Mashadi 2015

Applied Mathematical Modelling xxx (2015) xxx–xxx
Contents lists available at ScienceDirect
Applied Mathematical Modelling

journal homepage: www.elsevier.com/locate/apm
Driveline oscillation control by using a dry clutch system

Behrooz Mashadi ⇑, Milad Badrykoohi
School of Automotive Engineering, Iran University of Science and Technology, P.O.B. 16846-13114, Farjam st, Tehran, Iran
a r t i c l e i n f o a b s t r a c t
Article history: The reduction of vehicle driveline vibrations such as shuffle and shunt by controlling a dry
Received 7 March 2014 clutch system is studied in this paper. For this purpose a dynamic model for the driveline of
Received in revised form 19 December 2014 a typical passenger car including backlash, clutch friction model and tire slip is developed.
Accepted 19 January 2015
A linear controller is designed for the clutch control and damping of driveline vibrations.
Available online xxxx
The results of this research show that by using a dry clutch system, vehicle longitudinal
vibrations can be controlled effectively. Moreover, this method does not have the disadvan-
Keywords:
tages of other solutions including deteriorating air pollution and vehicle performance. In
Shuffle
Shunt
addition it involves low costs especially in cars with automated manual transmission sys-
Driveline vibration tem (AMT).
Dry clutch Ó 2015 Elsevier Inc. All rights reserved.
Friction model
1. Introduction
Drivability comprises a series of factors related to the driving comfort and depends on the feelings of a driver during driv-
ing and plays an important role in the commercial success of a vehicle. This factor mostly depends on the vehicle responses
during the transient driving conditions, for example gear shifting, accelerating and braking. As the vehicle driveline incor-
porates elastic elements, it vibrates during transient conditions. These vibrations that are transferred to the vehicle body
cause vehicle longitudinal vibrations and affects drivability and passenger comfort in a negative way. Major driveline vibra-
tions occur during tip-in, tip-out and gear shifting. The driveline vibrations are categorized as shunt, shuffle, judder, rattle
and clatter.
Shunt is the percent of the first overshoot and shuffle is the duration of vehicle longitudinal oscillations in the transient
response as graphically demonstrated in Fig. 1. Shuffle occurs during sudden changes in the transferred torque from engine
to the driveline because of the presence of the elastic elements and backlash in a driveline. These vibrations occur in the
range of 1–10 Hz and coincide with human body resonance frequency range. For example, shoulder frequency is between
4–6 Hz, stomach frequency is between 4–8 Hz and upper body frequency is between 3–6 Hz [1]. Thus, controlling these lon-
gitudinal vibrations is important.
The reduction of the driveline vibrations has been an important research area approached by many researchers. Engine
torque control during the transient conditions is a major concept for the driveline oscillation control. The control of spark
advance of engine is one of the methods in which controller uses the engine torque dynamics to actively control the driveline
vibrations. Berriri et al. [2], Fredriksson et al. [3], Lagerberg et al. [4] and Lefebvre et al. [5] are among several researchers
who have used this method. Another solution method is controlling the engine air/fuel ratio. In this method, the engine out-
put torque is controlled by adjusting the air/fuel mixture (Petterson et al. [6]). Third approach is the control of electronic gas
⇑ Corresponding author. Tel.: +98 (21)77240360; fax: +98 (21)73021560.

