This article was downloaded by: [University of California Davis]

On: 15 August 2014, At: 00:21

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

Measurement of contact forces

and patch features by means of
accelerometers fixed inside the tire to
improve future car active control
a b a a
F. Braghin , M. Brusarosco , F. Cheli , A. Cigada , S. Manzoni
a b
& F. Mancosu
a
Mechanical Engineering Department , Politecnico di Milano ,
Italy
b
Pirellli Tire System , Italy
Published online: 04 Apr 2007.

To cite this article: F. Braghin , M. Brusarosco , F. Cheli , A. Cigada , S. Manzoni & F. Mancosu
(2006) Measurement of contact forces and patch features by means of accelerometers fixed inside
the tire to improve future car active control, Vehicle System Dynamics: International Journal of
Vehicle Mechanics and Mobility, 44:sup1, 3-13, DOI: 10.1080/00423110600867101

To link to this article: http://dx.doi.org/10.1080/00423110600867101

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 Davis] at 00:21 15 August 2014
 Vehicle System Dynamics
Vol. 44, Supplement, 2006, 3–13

Measurement of contact forces and patch features

by means of accelerometers fixed inside the tire to improve
future car active control
Downloaded by [University of California Davis] at 00:21 15 August 2014

F. BRAGHIN†, M. BRUSAROSCO‡, F. CHELI*†, A. CIGADA†,

S. MANZONI† and F. MANCOSU‡
†Mechanical Engineering Department, Politecnico di Milano, Italy
‡Pirellli Tire System, Italy

Huge improvements and advances in sensor technology, together with the increasing demand for safety
and handling performances, pull research towards new control strategies. Robustness and promptness
of sensors used for active control are key requisites.
The main idea behind the research presented in this paper is to instrument the tire with appropriate
sensors in order to estimate several contact parameters ranging from the kinematic conditions of
the tire (the longitudinal slippage and the side slip angle [Fukada, Y., 1999, Slip-angle estima-
tion for vehicle stability control. Vehicle System Dynamics, 32(4–5), 375–388.]) to its dynamic
properties (the contact area shape and dimensions as well as the longitudinal, lateral and verti-
cal loads [Cole, D.J. and Cebon, D., 1989, A capacitative strip sensor for measuring dynamic
type forces. Proc. of the Second International Conference on Road Traffic Monitoring, London,
38–42; Cole, D.J. and Cebon, D., 1992, Performance and application of a capacitative strip tyre
force sensor. Proc of IEE Conference on Road Traffic Monitoring, London, 123–132.]) and to the
adhesion characteristics of the road (the surface roughness [Pasterkamp, W.R. and Pacejka, H.B.,
1997, The tire as a sensor to estimate friction. Vehicle System Dynamics, 27, 409–422; Pohl, A.,
Steindl, R. and Reindl, L., 1999, The ‘intelligent tire’ utilizing passive SAW sensorsmeasurement
of tire friction. IEEE Transactions on Instrumentation and Measurement, 48(6), 1041–1046; Ray,
L.R., 1997, Nonlinear tire force estimation and road friction identification: Simulation and experi-
ments. Automatics, 33(10), 1819–1833.]). Thus, the tire becomes a sensor of tire–road interaction.
Clearly, no other measuring device may ever be more robust nor prompt, and anyway closer to the
contact area.
After a preliminary research in which the pros and cons of the various sensor technologies were
taken into account, it was decided to put accelerometers inside the tire. These accelerometers were
fixed to the liner and on-board analysis of the acquired data was performed to reduce the amount of
data then sent via radio control to a central data processing unit. This unit, through post-elaboration
and recombination of the reduced signals coming from the various accelerometers in the tire, estimates
some of the above-described contact parameters. In the near future, these contact parameters coming
from all four tires and eventually other signals from sensors placed on the vehicle may be used to
develop innovative control strategies in order to increase vehicle performances as well as running
safety.

Keywords: Cyber wheel; Tire–road contact; Tire-road forces; Road roughness; Accelerometers

*Corresponding author. Email: federico.cheli@polimi.it

Vehicle System Dynamics

ISSN 0042-3114 print/ISSN 1744-5159 online © 2006 Taylor & Francis
http://www.tandf.co.uk/journals
DOI: 10.1080/00423110600867101
 4 F. Braghin et al.

