Sunteți pe pagina 1din 34

Modelling and control of automotive clutch systems

M.H.M. Dassen
Reportnumber 2003.73

Supervisor: A.F.A.Serrarens

Eindhoven, 22nd July 2003 Department of mechanical engineering TU/e Eindhoven

1 Introduction 1.1 Clutch control . . . 1.2 Literature overview 1.3 Problem definition 1.4 Objectives . . . . .

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

3 4 4 6

6
7 7 8 8 9 9 9 10 10 12 13 14 17 17 18 19 20 24

2 Ciutch modeiiing 2.1 Driveline Components . . . . . . . . . . . . . . . . 2.1.1 Clutch . . . . . . . . . . . . . . . . . . . . . 2.1.2 Engine . . . . . . . . . . . . . . . . . . . . . 2.1.3 Gearbox and axles . . . . . . . . . . . . . . 2.1.4 Wheels and external forces . . . . . . . . . . 2.2 Driveline model . . . . . . . . . . . . . . . . . . . . 2.3 Equations slipping and sticking case . . . . . . . . . 2.3.1 Differential equations with reduced matrices 2.3.2 State space form . . . . . . . . . . . . . . . 2.3.3 Karnopp's approach . . . . . . . . . . . . . 2.4 Simulation model . . . . . . . . . . . . . . . . . . .
3 Controller design 3.1 Requirements of the controller . . . . . . 3.2 Decoupling controller . . . . . . . . . . . 3.3 References for engine and slipping speed 3.4 Simulation with decoupling controller . . 3.5 Analyzing the decoupling controller . . .

. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . . . . . . . . . . .

4 Conclusions and recommendations 26 4.1 Conclusions . . . . . . . . . . . . . . . . . . . . . . . . . . . . 26 4.2 Recommendations . . . . . . . . . . . . . . . . . . . . . . . . . 26

A Sirnulink model

28

B Matlab switching function for Simulink model

ntroduct ion
In this report the dynamic behavior of an automotive clutch is analyzed using dynamical models. The engagement of the clutch is then simulated using the acquired model. Based on this analysis a control strategy is suggested and the designed control law is implemented. Finally, the performance of this controller is evaluated.

Figure 1.1: Automotive clutch (Luk)

1.1 Clutch control


Clutches in cars, trucks and other vehicles are gradually engaged after selecting the different gears, avoiding unpleasant shocks and jerks. A clutch has two plates that can be joined together by an actuator that exerts a force on one of the two plates, see figure 1.1. This plate is called the pressure plate, the other plate, the friction plate is connected to the engine crank shaft. The pressure plate is connected by an axle to the gear box and the remaining part of the powertrain. As the clutch engages the plates are pushed together by the actuator. When the plates touch, torque is transmitted from the engine to the drivetrain. The vehicle now starts to move. After a limited amount of time the speeds of the two plates will become equal. The plates are then sticking and the engine is directly connected to the drivetrain. To achieve a successful engagement, the right input force has to be applied by the actuator. This can be done by the driver through a foot pedal or automatically by a programmed actuator force. An advantage of controlling a clutch automatically is of course relieving the driver of the gear shifting task, and of the (dis)engagement of the clutch. But also an automatic clutch can be optimized further. For example wear can be predicted more accurately, because the forces acting on the clutch components are know. Also fuel consumption during engagement can be minimized and the time it takes t o engage the clutch can be shortened.

1.2

Literature overview

In the field of clutch control and drivetrain modelling a lot of research has been carried out recently. Next, several research articles on these subjects will be discussed briefly:
e

Optimal control of gear shift operations in automatic transmissions, A.Waj-Fraj, F.Pfeiffer (2001) In this paper an optimal control approach for gear shift operations is presented. On basis of an optimization criterion an analytical control law is derived and simulated.

H" control design for torque-converter-clutch slip system,


R-Hibino et al.

Hw control theory was applied systematically in this article to ensure


robust stability. This method allows for a detailed design in the frequency domain.