E-mail address: b_mashhadi@iust.ac.ir (B. Mashadi).
http://dx.doi.org/10.1016/j.apm.2015.01.061
0307-904X/Ó 2015 Elsevier Inc. All rights reserved.
Please cite this article in press as: B. Mashadi, M. Badrykoohi, Driveline oscillation control by using a dry clutch system, Appl. Math. Modell.
(2015), http://dx.doi.org/10.1016/j.apm.2015.01.061
2 B. Mashadi, M. Badrykoohi / Applied Mathematical Modelling xxx (2015) xxx–xxx
Nomenclatures:
Symbol Description
A frontal area
b damping coefficient
B damping
Cd drag coefficient
cs driveline damping coefficient
fslip coulomb friction
fstick stiction friction
fR rolling resistance coefficient
fv viscous coefficient of damping
g gravitational acceleration
F force
Fa aerodynamic force
Fr rolling resistance force
Ft traction force
Fz tire normal load
I Inertia
If flywheel inertia
It transmission inertia
Iw tire inertia
k stiffness coefficient
K gain
K stiffness
ks driveline stiffness
m vehicle mass
n overall gear ratio
Nc desired clutch force
rw tire rolling radius
rc clutch equivalent radius
s Laplace variable
S slip
t time
Tc clutch torque
Te engine torque
v velocity
Vrel relative velocity
Greek letters
a backlash angle
b road gradient
l friction coefficient
h rotation angle
q air density
x angular velocity
xe engine angular speed
xd clutch angular speed
pedal used by Northcote [7]. It is shown that this method has attenuated sudden changes in transferred torque to the drive-
line and has controlled the vibrations.
Driveability enhancement in hybrid electric vehicles (HEVs) is achievable by means of electric motor control. Kou and
Weslati [8] used a sliding mode control technique for attenuating vehicle shuffle. For the proposed control approach, refer-
ence states were computed for different transmission modes and the tracking errors were calculated using actual measured
states. Also Syed et al. [9] presented the design of an active damping wheel-torque control system to improve the drivability
of a power-split Ford Escape HEV.
Solution methods outlined above each has its own advantages and disadvantages. In the spark advance engine control
method, altering the spark timing against the original combustion settings will in general impair the engine pollution. Also
B. Mashadi, M. Badrykoohi / Applied Mathematical Modelling xxx (2015) xxx–xxx 3
Fig. 1. Shunt and shuffle phenomena.
there is limitation to produce the torque that controller demands. In both of the air/fuel ratio and electronic gas pedal control
methods, changes in the driver torque demand lead to reducing vehicle performance and is not acceptable for most drivers.
The concept of using a dry clutch for attenuating the driveline vibrations presented in this article overcomes the disad-
vantages of the existing methods. This concept can be implemented with relatively low cost and has no side effects such as
increasing engine pollution and reducing the vehicle performance.
2. Modeling
Vehicle driveline dynamics behavior depends on its subsystems and major components specifications. Clutch dynamics
and backlash are two influential phenomena that must be considered in this respect. Tire slip is also another important factor
that relates the vehicle longitudinal dynamics to the driveline dynamics. In this part these influential factors will be treated
and proper models will be introduced.
2.1. Driveline
The main components of the driveline system including engine inertia, clutch, transmission, driveshafts and tires are
shown in Fig. 2. Tf, Tc, Td and Tt represent the engine torque transmitted to the driveline, clutch torque transmitted to the
gearbox, differential drive torque acting on the wheels and wheel load torque acting against the vehicle motion. If represents
the sum of engine and flywheel moments of inertia, It is the equivalent inertia of transmission system and Iw denotes the
wheel inertia. hf, hc, ht and hw are engine, clutch, transmission and wheel angular displacements respectively. Also n is the
overall gear ratio between the engine and driveshafts (gearbox ratio multiplied by final drive ratio).
The equations of motion of the driveline system can be written by making use of the Newton’s second law of motion in
rotation:
If €hf ¼ T f T c ; ð1Þ
It €ht ¼ nT c bt h_ t T d ; ð2Þ
in which bt is the damping coefficient of transmission and Td is the driveshaft torque. The rotational dynamics of each wheel
is of the form:
Td Ft
Iw €hw ¼ rw ; ð3Þ
2 2
where F t is the traction force generated in the tire-road interface and in general is dependent on the tire normal load Fz and
longitudinal slip S (see Section 2.4 below).
Vehicle longitudinal motion that involves the interaction of tractive and resistive forces is related to the wheel dynamics
and driveline dynamics. According to the free-body-diagram of Fig. 3, the equation of motion reads:
mv_ ¼ F t F a F r mg sin b; ð4Þ
where F a and F r are aerodynamic and rolling resistance forces:
1
Fa ¼ :C d :A:q:v 2 ; ð5Þ
2
F r ¼ f R :m:g; ð6Þ
Fig. 2. Driveline model.
Fig. 3. FBD of vehicle longitudinal motion.
in which Cd, A, q, v, fR, b and m are drag coefficient, vehicle frontal area, air density, vehicle speed, rolling resistant coefficient,
road grade angle and vehicle mass.
2.2. Tire slip
The tire slip is influential in damping of the driveline oscillations especially on wet surfaces. The longitudinal tire force is
dependent on the normal force and slip, i.e.,
F t ¼ f ðF z ; SÞ: ð7Þ
The tire slip is defined as follows:
(
v rx if v > rx
v
S¼ r xv
; ð8Þ
rx
if rx > v
Flywheel Clutch plate

