/3 
Fz Fz 1/3 ρ ρ
b = w qrb2 + qrb1 ≈ w qrb2 + qrb1 , (10)
cz R0 cz R0 R0 R0

where w is the nominal width of the tyre. Since these expressions are functions of the tyre
deflection ρ, the effect of changing the tyre inflation pressure is taken into account. Lowering
the inflation pressure results in an increase in tyre deflection and thus an increase in contact
length.

3. Steady-state force and moment characteristics

For an introduction to Magic Formula tyre modelling, we refer to Pacejka [1, pp. 172–184].
Initially, the Magic Formula has been designed to describe the force and moment characteristics
of passenger car tyres within a limited camber range (±15◦ ). Beyond this range extrapolation
errors may occur and as such the formula proved not to be suitable for describing tyre char-
acteristics at large inclination angles. This resulted in a different Magic Formula equation for
motorcycle tyres, as proposed by De Vries [3], where the contributions of camber and sideslip
are fully separated, as is shown in Equation (11):
 
Cy arctan((1 − Ey )By αy + Ey arctan(By αy ))
Fy = Dy sin . (11)
+ Cγ arctan((1 − Eγ )Bγ γ + Eγ arctan(Bγ γ ))

This approach was subsequently adopted in a slightly modified form for the MF–MC Tyre
model, aimed specifically at motorcycle tyres. This model is described in Pacejka [1, pp. 579–
583]. One of the benefits of this approach is that the camber stiffness is defined explicitly,
which also proves to be convenient when new approaches to tyre measurements, like the
TIME procedure [4], are used. A logical step would be to also use the MF-MC Tyre model
for passenger car tyres, but the accuracy proved to be less compared with the normal Magic
Formula. An attempt was made to adapt the MF–MC Tyre model to suit passenger car tyre
characteristics better, but this model, also known as MF-Time [5], still suffers from a reduced
accuracy with respect to the normal Magic Formula for passenger car tyres.
 Vehicle System Dynamics 341
Downloaded by [University of California, San Francisco] at 10:33 03 December 2014

Figure 2. Creating the side force characteristic at an inclination angle by applying vertical and horizontal shifts.

It turns out that only limited modifications to the MF-Tyre 5.2 equations are required to
overcome these issues. First, the vertical shift in the lateral force SVyγ due to camber (and
vertical force) remains the same:

SV yγ = Fz (pV y3 + pV y4 dfz )γ , (12)

where dfz is the dimensionless load increment and pV y3 and pV y4 are some model parameters,
see appendix. The camber stiffness Kyγ is specified in exactly the same way as in the MF–MC
Tyre model:
Kyγ = (pKy6 + pKy7 dfz )Fz . (13)
Now the required horizontal shift (or sideslip angle) due to camber can be calculated using
the cornering stiffness Kyα :
Kyγ γ − SV yγ
SHyγ = . (14)
Kyα
This approach is illustrated in Figure 2. The side force characteristic is shifted vertically by a
magnitude SV yγ to account for the increase in side force due to an inclination angle for large
values of side slip. At zero side slip angle, we demand that the side force due to an inclination
of the tyre equals Kyγ γ . Since the slope of Fy versus α curve is known, it is equal to the
cornering stiffness Kyα , the magnitude of the required horizontal shift can be easily calculated
using Equation (14). In this simplified explanation, we ignore initial offsets for zero inclination
angle and changes of the shape of Fy versus α curve due to an inclination angle.
It appears that this straightforward modification enables to use a single Magic Formula
for large camber angles, without making sacrifices with respect to accuracy for normal pas-
senger car tyres. The modifications to the expressions for the self-aligning moment Mz are
also fairly limited. The main changes being an extended expression for the peak value of the
residual moment Mzr [Dr , Equation (A64)] and the fact that the side force at zero camber
is used in the expression for the self-aligning moment [Equation (A49)], which is similar to
MF–MC Tyre. Figure 3 illustrates the capabilities of this modified Magic Formula for repre-
senting the measured characteristics of a motorcycle tyre. The new equations have also been
tested on a large database with force and moment measurements, consisting of 55 passenger car
tyres, 14 motorcycle tyres, 27 racing tyres and 10 truck tyres. The fitting errors, representing
the difference between measurements and model, are presented in Table 1.
The following conclusions can be draw from Table 1:

• For normal tyres the difference between MF-Tyre 5.2 and MF-Tyre 6.1 is negligible and
the accuracy is approximately the same.
 342 I.J.M. Besselink et al.