Karnopp, D. C., Margolis, D. L., and Rosenberg, R. C., System Dynamics: A Unified Approach. (1990) In this book a method for analyzing dynamics using bond graphs is presented. Dynamic modelling and Control of a Car Transmission System, R.Zanasi et al. (2001) In this article a car transmission is modelled using Power oriented graphs. A simple control strategy is proposed and implemented in simulation. Smooth engagement for automotive dry clutch, F. Garofalo et al. (2001) Based on a reduced model a controller is designed which decouples the slip speed and engine speed control. With this controller clutch engagement is simulated. Model and control of a wet plate clutch, M.J.W.H. Edelaar (1997) This report describes modelling and control for a wet plate clutch. The model is validated with experiments.

Stop-Go with the ZI@-powertrain: a first glance, T.C.P. Verhagen (2000) The start and stop process of a vehicle with a zero inertia powertrain is analyzed. System models are developed. Dynamic modeling and control of hybrid electric vehicle powertrain systems, B.K. Powell, K.E. Bailey and S.R. Cikanek In this article a dynamic model of a powertrain is derived, control is also discussed. A study of shift control using the clutch pressure pattern in automatic transmission, W. Han and S.- J. Yi The shift transient characteristics of an automatic transmission with Ravigneaux planetary gears are investigated. Also a shift control strategy using engine torque reduction and optimum pressure trajectory is discussed. The influence of the interface coefficient of friction upon the propensity to judder in automotive clutches, D. Centea, H. Rahnejat and M.T. Menday This paper describes the causes of torsional vibrations in the driveline,

referred to as judder. Simulations of a nonlinear multibody dynamic model are found to agree experimental results.

1.3

Problem definition

The problem of controlling an automotive clutch can be stated as follows: Specify an input force that results in a smooth, but fast engagement of the clutch.

1.4
e

Objectives
Describe the engagement of an automotive wet and dry clutch in a dynarnical model.

The objectives of this exercise are:

a Simulate and analyse clutch engagement with the model.


o

Design an adequate controller for smooth clutch engagement based on this model. Using the derived input force, engage the clutch within a restricted time window.

In this chapter basic equations describing different parts of the drive line are presented. The drive line model is then written into three different forms. The structure of the model is depicted in figure 2.1.

Figure 2.1: Drive line model

2.1

Driveline Compbnents

To adequately control the clutch, a model that can accurately predict the response of the driveline to a specified clutch input is needed. In order to achieve this, a simplified dynamical model for the driveline is derived, as

done in [1,2]. For this model the various driveline components need to be specified. These include: the clutch itself, the engine, the axles, gearbox, the wheels. Also external forces on the vehicle and driveline need to be included in the analysis. First, a clear understanding of clutch operation is necessary.

2.1.1

Clutch

The function of an engaging friction clutch is to transmit torque gradually, to avoid high accelerations or jerks, when the engine is connected to the rest of the driveline. This torque is transferred from the engine through the pressure plate onto one or more friction plates connected to the transmission input shaft. These plates are pushed apart by a diaphragm springs. When the clutch is fully disengaged the clutch plates are not in mechanical contact and no torque is transmitted. When the clutch is engaged a normal force on the pressure plate pushes the clutch plates towards each other. The clutch now starts slipping and an increasing amount of torque can be transmitted. The velocities of the plates will become equal and the piates stick. The ciutch is now fully engaged and can transmit torque from the engine up to the maximum static friction torque. There are two main types of automotive clutches: wet and dry clutches. In wet clutches the plates are immersed in and actuated by oil, see figure 1.1; in dry clutches this oil is of course absent. In dry clutches actuation is mostly performed by mechanical mechanisms. To describe the clutch mathematically we take the direct normal force Fn on the pressure plate as input. transferred by the slipping clutch using the We then describe the torque TcL Coulomb friction model:

T C L = FnpRa sign(+^

@ , )

(2.1)

In which p is the friction coefficient and R, the active surface of the clutch plates. When the clutch sticks the maximal amount of torque that can be transmitted equals:
z~ma,

= Fnm,,NpRa s i g n ( ~ 1 - 9,)

P2)

2.1.2

Engine

The engine torque Te is assumed to be a quadratic function of the rotational speed of the crank shaft. This represents a crude approximation of the real engine map, with the maximum torque of the engine Tma,a t the The range for Te is given by: corresponding engine speed @l,,,,.