Tc Td
ωe Te ωd
To
Fn Fn driveshaf
Fig. 4. Schematic model of a dry-clutch.
Fig. 5. Characteristic of a surface in friction phenomenon [17].
where v and is the velocity of the wheel center in its heading direction and x is the angular speed of tire. The longitudinal
force model of a tire overwhelmingly used in the research works is Pacejka’s magic formula model. The details of tire model
used in this paper are taken from [10].
2.3. Clutch system
In Fig. 4, a schematic representation of a dry-clutch system is presented. In practice two friction surfaces namely fly-
wheel-clutch and pressure plate-clutch are acting in a clutch system. In the simple model of Fig. 4, however, an equivalent
single-surface is used. The clutch actuator produces the normal force Fn and with this force the clutch torque Tc that is trans-
ferred to the driveline is controlled. In Fig. 4, Te represents the engine torque, Td is the load torque reacting by the driveline, l
is the friction coefficient of the clutch-facings material with opposing surfaces, xe and xd are engine and clutch angular velo-
cities, and Tc is the clutch torque.
In a dry clutch system modeling the type of friction model is very influential. Several models including the basic coulomb
friction model, stick-slip friction model [11], static-kinetic friction model [12], polynomial approximation [13,14], hybrid
model [15] and neural network model [16] are proposed by several researches. Haessig and Friedland introduce bristle mod-
el which describes microscopic properties between two surfaces [17]. The contact points between surfaces are taken random
and chaotic. Each point is considered as a bristle with a given stiffness (see Fig. 5).
As this model is complicated and computationally inefficient, they introduced reset integrator model which converts bris-
tle model to a more computationally suitable model. This friction model has two parts: Stiction and sliding. In the stiction
state the friction force is increased according to relative displacement and velocity. The friction force is determined from:
F fric ¼ K stick X þ Brel :V rel ; ð9Þ
where K and B is the stiffness and damping of the surface. In the sliding zone, the friction force is mathematically expressed
as:
3:jVrelj
F fric ¼ f slip þ ðf stick f slip Þ eð astrib Þ þ f :v ;
v ð10Þ
in which fslip, fstick, astrib, Vrel and fv are coulomb friction, stiction friction, stribeck constant, relative velocity and viscous coef-
ficient of damping (see Fig. 6).
In this work reset integrator model is used for modeling of friction coefficient and parameters used for the model are
given in Table 1.
Fig. 6. Friction force versus relative velocity [18].
Table 1
Parameter values for Eq11.
Parameter Symbol Value

Coulomb friction fslip 0.24
Stiction friction fstick 0.3
Stribeck constant astrib 600
Relative velocity Vrel –
Viscous coefficient of damping fv 0
Fig. 7. Backlash between gears.
2.4. Backlash
One of the important factors in generation of unwanted vibration is clearance among driveline components that causes
backlash. This is a nonlinear phenomenon as in the backlash zone no torque transfer is occurred at no contact time and sud-
denly when the two parts of driveline come in contact with each other, a high torque is delivered together usually with an
impact. Some sources of backlash in driveline are clearances between gears in the gear box necessary for a gearbox to work
in high temperature (Fig. 7), couplings and engine mounts. The backlash can be modeled in different ways but a simplified
dead zone model with a clearance angle shown in Fig. 8 is used in this paper.
The torque transfer formulation that represents this phenomenon is:
8
_
< ks ðhrel aÞ þ cs hrel
> if hrel > a
T dz ¼ ks ðhrel aÞ þ cs h_ rel if hrel < a ; ð11Þ
>
:
0 if jhrel j 6 a
where ks and cs are the torsional elasticity and damping of shaft. a is the backlash angle and hrel, the relative angle between
two parts of the shaft is defined as:
hrel ¼ ht hw : ð12Þ
2.5. Model verification
In order to show that the model is reliable for the simulation purposes, it is verified with the Northcote model [7] that
uses information of 1.6 liter Volkswagen Jetta and is validated with measurement results. The parameters of vehicle driveline
used for this comparison are shown in Table 2. The driveline response for a tip-in with engine torque ramp from 10 to 90 Nm
in 0.1 seconds is illustrated in Fig. 9. A close resemblance seen for the two responses verifies the accuracy of response of the
current work.
2α
θt Td θw Td
Fig. 8. A simplified dead zone model for backlash.
Table 2
Vehicle parameters [7].
Name Symbol Value

Road Gradient b 0
Drag coefficient Cd 0.3
Transmission inertia It 0.01 kg.m2
Flywheel inertia If 0.17 kg.m2
Tire inertia Iw 1.00 kg.m2
Tire radius rw 0.32 m
Vehicle mass m 1400 kg
Rolling resistance coefficient fr 0.0136
Overall gear ratio (1st Gear) n 12.98
Overall gear ratio (2nd Gear) n 7.65
Overall gear ratio (3rd Gear) n 5.16
Car frontal area A 2.2 m2
Driveshaft stiffness coefficient ks 6420 N.m.rad1
Drivetrain damping coefficient cs 90 N.m.s.rad1
Backlash angle 2a 0.0785 rad
2.6. Model for controller design
For designing controller, one simple linear model is idealized from vehicle driveline model which has the real specifica-
tions of complete model. Linear control theory can be used with a liner model. The effect of nonlinearities on controller per-
formance is examined later.
The simplified driveline model shown in Fig. 10 consists of two inertias, the engine-flywheel and wheel inertia connected
to each other with a massless torsional spring/damper package. In this model the dynamics of engine and clutch, driveline
frictions and nonlinear effects such as backlash traverse and tire slip are neglected. In addition small inertias such as that of
transmission inertia are ignored in comparison with that of the flywheel inertia.
As a result of this simplification, the equations of motion (Eqs. (1)–(4)) reduce to:
Td
If €hf ¼ T f ; ð13Þ
n
!
hf h_ f _
T d ¼ ks : hw þ c s : hw ; ð14Þ
n n
Northcote model
Current model
Fig. 9. Comparison of model result with that of Northcote [7] for a tip-in input.
If Ic
θf Ratio=n
Tf Tc Td θd ks θw
cs
Tt
Transmission
Fig. 10. Simplified driveline model.
ð2Iw þ mr2w Þ€hw ¼ T d ðF roll rw F aero r w mgr w sin bÞ: ð15Þ