1. Introduction

At present, tires are passive elements. This means that, although carrying out the functions they
were designed for, they neither measure contact or working conditions nor do they perform any
controlled actions on the vehicle’s behaviour. However, being in a very privileged position (at
vehicle–road interface), it is easy to imagine the possible improvements in vehicle handling and
comfort characteristics that could be achieved by letting these vehicle components become
active: as a sensor, they could be used to measure actual contact forces as well as rapid
changes of working conditions thus providing existing electronic control units (ECUs) with
very important information; as an actuator, instead, they could be used to change vehicle–road
interaction properties according to some control strategies thus improving vehicle safety and
Downloaded by [University of California Davis] at 00:21 15 August 2014

performances.
A tire being able to measure contact forces and contacting conditions would allow existing
ECUs to be less complicated and expensive as measured data would be directly available (see
figure 1) and should not be estimated with complicated and uncertainty-adding procedures.
Moreover, even for quantities that still cannot be measured, such as body slip angle, their
estimation would be more precise and prompt. Finally, having information about contact
forces would allow the development of newer and faster control strategies thus leading to
safer vehicles with higher performances.
A tire being able to adjust its characteristics as a function of the working conditions and
of the vehicle’s performances could be able to optimize safety. For example, a tire being
able to actively change its inflating pressure could adjust slip stiffness as well as contact area
dimensions as a function of the contact conditions (high or low adhesion) and the tire wear.

Figure 1. General layout of a vehicle dynamic control; the improvements that can be achieved with tires as sensors
are pointed out.
 Measurement of contact forces and patch features 5

Moreover, in the case of tire failure, an active tire could send a warning message to the driver,
thus allowing to slow down in full safety.
Pirelli Pneumatici (Tire System Department), together with Politecnico di Milano (Mechan-
ical Engineering Department), is working in both directions – making the tire a sensor and an
actuator. This paper will focus on the tire as a sensor.

2. Measurements on the tire

A preliminary research has been devoted to the identification of which sensors to use in the
tire and what strategy to follow to get as much information as possible from the tire. The main
Downloaded by [University of California Davis] at 00:21 15 August 2014

idea was to get all the possible information from one sensor only, even if the same information
could be achieved from other external transducers. This makes the system easier to manage,
although lacking in redundancy. Before choosing the sensor to be mounted inside the tire, it
is necessary to specify the requirements for the measuring system as follows:

– it should need no external power supply;

– it should be able to transmit the measured data towards on-vehicle ECUs without the need
of cables (wireless transmission);
– it should not modify the tire performances such as its handling characteristics, comfort char-
acteristics, acoustic emission, braking performances, fatigue resistance, wearing properties,
etc.;
– it should be easily fitted into the tire;
– its life should be at least equal to that of the tire;
– it should be as simple as possible in order to have high reliability;
– it should be modular, i.e. easily adjustable to fit any kind of tire.

Because of their tiny dimensions and low weight, it was decided to use three-axial micro
electro mechanical systems (MEMS) accelerometers (figure 2). Another advantage of such
sensors is their low cost, wide passband, high reliability and robustness, necessary to withstand
the impulses occurring when the accelerometers enter and exit the footprint.
In order to identify the best locations (on the liner† ) to place the accelerometers, two aspects
have to be considered: the sensors’ outputs (i.e. the signal-to-noise ratio) have to be maximized
and the highest possible acceleration in any manoeuvre as a function of the rolling speed has
to be determined. Thus, an FE analysis was carried out. The simulation results allowed us to

Figure 2. The three-axial MEMS accelerometer.

† The inner liner is a good location to place sensors as neither dirt nor bumps on sidewalks nor other obstacles damage
the accelerometers. Another good location to place sensors is inside the tire tread elements. This location would have
reduced the rubber filtering effect (more prompt sensors) but would have posed many production problems.
 6 F. Braghin et al.
Downloaded by [University of California Davis] at 00:21 15 August 2014

Figure 3. Accelerometer positions inside the tire.

Table 1. Accelerometer characteristics.