2.1.3

Gearbox and axles

The gearbox and axles are connected t o the friction plate of the clutch through a torsional spring. These components are modelled as one inertia. The gear ratio selected within the gearbox is denoted by i,, the remaining reduction after the gearbox as id. The overal transmission ratio in the model is then written as:
Z

. = Zg

'

Zd

(2.4)

2.1.4

Wheels and external forces

The two powered wheels (usually the front whee1s)are described with a single inertia J3. A load torque acts on the vehicle inertia J,, which also includes the driven wheels inertia. This torque equals: Tl = FeZt&. In which R , is the wheel radius. The force Fext consists of the air drag F , : the rolling resistance of the driven wheels FTdrZven and the grade forces F,. Furthermore FTdrZvtng acts on the driving wheels icertia J3.

In these equations, y is the air density; Af the frontal surface area of the , the speed of the vehicle; 6 the angle of the vehicle with respect vehicle; V to the horizontal; m, the vehicle mass; ct: indicates the weight distribution on the front and back wheels respectively (0 < ct: < 1); g the gravitational constant; x, the arm of-the rolling resistance.

2.2

Driveline model

Combining the described parts a dynamical model of the driveline is obtained. See figure 2.1. The engine torque T, is directly applied on the engine inertia J1 with corresponding angle 91. This inertia is connected to the clutch pressure plate through an axle which is assumed to be rigid. The torque transferred The by the clutch Tclacts on the friction plate with inertia J, and angle 9,. torsional spring with stiffness kl attach the clutch to the gearbox with inertia J2and angle p2. The torsional stiffness of the rest of the driveline, i.e. the acts on drive shafts, is incorporated into k2. The rolling resistance FTdrZOZng

the inertia of the driving wheels J3with angle p3. The torque a t the wheels T , is equal to the force: kz (iy2 - 93). The inertia of the rest of the vehicle is projected onto J,. The load torque TLacts on this inertia. The traction of the tyres to the road is taken into account by a linear damping force with damping coefficient b acting between the wheel inertia and vehicle inertia. For this system the equations of motion can be derived as:

2.3

Equations slipping and sticking case

The equations presented in the previous paragraph can be formulated in different ways. Each description of the system aims a t uniting the slipping and sticking systems into a single system formulation. The difficulty that arises here is the apparent change in the number of degrees of freedom. When sticking occurs the inertias J1 and J, can be described by one coordinate instead of two in the slipping phase. This property complicates describing the system mathematically. In this section three possible forms are presented.

2.3.1

Differential equations with reduced matrices

To incorporate the two phases of slip and stick into one model the equations of motion can be manipulated using reduced matrices. As stated by Verhagen [3], the equations (2.10) to (2.14) are written in matrix form: from which this differential equations results for the slipping phase, - +,I > 0): With the generalized displacement column q = [cpl cp, p2 cps cp,]T. Herein g j represents the jth unit vector: el= [l 0 0 0 0IT. These unit vectors are used to insert the external torques into the model. The clutch torque is applied by the vector h:

When the clutch is engaged the degrees of freedom of the system are reduced, as cpl and 9, are now equal. This is can be denoted by:

This represents a kinematic constraint. A vector qT with four instead of five components is introduced to take into account that cp, is no longer a degree of freedom. The original displacement column is multiplied with the reduction matrix RT to obtain the reduced displacement vector.

The clutch is set to close at time t T :which implies RTjT (t,) . Let a matrix QT satisfy the condition:

hTq(t,) = 0 and Q ( t T = ) -

then
-T

Y (1,)

&Ti(&)

(2.20)

To determine q for t > t,, q is integrated. As the condition for stick is y will appear in the formulated in terms of velocity, an integration constant -T position q ( t ).

If we take -T q (t,) to be:

then
-T

[I - RTQT]q(tT)

(2.24)

Substituting qT into equation (2.12) and

Pre-multiplication with RT,

(RTh = 0 ) yields the equation:

The matrices RT and QT are found by making sure they reduce the displacement column correctly. And QT and R, should satisfy the conditions mentioned above in equations (2.18) and (2.19). In the actual case these matrices become:

The model obtained here transforms the slipping equations into the sticking system through matrix multiplications. This results in an actual reduction in the order of the model. In essence this is correct, as the actual system indeed looses a. degree of freedom, however it makes implementation in a simulation model less compact.