By defining:
Ic ¼ 2:Iw þ m:r2w ; ð16Þ
T t ¼ F roll :r w F aero :r w m:g:rw : sin b: ð17Þ

The above equations further simplifies to:
" !#
€ ks hf cs h_ f _
If hf ¼ T f hw þ hw ; ð18Þ
n n n n
!
hf h_ f _
Ic €hw ¼ ks hw þ cs hw T t : ð19Þ
n n
3. Controller
The control system must attenuate shunt and shuffle in order to improve the drivability. For this purpose the reference
input is the desirable acceleration that the vehicle finally should attain. The highest torque transmitted from driveline to the
wheel is in the first gear, so the worst shuffle and shunt phenomena will also occur in this gear. Fig. 11 compares the drive-
line vibrations in three different gears. The frequency of vibrations is larger for the higher gears whereas the amplitude is
smaller. The design of controller, therefore, will be considered for the first gear. Nonetheless the application of this approach
to other gears is similar.
3.1. Clutch control concept
The main idea is to control the clutch clamp force in a predictive way. When the shuffle vibration starts to occur, the con-
troller decreases the clutch clamp load in order to control the capacity of clutch torque transfer in one hand and on the other
hand to allow slip between the two sides of the driveline.
The prediction of possible vibrations is important for control action. In practice sensors such as wheel speed sensor, accel-
erator pedal sensor and road gradient sensor will be needed. These sensors are also used in other applications like anti-lock
Fig. 11. Vehicle acceleration response to engine torque variation from 10 to 90 Nm in different gears.
Tt 1
G(s)
Ic
u K Tc n y
S Ic
Fig. 12. Block diagram of the system with controller.
braking system, electronic stability program, adaptive cruise control and hill-start assist control. The vibration possibility can
be sensed by electronic gas pedal sensor as when the position of gas pedal changes suddenly (tip-in or tip-out process), the
control system will be activated. Also from the position of gas pedal (and its rate) the desired acceleration that a driver
demands can be estimated. From this acceleration, the maximum of desired transmitted torque Tc and by this torque the
maximum of desired clutch normal force is computed. The control system arranges the normal clutch force to Nc and the
capacity of clutch torque is adjusted:
Tc
Nc ¼ ; ð20Þ
lc :rc
where lc and rc are the clutch dry friction coefficient and equivalent radius respectively. Therefore the vibration is damped
without sacrificing the performance of vehicle.
3.2. Controller structure
As the shuffle, the system output is of acceleration type, the vehicle acceleration feedback is compared with the required
acceleration that the driver demands. The difference as an error is fed to a linear I-type controller shown in Fig. 12. This con-
troller generates a supposition torque T c from which the limitation of clutch normal force N c is estimated and produced by
the clutch actuator. For the transmitted driveline torques in excess of this limit the clutch slips and as a result the vibration is
prohibited.
The transfer functions between output and inputs are of the form:
y K:n
¼ ; ð21Þ
u K:n þ SIc
y S
¼ : ð22Þ
Tt K:n þ SIc
The root of characteristic equation, therefore, is:
K:n
S¼ : ð23Þ
Ic
Insofar as the gain K is positive, the root of the characteristic equation always lays in the left-hand side of the real axis,
also the root locus analysis have shown that the system roots are all in the left hand side and thus the whole system is stable.
3.3. Estimations
For calculating desired normal force applied to the clutch facings by the actuator, the desired acceleration should be esti-
mated first. It is the acceleration that driver demands and it occurs after transient response settles down. For estimating the
desired acceleration the simplified three DOF model (Section 2.6) is utilized. Governing equations in the form of state space
read:
2 3 2 1
32 3 2 3
X1 0 1 n X_ 0 0
6 7 6 76 1 7 6 7
4 X2 5 ¼ 6
ks
cIcs cs
74 X_ 2 5 þ 6 0 I1c 7; ð24Þ
4 Ic nIc 5 4 5
1
X3 nIks
f
cs
nIf
nc2sI
f
X_ 3 I
f
0
y ¼ X_ 2 ¼ €hw ; ð25Þ
Fig. 13. Engine torque variation from 10 Nm to 90 Nm in 0.1 seconds.
where:

hf
X1 ¼ hw ; ð26Þ
n
X 2 ¼ h_ w ; ð27Þ
X 3 ¼ h_ f : ð28Þ
By using the parameter values for the example vehicle, the transfer function between engine and output, also between
load torque and acceleration can be obtained as:
y s2 hw ðsÞ 0:2825s þ 20:15