• The MF-Time model is less accurate in comparison with the MF-Tyre 5.2, in particular the
representation of the self-aligning moment appears to be worse.
• MF-Tyre 6.1 can represent motorcycle tyre behaviour quite accurately: the fitting error of
the side force characteristic is even reduced in comparison with the MF–MC Tyre, though
the self-aligning moment is somewhat less accurate.

So a single set of Magic Formula equations has been developed which can represent passenger
car, truck, racing and motorcycle tyres with a similar accuracy as obtained with different magic
formula tyre models used by the industry today. The MF-Tyre 6.1 equations have the advantage
Downloaded by [University of California, San Francisco] at 10:33 03 December 2014

of including an explicit formulation of the camber stiffness and they can cope with large camber
angles. Next, the expressions have been extended to include the effect of tyre inflation pressure
changes on the tyre force and moment characteristics.
It is clear that for many passenger cars on the road the same tyres are used on the front and
rear axle, but the tyre pressure is different and may need to be adjusted for different vehicle
loading conditions. The Magic Formula equations published to date do not account for tyre
pressure changes, leading to multiple parameter datasets for different tyre pressures and an
additional measurement effort. Furthermore, only the inflation pressures tested can be selected
and it is not possible to perform an interpolation.
As the Magic Formula is a semi-empirical tyre model, each individual tyre characteristic has
to be analysed for the impact of changes to the tyre inflation pressure [6,7]. Both measurements

30
2000 γ = −5°
20 Fz = 1475 N
γ = 0°
γ = 5°
1000 10
γ = 20°
γ = 30°
Mz [Nm]

0
Fy [N]

0 γ = 45°
−10

−1000 −20

−30
−2000
−40
−5 0 5 −5 0 5
α [deg.] α [deg.]
Figure 3. MF-Tyre 6.1 fit of a motorcycle tyre, left: side force, right: self-aligning moment; markers are
measurements (source: TNO Tyre Test Trailer) and continuous lines are Magic Formula results.

Table 1. Comparison of average fitting errors for various tyre models and different tyre types.

Car/truck/race tyres (n = 92) Motorcycle tyres (n = 14)

Force/slip case MF-Tyre 5.2 MF-Time MF-Tyre 6.1 MF−MC Tyre 1.1 MF-Tyre 6.1

Fx pure (%) 4.17 4.16 4.17 4.11 4.06

Fy pure (%) 2.37 2.73 2.26 4.55 4.28
Mz pure (%) 6.47 9.53 6.23 8.76 10.19
Fx combined (%) 5.70 5.69 5.70 4.46 4.43
Fy combined (%) 8.58 8.81 8.53 11.06 11.01
Mz combined (%) 32.74 34.57 32.49 40.59 40.29
 Vehicle System Dynamics 343

2000 300
Magic Formula
Cornering stiffness [N/deg] 200
1500
100

M [Nm]
1000 0

z
Measurements:
−100
pi = pi0 − 0.6 bar
500
Downloaded by [University of California, San Francisco] at 10:33 03 December 2014

p =p
i i0 −200
pi = pi0 + 0.6 bar
0 −300
0 5000 10000 −5 0 5
Fz [N] α [deg]

Figure 4. Tyre pressure effects for a passenger car tyre; left: cornering stiffness and right: self-aligning moment.