Range ±2000 g

Sensitivity 0.2 mv/g

Frequency response 0–5 kHz
Resonance frequency 26 kHz
Mass 0.4 g
Dimensions 10 × 6.86 × 3.94 mm

select the proper positions (figure 3) and characteristics (table 1) of the three-axial MEMS
accelerometers. Three such accelerometers were fixed to the liner corresponding to the tire-
rolling surface, thus providing information about three different circumferential paths, at the
centre and at the two sides of the tire (inner and outer side, figure 3). It should be observed
that the connection of sensors to the liner is of crucial importance: it should not modify the
local stiffness characteristics of the tire, it should never break even in the presence of very
high static (centrifugal) and dynamic (contact impulse) stresses and it should not behave as a
filter of the accelerations coming from the tire–road contact.
At this preliminary stage, power supply or wireless transmission problems have been
neglected by connecting the sensors and the acquisition board through traditional cables using
a pneumatic valve on the rim and a slip ring. Figure 3 shows the final assembly that was
extensively tested both on MTS Flat Trac III (figure 4) and on a vehicle at Pirelli test track
in Vizzola (figure 5). The main focus has been put on the metrological performances of the
accelerometers checking the signal reliability and the data analysis procedures to reconstruct
tire kinematic and dynamic properties from acceleration signals.
The first check on data reliability has been made possible by the use of MTS Flat Trac III.
Figure 6 shows the comparison between numerical FEA results (dashed line) and experimental
acquisitions (solid line) during indoor tests performed at 20 km/h for radial as well as lateral
accelerations over the tire circumferential development (the zero circumferential coordinate
corresponds to the position of the accelerometer at the centre of the contact area). It can be
clearly seen that the FE model is able to correctly predict the time histories of the accelerations
inside the tire except for the higher frequency contributions.
Some further details concerning data acquisition are given. Asynchronous as well as syn-
chronous sampling was performed. Due to the higher complexity of synchronous sampling
 Measurement of contact forces and patch features 7
Downloaded by [University of California Davis] at 00:21 15 August 2014

Figure 4. MTS Flat Trac III machine.

Figure 5. The instrumented tire mounted on the test vehicle.

and to the fact that almost no difference between the two sampling methods was appreciable,
asynchronous sampling was finally adopted. The adopted sampling frequency is the best com-
promise between the capabilities of the data acquisition (DAQ) board and the need of fully
describing very quick transients (i.e. the interval just-before-contact to just-after-contact).
Thus, a sampling frequency of 10 kHz for each channel was used. This frequency corresponds
to a sample approximately every 3 mm along the tire revolution at the speed of 100 km/h.
Another important issue in data acquisition is resolution: both the low amplitude and low
frequency acceleration component, due to centrifugal forces (a function of the rotational
speed), and the high amplitude and high frequency acceleration component, due to the entering
and exiting the contact area have to be correctly measured. FEA simulations showed that these
acceleration peaks could reach 1500 g. With the adopted sensitivity of 0.2 mV/g (with ×10 of
amplification), the expected peak signal is around 3 V. Thus, a 5 V full-scale DAQ converter is
needed. With a 16-bit converter, the related least significant bit (LSB) value is around 1 m/s2 .
It is essential to compare this LSB value with the floor noise due to the connection between
the sensors and the data acquisition unit. Since the floor noise is approximately 10 m/s2 , a
satisfactory signal-to-noise ratio is obtained when medium to high vehicle speeds are reached.
 8 F. Braghin et al.
Downloaded by [University of California Davis] at 00:21 15 August 2014

Figure 6. Comparison between numerical FEA results (dashed line) and experimental acquisitions (solid line) for
(a) radial and (b) lateral accelerations over the tire circumferential development.

3. Data analysis

As already declared, the research work is aimed at extracting a series of kinematic and dynamic
tire parameters from the acceleration signals inside the tire. To assess the capabilities as well
as the reliability and robustness of the experimental set-up, a long lasting test campaign has
been carried out both indoor and outdoor. In particular, the advantage of indoor tests, carried
out on MTS Flat Trac III (figure 4), is that both imposed working conditions and hub forces
are measured with great accuracy thus allowing us to objectively evaluate the accuracy of the
developed methodology by comparing the measured quantities with the estimated ones.
A first approach was to consider the tire as a black box and to determine a relation between
input quantities (measured by instruments other than the accelerometers) and acceleration
signals. The drawback of this approach is that a set of tests as-complete-as-possible has to be
performed. Although lots of tests were performed, no good correlation between inputs and
outputs was achievable especially due to the fact that rolling speed seems to greatly influence
the acceleration signals.
A second approach was thus attempted on the basis of the extraction of synthetic parameters
from the acceleration signals. The deformation of the rolling tire, and consequently of the
contact area that is ‘measured’ by the acceleration signals, is influenced by the following
factors and working conditions of the tire:
• type and structure of the tire,
• rolling speed,
 Measurement of contact forces and patch features 9

• inflation pressure,
• normal load,
• camber angle,
• slip angle,
• slippage and
• wear of the tread elements.