¼ ¼ 2 ; ð29Þ
Tf T f ðsÞ s þ 3:766s þ 268:6
y s2 hw ðsÞ 0:006927s2 þ 0:02177s þ 1:553

¼ ¼ : ð30Þ
Tt T t ðsÞ s2 þ 3:766s þ 268:6
The steady state response in mathematical form is obtained as time approaches infinity. In Laplace domain this is
achieved by approaching s ! 0:
y
ðssÞ ¼ 0:075018; ð31Þ
Tf
y
ðssÞ ¼ 0:0057819: ð32Þ
Tt
Now by using these two coefficients the desired wheel rotational acceleration can be calculated as:
€hd ¼ u ¼ 0:075018 T f 0:0057819 T t : ð33Þ
The load torque Tt depends on road gradient and wheel speed. By utilizing the sensors and observers, the load torque can
be estimated [19–21]. Eq. (37) produces the desired acceleration.
Fig. 14. Vehicle response with and without controller for no-backlash case.
Fig. 15. Clutch torque variation.
The linear I-type controller generates a supposition torque T c . If the transmitted driveline torque is in excess of this torque
value, the clutch slips and as a result the vibration is prohibited. For calculating the clutch clamp force from Tc Eq. (20) is
used. After calculating this force, the clutch actuator is responsible for performing the control action. If the clutch is
disengaged equations for calculating angular acceleration is rewritten as:
€hw ¼ x_ 2 ¼ nT c T t ; ð34Þ
Ic Ic
Tt and Ic are introduced in Equations (16) and (17).
3.4. Controller response
To investigate the response of the controller, the tip-in and tip-out scenarios are considered. The first is the situation
when a driver is accelerating slowly at the beginning, and then decides to accelerate abruptly. The second scenario is when
the vehicle is accelerating and the driver fully releases the throttle pedal at once.
3.4.1. Tip-in
Two types of simulations one without a backlash and the other with a backlash will be presented. This situation can be
sensed by the rapid changes of the gas pedal position. In this situation the driveline does not pass through the backlash zone
so the shunt and shuffle oscillations are caused by driveline elasticity. This situation can be modeled as an engine torque
ramp from 10 to 90 Nm in 0.1 seconds as depicted in Fig. 13.
For demonstrating a preliminary system response it is considered that vehicle has travelled for 6 seconds and at this
moment the driver applies the accelerator input. For the first scenario the outputs including the vehicle response, clutch tor-
que, clutch force and clutch slip are provided in Figs. 14–17. The controller gain in this maneuver is adjusted to K = 250. It is
clear that the controller is able to adjust the clutch torque and prevent driveline oscillations. In fact when there is no control,
then the clutch is fully engaged and driveline is subjected to the oscillations. In case of the controlled clutch, the clamp force
is eased (Fig 16) to allow clutch to slip (Fig 17) while transferring the required torque (Fig 15). The clamp load is fully
returned when the slip is very small (clutch is reengaged).
The second case is when the vehicle is decelerating under the action of resistive loads and suddenly the driver decides to
accelerate quickly. In this situation the throttle is closed first and then the driver depresses the gas pedal rapidly. This can be
Fig. 16. Variation of clutch normal force during tip-in without a backlash.
Fig. 17. Clutch slip during tip-in without a backlash.
modeled as an engine torque ramp from 10 to 70 Nm in 0.1 seconds illustrated in Fig. 18. In this case the driveline encoun-
ters the backlash zone so the oscillations are caused by both backlash traverse and driveline elasticity.
For the second scenario the outputs including the vehicle response, clutch torque, clutch force and clutch slip are provid-
ed in Figs. 19–22. The controller gain in this maneuver is adjusted to K = 150. It is clear that the controller is able to adjust the
clutch torque and prevent driveline oscillations.
In practice driveline control system must be able to differentiate between the two cases. During the tip-in or tip-out pro-
cess if at the beginning the vehicle acceleration is positive it indicates that no backlash will be faced and the controller will
switch to the first control algorithm. When the acceleration is negative or zero before accelerator pedal is depressed, the con-
trol system can determine that the driveline is facing a backlash zone and it activates the second mode of controller.
3.4.2. Tip-out
In this case the throttle is fully open first and the driver releases the gas pedal rapidly. This situation is modeled as the
engine torque ramp from 90 to 10 Nm in 0.1 seconds as shown in Fig. 23. With this input the driveline encounters the back-
lash zone, so the oscillations are caused by both the backlash traverse and the driveline elasticity.
For this case the responses including the clutch torque, clutch force and clutch slip responses are provided in Figs. 24–27.
The controller gain in this maneuver is adjusted to K = 20.
4. Gain program
The clutch life is an important issue and it is necessary to account for it in controller design process. The clutch life is
inversely proportional to its slip. In fact the area under the clutch slip curve (e.g. Fig. 22) is an indication of the clutch wear.
Smaller the area means lesser the clutch wear. In the previous section the proportional gain was set at constant value and the
effect of gain variation during the clutch action will be demonstrated in this section.
4.1. Driveline without backlash
The response properties of the driveline without backlash for different gain values are shown in Figs. 28 and 29. Fig. 28
shows that for larger gains an overshoot is present and for smaller gains no overshoot is occurred. On the other hand, as can
be seen from Fig. 29 larger gains cause lower clutch slips. So the decrease of oscillations is achieved in expense of more clutch
wear.
Fig. 19. Vehicle response with and without controller for backlash case.
Fig. 20. Clutch torque variation during tip-in with a backlash.
Fig. 21. Variation of clutch normal force during tip-in with a backlash.
Fig. 22. Clutch slip during tip-in with a backlash.
Fig. 24. Vehicle acceleration response with and without controller for tip-out.
Fig. 25. Clutch torque for tip-out process.
Fig. 26. Variation of clutch normal force.
Fig. 27. Clutch slip in tip-out process.
Fig. 28. Vehicle acceleration at different gains for no backlash.
Fig. 29. Clutch slip at different gains for no backlash.
Table 3
Effect of gain variation on the driveline response properties.
Gain value Overshoot (%) Time of overshoot (s)