and a physical background model were used in this process. The main effects identified are
as follows:
• changes in longitudinal slip stiffness, cornering stiffness and camber stiffness;
• changes in peak friction coefficient, both longitudinal and lateral;
• a reduction of the pneumatic trail with increasing inflation pressure.
Details on the modified equations can be found in the appendix. As an example, the cornering
stiffness and self-aligning moment characteristics are shown in Figure 4. Since the effect
of tyre inflation pressure on, for example, the combined slip characteristics is small, it is
sufficient to measure this behaviour at a single inflation pressure and the total number of tests
can therefore be reduced. Details on the testing requirements can be found in the TNO tyre
model documentation [8].
As energy efficiency of road vehicles is becoming even more important, an accurate mod-
elling of the rolling resistance of the tyres should be addressed. In the SWIFT model the
increase in rolling resistance at high forward velocities has already been identified. Next, the
nonlinear dependency on the vertical force and inflation pressure has been added.
In Michelin [9], the following equation is given to adapt the rolling resistance for conditions
deviating from the ISO rolling resistance test:
 α  β
pi Fz
frr = frr,ISO , (15)
pi,ISO Fz,ISO
where frr is the rolling resistance coefficient, pi the tyre inflation pressure and Fz the vertical
tyre force. According to Michelin [9], the following coefficients are applicable: α = −0.4 and
β = 0.85 for a passenger car and α = −0.2 and β = 0.9 for a truck tyre. This equation has
been adopted in a slightly modified form and has been combined with the existing velocity
influence and reads (not including the camber effects) for the rolling resistance moment as
follows:
  4     
Fx Vx Vx Fz qsy7 pi qsy8
My = −R0 Fz0 λMy qsy1 + qsy2 + qsy3 + qsy4 .
Fz0 V0 V0 Fz0 pi0
(16)
Figure 5 shows that following this approach the rolling resistance can be modelled accurately
for a standard passenger car tyre.
 344 I.J.M. Besselink et al.

Rolling resistance
−10

−20 Fz = 1.8 kN, pi = 2.5 bar

F = 2.4 kN, p = 2.8 bar
z i
−30
Fx [N]

−40 Fz = 3.6 kN, pi = 2.0 bar

Downloaded by [University of California, San Francisco] at 10:33 03 December 2014

−50 F = 5.4 kN, p = 3.0 bar

z i

measurement
−60 Fz = 5.4 kN, pi = 2.0 bar
fit result

−70
0 20 40 60 80 100 120 140 160 180 200
Velocity [km/h]

Figure 5. Rolling resistance force as a function of vertical force, inflation pressure and forward velocity.

4. Tyre relaxation behaviour and belt dynamics

In the MF-Tyre 5.2 tyre model, an empirical relation is used to describe the relaxation length
dependency on vertical force. In the SWIFT model, the dynamics are modelled using a rigid
ring approach, residual stiffness and contact patch relaxation model. Ultimately, these two
different approaches could potentially result in simulating a different relaxation length for the
same tyre, which obviously is not acceptable. Depending on the application there will be a
need to be able to switch the model from a simple representation of the dynamics to a more
elaborate but more time consuming variant. Furthermore, the inflation pressure is considered
in the model, which also has an impact on the relaxation behaviour.
The solution for this combined set of requirements is to model the overall longitudinal
and lateral tyre stiffness and calculate the required parameters from these expressions. The
following expressions are used to describe the overall longitudinal cx and lateral stiffness cy
of the tyre at ground contact:



cx = cx0 1 + pcfx1 dfz + pcfx2 df2z 1 + pcfx3 dpi , (17)



cy = cy0 1 + pcfy1 dfz + pcfy2 df2z 1 + pcfy3 dpi , (18)

where cx0 and cy0 are the longitudinal and lateral stiffness of the tyre at the nominal vertical
force and inflation pressure. Using these stiffness and the longitudinal slip stiffness Kxκ and
cornering stiffness Kyα , the relaxation lengths for longitudinal and sideslip are, respectively,

Kxκ
σx = , (19)
cx
Kyα
σy = . (20)
cy

In principle, the stiffness cx and cy could be measured on a non-rolling tyre, but the preferred
approach is to measure the cornering stiffness and lateral relaxation length in a transient test for
a number of different inflation pressures and vertical loads. Subsequently, the lateral stiffness
equation can be fitted to these measurement points accordingly. Figure 6 gives an impression of
 Vehicle System Dynamics 345

Measured and calculated relaxation length σy to 1 deg. of side slip angle α

1

0.9 pi = 1.9 bar

p = 2.2 bar
i
p = 2.5 bar
0.8 i
p = 2.7 bar
i
pi = 3.0 bar
0.7

0.6
σy [m]

0.5
Downloaded by [University of California, San Francisco] at 10:33 03 December 2014

0.4

0.3 calculated
0.2 measured

0.1

0
0 1000 2000 3000 4000 5000 6000 7000 8000 9000
Fz [N]

Figure 6. Lateral relaxation length as a function of vertical force and inflation pressure.

the dependency of the lateral relaxation length on the vertical force and tyre inflation pressure
for a standard passenger car tyre.
Next to the steady-state representation, three approaches with increasing complexity are
possible to model the tyre transient behaviour and dynamics.

