Fuel Economy Benefits of Advanced

Techniques for CVT Actuation and

Control

P.H.W.M. Albers
Report number: DCT-2005-23

Supervisor:
Prof. Dr. Ir. J.J. Kok

Coach:
Dr. P.A. Veenhuizen

Master Thesis Committee:

Prof. Dr. Ir. J.J. Kok
Dr. Ir. W.J.A.E.M. Post
Dr. Ir. B.G. Vroemen
Dr. P.A.Veenhuizen

Technische Universiteit Eindhoven, the Netherlands

Department of Mechanical Engineering
Section Control Systems Technology

Eindhoven, February 28, 2005

Summary
Modern control techniques allow drive train components to operate under more optimal operating
conditions than previously possible. Powerful electric servo-motors and modern controllers are necessary
for the development and refinement of automated transmissions like the Double Clutch Transmission and
the Continuously Variable Transmission, in order to arrive at fuel consumption, comfort and performance
levels that are expected from today's power trains.

In order to benchmark the effect of various improvement measures on fuel economy at an early stage in
the development process based on above-mentioned technologies, detailed simulations can be made. This
paper describes the results of a simulation study of the effects on efficiency and fuel economy of a vehicle
equipped with a CVT, resulting from the introduction of the following new variator actuation and control
techniques [1]:

• Replacement of the high pressure hydraulic pump by a servo-electromechanical actuation system

(EMPAct) and
• introduction of variator slip control

For this study, the TNO-Advance simulation package is used, in which detailed models of all loss
contributions of the CVT were implemented. The fuel economy benefit is shown to be substantial, about
6% on the NEDC cycle.

2
Contents

Summary .........................................................................................................2
10th EAEC European Automotive Congress 2005, Belgrade ........................4
Abstract............................................................................................................5
1 Introduction .............................................................................................6
2 Continuously Variable Transmission ......................................................7
3 Reference transmission efficiency modeling ..........................................7
3.1 TRANSMISSION LOSS ANALYSIS METHOD.............................................. 8
3.2 PUMP LOSS ....................................................................................................... 8
3.3 VARIATOR LOSSES......................................................................................... 8
3.4 FINAL REDUCTION LOSSES........................................................................ 10
4 Reference transmission loss breakdown................................................10
5 Efficiency improvement options ...........................................................11
5.1 VARIATOR SLIP CONTROL ......................................................................... 11
5.2 THE SERVO-ELECTROMECHANICAL ACTUATION SYSTEM.............. 12
6 Efficiency modeling for a transmission with EMPAct system .............13
7 Measurement setup ................................................................................14
8 Modeling total EMPAct losses..............................................................16
9 Fuel consumption calculations with Advance.......................................17
10 Conclusions and recommendations .......................................................19
References .....................................................................................................20
Attachements .................................................................................................22
A Recommendations and Appendices.......................................................22
B Cross-section of the Jatco CK2 CVT ....................................................23
C Torque loss definition............................................................................24
D Matlab model .........................................................................................26
E Advance .................................................................................................29
F Modeling the CK2 in Advance..............................................................31
G Modeling the EMPAct system in Advance ...........................................32
Nomenclature & Acronyms ..........................................................................33
Table of figures .............................................................................................34
Table of tables ...............................................................................................35
Samenvatting .................................................................................................36
Acknowledgment...........................................................................................37

3
10th EAEC European Automotive Congress 2005, Belgrade

4
 Fuel Economy Benefits of Advanced
Techniques for CVT Actuation and Control
P.H.W.M. Albers*, P.A. Veenhuizen*

10th EAEC European Automotive Congress 2005, Belgrade

Abstract
Modern control techniques allow drive train components to operate under more optimal operating
conditions than previously possible. Powerful electric servo-motors and modern controllers are necessary
for the development and refinement of automated transmissions like the Double Clutch Transmission and
the Continuously Variable Transmission, in order to arrive at fuel consumption, comfort and performance
levels that are expected from today's power trains.

