Optimization of the damper settings

of the URE05e
S. van Rijk
DC 2010.051

Masters traineeship

Coach: Dr. Ir. I.J.M. Besselink, M. Raaijmakers

Supervisors: Prof. Dr. H. Nijmeijer

Eindhoven University of Technology

Department of Mechanical Engineering
Dynamics and Control Group
Eindhoven, December 7, 2010
Acknowledgements

In the curriculum of my study I have to complete an internship. With a fair interest and a modest
career in motorsport, I decided to take the possibility to do an interesting internship at University
Racing Eindhoven. Early 2010 I started to work on the optimization of the current dampers of
the URE05e, using the vehicle model developed by Marno Hopmans.

From the beginning of June URE started testing with their new car: the URE05e.
After reliability tests had been completed I became part of the test team. Frequently the car was
tested at a parking area in the vicinity of Ommel (Noord-Brabant).
With the use of a data-acquisition system I helped adjust the suspension of the car to find a good
balance.

First Id like to thank Igo Besselink for coaching me during this project, Jan Loof for all his
help with finding the right setup for the URE05e during the testing days and Marvin Raaijmakers
for introducing me to URE and making this internship possible. Id like to thank professional
race car driver and personal friend Chris Maliepaard for helping out. After working as his data-
engineer last year in the Benelux Racing League and winning the championship together, I have
much confidence in his ability to really feel what a race car does. Therefore I am glad that he
found the time to set some consistent laps to enable good comparison of different setups. Without
his help this wouldnt have been possible.

I
Summary

University Racing Eindhoven participates in the Formula SAE competitions. The team engineers,
manufactures, tests and races a formula race car. This year URE competes with the URE05e, a
full electrical vehicle.

The suspension system of the URE05e consists of a multi link suspension that is connected to
a damper through a pull rod.
The damper, a Koni 2612 series, is originally designed for formula cars that make use of high
downforce. The damping ratio at low damper velocities is therefore far above one and so the car
is well overdamped to control body movement.
This harsh damping is unfortunately not good for roadholding performance, therefore a new
damping curve is designed. To do so, the vehicle model of Marno Hopmans is used to simulate
the behavior during 3 specific maneuvers: straight line driving, skidpad and j-turn.
The results of the simulation show that an optimal damper setting can be reached when the re-
bound damping at low speeds is adjustable between 3600 and 6400 Ns/m and the compression
damping between 1200 en 3200 Ns/m.
The high speed damping both in rebound and compression is optimal at 1500 Ns/m and the tran-
sition from low- to highspeed damping should start at 50 mm/s.

After a period of building the car, the team tests and enhances the reliability of the car. When
the car is reliable, much work is done on adjusting suspension settings to get a good balance.
Until now most of the adjustments were done directed by the driver; what he felt while driving
was used as a guideline to set up the balance of the car.
Because this method is subjective, it is better to use a data-acquisition system to measure various
vehicle parameters. With the help of a couple of mathematical equations a good indication can
be given of the handling of the vehicle. The result of these equations have been compared to the
feedback of a professional race car driver. It is concluded that a good balance can be found with
the use of the data-acquisition system.

III
Nomenclature

Symbol Description Unit

MR Motion ratio [-]

CD Linear damping coefficient [Ns/m]
eD Asymmetry factor [-]
Progressivity factor [-]
v Damper velocity [m/s]
k Factor for high damper velocities [-]
l Damper transition point [m/s]
Ccrit Critical damping [Ns/m]
Ks Spring rate [N/m]
Kt Tire spring rate [N/m]
ms Sprung mass [kg]
mu Unsprung mass [kg]
Damping ratio [-]
C Damping coefficient [Ns/m]
Fz Dynamic tire load [N]
zs Vertical acceleration of sprung mass [m/s2 ]
z Spring travel [m]
Szr Road spectral density [m3 ]
Phase difference [m]
n Road frequency [m1 ]
xs Body travel [m]
x0s Body velocity [m/s]
g Gravitational acceleration [m/s2 ]
mwheel Wheel mass [kg]
R Corner radius [m]
V Vehicle velocity [m/s]
ay Lateral acceleration [m/s2 ]
Vehicle angle [ ]

V
Contents

Preface I

Summary III

Nomenclature V

1 Introduction 1

2 Literature survey of damping characteristics 2

2.1 Damper characteristics . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2
2.2 Current dampers . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3

3 Optimization of the suspension 7

3.1 Road profile spectrum . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 7
3.2 Quarter car model . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 8
3.3 Full car model . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 9
3.3.1 Straight line simulation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 9
3.3.2 Skidpad simulation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 11
3.3.3 Step steer simulation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 14

4 New valving design 18

4.1 Damping characteristics . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 18
4.2 Damper performance . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 20

5 Test procedure 21
5.1 Sensor calibration . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 21
5.2 Motec . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 22
5.2.1 Roll angles . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 22
5.2.2 Tire load . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 23
5.2.3 OSUS . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 24
5.2.4 Damper velocity histogram . . . . . . . . . . . . . . . . . . . . . . . . . . . 25

6 Conclusions and recommendations 26

A Testtrack 28

B Skidpad 31

VII
Chapter 1

Introduction

Formula SAE is a competition where universities from all over the world compete against each
other. Students engineer, manufacture, test and race their own formula race car. University Rac-
ing Eindhoven (URE) participates in the Formula Student competition. For 2010 URE focuses
on an electric race car to compete in the low carbon emission category, known as category 1A.
This category contains contenders with electrical and hybrid vehicles.

The URE05e is equipped with a multi-link suspension. Pull rods at the uprights actuate the
dampers and anti-roll bar through a rocker. The main component of the suspension system is
the damper. On the URE05e this component is supplied by Koni. The damper is designed for
Formula-racing and the question arises if the damper is suitable for application on a Formula
student type racing car. Therefore the performance of the vehicle is studied with the use of a
simulation model.