4.1. Linear transients

The relaxation length is determined using Equations (19) and (20) and is subsequently used
in the next two differential equations to calculate the (transient) slip quantities:

σx κ̇ = −Vx κ − Vsx , (21)

σy α̇ = −Vx α + Vsy . (22)

4.2. Nonlinear transients

In this approach, the tyre-road contact is separated into two parts and can be considered as a
series connection of a spring-damper with a relaxation system (Figure 7):

kcy ε̇y + ccy εy = Fy , (23)

σc α̇ = −Vx α + Vsy + ε̇y , (24)

where σc equals half of the contact length a and εy is the lateral carcass deflection. The carcass
stiffness ccy is calculated as follows:
Kyα
ccy = cy . (25)
Kyα − cy a
The same method can be used in the longitudinal direction. The benefit of this approach is the
resulting decreasing relaxation behaviour with increasing sideslip angle. Also, the relaxation
behaviour of the tyre on changes in vertical force is more accurately captured in this approach.
 346 I.J.M. Besselink et al.
Downloaded by [University of California, San Francisco] at 10:33 03 December 2014

Figure 7. Schematic overview of different ways to model the contact transients/dynamics.

4.3. Rigid ring dynamics

When including rigid ring dynamics, the spring-damper system is now subdivided into various
components as shown in Figure 7. In this case, the eigenfrequencies of the tyre belt determine
the stiffness between rim and belt; still the overall stiffness is specified by Equations (17) and
(18). This implies, for example, that the residual stiffness cry is calculated from
1 1 R2 1 a
= + l + + , (26)
cy cby cbγ cry Kyα
    
carcass contact patch

in which cby and cbγ are the lateral stiffness and rotational stiffness about the longitudinal
axis between belt and rim. As is to be expected, the eigenfrequencies of the tyre belt will be
higher when the tyre inflation pressure is increased. Empirical relations have been developed
to account for this effect using a physical background model [7]. Some other aspects have to
be taken into account, but are not discussed here due to space limitations:
• correction of the tyre sideslip angle for the twist of the contact patch;
• the transient dynamics of the self-aligning moment;
• combining the radial belt stiffness with the loaded radius equation.

5. Enveloping behaviour

To describe the tyre behaviour accurately on road unevenness with short wavelengths an
enveloping model using elliptical cams has been developed by Schmeitz [10]. Research has
shown that the shape of the elliptical cams does not change with tyre inflation pressure and
that the tyre stiffness and contact length change cause the main effect [6]. Figure 8 shows
the results of a low speed enveloping test with fixed axle height and initial vertical force of
4000 N. As the tyre pressure is increased, the stiffness increases and the contact length becomes
smaller, resulting in larger forces and a shorter response. This is represented quite accurately
by the model. The results for a high speed cleat test are shown in Figure 9. This figure clearly
shows different responses for different inflation pressure: with increasing inflation pressure
the frequency of the vertical mode increases, the peak loads increase and the excitation level
of the vertical mode reduces.
 Vehicle System Dynamics 347

pi = 1.8 bar pi = 2.2 bar pi = 2.6 bar

Measurements Simulations
6000

F 5000
z
[N] 4000

3000
500
250
ΔFx
0
Downloaded by [University of California, San Francisco] at 10:33 03 December 2014

[N]
−250
−500
−0.1 −0.05 0 0.05 0.1 −0.1 −0.05 0 0.05 0.1
x [m] x [m]

Figure 8. Low speed enveloping with fixed axle height (10 × 50 mm cleat).

Measurements Obstacle: 10x20mm cleat

Simulations F = 4.8 kN; V = 30 km/h
z0
100
2000 pi = 2.5 bar
ΔF √PSD
z 50
[N] 0 [N/√Hz]

−2000 0
100
2000 pi = 3.0 bar
ΔFz √PSD
50
[N] 0 [N/√Hz]

−2000 0
0.1 0.15 0.2 0.25 0 50 100 150
tim e [s] f r e q u e n c y [H z ]
Figure 9. Cleat test of a passenger car tyre at a higher forward velocity. The dashed lines indicate the maximum
vertical force and resonance frequency at 2.5 bar inflation pressure.

