TECHNICAL TRANSACTIONS 8/2019

MECHANICS
DOI: 10.4467/2353737XCT.19.085.10864
SUBMISSION OF THE FINAL VERSION: 2/07/2019

Aleksander Kuranowski orcid.org/0000-0002-1688-7027

[email protected]
Faculty of Mechanical Engineering, Cracow University of Technology

Electrical power steering – modelling and bench testing

Elektryczne urządzenie wspomagające kierownicę –

modelowanie i badania stanowiskowe

Abstract
The article presents a dynamic model of an electric device supporting the steering system of a passenger car;
the model was verified through a series of bench tests. The research object was an integrated electric power
steering system (EPS) mounted on a steering column and cooperating with a steering gear. The results of
the theoretical analysis were compared with the results of tests performed on a specially built research bench
fully reflecting the work of assistance in the car. A satisfactory level of agreement between the results of the
model tests and the bench tests was obtained.
Keywords: EPS, modelling, bench tests

Streszczenie
W artykule przedstawiono zweryfikowany testami laboratoryjnymi dynamiczny model elektrycznego urzą-
dzenia wspomagającego kierownicę. Przedmiotem badań było zintegrowane elektryczne urządzenie wspo-
magające kierownicę (EPS) zamontowane na kolumnie kierownicy. Wyniki analizy teoretycznej porów-
nano z wynikami testów przeprowadzonych na specjalnie zbudowanym stanowisku badawczym w pełni
odzwierciedlającym pracę urządzenia w samochodzie. Uzyskano zadawalającą zgodność między wynikami
badań stanowiskowych a wynikami analiz teoretycznych.
Słowa kluczowe: EPS, modelowanie, stanowisko badawcze

143
 1.  Introduction

The subject of this work is the development of an EPS device model and its verification
with the results of bench tests. Workplace tests of an electrically supported passenger car
steering system were conducted in order to determine the values of the stiffness, damping
and friction parameters occurring in the assistance system. Harmonic measures were applied
from the steering wheel and wheels.
The study of the EPS system includes the examination of both the control objectives
and the control strategies. Control objectives relate to the size of the support system and to
the feeling of the road. Control strategies are divided into classic strategies resulting from
the drive automation and strategies using models developed separately.
The article [5] focuses on the development of a control algorithm and analyses the choice
of simplifications applied when developing a control system to approximate the mechanical
characteristics of a vehicle.
Articles [7, 9] present a simulation of the EPS control system integrated with the overall
dynamic vehicle model. Co-simulation was used, i.e. the simultaneous cooperation of two
software environments (ADAMS and Matlab) by means of a subprogram specifying data
channels and running the ADAMS environment in each simulation clock. Using co-simulation,
a multi-mass dynamic model of a vehicle cooperating with the EPS control algorithm was
simulated on the smooth road with one inequality [7]. The author of article [9] presented
a method of modelling the EPS system and its co-simulation in combination with the control
system. The mechanical model was calibrated using the experimentally measured torsional
stiffness of the torsion bar and the friction of the mechanism. The EPS model optimised in
ADAMS was converted into Matlab/Simulink, and was then combined with the Simulink
driver model. Co-simulation was performed and described for the frequency response test of
the EPS device. The proposed method enables a good prediction of the dynamics of the EPS
system and can be an effective way to optimise the EPS control system.
The authors of article [8] proposed two indicators – steering sensitivity and road feel – in
order to assess the performance of the steering system. It was found that steering sensitivity
depends on the following factors: stiffness of the torque sensor, the electrical characteristics
of the propulsion engine, its moment of inertia, the transmission ratio of the power steering,
the boost control and the vehicle parameters.
In [1], the authors considered the possibility of adjusting steering sensitivity; they used
a simplified model of the steering system, containing both a reduced mechanical and an
electrical model. The effectiveness of the adopted simplification has been proven by computer
simulation.
Of the many methods of torque control of an electric motor used in EPS systems, the
authors of article [2] drew attention to the method of selecting the moment consisting of
minimising the difference between the values of the standard moment and the measured
moment. The process of designing a controller which realises the algorithm of minimising the
mentioned difference is presented. During the tests, the discontinuity of the steering torque
was analysed due to the fact that this moment occurs both with and without assistance. It has

144
been surmised that the cause of discontinuity is the occurrence of dry friction in the steering
system. After formulating the requirements for the steering torque controller by specifying
the target values for the characteristic parameters of the control device, a H∞ type controller
was developed which was implemented as a steering torque controller using frequency-
-dependent weighted functions.