After an extensive period of building the URE05e, the car has to be tested and suspension settings
have to be optimized. For this purpose URE tests at a nearby trackside in Ommel. Adjustments
are made using comments from the driver, but while the drivers are volunteers and not professional
race car drivers this method is not fully reliable.
The vehicle is equipped with a large amount of sensors and a data-acquisition system. This system
can give a good indication of the performance and balance of the car.

The objective of this study is to find a damper characteristic that is optimized for the URE05e.
In a later stage, beyond the scope of this study, this damper characteristic has to be achieved by
the Koni-dampers by revalving them.
Another objective is to describe a method, using the vehicle data-acquisition system, to quantify
the behavior of the car and formulate suspension c.q. geometry adjustments to improve the bal-
ance of the car.

In chapter 2 the general damper characteristic is described and the performance of the current
dampers is examined. An optimization will be done in chapter 3 using vehicle models and the
new optimal damper characteristic will be discussed in chapter 4. In chapter 5 a procedure will
be presented that will improve the setup of the car. Finally in chapter 6 all conclusions and
recommendations will be summarized.

1
Chapter 2

Literature survey of damping

characteristics

The URE05e is a full electric powered vehicle. The monocoque and rear frame are connected to
the wheels through a multi-link suspension.
On the front of the car four of the links connect the upright to the monocoque, the steering link
is used to rotate the upright and the pull rod is used to activate the damper through a rocker
system.
The rear suspension has the same four links to connect the upright to the rear frame, a link to fix
the toe angle and a push rod to activate the rear dampers through a rocker. Both sides of the car,
front and rear, furthermore have an anti-rollbar that is also connected to the rockers and controls
the amount of body roll.

The URE05e has a motion ratio of 2, meaning that the travel of the wheel is twice as big as
the travel of the damper. The motion ratio can be expressed as:

wheeltravel
MR = (2.1)
suspensiontravel

2.1 Damper characteristics

The important characteristics of the damper have been described by John C. Dixon [1]. The
Force-velocity equation or graph consists of 3 parts:

An overall mean damping coefficient Cd

The linear damping coefficient characterizes the damper, CD defines a linear relationship
between the damper velocity and the forces.

An asymmetry factor eD
The eD -factor defines the relation between the slope of the extension and compression damp-
ing. There are various theories why the rebound damping coefficient should be higher than
the compression damping coefficient.
According to John C. Dixon [1] rebound damping should be a factor 3 to 4 higher than
compression. Claude Rouelle [9] claims that forces in rebound are normally two times higher
than forces in bump. This results in a eD -factor of approximately 12 to 53 .

A progressivity factor
The progressivity factor represents the shape of the force-velocity curve (e.g. constant,
linear, parabolic). Progressivity values are as follows:

= 0 constant Coulomb - type friction

2
 = 1 linear viscous - linearly increasing with velocity
= 2 quadratic damping

Judging from the Koni valving diagrams [6] over the separate ranges the force-velocity curve is
close to proportional, so equals 1.

The physical design of the damper features a blow-off or main valve, which makes it possible
to have a different force-velocity characteristic for controlling body movements (low damper ve-
locities) and road input (high damper velocities).
The point where the main valve opens is the point in the force-velocity graph where the curve
changes from low speed damping to high speed damping. This point, the velocity transition
point, is also referred to as the knee of the curve. The damper forces as a function of damper
velocity can be characterized by the following equations:

FC = CD, (1 eD ) v 0<v<l (2.2)

1
FC = CD, (v l) + CD, (1 eD )l vl (2.3)
k
FE = CD, (1 + eD ) v l <v <0 (2.4)
1
FE = CD, (v + l) CD, (1 + eD )l v l (2.5)
k
FC = Damping force in compresion [N]

FE = Damping force in extension [N]

v = Damper velocity [m/s]

k = Factor for high damper velocities [-]

l = Transition point [m/s]

Figure 2.1: Force-velocity curve, from [3]

2.2 Current dampers

The URE05e is equipped with the same dampers as the URE05, the Koni 2612.
The 2612 series is designed by Koni with requirements on weight, aerodynamics and friction.
In special the damper is designed for Formula 3 series and other Formula racing series that make
use of high aerodynamic downforce.

3
 Figure 2.2: damper characteristic of the 2612 at mid-adjustment range

The critical damping is given by:

p
Ccrit = 2 Ks ms (2.6)

From equation 2.6 the critical damping of the URE05e is 4131 Ns/m.

C
= (2.7)
Ccrit

Evaluating the damper characteristic shows that at lowspeed the damping ratio for rebound and
compression are respectively 3.4 and 1.96.
Judging from literature [8], these values are too high. This causes body motion of the URE05e
to be well overdamped and therefore slow reacting to changes (e.g. roll, heave, pitch). The main
reason for high damping ratios at low damping speeds is to maintain constant ride height, so to
control aerodynamics. In normal Formula racing aerodynamics is a very important factor, but
this does not apply to Formula student. Here the focus is on mechanical grip.

These specifications of the damper do not really match the requirements of URE, therefore the
damper settings need to be tested and most likely be modified.
The dampers have been tested at the track of Ommel. Professional race driver Chris Maliepaard
executed the test session to set a consistent number of laps to enable comparison of different
setups.

4
 Figure 2.3: Measurement data of a serie of laps

Each subplot of figure 2.3 shows four identical laps at the test track. The red and purple line
represent the case where both compression and rebound damping are adjusted to deliver a mini-
mum amount of damping, the light and dark blue lines represent the case where the damping is
adjusted to deliver a maximum amount of damping.
From the speed plot it can be seen that the four laps are practically identical: the brake is initiated
at the same distance and all apexes lie at the same distance. The steer angle plot also shows that
the turn-in action also starts at the same position each lap.
This set of laps is thus representative enough to draw some conclusions regarding the current
suspension.