In order to benchmark the effect of various improvement measures on fuel economy at an early stage in
the development process based on above-mentioned technologies, detailed simulations can be made. This
paper describes the results of a simulation study of the effects on efficiency and fuel economy of a vehicle
equipped with a CVT, resulting from the introduction of the following new variator actuation and control
techniques [1]:

• Replacement of the high pressure hydraulic pump by a servo-electromechanical actuation system

(EMPAct) and
• Introduction of variator slip control

For this study, the TNO-Advance simulation package is used, in which detailed models of all loss
contributions of the CVT were implemented. The fuel economy benefit of the EMPAct system in
comparison to the conventional CVT is shown to be substantial, about 6% on the NEDC cycle.

Keywords: continuously variable transmission, simulation, power train modeling, efficiency, CVT

* Technische Universiteit Eindhoven, Department of Mechanical Engineering, P.O. Box 513, 5600
MB Eindhoven, The Netherlands

5
1 Introduction
Since its introduction over 15 years ago, the CVT has contributed to the improvement of fuel economy of
passenger cars. It gives a better fuel economy than the stepped automatic transmission (AT) due to its
wider ratio coverage and its ability to shift ratio without compromising comfort. Another advantage is that
it allows for the torque converter to lock up at lower vehicle speeds than when an AT is applied.
Unfortunately, the metal V-belt CVT does not have a much better efficiency than the AT. An important
reason for this is that it needs significantly higher hydraulic pressure for variator actuation purposes,
(typically up to 50 [bar]) than an AT does, which leads to a larger oil pump drive torque. Since the oil
pump usually rotates at a speed proportional to engine speed, quick ratio shifts at low engine speed dictate
rather large pump displacement. For a typical CVT, rated at 200 [Nm], the pump displacement amounts to
about 19 [cc/rev]. Making use of the double piston principle in combination with a two stage torque
sensor, allows for a more favorable pump displacement of 11 [cc/rev] at 310 [Nm] rated torque [2].

Additionally, the power loss in the variator (defined as belt, pulleys and shaft bearings) is not small. This
leads to transmission efficiencies which barely exceed 90 [%] at rated load. Under part load conditions,
efficiency drops even further. This is caused by limitations in the actuation system, preventing the
clamping force to drop to sufficiently low values, causing over-clamping. In order to prevent variator slip,
over-clamping is applied, causing increased variator torque loss. It is mainly by these causes that the
efficiency may have dropped to values below 75 [%] at 25 [%] of rated load. On various driving cycles
used for fuel consumption benchmarks, a large time fraction is spent in part load conditions. Therefore,
improving part load efficiency receives considerable attention. Improving efficiency at rated load will be
beneficial for high torque situations like top speed, launch and fast acceleration maneuvers.

In order to reduce actuation system power consumption, Bradley and Frank [3] proposed a servo-electro
hydraulic system, largely eliminating excessive oil flow. The pumps need to deliver enough flow to
compensate for seal leak. A reduction of actuation power of about one order of magnitude was reported.

By using variator slip as a control variable for determining the distance between actual clamping force and
the slip threshold at which the variator is irreversibly damaged, Faust et. al. [2] indicate a transmission
efficiency improvement of almost 2 [%].

This paper first shows the principle of the CVT. Next, the modeling of the efficiency of the reference
transmission will be presented. The most important models, describing variator torque and slip loss and
pump power loss, were based on measurements on a commercially available CVT. Losses in the final
drive and clutch drag losses are based on literature data. Next the transmission efficiency improvements
that can be expected after replacing the hydraulic pump with a servo-electromechanical actuation system
(EMPAct) and after introduction of variator slip control will be presented. Simulations will show the fuel
consumption benefit of these measures.

For the EMPAct system simulations are made in Matlab/ Advance. Advance is a simulation package
developed by TNO Automotive. The EMPAct system strongly reduces the power needed for clamping and
shifting the variator. It avoids the situation where an engine driven pump delivers excess oil at high
pressure at elevated engine speeds, thereby not only improving part load but also full load transmission
efficiency. Slip control allows the variator clamping force to be set to very low values, thereby avoiding
losses associated with over-clamping. Slip losses, which are usually neglected in a CVT efficiency
assessment, are seen to increase somewhat when the clamping force is lowered and must therefore be
taken into account. The models for the EMPAct system and the slip controller are implemented in the
Advance simulation and the results are compared to the measurements.