2.  Model of the EPS support device

Figure 1 shows an EPS device that has been removed from the vehicle and connected to
the rack steering gear, and Fig. 2 shows the dynamic model of the system.

Fig. 1. The actual steering system with the electric assist device;
1 – steering wheel, 2 – EPS, 3 – steering gear

Fig. 2. The dynamic model of a steering system with electric assist device

145
 Notation:

cdk – angular stiffness of the lower part of the column

cdr – angular stiffness of the torsion bar
cgk – angular stiffness of the upper part of the column
cm – angular stiffness of the electric motor clutch
cz – stiffness of the elastic housing of the rack
Ddk – damping factor the lower part of the steering column
Ddr – damping factor of the torsion bar
Dgk – damping factor of the upper part of the steering column
DL – damping factor of the rack
Dm – damping factor of the electric motor
FL – external force of the steering rods imported from the axle rack
Fl – force in the left tie rod
Fp – force in the right tie rod
Fz – circumferential force on the pinion gear
ip – steering gear ratio
is – worm gear ratio
Jdk – moment of inertia of the lower part of the column
Jgk – moment of inertia of the upper part of the column
Jk – moment of inertia of the steering wheel
Jk+d – moment of inertia of the steering wheel including mass of the hand
Jm – moment of inertia of the rotor of the electric motor
Js – moment of inertia of the worm wheel
Jz – moment of inertia of the pinion
ke – electromechanical constant
kws – constant assisted
Mdr – torque occurring on torsion rod
Mk – torque applied to the steering wheel
Mm – the electric motor torque assist device
MTs – friction torque of the worm gear
mL – mass of the rack
T – friction force of the rack
y – shift of the rack
φdk – angle of rotation of the lower section of the steering column
φgk – angle of rotation of the upper part of the column
φk – angle of rotation of the steering wheel
φm – angle of rotation of the electric motor
φs – angle of rotation of the worm wheel
φz – angle of rotation of the pinion

146
 The equations of the system dynamics shown in Fig. 2 are recorded as follows:

Jk     
k  D gk  k   gk  c gk k   gk  M k (1)

In view of the kinematic forced rotation of the steering wheel it was assumed that φk = φgk.
  
 s  cdr  s  k   cdk  s  z   cm is2  s  m
Js   0 (2)
 is 

m  cm  m  s is   MTm  M m (3)

Jm 

J dk  ml rz2    
 z  Dl rz2  z  cdk  z  s   Trz  c z rz2 z  Fp  Fl rz (4)

Damping of the rack described by the Dl rz2 ϕ z member was omitted. The engine torque
depends on the stiffness and the angle of deflection of the torsion bar placed in series with the
gear pinion described by constant kws:

M m  kws cdr  k  s  (5)

Differential equations of motion of the booster are linear, thus it was possible to apply the
principle of superposition. Therefore, differences in the variable waveforms with the power off and
the variable waveforms corresponding to the system with the power on are caused by influence of
unstable torque due to the introduction of additional torque of assist motor Mm. The numerical
analysis results discussed later were verified experimentally on a specially built test bench.
The model of isolated steering system simulating the conditions of the bench measurements
was formulated on the basis of the following assumptions:
▶▶ Sinusoidal kinematic input function caused changes of the angle φk of the steering
wheel rotation with amplitude ±15° and frequency 0.5 Hz.
▶▶ External loading of the system is caused by the deformation of the spring elements
having a coefficient of stiffness cz connected in series with the gear rack.
▶▶ Assist torque from the electric motor is proportional (P-controller) to the torsional
moment on the torsion bar mounted on the steering column.
▶▶ Friction forces are described using the model of Coulomb.
▶▶ Omitted from the model are:
▶▶ the impact of universal joints;
▶▶ flexibility of the rack gear housing;
▶▶ play in the moving joints;
▶▶ dynamics of the control system (sensors, transducers).
The mechanical model of the analysed system had three degrees of freedom: φk; φm; φz.
Differential equations describing the dynamics of the system were solved in Matlab. As in
[3], the torque of the EPS electric motor was treated as the product of the current i and the
electromechanical constant ke.