Especially at 350 meters at turn-in, with maximum damping, the dampers react very slow to
a transition. Even at turn-exit the dampers are still not at a new stable position.
At the end of the straight, while braking at 25 meters, the front suspension compresses while the
rear suspension goes into extension. While the brake is applied with a constant pressure, the front
suspension keeps compressing even until the brake is released. The laps with minimum damping
show some situations where a stable situation is reached, but during the laps at maximum damp-
ing the suspension is constantly reacting too slow: a new input is given even long before a stable
situation has been reached after the previous input.
The response of the front right, rear left and rear right wheel can be seen in Appendix A.

A large part of the combinations of the compression and rebound adjustment range of the dampers
have been simulated.
In the simulations the vehicle drives over a relatively smooth road at a speed of 20 m/s. The
root mean square value of the vertical tire load is calculated for every simulation and is plotted in
figure 2.4.
Figure 2.4 shows that the fluctuations in vertical tire load decrease with damper adjustment (from
click 8 to click 1, both in rebound and compression). From click 8 to click 1 the fluctuation keeps
decreasing, so the minimal value is not reached within the adjustment range of the damper.
Looking at a time history (figure 2.5) of the current damper in three different positions it can be
seen that for the hardest damping the suspension does not go back to its original position, it keeps
compressing. According to W.F. Milliken [8] this effect is caused by a high amount of rebound
damping in comparison to the compression damping. This effect is called jacking down.

5
 Figure 2.4: RMS-values of Fz for different combinations of compression and rebound

Figure 2.5: Simulation for different damper values: for high damper values the vehicle is jacking
down

6
Chapter 3

Optimization of the suspension

3.1 Road profile spectrum

The vehicle is analyzed while driving over a smooth tarmac road. The road profile is given as two
tracks (left and right), each with the road height given as a function of the travelled distance. The
power spectral density of the road profile is depicted in figure 3.1

Figure 3.1: PSD of road profile

The spectrum can be approximated by the simplified description

Szr (n) = (3.1)
n2

From figure 3.1 it can be seen that 2 106 m, which represents a very smooth road. The
Formula Student competition is ran on the paddock area of some international circuits. The track
area used by FS consists of a large region of relatively fine tarmac, there are no potholes or large
obstacles.
Therefore the used road profile gives a good estimation for the road surface seen at the competition.

7
3.2 Quarter car model

To get a better understanding of the vertical dynamics of the URE05e, a quarter car model is
made as described by W. Lamers [7]. With this model the response to a vertical road input can
be evaluated. Furthermore the response can be optimized by fine-tuning damper values.

The quarter car model consists of two masses, one unsprung and one sprung mass, one damper
and two springs. The springs represent the spring between sprung and unsprung mass and the
spring rate of the tire. The values are given in table 3.1

front rear
ms [kg] 69.95 69.15
mu [kg] 8.5 8.75
K [N/m] 61000 61000
Kt [N/m] 124496 124496
D [Ns/m] 9900 9900

Table 3.1: Parameters for quarter car model

Furthermore the motion ratio between wheel and spring is implemented. The spring-damper ele-
ment between sprung and unsprung mass is built up in the same way as described by equations
2.2, 2.3, 2.4 and 2.5; there is a linear damping coefficient Cd , an asymmetry factor eD and a
lowspeed/highspeed point l.

There are three RMS-values that are evaluated in this process. The vertical acceleration is a
criterion to evaluate the ride comfort of the vehicle. In the case of a race vehicle comfort is of
course of secondary concern.
The primary concern is to keep the fluctuation of the vertical tire load to a minimum. By doing
so the tire is better able to transmit the lateral and longitudinal tire forces to the road, giving rise
to higher and more constant acceleration capabilities.
The third factor, suspension travel, is about keeping the travel to a technical achievable limit. For
instance at a minimum of the vertical tire load fluctuation, the suspension travel can not exceed
the maximum amount of travel of the damper.

The quarter car model consists of the vehicle masses, spring rates and motion ratio. The damping
coefficient is varied between 1000 and 8000 Ns/m, while the vehicle drives over a road profile at
20 m/s.
At 1800 Ns/m there is an optimum for the vertical acceleration, which means that for this damping
value the driver will experience the best comfort level.
Using a higher damping coefficient will cause the suspension to be damped out quicker and thus
travel of the damper will be smaller as can be expected. In figure 3.2 it can be seen that for
increasing damping coefficients the RMS-value of suspension travel keeps dropping.
For the tire load fluctuation there appears to be a minimum at 3200 Ns/m. As the main target
for the racing vehicle is to optimize ground-tire interaction, this is the optimum average damping
coefficient.

8
 Figure 3.2: normalised RMS values for different values of Cd

3.3 Full car model

For simulation of the URE05e, a full vehicle model is used that has been developed by Marno
Hopmans [4] .
This simulation model contains all the dynamics of the vehicle including all tire parameters.

Evaluation of the vehicle dynamics of the URE05e consists of three simulations:

Straightline driving at constant speed

Skidpad simulation

Stepsteer or J-turn simulation

3.3.1 Straight line simulation

For the straight line simulation the car accelerates to 20 m/s, then holds this speed until a total
time of 60 seconds is reached. For optimization of the road-tire interaction, the fluctuations in the
vertical tire load have to be kept to a minimum. The RMS values of the fluctuation in vertical tire
load are calculated for simulations with different CD -coefficients, the results are shown in figure 3.3.

While front and rear wheels experience a different static load due to weight distribution, this
results in different ride frequencies and damping ratios. Therefore a difference in optimum values
between front and rear damping can be expected.