It should be noted that in principle for the parameterisation no additional cleat tests are
required for the different inflation pressures. The shape of the cams remains constant and the
eigenfrequencies of the tyre belt are modified using an empirical formula. Only the stiffness
and the contact length depend on the tyre inflation pressure.

6. Concluding remarks

This paper shows the steps taken to enhance the Magic Formula model and to integrate various
developments into a single tyre simulation model, suitable for different tasks: vehicle handling,
ride and calculation of suspension loads. Though certainly open issues exist and research will
continue, it can be seen as another step forward to an accurate digital representation of the
physical tyre. The tyre model described in this paper has already been implemented in various
multi-body software packages and can also be obtained directly from TNO Automotive. In
addition to the MF-Tyre/MF-Swift 6.1 tyre model, the software package MF-Tool 6.1 is
available for parameter identification, for more information we refer to the TNO Delft-Tyre
website [11].
 348 I.J.M. Besselink et al.

References

[1] H.B. Pacejka, Tyre and Vehicle Dynamics, 2nd ed., Butterworth-Heinemann, Oxford, UK, 2006, ISBN-13:
980-0-7506-6918-4.
[2] TNO, MF-Tyre User Manual Version 5.2, 2001
[3] E.J.H. de Vries, Motorcycle Tyre Measurements and Models, Proceedings of the 15th symposium Dynamics of
Vehicles on Road and Tracks, IAVSD, Budapest, August 1997.
[4] J.J.M. van Oosten, C. Savi, M. Augustin, O. Bouhet, J. Sommer, and J.P. Colinot, TiMe, TIre MEasurements
forces and moments: a new standard for steady state cornering tire testing, EAEC Conference, Barcelona, 30
June to 2 July, 1999.
[5] J.J.M. van Oosten, E. Kuiper, G. Leister, D. Bode, H. Schindler, J. Tischleder, and S. Kohne, A new tyre model
for TIME measurement data, Tire Technology Expo 2003, Hannover, Germany, 2003.
Downloaded by [University of California, San Francisco] at 10:33 03 December 2014

[6] A.J.C. Schmeitz, I.J.M. Besselink, J. de Hoogh, and H. Nijmeijer, Extending the magic formula and SWIFT tyre
models for inflation pressure changes, Reifen, Fahrwerk, Fahrbahn – VDI Conference, Hannover, Germany,
2005, pp. 201–225.
[7] I.B.A. op het Veld, Enhancing the MF-Swift tyre model for inflation pressure changes, DCT Rep. 2007.144,
Eindhoven University of Technology, 2007.
[8] TNO, Measurement requirements and TYDEX file generation for MF-Tyre/MF-Swift 6.1. Available at www.delft-
tyre.nl
[9] Michelin, The Tyre – Rolling Resistance and Fuel Savings, Société de Technologie Michelin-Ferrand, France,
2003, p. 84.
[10] A.J.C. Schmeitz, A semi-empirical three-dimensional model of the pneumatic tyre rolling over arbitrarily uneven
road surfaces, Dissertation, Delft University of Technology, The Netherlands, 2004.
[11] TNO Delft-Tyre website. Available at www.delft-tyre.nl

Appendix. TNO MF-TYRE 6.1 Magic Formula equations

The Magic Formula can be considered as a nonlinear function with multiple inputs and outputs, as is shown in
Figure A1. The model parameters, typically starting with the character p, q, r or s, are determined in a numerical
optimisation process minimising the difference between the output of the Magic Formula and the measured forces
and moments. In this process, the scaling coefficients, starting with the character λ, will remain equal to one. Please
note that the difference between αF and αM disappears when the additional transient behaviour of the self-aligning
moment is not taken into account, so α = αF = αM .
When the inputs are far outside the measurement range (e.g. extremely high vertical loads or very large inclination
angles), the extrapolation capabilities of the model can possibly fail. To prevent this from happening, the inputs to
the Magic Formula are bounded and the Magic Formula is not evaluated outside this range, so

κmin < κ < κmax , (A1)

αmin < αF < αmax and αmin < αM < αmax , (A2)
γmin < γ < γmax , (A3)
pi,min < pi < pi,max . (A4)

Figure A1. Inputs and outputs of the Magic Formula.