147
 M m  ke  i (6)

Values of the parameters of the examined model are shown in Table 1.

Table 1. Values of the parameters of the tested EPS model
Unit
Values that were determined Notation Value
of measure
moment of inertia of the steering wheel Jk 33 · 10-3 kgm2
moment of inertia of the upper part of the column Jgk 4.62 · 10-5 kgm2
moment of inertia of the worm wheel Js 0.83 · 10-3 kgm2
moment of inertia of the rotor of the electric motor Jm 0.0321 · 10-3 kgm2
moment of inertia of the lower part of the column Jdk 0.002 · 10-3 kgm2
moment of inertia of the pinion Jz 5.48 · 10-6 kgm2
angular stiffness of the upper part of the column cgk 9800 Nm/rad
angular stiffness of the torsion bar cdr 91 Nm/rad
angular stiffness of the electric motor clutch cm 3.85 Nm/rad
angular stiffness of the lower part of the column cdk 2400 Nm/rad
stiffness of the elastic housing of the rack cz 1.7 · 10 6
N/m
damping factor of the upper part of the steering column Dgk 0.0275 Nms/rad
damping factor of the torsion bar Ddr 0
damping factor of the electric motor Dm 0.0035 Nms/rad
damping factor the lower part of the steering column Ddk 0
damping factor of the rack DL 0.0275 Ns/m
worm gear ratio is 0.0455
steering gear ratio ip 20.5 rad/m
mass of the rack mL 1.45 kg
friction force of the rack T 175 N

3.  Workplace tests

A steering system with a rack gear and an EPS supporting device was installed on a special
measurement and research stand (Fig. 3). The advantage of this is stand is that it was created
from the deconstruction of this part of the CAN communication, which is essential in cars
for the operation of steering assistance. The amount of assistance was dependent upon the set
driving speed. Two types of steering rotation were generated.
In the first type of steering rotation (test I), a direct-current (DC) motor with two
different rotational speeds was used to drive the steering wheel, allowing a steady rotation

148
of the steering wheel. The input torque was transferred to the steering column via a worm
gear and a toothed belt. This enabled testing at two rotational speeds of the steering wheel:
0.2  rpm and 0.4 rpm. The drive system of the steering wheel was equipped with circuit
breakers in order to protect against overload resulting from the extreme positions of the
rack. These breakers allowed the reversible operation of the stand with steering wheel
rotation in the range of ±500°.
In the second type of extortion (test II), harmonic kinematic exclusion was first performed
from the steering wheel side and then from the rack side using a connecting rod machine
originally used to determine the characteristics of the shock absorbers. This machine was
powered by a 40 kW DC motor. Variable stroke adjustment in the range of 0–100 mm and
smooth regulation of the frequency of excitation were possible. The uniformity of the drive
was ensured by the flywheel with an inertia moment of 17.0 kgm2.

Fig. 3. View of the test bench for the EPS device with equipment

Measurements of forces and displacements were performed by equipping the stand with
additional external devices, specifically, HBM force sensors and wire sensors to measure
displacements (turn of the steering wheel, shift of the rack).

149
 Table 2. Listing of external measuring equipment
Sensor type Type Measurement range Accuracy
force sensor HBM U2B 20 kN ±0.02%
position sensor Kubler DB3A1 500 mm ±0.1%

For the driver’s input function, the steering wheel was connected to the machine connecting
rod fastener using a 1.5-m-long connector with ball joints. The wire sensors were used to
measure the angle of rotation of the steering wheel and the displacement of the toothed bar.
The cable for measuring the angle of the rotation of the steering wheel was wound on a wheel
mounted on the steering column shaft, which was coaxial with the steering wheel.
At the input function from the side of the steering wheel, the load on the rack was obtained
by elastic elements of a stiffness, connecting the gear rack to the housing.