500 27.0 0.264
350 21.0 0.220
330 18.0 0.210
290 11.0 0.192
270 7.0 0.183
260 2.3 0.180
235 – –
Fig. 30. Vehicle acceleration with the gain scheduling controller (first scenario).
Fig. 31. Variation of the gain with time (first scenario).
In order to minimize the clutch wear, controller gains during clutch engagement control can be scheduled. By making use
of a switching controller the gain can be kept small at the beginning in order to prevent the shunt and it is increased as time
elapses. Various simulations for the example vehicle showed that 0.3 second duration for the low gain portion suffices
because the shunt phenomenon occurs in less than 0.3 seconds for the case without backlash zone. According to Table 3,
the gain value as large as K = 235 is found to be appropriate for this duration.
After this duration the gain can rapidly increase to decrease the clutch wear. Variation of the gain during this period can
be achieved with different functions. In this paper the linear function is found to be equally appropriate. The gain program,
therefore, is of the form:

K ¼ 235 for t 6 0:3
: ð35Þ
K ¼ 235 þ 400 ðt 0:3Þ for t > 0:3
The slope 400 was obtained by trial and error. Simulation results by the application of this gain program are illustrated in
Figs. 30–32 for the first scenario. It can be seen that this technique has caused the clutch wear to decrease to a large extent.
Compared to Fig. 22, Fig. 30 shows that the area under the curve is reduced almost 10 times. This means a significant reduc-
tion in the clutch wear.
Fig. 32. Clutch slip with gain scheduling controller (first scenario).
Fig. 33. Vehicle acceleration with active switching and gain scheduling controller (second scenario).
Fig. 34. Variation of the gain with time (second scenario).
Fig. 35. Clutch slip (second scenario).
4.2. Driveline with a backlash
It was shown in previous section that the backlash traverse intensifies the driveline vibration. The controller designed for
the case without the backlash is not able to eliminate the shuffle for this case. In this part, too, a switching controller is used
in which when system enters into the nonlinear backlash zone, the controller keeps the gain small so that a smooth transfer
is achieved. After the backlash zone is traversed the controller increases the gain and permits more torque passing.
Examinations for the example vehicle indicated that the time period driveline encounters the backlash zone is always less
than 0.05 seconds. The low gain period is extended 0.15 seconds then it is increased linearly:

K ¼ 125 for t 6 0:15
: ð36Þ
K ¼ 125 þ 300 ðt 0:15Þ for t > 0:15
Fig. 36. Vehicle acceleration with and without the actuator delay.
Fig. 37. Clutch slip with and without the actuator delay.
The responses of vehicle driveline when a backlash is present with application of the switching controller during a tip-in
are shown in Figs. 33–35. For this case, too, responses are desirable in terms of both oscillation elimination and clutch wear.
5. Discussions
In this section the sensitivity of the designed system to the two issues namely the actuator dynamics and the uncertainty
of clutch friction is investigated.
5.1. Actuator dynamics
The delay associated with the actuator dynamics may impair the system response. For this reason the response of actua-
tors must be faster than the system itself. In the controller design process it was assumed that the actuator is fast enough
that is has no effect on the overall system response. For the simulation, however, a first order delay in the form of:
y ¼ y0 ð1 eTt Þ; ð37Þ
Fig. 38. Acceleration response to uncertainty in the clutch friction.
Fig. 39. Clutch slip response to uncertainty in the clutch friction.
with y0 as the control command and y as the actuator output, was considered. A large time constant of T = 0.1 second was
shown to have no significant effect on the system output for the tip-in case, as may be seen in Fig. 36. The clutch slip shown
in Fig. 37 becomes larger due to the actuator delay.
5.2. The clutch friction uncertainty
Some parameters such as the value of the clutch friction are uncertain and this may lead to improper system response. In
order to investigate the effect of change of the clutch friction on the system response, a 10% variation in this property is con-
sidered in the tip-in case and the results are illustrated in Figs. 38 and 39. It can be seen that the system output is not chan-
ged noticeably due to the uncertainty of the clutch friction. In fact the clutch slip and in turn the clutch force are adjusted to
the available friction.
6. Conclusions
The concept of a dry clutch control for the reduction of driveline vibrations was studied. A model incorporating the dry
clutch, nonlinear backlash traverse and tire slip was developed and verified for the simulation of vehicle driveline dynamics.
A linear controller was designed and implemented in the simulation package. Information inputs to the controller are the
wheel speed, accelerator pedal and road gradient. It was shown that a constant gain was effective in the elimination of large
driveline oscillations but in expense of large clutch wear. A gain program was then introduced in the controller in order to
adjust the gain during the clutch action. It was shown that such gain program reduced the clutch wear to a large extent and
increased the clutch lifetime. The application of this system was examined for two cases of driveline dynamics with and
without backlash. For both cases it was shown that the control system was successful in eliminating the vehicle longitudinal
vibrations. The advantages of this approach compared with the other solutions are lower cost, low development time and
easy implementation, especially in cars with automated manual transmission (AMT) system.
References
[1] A. Balluchi, L. Benvenuti, C. Lemma, P. Murrieri, A. L. Sangiovanni-Vincentelli, Hybrid models of an automotive driveline, Tech. rep., Parades, Rome, I,
Dec 2004.
[2] M. Berriri, P. Chevrel, D. Lefebvre, Active damping of automotive powertrain oscillations by a partial torque compensator, Control Eng. Pract. 16 (2008)
874–883.
[3] J. Fredriksson, H. Weiefors, B. Egardt, Powertrain control for active damping of driveline oscillations, Vehicle Sys. Dyn. 37 (2002) 359–376.
[4] A. Lagerberg, B. Egardt, Evaluation of control strategies for automotive powertrains with backlash, in: AVEC ‘02, Intl. Symp. on Advanced Vehicle
Control, September 2002, Hiroshima, Japan.
[5] D. Lefebvre, P. Chevrel, S. Richard, An H-infinity-based control design methodology dedicated to the active control of vehicle longitudinal oscillations,
IEEE Trans. Control Sys. Technol. 11 (2003) 948–956.
[6] M. Pettersson, L. Nielsen, Diesel engine speed control with handling of driveline resonances, Control Eng. Pract. 11 (2003) 319–328.
[7] N.M. Northcote, The modeling and control of an automotive drivetrain (Ms thesis), Department of mechanical engineering, University of Stellenbosch
April 2006.
[8] Y. Kou, F. Weslati, Development of a hybrid powertrain active damping control system via sliding mode control scheme, SAE Technical Paper 2013–01-
0486, 2013.
[9] F.U. Syed, M.L. Kuang, Y. Hao, Active damping wheel-torque control system to reduce driveline oscillations in a power-split hybrid electric vehicle, IEEE
Trans. Vehicular Technol. 58 (2009) 4769–4785.
[10] H.B. Pacejka, Tyre and Vehicle Dynamics, second ed., Butterworth-Heinemann, 2006.
[11] G. Lucente, M. Montanari, C. Rossi, Modelling of an automated manual transmission system, Mechatronics 17 (2007) 73–91.
[12] G. Naus, M.A. Beenakkersa, R.G.M. Huismanb, M.J.G. van de Molengrafta, M. Steinbuch, Robust control of a clutch system to prevent judder-induced
driveline oscillations, Vehicle Sys. Dyn. 48 (2010) 1379–1394.
[13] D. Centea, H. Rahnejat, M.T. Menday, Non-linear multi-body dynamic analysis for the study of clutch torsional vibrations (Judder), Appl. Math. Model.
25 (2001) 177–192.
[14] M. Saeedi, M.A. Saeedi, R. Kazemi, M. Rafat, A.H. Pasdar, Real World Modeling and Nonlinear Control of an Electrohydraulic Driven Clutch, Int. J.
Automot. Eng. 2 (2012) 107–116.
[15] R. Nouailletas, E. Mendes, D. Koenig, Hybrid modeling and identification of dry friction systems, application to a clutch actuator, Control Eng. Pract. 18
(2010) 904–917.
[16] A. Senatore, V. D’Agostino, V. Petrone, Experimental investigation and neural network prediction of brakes and clutch material frictional behaviour
considering the sliding acceleration influence, Tribol. Int. 44 (2011) 1199–1207.
[17] H. Olsson, K.J. Åström, C. Canudas de Wit, M. Gäfvert, P. Lischinsky, Friction models and friction compensation, Eur. J. Control 4 (1998) 176–195.
[18] LMS Imagine.Lab AMESim, <http://www.lmsintl.com/LMS-Imagine-Lab-AMESim>.
[19] K. Jo, J. Kim, M. Sunwoo, Real-time road-slope estimation based on integration of onboard sensors with GPS using an IMMPDA filter, IEEE Trans. Intell.
Transp. Syst. 14 (4) (2013).
[20] H. Ohnishi, J. Ishii, M. Kayano, H. Katayama, A study on road slope estimation for automatic transmission control, JSAE Rev. 21 (2000) 235–240.
[21] Y. Sebsadji, S. Glaser, S. Mammar, J. Dakhlallah, Road slope and vehicle dynamics estimation, in American Control Conference, 2008, pp. 4603–4608.