Between the left and right wheels there is also a difference in optimum visible.
This is due to the phase difference between the left and right road profile. It emphasizes the need
for a reasonable adjustment range, so that e.g. banking or other non-symmetrical road irregulari-
ties can be taken into account. It is validated that with equal left and right road profiles the left
and right optimum is the same.

From figure 3.3 it can be seen that almost all wheels have an optimum around 2400 Ns/m. This
is therefore the optimum linear damping coefficient for the straight line simulation.

9
 Figure 3.3: normalised RMS values for different values of CD

The influence of asymmetry is studied for a linear damping of 3000 Ns/m and a asymmetry factor
varying from eD is -1 ( no rebound damping ) to 1 ( no compression damping ).
Figure 3.4 shows that the whole eD range left of zero has a negative influence on tire load fluctua-
tion. The optimum can be found in the righthand side of the eD range. The optimum at the rear
wheels appear to be between 0.25 at the left rear wheel and 0.15 at the right rear wheel, which is
an acceptable difference.
The optimum at the front wheels lie further apart with an optimum left front at 0.2 and right
front at 0.45, so the front wheels are much more sensitive to adjustments in asymmetry factor. To
optimize the suspension a front eD factor of 0.3 is chosen and a rear eD factor of 0.2.

Figure 3.4: normalised RMS values for different values of eD

The last parameter studied is the damper velocity transition point. The objective of this parameter
is to get a compromise between harsh body control and smooth ride over bumps. At low damper
velocities more damping to control body movements is wanted, at higher velocities we want less
damping to have a smooth ride over bumps and reduce tire load fluctuations.
Overall the damper velocities do not exceed 200 mm/s. This is the starting point of the study;
from 0 to 200 mm/s the damper velocity is varied in steps of 50 mm/s with a linear damping
coefficient of 3000 Ns/m and an asymmetry factor of 0, this to solely study the influence of the
transition point. The damping coefficient at high damper velocities are taken a factor 2 lower
than the linear damping coefficient. This choice was made after reviewing literature from W.F.

10
Milliken [8] and Claude Rouelle [9].
From these simulations it is concluded that above 100 mm/s the transition point does not really
have a major effect on tire load fluctuation.
The transition point is now investigated from 0 to 100 mm/s in steps of 20 mm/s to find an
optimum value in this range. Figure 3.5 shows that the left wheels have an optimum at around
40 mm/s and the right wheels at 60 mm/s. Because the skidpad is run both clockwise and
counterclockwise, the transition point is taken where the lines of left and right track intersect.
The optimum transition point is therefore for the front track approximately 48.5 mm/s and for
the rear track 47.5 mm/s, so a common optimum of 48 mm/s is taken.

Figure 3.5: normalised RMS values for different values of l

3.3.2 Skidpad simulation

A normal FS skidpad consists of two full circles with a diameter of 8.5 meters. During a skidpad
run the vehicle rotates two times the circle clockwise and then counterclockwise.
This test is all about steady-state vehicle dynamics, so for simulation purposes the entry and exit
from the constant-radius circle can be left out of the analysis.
For simplicity only rotation to one direction is simulated at the time. Because the car is symmet-
rical about its longitudinal axis, the results for clockwise and counterclockwise rotations should
be the same. This will be validated in a later stage.

First the clockwise rotation is simulated at a speed of 11 m/s or about 40 km/h. For the simulation
the same road profile is used as for the straight line analysis, although the inner and outer track
profiles are matched so that the phase difference stays intact.
At t=0 s the vehicle starts accelerating and follows the skidpad for 90 seconds. From approx-
imately 50 to 90 seconds the vehicle negotiates the constant-radius turn at a constant velocity.
This is therefore the time frame for the analysis.

11
 Figure 3.6: RMS values of the vertical tyre load for different values of the damping coefficient

For a clockwise turn it can be expected that the outside wheels, the left track, will have the highest
mean vertical tire load, due to load transfer while cornering. These two tires also experience the
biggest absolute fluctuation in vertical tire load.

Figure 3.7: normalised RMS values for different values of CD

Figure 3.6 shows the resulting RMS-values of Fz for simulations with different linear damping
coefficients CD .
There is a distinct difference between the left and right track in optimum damping coefficient.
At the rear axle there is also a larger difference between the left and right wheel in tire load
fluctuation and optimum of CD . This indicates that the rear suspension is more roll stiff.
The vertical tire load fluctuation of the front track is smaller and the optimum of left and right
wheel also lie closer together.

Both clockwise and counterclockwise rotation simulations show that the outer tires have the most
tire load fluctuation.
Because the simulations show symmetrical results, the optimum damping coefficient for the skid-
pad can be found by finding the intersection of the lines of the left and right wheels for the front
and rear axle in figure 3.7.

In figure 3.6 the RMS-values of the front left and front right wheel intersect at 3950 Ns/m,

12
the RMS-values of the rear left and rear right wheels intersect at 3400 Ns/m. So for the front
suspension there is an optimum at a linear damping coefficient of 3950 Ns/m and for the rear
suspension at 3400 Ns/m.

To study the effect of asymmetry, the vehicle is considered with a linear damping coefficient
of 3000 Ns/m. The asymmetry-factor eD is simulated from -1 (compression damping only) to 1
(rebound damping only), where a negative factor means that the compression damping increases
and rebound damping decreases.

Figure 3.8: normalised RMS Fz values for different values of eD

The results of the simulations are plotted in figure 3.8. It is clear from the figure that a negative
eD -factor has a negative result on vertical tire load fluctuation.
Also there is an optimum visible in the positive eD -range. The optimum is at an eD factor of 0.6,
which represents a ratio of 4 between rebound and compression damping coefficient.