3.1.  Tests of the input function of the steering wheel

Two types of tests were performed on the input function of the steering wheel:
▶▶ rotation of the steering wheel by an angle of ±500°,
▶▶ harmonic excitation in the range of ±15°.
During tests with a steering wheel rotation of ±500°, the transmission worked without
a load and the steering wheel rotation was at one of two values: either the speed of around
0.2 rpm or 0.4 rpm.

Fig. 4. Bench tests with harmonic input function from the steering wheel in the range of ±15°

Taking into account the transmission ratio of the rack gear, the harmonic input function of
the steering wheel caused the displacement of the rack by approx. 2.5 mm. The investigations
concerned two cases: a free (unloaded) rack; a rack supported by elastic elements with
a stiffness of around 1.7 · 106 N/m.

150
 Fig. 5. Work bench tests with kinematic input function on the steering wheel side
within the range of ±500° rotation

The bench tests included measurements of the torque at a ±15º steering wheel rotation in
which the movements of the steering wheel are performed during the correction that takes
place during the process of driving.
In order to determine the torque developed by the motor Mm at different booster operating
states, the characteristics Mk = f(φk) were first determined with the booster switched off,
followed by the same input parameters (amplitude φ = ±15° and frequencies f = 0.5, 1.0,
1.5 Hz) with the assistive device switched on.
The constructed test stand proved to be sufficiently functional and the registered
characteristic curves showed repeatability.

3.2.  Tests at the input function from the rack

The tests of the steering system at the kinematic input function of the gear rack movement
were performed on the modernised stand (Fig. 6).

Fig. 6. Research on kinematic input function from the rack mounted in a vertical position;
a – force sensor, b – position sensor

151
 The adaptation of the stand consisted of the articulated connection of the gear rack with
the machine slider for, among other purposes, testing the shock absorbers using the steering
rod (with factory ball joints) on which the force sensor was mounted. The amplitude of the
input function had a fixed value of ±15 mm. Such a displacement of the rack caused rotation
of the steering wheel in the range of approximately ±175°. The frequency of the input function
excitation was either 0.1 or 1.0 Hz.
The tests concerned transmission with an assistance mounted with different loading
of the system: free steering wheel without additional weight; steering wheel loaded on the
circumference with weight 0.8 kg multiplied by 2, replacing the weight of the driver hand and
forearms placed freely on the steering wheel.
A very strong dependence of the dynamic load of the rack on the size of additional
masses attached to the steering wheel was observed. In the case of the steering wheel without
additional weight, the force in the rack did not exceed 0.75 kN. There were noticeable
vibrations related to the torsional elasticity of the torsion bar. With mounted masses of 0.8 kg
multiplied by  2, and a similar kinematic input function – the force loading the sprocket
increased around four times up to nearly 3 kN.

3.3.  Comparison of the results of bench and simulation tests

Two simulation tests were performed.

Test I referred to forcing the rotary-oscillating motion of the steering wheel with
a  sinusoidal waveform and is described by the following parameters: frequency of input
function 0.5 Hz, the absence of a load on the side of the rack (cz = 0), and simulated car speed
of 0 km/h. Figure 7 shows the time courses of changing the angle of rotation of the steering

Fig. 7. Comparison of time courses from bench measurements and simulations during Test I

152
wheel φk and the displacement of the rack relative to the housing yz,o. Figure 8 presents the
torque courses of the torsion bar Mdr equal to the torque at the steering wheel Mk.
When forcing a change in the angle φk with a sinusoidal pattern, the occurring displacement
of the rack yz,0 is similar and only slightly changed by the friction forces at the extreme
positions. A smaller displacement of the yz,o racks was obtained with the model in comparison
with the bench tests. The mapping of moment changes of the Mdr torsion bar is satisfactory
with regard to the phase and amplitude of the response, although according to the model, the
changes do not show vibrations at the vertices of the waveform. The above statements are also
confirmed by the torque characteristics as a function of the angle of rotation of the steering
wheel, which is shown in Fig. 8.