2.3.2

State space form

The system can also be described in the state space form x = Ax Bu, as described in 141. The state vector x is defined in terms of the generalized coordinates. The term k2 (ip2- p3) represents the torque transmitted by the axle stiffness k2, the torque acting on the wheels. The input u for this and the engine torque Te: clutch model is the clutch torque TcL

Writing the system in the form x = Ax+ B u allows for a model with switching parameter d with d = 1 if (c.-, -kc > 0 and d = 0 if the slipping speed becomes zero, i.e. - (c., = 0

x = d . A,LX-+ (1- d) . A,,x

+ d . BsLu+ (1 - d) . BStu$ f (t)

(2.29)

The matrices ASLand BsI represent the system when the clutch is slipping while, Ast and BSt are the system matrices for the sticking system. The The matrices for both cases were column f (t) contains the load torque found t o be:

r.

The column of forces f ( t )acting on the system only contains resented as follows:

and is rep-

This model adequately describes the system in state space form. A disadvantage of this method is that essentially the same equation is solved twice as pl and 9, are equal in the sticking case. Another formulation of the model that does not have this drawback will be discussed next.

2.3.3

Karnopp's approach

The previous two formulations described two systems within one mathematical description. The sticking and slipping system however can also be described within one expression according to Karnopp [5]. As the system equations (2.10) t o (2.14) are also valid for the sticking system, this expression can be used. However when the clutch locks, increasing Fn does no longer influence the torque Tcltransmitted through the clutch, so the torque transmitted at the interface of J1 and J, has to be substituted for Tcl. The clutch system is seen as the two migrated inertias J1 and J,. The torque TcI, that has to be substituted can then be found from equations (2.10) and (2.11):

Consequently when the clutch sticks, the torque Tclabruptly changes from the torque as result of Fnto the torque in sticking case derived above. The advantage in using this formulation is that the same set of equations is used for both the slipping and the sticking phases. Consequently there is no

Englne and clutch dlsk speed

time [s] Silpplng speed LO$,


I
I

Figure 2.2: Engine and clutch disk speed during engagement switching required within this set. Only the torque Tcltransmitted by the clutch changes as the clutch sticks.

2.4

Simulation model

In order to get a better understanding of the properties of the clutch model, a simulation of this system was carried out in Matlab Simulink, see appendices A and B. In this model the system formulation according to Karnopp, as presented in section 2.3, is used. These equations are implemented in a function that incorporates switching behavior of the torque at the clutch interface TCI. Stick is detected by observing the difference between the engine and clutch speeds. If this difference becomes smaller than 0.01 rad/s the clutch is assumed to stick. To engage the clutch in the simulation model, this torque is implemented as a step function after 0.5 seconds with a magnitude of 52 Nm. The engine torque T,is taken to be constant a t 50 Nm from 0.4 seconds, the initial engine speed is assumed to be 100 rad/s. The vehicle has no initial velocity. Engaging first gear with the clutch is now simulated. The resulting engine and clutch disk speed are plotted in figure 2.2. As can be

Torque transmitted fhroug the clutch during engagement

Figure 2.3: Torque at the clutch interface (Tcl) observed in figure 2.1 the difference between the engine speed and the clutch disk speed decreases to zero. The clutch is then assumed to be engaged. The specified value of 52 Nm is chosen small because larger values would decrease the engine speed too much, resuIting in an engine stall. This low value for TcL also results in a rather slow engagement of the clutch. The torque a the clutch interface is plotted in figure 2.3. It clearly switches from the specified step function to the torque in the sticking phase derived in equation (2.33). This torque level is of course lower then the torque in the slipping phase otherwise the clutch is not able to engage and stay engaged. It is also clear that the torque in the sticking phase is lower than the engine torque T,, as kinetic energy is stored by the engine and clutch disk inertias J1 and J,. The oscillation occurring in the torque is caused by the clutch disk springs kl and the axle stiffness kz. In the actual drive line this vibration will damp out much faster as axles will have some damping. This damping is not modelled, only damping in the tyres is taken into account. The torque at the driving wheels is plotted in figure 2.4. It reaches a maximum value in the slipping phase, as at this time the maximal torque is transmitted by the clutch. The torque then drops to a lower level after the clutch sticks and then remains constant. Again the vibrations caused by the springs damp out very slowly.

Tomue at the driving wheels

3
time Is]

Figure 2.4: Torque at the wheels (T,)

ontroller
In this chapter a controller for clutch engagement will be designed, on basis of the clutch model presented in the previous chapter. The difficulty in designing a controller for this system lies in the variable degree of freedom of the system.