6
2 Continuously Variable Transmission
Applying a Continuously Variable Transmission (CVT) in an automotive driveline has several advantages.
A CVT can operate at a wider range of transmission ratios, therefore the engine can be operated more
efficiently than with a stepped transmission. Also, a CVT does not abruptly change the torque level when
shifting. This gives a smoother ride than a stepped transmission does. A V-belt based CVT uses a belt or a
chain to transmit torque from a driving side to a driven side by means of friction. The layout of the CVT
and the V-belt are shown in Fig. 1 and Fig. 2.

The variator consists of two pulleys, which are wedge shaped. Both pulley sets have a fixed and a
moveable pulley, in opposite position. The movable pulleys are actuated by hydraulic pressure cylinders.
By changing the axial position of the pulley sheaves the ratio of the CVT can be adjusted. The variator can
cover any ratio between two extremes as shown in Fig. 1, low and overdrive.

The V-belt consists of blocks, which are held together by two rings. These rings consist of a set of bands,
see Fig. 2. To achieve torque transmission, sufficiently high clamping force levels are needed to prevent
slip in the variator. Because the torque level is not exactly known at all times, since no torque sensor is
used due to cost and packaging considerations, a safe clamping force level based on the maximum
possible load is maintained at all times. This safety level is based on the assumed maximum shock load
levels from the road and the engine torque.

Laminated steel bands

V-shaped steel elements

Figure 1: Low ratio and Overdrive Figure 2: V-belt with steel bands

3 Reference transmission efficiency modeling

For benchmark purposes, the 2 liter class Jatco CK2, manufactured by Jatco Transtechnology Co. Ltd. is
selected. This type of transmission will also be used as a carrier for the servo-electromechanical actuation
system that is currently in the realization phase. The reference transmission comprises a torque converter
as drive-off element, a DNR set, a hypotrochoid gear pump, a push-belt type variator, followed by the
final reduction and differential gear [4]. The majority of the presented data will be limited to the over-
drive ratio, since this ratio is most relevant for fuel economy evaluations.

7
3.1 TRANSMISSION LOSS ANALYSIS METHOD

In order to assess the efficiency improvement potential of the proposed modifications, first the power loss
occurring in the reference transmission were carefully modeled. The main power losses are caused by the
hydraulic pump and by the variator and were obtained by means of measurements. Losses in the final
reduction were calculated by means of models obtained from literature [5]. These loss components will be
treated separately in the following sections.

3.2 PUMP LOSS

One of the major loss sources within the transmission is the power absorbed in driving the hydraulic
pump. This loss was measured on a spin loss test rig. This test rig allows driving the transmission with
unloaded output. For these measurements, the clutch plates of the DNR set were removed to eliminate the
torque loss due to friction between clutch plates. In order to control pump torque independent from the
original transmission controller, a separate pressure control system was developed. The dependence of
pump power loss on input speed (equal to engine speed), pump output pressure (equal to the clamping
pressure in the secondary variator cylinder) and temperature, was measured in a wide operating range and
mapped into lookup tables.

3.3 VARIATOR LOSSES

Contrary to form closed transmission components like gears, the variator not only loses power by torque
loss, but also slip loss occurs. Although slip loss is generally much smaller than torque loss in a V-belt
variator, it can not be neglected, since slip loss tends to increase when over clamping is reduced. Next the
methods will be described by which these losses were analyzed.

Variator slip loss may be defined as follows:

ωsec
ωloss = ω pri − (1)
rω 0

A relative slip number sω can be defined by

ω loss r ω
sω = = 1 − ω , with rω = sec (2)
ω pri rω 0 ω pri

Here rω0 is defined as the speed ratio rω at zero variator output load. Furthermore ωpri and ωsec are defined
as the angular speed of the primary and secondary variator shaft, respectively. This definition ensures that
sω equals zero at zero variator output load.

Power is transmitted by means of friction between the belt and the pulleys. The torque that is transmitted
through the variator can be calculated using the force balance on a pulley, according to [6]:

2 Fs R p µ eff (sω )
T pri = (3)
cos λ

8
where Tpri is the torque on the primary shaft, Fs is the secondary clamping force, which is the axial force
applied by the pulleys onto the belt, Rp represents the primary running radius of the belt, λ is half the
pulley wedge angle and µeff is the effective traction coefficient between belt and pulley. The traction
coefficient µeff is not constant but depends on the relative slip between the belt and the pulleys, defined by
Eq. 2.