Fig. 8. Comparison of example runs from bench measurements and simulations of torque changes
on the steering wheel (Mdr = Mk) as a function of the angle of rotation of the steering wheel

Test II consisted of forcing a rotational-oscillating motion of the steering wheel with

a sinusoidal waveform, where the frequency of input function is 0.5 Hz, loads on the side
of the rack with stiffness cz = 1.7 · 106 N/m, and the simulated car speed = 0 km/h. Figure 9
shows the changes of the steering wheel rotation angle φk, the displacement of the rack with
respect to the housing, yz,o and the moment of the Mdr torque rod equal to the torque at the
Mk steering wheel.
Due to an applied sinusoidal input function, φk moves along a similar course as yz,o, which
is slightly deformed by friction forces in extreme positions. In the model’s response, a smaller
amplitude of displacement of the yz,o racks was obtained. The mapping of torque changes on
the Mdr torsion bar is satisfactory with regard to the phase and amplitude of the response,
although the variables determined through the use of the model do not show interference at
extreme positions.
Details of the results obtained from the model in the form of a torque graph as a function
of the angle of rotation of the steering wheel are shown in Fig. 10.
The waveforms of the variables determined by the model are satisfactorily represented in
relation to the hysteresis loop width and the average slope of the characteristic, although this
is not very similar to the S-letter profile. The linear course without accentuating the S-letter

153
 profile can be explained by the linear dependence of the torque of the auxiliary motor from
the steering angle of the control rod with the torque value assumed in the model. In fact, as
research shows, the dependence is not linear and it is progressive.
The adopted simulation model of the electrical booster satisfactorily describes the real
object despite the introduced simplifications. There is a noticeable influence of friction,
which causes a hysteresis loop effect with a width similar to the previously measured one. The
vibrations, which are small but noticeable at higher frequencies in the study, are not true to
a simulative mapping. It could be caused by the irregular torque of the DC motor.

Fig. 9. Comparison of time courses from bench measurements and simulations during Test II

Fig. 10. Comparison of example runs from bench measurements and simulations of torque changes
on the steering wheel (Mdr = Mk) as a function of the angle of rotation of the steering wheel

It can be assumed that a better representation of the actual characteristics of the

transmission can be obtained if instead of the linear relationship between the torque of the
motor and the load of the transmission, the non-linear function is introduced.

154
 The EPS current characteristics were determined on the same bench. The waveform
of changes in current values as a function of the torque on the steering wheel at different
speeds is shown in Fig. 11. It was found that the value of EPS current as a function of the
torque for an assumed velocity increases to a certain value in a non-linear manner, and after
reaching the value imposed by the limiter, it is intentionally limited and takes an almost
constant value.
The equation approximating a non-linear increase in the current can be written as:

I( M )  a  M  eb M (7)

where a, b, and e are parameters dependent upon speed.

Estimation of parameters a and b by the method of least squares at a confidence level of 95%
enabled determining the function approximating the results of measurements in a range of
non-linear dependence. Based on the measurements, it was assumed that in the range of
linear dependence, the current is constant.

I (= = c (8)
M ) const

Theoretical characteristics (Fig. 12) of the current intensity obtained at different speeds
were similar to obtained results of the measurements (Fig. 11). Appropriate values of the
parameters a,  b and c were determined. The values of these parameters at speeds within
the range of 0–160 km/h with increments of 20 km/h are shown in Table 3.
Table 3. Values of the parameters of the equation (8) describing the EPS current
v km/h a b c
0 0.74 0.33 37.5
20 0.24 0.43 22.5
40 0.24 0.31 17.5
60 0.16 0.30 14.8
80 0.07 0.38 14.6
100 0.11 0.31 14.8
120 0.10 0.33 15.0
140 0.09 0.34 15.0
160 0.09 0.33 15.0

Figure 11 shows changes in the intensity of the current supplying EPS as a function of
the torque on the steering wheel. Changes in the intensity of the EPS current as a function
of the steering wheel torque when simulated at different driving speeds are shown in Fig. 12.
On account of ensuring legibility of the drawing, the number of speeds plotted was limited to
only six. Figure 13 shows changes in the intensity of the EPS supply current when the vehicle
is at a standstill.