The damper velocity transition point is simulated for the range of 0 to 100 mm/s, in steps of
20 mm/s. The curves in figure 3.9 show a logarithmic shape, so when the transition point in-
creases above 40 mm/s the tire load fluctuation does not significantly decrease anymore.
The left rear tire shows an increase in tire load fluctuation after 40 mm/s. The lines left and right
rear tire intersect just before 50 mm/s. A velocity transition point of 50 mm/s is therefore chosen.

13
 Figure 3.9: normalised RMS Fz values for different values of l

3.3.3 Step steer simulation

The stepsteer maneuver is performed to analyze the dynamic behavior of the vehicle. This ma-
neuver is described in ISO 7401 [10].
The vehicle is driving over a perfectly even road with no disturbances at a speed of 58 km/h. At a
certain point in time a steering wheel angle of 90 degrees is introduced and the car starts rotating
around its vertical axis.

The dynamic behavior of the vehicle in time is studied, especially the rise time, settling time
and overshoot of yaw and lateral acceleration.
These factors indicate how fast the vehicle responds to a steer input and the time that is needed
for the car to reach a new steady state situation.

This also indicates how the driver experiences the car. A car that is well damped reaches a
new steady-state condition much faster and is felt by the driver to be a stable car.
Although the car feels stable with a high overall damping, the reaction to a steering wheel input
delayed.
A softer damping gives rise to a quicker rise time, but there is a fair amount of overshoot. This
overshoot gives the driver the idea that the steering wheel angle he inputs to the car is too large
and he will decrease the steering wheel angle.
The decrease in steering wheel angle will decrease the yaw of the vehicle and thus the driver will
have troubles negotiating the turn.

14
 Figure 3.10: Yaw response to a stepsteer input

CD 1000 2000 3000 4000 5000 6000

overshoot [%] 20.16 16.34 13.90 12.45 11.65 11.26
rise time [ms] 92 92 93 93 94 94
settling time [s] 1.067 0.649 0.651 0.652 0.652 0.651

Table 3.2: Yaw response for different dampingcoefficients CD

According to W.F. Milliken [8] an acceptable limit to the amount of overshoot for the yaw response
is 15 % and the vehicle should reach 90% of the steady state value within 0.2 seconds and should
find a new stable situation (value lies between 90% and 110% of the steady state value) within 0.7
seconds.
For the lateral acceleration response these values are 0.4 seconds for the rise time, 20 % for the
overshoot and 0.7 seconds for the settling time.
Looking at table 3.2 and 3.3 this limits the damping to the domain of 3000 to 6000 Ns/m.

CD 1000 2000 3000 4000 5000 6000

overshoot [%] 21.02 18.40 17.23 17.07 17.53 18.08
rise time [ms] 179 127 127 127 127 128
settling time [s] 0.743 0.710 0.687 0.667 0.336 0.328

Table 3.3: Lateral acceleration response for different damping coefficients CD

15
 Figure 3.11: Lateral acceleration response to a stepsteer input

For a linear damping of 3000 Ns/m the influence of asymmetry eD is studied.

Stepsteer simulations are done for eD factors -1 (no rebound damping) to +1 (no compression
damping).
The stepsteer maneuver starts with the compression of the outer wheels. The increase in com-
pression damping will slow down the roll of the vehicle, this will thus reduce the body roll.
Therefore the amount of overshoot, rise time and settling time of the lateral acceleration response
will be reduced.
Also the overshoot of the yaw-response is reduced. An increased compression damping will slow
down the rotation of the vehicle around its axis and this has a negative influence on rise- and
settling time.

Increasing eD has a positive influence on rise- and settling time of the yaw-response, but de-
lays rise and settling time of the lateral acceleration response. Overshoot of the yaw response
decreases when eD approaches zero, the lateral acceleration response has its minimum overshoot
just before zero. These results can be seen in the tables below.

eD -1 -0.6 -0.3 0 0.3 0.6 1

overshoot [%] 23.33 13.57 13.52 13.90 14.89 16.94 22.54
rise time [ms] 97 99 95 93 90 86 81
settling time [s] - 654 653 652 650 653 672

Table 3.4: Yaw response for different asymmetry-factors eD

eD -1 -0.6 -0.3 0 0.3 0.6 1

overshoot [%] 21.81 16.91 16.87 17.23 18.02 19.5 23.45
risetime [ms] 121 123 125 128 130 192 206
settling time [s] 7269 670 679 688 698 709 735

Table 3.5: Lateral acceleration response for different asymmetry-factors eD

16
 Figure 3.12: Yaw response to a stepsteer input for different values of eD

With the requirements earlier mentioned in this paragraph, asymmetry-factors from -0.6 to 0.3
are valid.

For the range of 0 to 100 mm/s the damper velocity transition point is simulated. The tables
below show that for a transition point at a higher velocity the amount of overshoot for yaw and
lateral acceleration response decreases for both rebound and compression. This is because the
damper sees a higher damping coefficient for a longer time.
Increasing the transition point has a negative effect on the settling time of the yaw response, but
has a positive effect on the settling time of the lateral acceleration response.

l [mm/s] 0 20 40 60 80 100
overshoot [%] 18.05 17.19 16.27 15.61 15.04 14.56
rise time [ms] 93 93 93 93 93 93
settling time [s] 651 650 650 652 653 653

Table 3.6: Yaw response for different transition points l

l [mm/s] 0 20 40 60 80 100
overshoot [%] 19.51 19.37 19.15 18.61 18.00 17.68
rise time [ms] 128 129 130 130 130 130
settling time [s] 726 713 704 700 697 694

Table 3.7: Lateral acceleration response for different transition points l

Looking at the requirements stated earlier in this section, the transition points 0-40 mm/s are not
suitable because they have too much yaw response overshoot. A suitable domain for the transition
point is 40 < l 80 mm/s.

17
Chapter 4

New valving design

Chapter 3 deals with the simulation and damper optimalization of the URE05e for different con-
ditions.
The simulations show different optimal results for straight line driving, a skidpad and a j-turn.
The next step is to combine these results into one set of damper characteristics.