The relation between de traction coefficient µeff and the slip sω is shown for three ratios in Fig. 3 [8]. It can
be seen that the slope of the curves and the maximum traction coefficients clearly depends on the ratio, but
all curves show the same distinct shape. At first, for low slip values the traction coefficient increases with
increasing slip, until a maximum value is reached. This region is called the microslip region. When the
maximum value of the traction coefficient is reached, increasing slip will result in a slow decrease of the
traction coefficient. This region is known as the macroslip region.

Figure 3: Traction coefficient µeff as a function of the slip sω measured with an input
speed of 225 rad/s for ratios low (0.43), medium (1) and overdrive (2.25)

For normal operation, where only the micro-slip region is used, the dependence of torque on slip can
effectively be approximated by a linear relationship. The dependence of the slope of this linear
relationship on ratio and clamping force was captured in a lookup table.

Variator torque loss may be defined by:

Tloss = T pri − rTo Tsec (4)

where Tpri and Tsec are defined as the torque on the primary and the secondary shaft respectively and rTo is
defined as the ratio of primary torque and secondary torque at zero torque loss.

Measurements of Tloss have been carried out on a variator test rig. Generally, the variator torque loss is
considered to be a function of four variables, i.e. secondary pressure psec, primary speed ωpri, speed ratio
rω, and input torque Tpri. According to measurements results shown in Fig. 4, the torque loss of the variator
shows very small dependency on Tpri. This is in line with the literature data [7]. The solid line represents a
fit of Eq. 4 to the measured data. It was therefore decided to measure torque loss under unloaded
conditions (Tsec = 0 [Nm]) on a CK2 transmission on the same test rig that was used for pump loss
determination.

9
 Figure 4: Variator input torque versus variator Figure 5: Variator efficiency versus safety factor
output torque at fixed secondary clamping force at Sf (over-drive, 1500 [rpm])
low and OD ratio. The line represents a fit of Eq. 4
to the data.
By removing the intermediate shaft, power loss in the final reduction was eliminated from the
measurement. By switching the transmission into drive and applying torque converter lock-up and by
subtracting pump torque loss from the measured data, variator torque loss as a function of input speed,
ratio and clamping force could be determined over a wide operating range. A result from these
measurements is shown in Fig. 5. Here, the data is used to calculate variator efficiency as a function of
safety factor Sf, which is defined as the ratio of actual clamping force and the minimum clamping force,
required for transfer of a specific input torque at a specific variator ratio. The graph shows that Sf must be
reduced to the lowest possible value in order to obtain the best variator efficiency. Usually, lowering of
the safety factor is severely limited, especially for part load conditions. Very often, the secondary or line
pressure, which is directly related to secondary clamping force, cannot be lowered to values below 5-7
[bar]. This is caused by the fact that auxiliary pressures, which should not drop below 5-7 [bar], are
derived from line pressure. Low safety value at low nominal torque increases the risk of variator slip
caused by torque peaks originating from road irregularities. This forms another reason why safety values
are seen to be considerably higher than 1 in practice.

3.4 FINAL REDUCTION LOSSES

The final drive of the CVT is similar to that found in any conventional transmission and can be modeled in
a similar way. The calculations have been based on literature data [5] and references therein. Bearing
losses, oil churning losses and gear meshing losses have been taken into account. During measurements it
was noticed the torque losses in the final reduction depends on the applied torque to the primary shaft.
This is in line with the literature data [10].

4 Reference transmission loss breakdown

The efficiency analysis mentioned in the previous sections was used to generate Fig. 6, showing the power
loss breakdown of the reference transmission in overdrive at input speed of 1500 [rpm].

10
 Figure 6: Loss breakdown of the reference transmission (over-drive, 1500 [rpm])

Obviously torque loss in the variator and pump loss are the most important loss sources. The change in
slope occurring at about 120 [Nm] is caused by the limitation in hydraulic line pressure (equal to pressure
in secondary cylinder) to drop below psec_min = 6.6 [bar]. This leads to rather severe over-clamping at low
torque levels, causing both pump driving loss and variator torque loss to be high.