3.1

Requirements of the controller

To guarantee successful clutch engagement two conditions, as seen in [ 4 ] , have t o be satisfied:


e

The engine speed during engagement must satisfy the no-kill condit i o n . i.e. The engine should not stall. If the force F, applied to the clutch disk, and consequently the transferred torque Tcl is very high, the engine speed could drop too severely resulting in engine stall. This condition can be stated as :

In order to maintain a decent level of driver comfort, clutch engagement should be as smooth as possible. The main source of driver discomfort are oscillations in the drive train. These oscillations depend on the time derivative of the slip speed. The slip speed is defined as:

If for now, we assume an constant engine torque T,, we can formulate the no-lurch condition as follows:

With C a positive constant. These requirements however are in conflict with a fast engagement of the clutch. If we want to engage the clutch within a limited amount of time, we need to apply an amount of torque that could cause a high deceleration of the clutch disk (;j, that induces unwanted oscillations in the drive train. A l s ~ a too fast engagement could cause an engine stall. A compromise between the different requirements and the desired engagement time has to be made. The input variable used to control the clutch is first of all the torque transmit, on the pressure ted by the clutch Tcl.Created by applying the input force F plate. Secondly the engine torque Te can be seen as an input variable.

3.2

Decoupling controller

To avoid vibrations in the system the derivative of the slip speed (F, - F2) should be continuous. As described by Garofalo et al. 1 4 1 a controller that w,l can be derived. independently controls the engine speed and slip speed With this controller it should be possible to prevent discontinuities in the difference between F, and Fl. To derive the controller the axes will be assumed to be perfectly rigid, the stiffness terms will therefore be discarded. The following equations are found:

After transformation of these equations to the Laplace domain the engine and clutch speed can the be written as:

With Jd the combined inertias of the driveline:

NOWthe slipping speed w S l ( s ) i.e. the difference between the engine and clutch disk velocities is found to be:

With this in mind the system can be written as follows:

By choosing the following controller the two speeds can be decoupled and controlled separately:

The obtained decoupling controller can now be substituted into equation (3.11). This results in the following expressions for the engine and slipping velocities:

in which

This system represents a decoupled system, in which independent control of both engine and slip speed should be possible by substituting for example PID controllers for u1 and uz. A simple representation of this approach in schematic form is given in figure 3.1.

3.3

References for engine and slipping speed

The references for the engine speed a linear function was chosen starting at 100 rad/s and rising with 75 rad/s. As reference for the slipping speed w,l a function that goes to zero within a certain time frame and then remains zero has to be chosen. Furthermore the initial value of this reference has to be equal to the initial engine speed. This ensures that there is no step in the reference and it matches the initial difference between engine and clutch disk speed. Another important factor in choosing this function is the time in which it reaches zero, because this represents the desired time in which the clutch should stick. Subsequently as reference function for the slipping speed

Figure 3.1: Scheme decoupling controller


wsl

a function is chosen that goes exponentially to zero and has a value of 100 rad/s at, t = 0: W,L = 100 . Ct2 (3.15)

3.4

Simulation with decoupling controller

To test the designed controller, simulations were carried out using the simulink l and model. As controllers for the decoupled systems a PID controller for u v2 were chosen. To find out which value for the time constant 7- is the most three values were tested. For constants 7- = 2 , 5 , 8 simulations were carried q and u2 specifically for every constant. The out, with tuned controllers , resulting slipping speeds are plotted in figure 3.2. Clearly the differences between the different values for r and thus wslref are small. What does become clear however is that the shortest time in which the clutch can be engaged without violating the no kill condition (3.1) is about 1.5 seconds. With this knowledge it seems relevant to take a wslref which reaches zero in 1.5 seconds, this translates into a r of 5. So further simulations were carried out with the 7- = 5.

sl~p speeds for; = 2, 5,8

2
15 25

35

tune [s]