The used damper remains the Koni 2612 series, only the valving is adjusted. The dampers con-
sist of 2 cartridges and a main valve. One cartridge regulates the compression damping and one
cartridge regulates the rebound damping, a main valve opens when the damper velocity passes a
certain limit. By selecting the right cartridges and valve, the damping character can be matched
with the required character.

Figure 4.1: Schematic overview of the working principle of the 2612, from [5]

4.1 Damping characteristics

From sections 3.3.1, 3.3.2 and 3.3.3 the optimum damping coefficient is known. The values are
given in table 4.1

CD eD l
straightline 2400 0.2-0.3 48
skidpad 3950 front, 3400 rear 0.6 50
jturn 3000-6000 -0.6 to 0.3 40 < l 80

Table 4.1: Damper coefficients for different tests

18
The results show that driving over a flat road at constant speed requires a lower damping coeffi-
cient than negotiating a skidpad or a more dynamic maneuver (j-turn).This is because the straight
line simulation mainly covers tire load fluctuations and j-turn mainly considers vehicle handling
performance. The first can also be seen as the road holding capability.

Taking into account the results from the simulations above and balancing one another, a re-
quirement for the new damper valving can be formulated.
The numbers in table 4.1 indicate that for maximum body control the linear damping coefficient
should be around 4000 Ns/m. To be able to adjust the damping to a little bit less harsh setting
for roadholding, the linear damping coefficient should be adjustable down to 3000 Ns/m.
So the linear damping coefficient should be adjustable between approximately 3000 and 4000
Ns/m.
The J-turn-simulation indicates that a negative eD -factor is valid for good body control, but in
every other simulation it is shown that a negative value has a negative effect on tire load fluc-
tuations. Therefore only positive values are taken into account, furthermore best results can be
obtained if the eD factor is adjustable between 0.2 and 0.6. With an eD factor between 0.2 and
0.6 the rebound damping coefficient has to be between 3600 and 6400 Ns/m. The compression
damping has to be between 1200 and 3200 Ns/m. This new damper curve is plotted in figure 4.2
together with the curve, the Koni 2612 is currently fitted with.

Figure 4.2: Desired damper characteristic

From Chapter 2 the critical damping of the vehicle is known. With the new damping coefficients
the damping ratios can be calculated.

compression rebound
min lowspeed damping coeff [Ns/m] 1200 3600
max lowspeed damping coeff [Ns/m] 3200 6400
highspeed damping coeff [Ns/m] 1500 1500
min lowspeed dampingratio [-] 0.29 0.87
max lowspeed dampingratio [-] 0.77 1.55
highspeed dampingratio [-] 0.36 0.36

Table 4.2: Desired damper coefficients and ratios

The rebound damping can be adjusted from just underdamped to overdamped, the compression
damping is always underdamped.
The damping ratio at high speeds is about one third, which is a right value to accomplish good
roadholding performance.

19
4.2 Damper performance
With simulations of the current dampers as a reference, the performance of the new damper
characteristics are evaluated. Table 4.3 shows that the new damper curve has an improvement in
roadholding for both straight line and skidpad of more than 10 percent. Because, as was noted
earlier, the current dampers have a high damping ratio at low damper velocities, it is expected
that the increase in roadholding performance brings a decrease in handling.
The table shows that the overshoot is increased for both yaw and lateral acceleration response.
The rise and settling time of the yaw response improves, but the rise and settling time of the
lateral acceleration response show a negative result. Although the settling time of the lateral
acceleration response increases considerably it still stays within the requirements stated earlier.
So the settling time of the current dampers is faster than necessary.

original dampers new dampers improvement [%]

straightline Fz [N] 162 134.17 17.18
skidpad Fz [N] 191.33 165.11 13.70
j-turn (yaw) overshoot [%] 11.42 13.74 -20.32
risetime [ms] 103 93 9.71
sett. time [ms] 659 654 0.76
j-turn (lat acc) overshoot [%] 17.26 18.44 -6.84
risetime [ms] 124 130 -8.95
sett. time [ms] 326 682 -109.20

Table 4.3: Comparison between old and new damper characteristic

20
Chapter 5

Test procedure

In preparation of the 2010 Formula Student competition, URE launched an extensive test pro-
gram.
This test program consisted of 12 test days before the first race. Every test day consisted of
effectively 6 to 8 hours.
When the test sessions for reliability purposes are deducted from this number, there are at least
8 days left that can fully or partially be used for suspension testing.

At eight days with approximately 8 hours, there are 64 hours to be used for suspension test-
ing. In normal motorsport testing is limited to 2 to 4 days at 4 one hour sessions. When it comes
to setting up suspension during a race weekend, time is even limited to 2 one hour sessions.
Furthermore UREs test sessions have a very open-pitlane policy, at any point in time modifi-
cations can be made and the modification can be tested directly. There are no session times or
pitlane-opening times.
Therefore these test sessions feature a very open character.

Test days should be organized wisely so that a maximum amount of suspension changes can
be tested and data can be collected and analyzed at a later stage.

5.1 Sensor calibration

The most important measurements for vehicle handling performance are:
lateral, longitudinal and vertical acceleration [g]
yaw rate [deg/s]
rocker rotation [deg]
steering wheel angle [deg]
wheel speeds [km/h]
These sensors are very important for evaluating the vehicle behavior. To draw any conclusions
from the measurements it must be clear that the measurements are reliable.
The calibration of the sensors therefore must be correct and must be carefully checked.