In this paper, it is assumed that the tire does not change during the tests† . Thus, the influence
on the contact area dimensions of the type and structure (carcass and belts) of the tire are
neglected. Also, the influence of tire wear on the measured accelerations has been neglected
at the present stage of the research but it is clear that further investigations are necessary. The
influence of all other parameters, instead, has been taken into account.
Downloaded by [University of California Davis] at 00:21 15 August 2014

At increasing rolling speed and equal inflation pressure, normal load, etc., the centrifugal
forces increase. This is equivalent to a stiffer tire. Thus, the contact area dimensions decrease.
However, the low amplitude and low frequency radial acceleration component increases. The
inflation pressure, instead, does not change the low amplitude and low frequency radial accel-
eration component but modifies the contact area dimensions: a higher inflation pressure means
a smaller contact area. Similarly, the normal load does not change the low amplitude and low
frequency radial acceleration component but modifies the contact area dimensions. Moreover,
at increasing normal load also the opposite transversal accelerations, measured by the inner
and outer accelerometers when contact occurs, increase. Camber angle and slip angle are
easily determined by looking at the tire contact area shape: in the absence of any camber or
slip angle, the contact area is symmetric with respect to the central section of the tire; in the
presence of these angles, instead, a pear contact shape is obtained. The difficulty is to separate
the contribution due to the camber angle and that due to the slip angle. Finally, slippage may be
determined by looking at the longitudinal acceleration measured when the sensors are inside
the contact area.
From the measured acceleration time histories, a number of synthetic parameters have
been extracted. These parameters depend only on one (or few) tire working condition(s) thus
allowing the go back from the acceleration signals to the tire kinematic and dynamic properties.
The synthetic parameters can be subdivided into two different families: those that are referred
to an amplitude and those that are related to a time lag (figure 7).

Figure 7. Extraction of synthetic parameters related to the amplitude and time lag of acceleration signals.

† If the tire is changed, a new calibration of the synthetic parameters has to be performed.
 10 F. Braghin et al.

The most significant parameters of the first family are:

• maximum radial acceleration difference (difference between the minimum value at the
entrance and the maximum value at the exit of the contact patch);
• maximum tangential acceleration difference (difference between the minimum value at the
entrance and the maximum value at the exit of the contact patch);
• mean radial acceleration when the sensor is out of contact;
• RMS of radial acceleration when the sensor is out of contact.

The most significant parameter of the second family is the time interval between the mini-
mum and maximum tangential acceleration peaks (this interval is strictly correlated with the
Downloaded by [University of California Davis] at 00:21 15 August 2014

contact area length).

In the following paragraphs, it will be shown how some synthetic parameters can be used
(separately or in combination) to determine the tire working conditions.

3.1 Identification of road roughness

To determine a correlation between acceleration signals (i.e. synthetic parameters) and road
roughness, only outdoor tests are of use since on MTS Flat Trac III just the micro-texture of
the abrasive paper on which the tyre rolls can be changed. Figure 8(a) shows two different
asphalts, characterized by completely different micro- and macro-texture, present in Vizzola
test track. Figure 8(b), instead, shows the corresponding measured (using a profilometer)
roughness profile.
The filtered RMS value of the radial acceleration sensors when the sensor is out of contact
is able to discriminate between the two different kinds of asphalts as shown in figure 9: the
higher curve is related to the rougher road while the lower curve, as expected, is referred to the
smoother road. The separation becomes more and more evident as the rolling speed increases.
Thus, the determined synthetic parameter is a function of the road roughness and of the tire
rolling speed.
It has been verified that the shown curves do only marginally depend on the tire size and
type. However, these data will not be displayed as it is out of the scope of the present paper.

Figure 8. Rough (upper) and smooth (lower) asphalts in Vizzola test track: (a) photo and (b) measured roughness
profile.
 Measurement of contact forces and patch features 11
Downloaded by [University of California Davis] at 00:21 15 August 2014

Figure 9. Filtered RMS value of the radial acceleration sensors when the sensor is out of contact for rough (red),
intermediate (green) and smooth (blue) asphalts.

Figure 10. FL parameter vs. slip angle: (a) inner sensor and (b) outer sensor as a function of tire load and camber.