Figure 3.2: Comparison between different references

w,lref

The system is simulated with parameters P = 0.72 D = 0.021 and I = 0.01 for the engine speed controller .qand P = 1.57, D = 0.018, I = 0.01 for the slipping speed controller v2, the behavior shown in figure 3.3 results. The controller is tuned iteratively by starting at safe values for both controllers, for example 1 for both proportional values. As clutch engagement is slow the slipping speed controller P-value is slightly increased. It was observed that it is possible to engage the clutch in 1.8 seconds without the engine speed dropping lower than the starting value of 100 rad/s. At this setting the engine torque T, is already saturating a short time. This causes a drop in the engine speed, although it remains above 100 rad/s. The drop is caused . q , which is not not only by the limited engine power, but also the value for z limited. This value is added to v:,to obtain the input value for the clutch This means increasing the output ulof the engine speed controller torque TcI. the engine speed before will worsen the problem. Because by increasing TCI engagement will be reduced further. To obtain a fast engagement but also limit the drop in engine speed a compromise between them has to be made. If we set the lower limit for the engine speed, the no-kill condition, at 90 rad/s the controller can be tuned to engage in about 1.5 seconds, as seen in figure 3.3.

engine and clutch disk speed

0.2

0.4

0.6

0.8

1 time [s]

1.2

1.4

1.6

1.8

slippmg speed as,

100 80 -

C & ---

--. . ._

- 60 r u Z 40" 20m

,'.-.

', . .
1 '

' . .5 .

--.
' \ .
12

0-20 0 02 04 06 08
1 time [s]

--._
14 16 18

Ii
2

Figure 3.3: Engine and clutch disk speed during engagement For the transmitted torque through the clutch TcI we find an interesting result as seen in figure 3.5. The signal is smooth before engagement and shows vibrations after the clutch sticks, caused by the discontinuity in w, occurring. The other control signal is the engine torque T,, plotted in figure 3.4. This input is limited at 100 Nm and also shows a smooth curve. However it does saturate for a significant time, for reasons discussed earlier, which means we have no direct control over the engine speed during this period. The torque at the wheels T , resulting with this controller is much smoother than the torque curve achieved earlier with the step input, see figure 3.6. This curve determines the driver comfort, i.e. T , should be as smooth as possible, preferably showing no oscillations.

Figure 3.4: Engine torque T, during engagement

Figure 3.5: Transmitted torque Td during engagement

Torque at the driving wheels

06

08

1 time [s]

12

14

Figure 3.6: Torque a t the driving wheels T, during engagement

3.5

Analyzing the decoupling controller

To get a better understanding of the working of the decoupling controller its frequency response is evaluated. The function G(s) (3.14) represents the decoupled system. If we evaluate this transfer function for s = j w , the bode plot in figure 3.7 is obtained: As seen in this bode plot, for low frequencies the system has a gain of -7.28 dB and no phase delay. For higher frequencies the gain falls with a slope of -1 and the system has a phase delay of 90 degrees. So the decoupled system represents a first order system. Of course when the model is not exact an error in the decoupling will be made, i.e. the matrix of (3.13) is not diagonal. The slipping and engine speeds are in that case not completely decoupled. If the control inputs are examined it seems that, except for the term JIG(s)s, the decoupling controller simply represents a combination of the PID-controllers 4 for we and v2 for w,~. For low frequencies the controller derived from equation (3.12) becomes:

remains the same and T, becomes: And for high frequencies the TCL

10'

10:

Frequency (radk68)

Figure 3.7: Bode diagram G ( s )

An interpretation of this is that the absolute value of engine torque must be smaller or equal to the torque transmitted trough the clutch plus a contrbution of the driveline and engine inertia. This then also states that the torque has a minimal value, below which the clutch cannot transmit transmitted TCL enough troque to engage. From the control point of view this controller is basically a summation of the two PID-controllers plus a small constant. This means the decoupling controller will weigh the error of the slipping speed as well as the engine speed according to the chosen PID parameters of each speed. By adjusting the PID parameters a compromise between fast engagement and a drop in the engine speed can be achieved. However as soon as the control input of the engine torque T, saturates, control over the engine speed is lost as T, cannot compensate for the drop in engine speed we anymore. Obviously if saturation of the controllers is not allowed, it can be avoided by selecting smaller values for the P and D parameters of the controllers. This will result in a slower engagement.

ns and

The clutch and drive line were modelled and written into three different descriptions. The model was implemented in Matlab Sirnulink and simulations were carried out.

On basis of a reduced model a decoupling controller was derived.


This controller was tested by simulation and clutch engagement was achieved within 1.5 seconds, without significant oscillations in the torque at the wheels. The decoupling controller works by summarizing the PID-controllers for slipping and engine speeds.