The construction of the rocker rotation measurement is not optimal to measure damper or wheel
displacement.
The sensors are made for a 360 degrees rotation, while the rockers will at most rotate 40 degrees.
Furthermore there is some play in the connections between the rockers and the dampers.
The rotation sensors operate between 0.5 and 4.5 Volts, when the sensor rotates more than 360

21
degrees the voltage suddenly drops from 4.5 back to 0.5 Volts. Thus if the sensor is mounted
incorrectly the measurement will suddenly decrease when rotation is increasing.

At the beginning of each session the main sensors should be zeroed. This includes rocker ro-
tation, all acceleration and yaw rate sensor. Zeroing is done by connecting the Motec software
with the data-acquisition system in the car. This should be done when the car is standing on all
four wheels with the driver seated.
The steering wheel angle sensor should be zeroed when the vehicle is at a flat surface, with the
steering wheel exactly at the center.

5.2 Motec
5.2.1 Roll angles
The front and rear vehicle roll angles are calculated from the damper positions.
With the following formula the front roll angle can be calculated, this formula also holds for the
rear roll angle.
This method is also explained in Austin Dvorak et al. [2] and Jorge Segers [11]
 
180 M R (Dampertravelf ront,lef t Dampertravelf ront,right )
Rollf ront = atan (5.1)
trackwidthf ront

Figure 5.1: Rear versus front rollangle for a testsession

In figure 5.1 the rear roll angle is plotted versus the front roll angle.
First the vehicle drove with the setup indicated by the orange point cloud, the front rollbar was
almost in its least stiff position. At the front of the car the roll angles were 0.4 degrees at max.

22
An improvement was made at the front rollbar to increase the amount of roll. The green points
indicate the improvement, but the front roll angle has only increased to 0.6 degrees.
To indicate what influence the rollbar has on the roll it is totally disconnected. The purple point
cloud shows the new roll angles; the front roll angle increases to 1.8 degrees at max, while the rear
roll angle does not increase very much.
While we desire a front rollangle of about 1 - 1.2 degrees, it can be concluded that with the
current adjustment range it can not be established. Between 0.6 (softest rollbar adjustment) and
1.8 degrees (no rollbar) there is no adjustment possibility.
From this mathchannel it can be concluded that a better combination of springs an rollbar should
be sought.
The roll calculation of the vehicle is therefore a very important tool to consider the roll balance
of the car.

5.2.2 Tire load

Measurement of the vertical tire load is a very powerful tool to address problems with grip.
Unfortunately load cells are very expensive or acquire heavy amplifiers to be mounted inside the
car. Nonetheless these values are important for suspension optimization. The vertical tire load
can be estimated with formula 5.2.

F = (Ks (xs ) + c (xs )) M R (5.2)

The damping coefficient is implemented in Motec as a table of damper velocity versus coefficient.
When the spring rate is entered in Motec as a constant, the damper force can be calculated using
the measurement of the damper travel.
The static weight on each wheel should be known from static measurement on the scales to cal-
culate the load on each wheel, the formula now becomes:

Fwheel = (Ks (xs ) + c (xs )) M R + mwheel g (5.3)

Figure 5.2: Tire load calculation for measurement data

Figure 5.2 shows the velocity of the rear wheels. At certain points a problem occurs: one wheel
locks under braking or one wheel spins up while exiting a corner. These problems can be seen in
the figure as a velocity trace going to zero or speeding up rapidly.

23
This problem can be explained by the calculated tire loads in the bottom plot of figure 5.2. When
the vehicle brakes at turn 3, the left rear wheel locks up because vertically there is only 300 N or
30.6 kg of force on the wheel. This causes the wheel to lose grip and lock up.

5.2.3 OSUS
The oversteer/understeer-gradient (OSUS) compares the steer input of the driver to the amount
of steer that is needed to negotiate the corner. To do so, first the radius of the corner is calculated:

R = V 2 /ay (5.4)

The amount of steer that is needed to negotiate the corner is then calculated from:
l ay mg/l
= + b
(5.5)
R g C1 Ca2

For simplicity this equation reduces to:

l
= (5.6)
R
The steering wheel angle can be found by multiplying the vehicle angle by 5.3. The oversteer/understeer-
gradient is calculated by subtracting the theoretical steer from the steer input. If the resultant is
positive, then the driver steers more than is theoretically necessary.

Figure 5.3: Motec data of a lap at the test circuit

The upper graph of figure 5.3 shows the speed of the vehicle in km/h. From this graph it can
clearly be distinguished when the vehicle accelerates, brakes and negotiates a turn at near constant
velocity.
The second graph of the figure shows the steering wheel input of the driver and the theoretical
needed amount of steer to negotiate the turn. The third graph indicates the amount of understeer
or oversteer by subtracting the needed amount of steer from the amount of steer the driver applies.
Looking at the graph it can be noticed that in turn 3 and turn 8 there is a lot of understeer, the
driver needs to apply approximately 60 degrees of steering wheel angle more than theoretically
necessary.

24
The apex or clipping point of turn 3 can be considered as the point where the speed line changes
from decreasing to increasing. The understeer seen is mostly visible before the apex, indicating
turn in understeer.
Also the understeer starts increasing when the rate decreases with which the vehicle slows down.
So understeer arises at the point where the driver releases the brake.
The measured data here indicates that the driver is going too fast when entering the corner. When
he tries to steer the vehicle into the turn he notices that he cannot get the car to the apex.
The same holds for turn 8. Looking at the rest of the lap the theoretical and real steering wheel
angle lie on top of each other, indicating that the balance of the car is good and the car is neutral
steered.
So most of the understeer is caused by the driver and therefore in this case there is no direct cause
to change the balance of the car.

5.2.4 Damper velocity histogram

The velocities of the damper, both in compression and rebound, can be plotted in a histogram.
The histogram indicates the probability a certain velocity is seen during a run.
Figure 5.4 shows the damper histogram of two laps at the test circuit. The blue bars represent a
lap with maximum damping and the red bars represent a lap with minimum damping.