Mashadi 2015

Încărcat de

Informații document

Titlu original

Drepturi de autor

Formate disponibile

Partajați acest document

Partajați sau inserați document

Opțiuni de partajare

Vi se pare util acest document?

Este necorespunzător acest conținut?

Drepturi de autor:

Formate disponibile

Mashadi 2015

Încărcat de

Drepturi de autor:

Formate disponibile

Applied Mathematical Modelling xxx (2015) xxx–xxx

Contents lists available at ScienceDirect

Applied Mathematical Modelling

Driveline oscillation control by using a dry clutch system

⇑ Corresponding author. Tel.: +98 (21)77240360; fax: +98 (21)73021560.

Fig. 1. Shunt and shufﬂe phenomena.

Fig. 2. Driveline model.

Fig. 3. FBD of vehicle longitudinal motion.

2.2. Tire slip

Flywheel Clutch plate

Fig. 4. Schematic model of a dry-clutch.

Fig. 5. Characteristic of a surface in friction phenomenon [17].

2.3. Clutch system

F fric ¼ K stick X þ Brel :V rel ; ð9Þ

Fig. 6. Friction force versus relative velocity [18].

Parameter Symbol Value

Fig. 7. Backlash between gears.

2.5. Model veriﬁcation

Fig. 8. A simpliﬁed dead zone model for backlash.

Name Symbol Value

2.6. Model for controller design

ð2Iw þ mr2w Þ€hw ¼ T d ðF roll rw F aero r w mgr w sin bÞ: ð15Þ

Ic ¼ 2:Iw þ m:r2w ; ð16Þ

T t ¼ F roll :r w F aero :r w m:g:rw : sin b: ð17Þ

3.1. Clutch control concept

Fig. 12. Block diagram of the system with controller.

3.2. Controller structure

Fig. 13. Engine torque variation from 10 Nm to 90 Nm in 0.1 seconds.

y s2 hw ðsÞ 0:2825s þ 20:15

y s2 hw ðsÞ 0:006927s2 þ 0:02177s þ 1:553

Fig. 15. Clutch torque variation.

Tt and Ic are introduced in Equations (16) and (17).

3.4. Controller response

Fig. 17. Clutch slip during tip-in without a backlash.

4.1. Driveline without backlash

Fig. 18. Engine torque variation from 10 Nm to 70 Nm in 0.1 seconds.

Fig. 20. Clutch torque variation during tip-in with a backlash.

Fig. 22. Clutch slip during tip-in with a backlash.

Fig. 23. Engine torque variation from 90 Nm to 10 Nm in 0.1 seconds.

Fig. 25. Clutch torque for tip-out process.

Fig. 26. Variation of clutch normal force.

Fig. 27. Clutch slip in tip-out process.

Fig. 28. Vehicle acceleration at different gains for no backlash.

Fig. 29. Clutch slip at different gains for no backlash.

Gain value Overshoot (%) Time of overshoot (s)

Fig. 31. Variation of the gain with time (ﬁrst scenario).

Fig. 34. Variation of the gain with time (second scenario).

Fig. 35. Clutch slip (second scenario).

4.2. Driveline with a backlash

5.1. Actuator dynamics

Fig. 38. Acceleration response to uncertainty in the clutch friction.

Fig. 39. Clutch slip response to uncertainty in the clutch friction.

5.2. The clutch friction uncertainty

S-ar putea să vă placă și