 Vehicle System Dynamics 349

For the vertical force a range is also defined. When the vertical force Fz is outside this range, the Magic Formula is
evaluated for the corresponding boundary (Fz,min or Fz,max ) when the vertical forces is below Fz,min the resulting
forces and moments are scaled with the actual value of the vertical force. A simple example, if the vertical force equals
0.5 times Fz,min , then the Magic Formula is evaluated for Fz,min and the resulting forces/moments are multiplied with
a factor 0.5. No scaling is applied when the vertical force exceeds forces.
To make the Magic Formula equations dimensionless, the following parameters are introduced:
• unscaled free tyre radius of the non-rolling tyre R0 ;
• nominal vertical force Fz0 ;
• reference forward velocity V0 ;
• nominal tyre inflation pressure pi0 .
To account for changes in the vertical force and tyre inflation pressure, dimensionless increments are introduced:
Downloaded by [University of California, San Francisco] at 10:33 03 December 2014

Fz − Fz0
dfz = , (A5)
Fz
pi − pi0
dpi = . (A6)
pi0
Many parameters of the MF-Tyre 6.1 tyre model are unchanged and have the same name as in the MF-Tyre 5.2 model.
The additional parameter FITTYP has been introduced by TNO to uniquely identify equations. When FITTYP equals
61, we are dealing with a MF-Tyre 6.1 dataset.

Al Longitudinal force Fx

Fx = (Dx sin[Cx arctan{Bx κx − Ex (Bx κx − arctan(Bx κx ))}] + SV x )Gxα . (A7)

A1.1 Pure slip

κx = κ + SH x , (A8)
Cx = pCx1 λCx , (A9)
Dx = μx Fz , (A10)



μx = (pDx1 + pDx2 dfz ) 1 − pDx3 γ 2 1 + ppx3 dpi + ppx4 dp2i λμx , (A11)


Ex = pEx1 + pEx2 dfz + pEx3 df2z (1 − pEx4 sgn (κx )) λEx , (A12)


Kxκ = (pKx1 + pKx2 dfz ) exp (pKx3 dfz ) 1 + ppx1 dpi + ppx2 dp2i Fz λKxκ , (A13)
Kxκ
Bx = , (A14)
Cx Dx
SH x = (pH x1 + pH x2 dfz )λH x , (A15)
SV x = (pV x1 + pV x2 dfz )Fz λV x λμx . (A16)

A1.2 Combined slip

cos[Cxα arctan{Bxα αs − Exα (Bxα αs − arctan(Bxα αs ))}]

Gxα = , (A17)
cos[Cxα arctan{Bxα SH xα − Exα (Bxα SH xα − arctan(Bxα SH xα ))}]
αs = αF + SH xα , (A18)
Bxα = (rBx1 + rBx3 γ ) cos{arctan[rBx2 κ]}λxa ,
2
(A19)
Cxα = rCx1 , (A20)
Exα = rEx1 + rEx2 dfz , (A21)
SH xα = rH x1 . (A22)

When combined slip is not used, Gxα = 1.

 350 I.J.M. Besselink et al.

A2 Lateral force Fy (input αF )

Fy = Gyκ Fyp + SV yκ . (A23)

A2.1 Pure slip

 


Fyp = Dy sin Cy arctan By αy − Ey By αy − arctan By αy + SV y , (A24)
αy = αF + SHy , (A25)
Cy = pCy1 λCy , (A26)
D y = μ y Fz ,
Downloaded by [University of California, San Francisco] at 10:33 03 December 2014

(A27)




μy = pDy1 + pDy2 dfz 1 − pDy3 γ 2
1 + ppy3 dpi + ppy4 dpi λμy ,
2
(A28)





Ey = pEy1 + pEy2 dfz 1 + pEy5 γ − pEy3 + pEy4 γ sgn αy λEy ,
2
(A29)
  

 Fz

Kyα = pKy1 Fz0 1 + ppy1 dpi sin pKy4 arctan

 1 − pKy3 |γ | λKyα ,
pKy2 + pKy5 γ 2 1 + ppy2 dpi Fz0
(A30)



Kyγ = pKy6 + pKy7 dfz 1 + ppy5 dpi Fz λKyγ , (A31)
Kyα
By = , (A32)
Cy Dy
SHy = SHy0 + SHyγ , (A33)


SHy0 = pHy1 + pHy2 dfz λHy , (A34)
Kyγ γ − SV yγ
SHyγ = , (A35)
Kyα
SV y = SV y0 + SV yγ , (A36)