155
 Fig. 11. Chart of assist-motor current intensity depending upon
the steering wheel torque

Fig. 12. Changes in EPS current intensity as a function of the torque on the steering
wheel different speeds simulated on the bench

Fig. 13. Changes in current intensity as a function of the torque on the steering wheel; velocity v = 0 km/h

The method of controlling the current intensity as a function of the steering wheel torque
and dependent upon the driving speed is shown in the spatial diagram (Fig. 14).

156
 Fig. 14. Changes in EPS current intensity as a function of the torque on the steering wheel and variable speed

The torque developed by the engine of the booster is dependent upon the current value
of the electromechanical constant, hence the fact that the shape of the shell I = f(Mk) maps
the shape of the support characteristic. The current intensity value is the resultant value of the
EPS device controller and it determines the course of the characteristics.

4.  Conclusions

Waveforms determined by the model are satisfactorily well represented in relation to the
width of the hysteresis loop and the average slope of the characteristic (Fig. 10), again this
seems very vague – you should clearly state which characteristic you are referring to although
this does not demonstrate a clear similarity with the S-profile. The linear course without
emphasising the S-profile can be explained by the linear dependence of the torque between
the auxiliary motor and the steering angle controlling the torque value in the model. In fact, as
this research shows, the dependence is not linear; it has a progressive character.
The adopted simulation model of the electrical booster, despite the introduced
simplifications, satisfactorily describes the real object. There is a noticeable influence of
friction, which causes a hysteresis effect with a width similar to the previously measured one.
The vibrations, which are small but noticeable at higher frequencies in the study, are not true
to a simulative mapping.
It can be assumed that a better mapping of the actual transmission characteristics can be
obtained if instead of the linear relationship between the engine torque and the transmission
load, a non-linear function is introduced.

The author is also an employee of the Institute of Forensic Research (Instytut Ekspertyz Sądowych im. prof. dra Jana Sehna).

157
 References

[1] Badawy A., Zuraski J., Bolourchi F., Chandy A., Modeling and Analysis of an Electric
Power Steering System, Delphi Automotive Systems 2009, 25–31.
[2] Dannöhl C., Müller S., Ulbrich H., H∞-control of a rack-assisted electric power steering
system, Vehicle System Dynamics 2011, 1–18.
[3] Data S., Pesce M., Reccia L., Identification of steering system parameters by experimental
measurements processing, Proc. Instn. Mech. Engrs 218, Part D: J. Automobile
Engineering 2004, 783–792.
[4] Govender V., Khazaridi G., Weiskircher T., Keppler D., Müller S., A PID and state space
approach for the position control of an electric power steering, 16th Stuttgart International
Symposium Automotive and Engine Technology, March 2016, 755–770.
[5] Kim J.-H., Song J.-B., Control logic for an electric power steering system using assist motor,
Mechatronics 12, 2002, 447–459.
[6] Kuranowski A., Effect of variable input and output on the characteristics of a steering
system with EPS (Electric power system), Journal of KONES, 19, No. 1, Warszawa
2012, 215–226.
[7] Liao Y.G., Du H.I., Cosimulation of multi-body-based vehicle dynamics and an electric power
steering control system, Proc. Instn. Mech. Engrs. 215, Part K, 2001, 141–151.
[8] Liu Z., Yang J., Liao D., The Optimization of Electric Power Assisted Steering to Improve
Vehicle Performance, Proc. Instn. Mech. Engrs. 217, Part D, 2003, 639–646.
[9] Qiu Y., Electrical power steering mechanical modelling and co-simulation with control system,
Institute of Sound and Vibration Research University of Southampton, 2010.
[10] Xue-Ping Z., Xin L., Jie Ch., Jin-Lai M., Parametric design and application of steering
characteristic curve in control for electric power steering, Mechatronics 19, 2009, 905–911.

If you want to quote this article, its proper bibliographic entry is as follow: Kuranowski A., Electrical power steering – modelling
158 and bench testing, Technical Transactions, Vol. 8/2019, pp. 143–158.