3.2 Identification of normal load

As mentioned earlier, the contact area dimension is greatly influenced by the normal load. The
problem is that the length of the contact patch is also influenced by the inflation pressure, the
rolling speed, the camber angle and the slip angle. The problem is therefore to determine a
synthetic parameter that is sensitive only from the normal load. Figure 10 shows the deter-
mined synthetic parameter (called FL parameter, footprint length) for the inner and central
accelerometers as a function of the slip and camber angles during a test performed on MTS
Flat Trac III at 40 km/h. Similar tests have been performed at different speeds and inflation
pressures. It can be clearly seen that the FL parameter for the central accelerometer does not
depend on the slip and camber angles but is only influenced by the normal load. The same
does not apply to the external sensor.
 12 F. Braghin et al.

4. Validation of the data analysis methodology

The first validation of the data analysis methodology was carried out on MTS Flat Trac III
since, in these tests, both input (imposed tire working conditions) and output (acceleration)
quantities are known. As an example, figure 11 shows the time history of the applied (solid line)
and estimated (dashed line) vertical load during a double lane change and during a sweep sine.
Besides a small vertical shift (of approximately 100 N), the vertical force is correctly estimated
in both manoeuvres. Moreover, since at present the estimation of the contact parameters is
carried out only once per wheel revolution, a small time shift is visible.
Outdoor tests to validate the data analysis methodology, instead, were carried out using
an instrumented car (figure 12) having all four tires equipped with three three-axial MEMS
Downloaded by [University of California Davis] at 00:21 15 August 2014

accelerometers whose data were transmitted wireless to the vehicle body. These tests were used
to examine the influence of environmental conditions as well as the reliability and robustness
of the measuring system.
As shown in figure 13, the developed methodology allowed us to correctly estimate the
time history of the side slip angle (dashed line) for the front left wheel after almost one month
of continuous outdoor testing. The real side slip angle (solid line) is measured through a
DATRON device. It can be clearly seen that the proposed methodology is able to correctly
estimate the side slip angle. When this angle becomes large (above 2◦ ), however, a saturation
effect seems to occur. Further investigations on this problem are currently being carried out.

Figure 11. Time history of the applied (solid line) and estimated (dashed line) vertical load during (a) double lane
change and (b) a sweep sine. Tests are carried out on an MTS Flat Trac III.

Figure 12. Layout of the test vehicle used to verify the described data analysis methodology during outdoor
long-term tests.
 Measurement of contact forces and patch features 13
Downloaded by [University of California Davis] at 00:21 15 August 2014

Figure 13. Time history of the applied (solid line) and estimated (dashed line) side slip angle during an outdoor
manoeuvre for the front left wheel.

To check whether the measuring device was flexible enough, tests were carried out with
tires having different structure as well as size. Changing the calibration of the data analysis
methodology, good results were achieved for any tire tested, showing that the developed
methodology is robust enough.

5. Conclusions and future developments

The paper deals with a measurement and elaboration methodology of acceleration signals
coming from sensors fixed to the tire liner to determine both contact forces and contact patch
features directly at the tire level. Good results were obtained both during indoor and outdoor
tests for several kinematic and dynamic quantities.
The research activities that are currently being carried out deal with the identification of the
kinematic conditions of the tire (the longitudinal slippage and the side slip angle), of other
dynamic properties (the longitudinal and the lateral contact force components) and of the
adhesion characteristics of the road (the surface roughness). Good results have already been
obtained for most of these parameters but a thorough validation is still missing.

References
[1] Fukada,Y., 1999, Slip-angle estimation for vehicle stability control. Vehicle System Dynamics, 32(4–5), 375–388.
[2] Cole, D.J. and Cebon, D., 1989, A capacitative strip sensor for measuring dynamic type forces. Proc. of the
Second International Conference on Road Traffic Monitoring, London, UK, 38–42.
[3] Cole, D.J. and Cebon, D., 1992, Performance and application of a capacitative strip tyre force sensor. Proc of
IEE Conference on Road Traffic Monitoring, 123–132.
[4] Pasterkamp, W.R. and Pacejka, H.B., 1997, The tire as a sensor to estimate friction. Vehicle System Dynamics,
27, 409–422.
[5] Pohl, A., Steindl, R. and Reindl, L., 1999, The ‘intelligent tire’ utilizing passive SAW sensors measurement of
tire friction. IEEE Transactions on Instrumentation and Measurement, 48(6), 1041–1046.
[6] Ray, L.R., 1997, Nonlinear tire force estimation and road friction identification: Simulation and experiments.
Automatics, 33(10), 1819–1833.