Figure 5.4: Motec data of a lap at the test circuit

On the x-axis the velocity of the damper is drawn in block sizes of 4 mm/s, the y-axis represents
the percentage.
The main difference between the red and blue is that the blue bars have a larger percentage at low
velocities. This indicates that there is more low speed damping in both compression and rebound.
The red bars have a wider spread which means that a wider range of the damper is used and the
body is not (too much) overdamped at low frequencies.
It can also be seen in the graphs that there is slightly too much rebound damping, this can be
compensated by lowering rebound damping coefficient or raising compression damping coefficient
through the damper adjustment range.
In this manner the damper histogram forms a good tool to set up the dampers at the track.

25
Chapter 6

Conclusions and recommendations

With the help of a very comprehensive vehicle model different ride conditions have been simulated.
Simulations with the current dampers show that the damping coefficients are too high.
By taking RMS-values of different signals of the simulations the results are analyzed and an opti-
mum is found for the linear damping coefficient CD , asymmetry-factor eD and velocity transition
point l.
For the straight line simulation the optimum CD -coefficient lies a little bit lower than for the
skidpad and stepsteer, but finally a damper characteristic is formulated that performs well for all
three experiments.
The damper characteristics features a low damper velocity range up to 50 mm/s where the com-
pression damping is adjustable from 1200 to 3200 Ns/m and the rebound damping from 3600 to
6400 Ns/m.
Above 50 mm/s the damping coefficient is 1500 Ns/m both for compression and rebound. With
this adjustment range the required damping force and asymmetry between rebound and compres-
sion can be realized.
The velocity transition point makes it possible to have good body control at low damper velocities
and a good roadholding capability at higher damper velocities. With the new damper character-
istic the dynamic wheel load is reduced with 13 to 17 percent. Because the current dampers are
designed to give an excellent handling at the expense of roadholding, the handling performance
decreases with the new damper curves. Overall the performance is still expected to increase.

To optimize testing four math channels in the data-acquisition analysis have been introduced:
body roll angles, tire load, steer behavior and damper histograms.
The results of these channels have been compared to the findings of a professional race car driver.
It is concluded that the math channels are a good tool to evaluate the vehicles balance and pro-
pose adjustments.

The first recommendation concerns the dampers: it is demonstrated that the new damper curve
gives an improvement in vertical tire load fluctuation. Therefore it is recommended that the cur-
rent dampers will be fitted with this new curve. Secondly test days are limited, so it is wise to
make good use of the available time. A good structure for improvement of the setup will speed
up the process. During the test days a large amount of data is collected. For the near future it is
advisable to use this measurement data to validate the full vehicle simulation model.

26
Bibliography

[1] J. C. Dixon. The Shock Absorber Handbook. Wiley-Professional Engineering Publishing Series,
2007.

[2] A. Dvorak and P. Fitzhorn. State transition diagrams of transient roll and pitch. Motorsports
Engineering Conference Concord, North Carolina, 2008.
[3] S. P. Gacka and C. G. Doherty. Design analysis and testing of dampers for a formula sae race
car. 2006.

[4] J. Hopmans. Optimization of the tyres used on the ure04/05 race car. 2009.
[5] Koni. Koni 2612 technical manual v1.2.1.
[6] Koni. Technical manual 28 series racing damper v5.0, june 2005.
[7] W. Lamers. Development and analysis of a multi-link suspension for racing applications.
2008.
[8] D. Milliken, W.F. & Milliken. Racecar Vehicle Dynamics, pages 260264. Society of Auto-
motive Engineers, SAE, tenth edition, 1998.
[9] C. Rouelle. OptimumG seminar binder, pages 471472. OptimumG.

[10] SAE. Lateral transient response test methods - open loop test method.
[11] J. Segers. Analysis Techniques for Racecar Data Acquisition. Society of Automotive Engineers,
SAE, 2008.

27
Appendix A

Testtrack

Test sessions are performed at the test track at Ommel (Noord-Brabant). The track layout that
was used during the test sessions with Chris Maliepaard is depicted in figure A.1.

Figure A.1: Layout of the testtrack at Ommel

The measurement data of the front right, rear left and rear right wheels is plotted in figures A.2,
A.3 and A.4.

28
Figure A.2: Measurement data of the damper response of the front right wheel

Figure A.3: Measurement data of the damper response of the rear left wheel

29
Figure A.4: Measurement data of the damper response of the rear right wheel

30
Appendix B

Skidpad

The road profile definition is measured for a vehicle driving in a straight line. When a skidpad is
driven, the phase between the left and right track is disrupted. The road profile therefore needs
to be adjusted to suit the skidpad.
The inner circle of the skidpad is 15.25 m in diameter, the outer circle 21.25 m. Because the
performance on the skidpad is measured as lap time, the distance traveled should be minimal.
Therefore the inner wheel of the car should be as close as possible to the inner radius cones: the
radius driven by the center of the car is 8.3 m.

Figure B.1: Skidpad lay-out

Ri = 7.6875 m
R = 8.3 m
Ro = 8.9125 m
The angle that is driven of the skidpad circle
 
0.5 dx
= 2 asin (B.1)
R

31
The displacement of the outer wheel is
 

xo = 2 Ro sin (B.2)
2

The displacement of the inner wheel is

 

xi = 2 Ri sin (B.3)
2

The displacement of the middle circle is

 

x = 2 Rsin (B.4)
2

The new inner x-coordinate is x RRi and the new outer x-coordinate is x RRo . From these x-
coordinates and the original Z-coordinates, the Z-coordinates for a common x-coordinate can be
calculated.

The matlab-code to transform the current road profile is: x = 0 : 0.05 : 1234;
zi = interp1(X(:,1)*Ro/R,X(:,2),x);
zo = interp1(X(:,1)*Ri/R,X(:,3),x);

32