5 Efficiency improvement options

From the previous sections it may be concluded that for the optimization of transmission efficiency by
actuation and control measures, the following aspects should be taken into account:

• Over-clamping should be avoided or, equivalently, the safety factor Sf should be kept close to 1, also at
low torque levels.
• In order to realize this, hydraulic limitations for lowering clamping force should be avoided
• Actuation power must be reduced

In order to realize this, a servo-electromechanical actuation system with variator slip control has been
proposed, which will be described in the following sections.

5.1 VARIATOR SLIP CONTROL

As shown in Fig. 5, variator torque loss can be reduced by minimizing over-clamping. However, this
increases the risk of variator slip. In order to limit variator slip as defined before, variator slip control has
been proposed [8]. With this type of control, safety values very close to Sf ≅ 1 can be realized. Road
irregularities may cause short torque peaks in the drive train, causing a temporary macro-slip condition.
Recent experimental results [9] have shown however, that the push-belt variator is much more robust
against macro-slip than previously assumed. It will be the subject of further research to what extent this
property of the push-belt variator can be used when applying slip control.

11
5.2 THE SERVO-ELECTROMECHANICAL ACTUATION SYSTEM

In order to assess the apparent advantages of an all-mechanic system, an electrically driven variator
actuation system has been proposed by van de Meerakker et al. [1]. This system is schematically shown in
Fig. 7 and is characterized by the following features:

• Both moveable sheaves are actuated by means of electric motors (Mp and Ms in Fig. 7)
• Both motors are stationary when both clamping force and variator ratio are constant. This is
accomplished by applying two differential gear sets at each shaft, serving as a decoupling mechanism
between the rotating transmission shafts and the transmission housing.
• By coupling two moving ring gears of the adjustment mechanisms, shifting energy can be exchanged
between one movable sheave which is shifted with positive power and the other shifted with negative
power.
• It is important to note that the EMPAct system calls for a novel way of estimating the clamping force,
since a hydraulic pressure signal is absent. Slip control seems therefore to be a logical approach for
assessing the clamping status of the variator.
• Clamping force loops are closed on the rotating shaft and not supported via the transmission housing,
thereby avoiding highly loaded thrust bearings.
• A separate low pressure pump for auxiliary hydraulic functions is required.

Figure 7: Schematic representation of the EMPAct system

12
6 Efficiency modeling for a transmission with EMPAct
system
The efficiency analysis mentioned in the previous sections, was used to generate Fig. 8, showing the loss
breakdown of the transmission with EMPAct system and slip control in over-drive at an input speed of
1500 [rpm]. For the torque loss calculation see also section 8. In comparison to Fig. 6 the following
differences can be noted:

• The change in slope at about 120 [Nm] has disappeared. This is associated with the ability of the
EMPAct system to reduce clamping force to low levels.
• Especially the power loss at part load is strongly reduced.
• Actuation power is strongly reduced.
• Part load variator torque loss is reduced considerably, which must be attributed to the reduction of over-
clamping.
• Variator slip loss has increased slightly by the reduction of the safety factor to values close to 1. This is
most clearly seen at high input torque levels.

Figure 8: Loss breakdown of a transmission with EMPAct

system and variator slip control (over-drive, 1500 [rpm])

In order to appreciate the contribution of the various measures to the reduction of power loss, they were
analyzed separately. Table 1 defines the various improvement measures.

Table 1: Improvement measures

Variant A Reference transmission with Sf = 1.3 and psec_min = 6.6 [bar]
Variant B Reference transmission with slip control, Sf ≅ 1 and psec_min = 6.6 [bar]
Variant C Reference transmission with Sf = 1.3 and decreased secondary pressure psec_min = 1.35 [bar]
Variant D Slip control, Sf ≅ 1 and psec_min = 1.35[bar]
Variant E EMPAct system, slip control Sf ≅ 1 and Fsec_min = 1.9 [kN]

13
From Fig. 9 the effects of various measures on power loss can be estimated. Curve A and curve B both
have high torque losses for low input torque because of the minimum secondary pressure psec_min = 6.6
[bar]. Curve C indicates the very important effect of decreasing the clamping force boundary to a
secondary pressure of psec_min = 1.35 [bar]. Curve D represents a further lowering of safety to Sf ≅ 1,
meaning that variator slip control must be applied, in order to avoid gross slip. Curve E (EMPAct system),
with a minimum clamping force of 1.9 [kN] which represents a minimum line pressure of psec_min = 6.6
[bar], shows the most important loss reduction at higher torques, due to the reduction of actuation power.