SV y0 = Fz pV y1 + pV y2 dfz λV y λμy , (A37)


SV yγ = Fz pV y3 + pV y4 dfz γ λKyγ λμy . (A38)

A2.2 Combined slip


SV yκ = DV yκ sin rV y5 arctan rV y6 κ λV yκ , (A39)




DV yκ = μy Fz rV y1 + rV y2 dfz + rV y3 γ cos arctan rV y4 αF , (A40)
 


cos Cyκ arctan Byκ κs − Eyκ Byκ κs − arctan Byκ κs
Gyκ =  

 , (A41)
cos Cyκ arctan Byκ SHyκ − Eyκ Byκ SHyκ − arctan Byκ SHyκ
κs = κ + SHyκ , (A42)
Byκ = (rBy1 + rBy4 γ ) cos{arctan[rBy2 (αf − rBy3 )]}λyκ ,
2
(A43)
Cyκ = rCy1 , (A44)
Eyκ = rEy1 + rEy2 dfz , (A45)
SHyκ = rHy1 + rHy2 dfz . (A46)

A3 Overturning moment Mx


 Fy
Mx = R0 Fz λMx qsx1 λV Mx − qsx2 γ 1 + ppmx1 dpi + qsx3
Fz0
      
Fz 2 Fy
+ qsx4 cos qsx5 arctan qsx6 · sin qsx7 γ + qsx8 arctan qsx9
Fz0 Fz0
  
Fz
+ qsx10 arctan qsx11 γ . (A47)
Fz0
 Vehicle System Dynamics 351

A4 Rolling resistance moment My

  4 
Fx Vx Vx Fz 2
My = −R0 Fz0 λMy qsy1 + qsy2
+ qsy3 + qsy4 + qsy5 γ + qsy6
2
γ
Fz0 V0 V0 Fz0
   
Fz qsy7 p qsy8
· (A48)
Fz0 p0

When combined slip is not used, SV yκ = 0 and Gyκ = 1.

Downloaded by [University of California, San Francisco] at 10:33 03 December 2014

A5 Self-aligning moment Mz (input αM )

Mz = −tFyp0 Gyκ0 + Mzr + sFx , (A49)

where Fyp0 Gyκ0 is the combined slip side force with zero inclination angle (γ = 0).

αt = αM + SH t , (A50)
SH t = qH z1 + qH z2 dfz + (qH z3 + qH z4 dfz ) γ , (A51)
SV y
αr = αM + SHy + , (A52)
Kyα

A5.1 Pure slip

αt,eq = αt , αr,eq = αr , s = 0. (A53)

A5.2 Combined slip

  2
Kxκ
αt,eq = arctan tan2 (αt ) + κ 2 · sgn(αt ), (A54)
Kyα
  2
Kxκ
αr,eq = arctan tan2 (αr ) + κ 2 · sgn(αr ), (A55)
Kyα
   
Fy
s = ssz1 + ssz2 + (ssz3 + ssz4 df z ) γ R0 λs , (A56)
Fz0

A5.3 Pneumatic trail t

 


t = Dt cos Ct arctan Bt αt,eq − Et Bt αt,eq − arctan Bt αt,eq cos (αM ) , (A57)

 λKyα
Bt = qBz1 + qBz2 df z + qbz3 df 2z (1 + qBz4 γ + qBz5 |γ |) , (A58)
λμy
Ct = qCz1 , (A59)


 R0
Dt = (qDz1 + qDz2 df z ) 1 − ppz1 dpi 1 + qDz3 γ + qDz4 γ 2 Fz λt , (A60)
Fz0
   

 2
Et = qEz1 + qEz2 df z + qEz3 df 2z 1 + (qEz4 + qEz5 γ ) arctan (Bt Ct αt ) . (A61)
π
 352 I.J.M. Besselink et al.

A5.4 Residual moment Mzr

Mzr = Dr cos[arctan(Br αr,eq )] cos(αM ), (A62)

λKyα
Br = qBz9 + qBz10 By Cy , (A63)
λμy
 
(qDz6 + qDz7 df z )λr + (qDz8 + qDz9 df z )(1 + ppz2 dpi )γ λKzγ
Dr = Fz R0 λμy . (A64)
+(qDz10 + qDz11 df z )γ |γ |λKzγ
Downloaded by [University of California, San Francisco] at 10:33 03 December 2014