4.2
e

Recommendations
The references chosen for the engine and slip speed are arbitrary. With other types of control, for instance optimal control, the optimal trajectories for these speeds could possibly be calculated. The robustness of this type of controller is limited. If an inaccurate model is used, decoupling will fail.

Although the decoupling controller performs well, it has some disadvantages:

The tuning of the PID controllers for this system seems very sensitive, un-modelled disturbances will affect performance.

function [output]=switchfunc(u) %switchfunc(sh,Fn,we,wc,w2,w3,Te,Tkl,Tk2,Tllstick,Tc) sh = u ( l ) ; Fn = u(2) ; we = u ( 3 ) ; wc = u ( 4 ) ; w2 = u ( 5 ) ; w3 = u ( 6 ) ; wv = u ( 7 ) ; Te = u ( 8 ) ; Tkl = u ( 9 ) ; Tk2 = ~(10); T1 = ~(11); stick = ~(12); Tc = ~(13);

% inertias Jl=O.156; Jc=0.5; J2=0.2; mv=1200; Rw=.31; J3=l.7; Jv=mv*Rw*Rw;

% stifness & damping

% gear ratios is = [1/3.94,1/2.14,1/1.41,1/1.12,1/0.891; ig = l/3.73; i = ig*is(sh) ;

% clutch parameters
N

=I;

Ra = 1; mu =I; cslip = mu*Ra*N; cstick = 2*mu*Ra*N; eps


=

Le-2;
== 1 = Fn*cstick*sign(Tc); =

if stick Tcfn else Tcfn end

Fn*cslip*sign( we - wc ) ;

if abs (we - wc) <= eps & abs(Tcfn) >= abs ((Jl*Tkl+Jc*Te) / ( J1+Jc)) Tc = (Jl*Tkl+Jc*Te)/(Jl+Jc); stick = 1; else Tc = Tcfn; stick = 0; end Tkld = kl* (wc-w2); Tk2d = k2*(i*w2-w3) ; Tb=b*(w3-wv) ; wedot wcdot w2dot w3dot
=
=

= =

l/Jl*(Te -be*we - Tc); 1/Jc*(Tc - Tkl) ; 1/J2*(Tk1 - i*Tk2); I/J3*(Tk2 - Tb) ;

wvdot

= =

l/Jv*(Tb

Tl);

output

[wedot,wcdot,w2dot,w3dot,wvdot,Tc,Tkld,~k2d,stick,i];

Dynamic modeling and control of hybrid electric vehicle powertrain systems, B.K. Powell, K.E. Bailey and S.R. Cikanek, IEEE Control systems, pp 17-33, October 1998. Coordinated control of the Zero Inertia Powertrain, Serrarens, Alexander F. A., Proefschrifl Technische Universiteit Eindhoven, ISBN 90-3862583-9, 2001. Stop-Go with the ZI@-powertrain: a first glance, T.C.P. Verhagen, W F W Report 2000.27, 2000. Smooth engagement for automotive dry clutch, F. Garofalo et al., Proceedings of the 40th IEEE Conference on Decision and Control, December 2001. Karnopp, D. C., Margolis, D. L., and Rosenberg, R. C., System Dynamics: A Unified Approach. (1990), Wiley-Interscience, New York, 1990. Optimal control of gear shift operations in automatic transmissions, A.Haj-Fraj, F.Pfeiffer, Journal of the Franklin Institute 338, pp 371390, 2001.

H" control design for torque-converter-clutch slip system, R.Hibino et


al., Proceedings of the 35th IEEE Conference on Decision and Control, December 1996. Dynamic modelling and Control of a Car Transmission System, R.Zanasi et al. (2001), Universita di Napoli Federico II, Universita deE Sannio.

A study of shift control using the clutch pressure pattern in automatic transmission, W. Han and S.- J. Yi, Proc. Instn Mechanical Engineers Vol. 21 7 Part D: J,Automobile Engineering.

10. The influence of the interface coefficient of friction upon the propensity to judder in automotive clutches, D . Centea, H. Rahnejat and M.T. Menday, Proc. Instn Mechanical Engineers Vol. 213 Part D.
11. Model and control of a wet plate clutch, M.J.W.H. Edelaar, Afstudeerverslag WFW 97. Of 8.

S-ar putea să vă placă și