Figure 9: Loss reduction by various measures (Variants B through E)

compared to the reference transmission (Variant A).

7 Measurement setup
To validate the results of the reference model a test rig is available (see Fig. 10). This test rig is designed
to perform realistic drive train experiments. The rig consists of two electric Siemens motors (engine and
brake), two torque sensors (Te and Ts), a Jatco CK2 CVT and a manual gearbox. The Siemens motors
both include an encoder to measure the rotational speed. With the electric motors the combustion engine
of a passenger car and the road load and brake torque can be simulated respectively. The torque sensors
are used to measure the ingoing and outgoing torque of the CVT on the primary shaft and the drive shaft.
The gearbox is used to increase the speed applied to the Siemens brake motor.

Gear SIEMENS 2
Ts
box Brake
Jatco
CK2
SIEMENS 1 Te
Engine

Figure 10: Overview of the test rig used for loss measurements

Normally the Jatco CK2 will be controlled by the Transmission Control Module (TCM). On this test rig
no TCM is available so the clamping force to transmit engine torque has been calculated using an
implemented safety 1.3 controller (Variant A from table 1) or with the slip controller (Variant B). To

14
simulate a driveline both electric motors should be controlled. The test rig is controlled with Matlab/
Simulink in combination with a dSpace data acquisition system.

To track a reference from a driving cycle, a driveline controller is implemented which calculates the
necessary throttle position and brake torque. To obtain a stable behavior which minimizes the tracking
error, a PI controller is used with anti-windup. A variogram is used to get a reference for the CVT input
speed, given the vehicle speed and the throttle position. The throttle position has been calculated with an
inverted engine torque map. Here the measured engine torque and engine speed result in a throttle
position. With the test rig the NEDC cycle and the FTP-72 cycle are simulated. Both cycles are shown in
Fig 11.

Figure 11: Speed cycles: NEDC and FTP-72

Now the results of the measurements can be compared with the results of the reference model described in
chapter 3. This comparison is done in order to check the correctness of the implementation of various
component loss models and also to ascertain that these models can be used also under loaded conditions.
The engine speed, primary speed, secondary speed and engine torque of the measurements are the inputs
for the reference model. In Fig. 12 and Fig. 13 the comparison of secondary pressure and torque loss
between the measurements and the reference model (Variant A) is given. What can be noticed is the
difference in secondary pressure. This difference is caused by the difference in the modeling of the torque
converter. In the model only an estimated clamping force multiplier is used for take off. On the test rig the
model of the torque converter has a safety factor which assures there will be enough clamping force on the
secondary pulley to prevent slip. The differences in secondary pressure are shown not to have a large
influence on the torque losses. Fig. 13 shows that the model predicts the torque loss accurately. Only for
higher speeds and therefore higher input torque the model has some difference compared to the
measurements. The most likely explanation for this discrepancy is that the final reduction exhibits more
torque loss than what was calculated on the basis of models.

15
 Figure 12: Secondary pressures for the NEDC and FTP-72 cycle

Figure 13: Torque loss for the NEDC and FTP-72 cycle

8 Modeling total EMPAct losses

For analysis, control design and testing of the electro-mechanically actuated CVT, a simulation model is
built [11]. The model incorporates all major driveline components and the proposed actuation system with
servomotor actuation. The clamping forces in the variator are calculated using an explicit formulation of a
model based on Coulomb friction. The model also includes slip and shifting losses based on transient
variator models.

For torque loss calculation due to the friction forces in the bearings, gears and spindles, Lugre friction
models are incorporated in the model [11]. To calculate the power consumed by the actuation system,
electromotor models and characteristics are taken into account. Also the power loss due to the external oil
pump used for auxiliaries is taken into account.

16
9 Fuel consumption calculations with Advance
In order to assess the potential for fuel economy improvement, the data described in section 3 was used in
the simulation package Advance, developed by TNO Automotive [12]. This package allows for detailed
fuel consumption simulations.

With the simulation package first a comparison has been made between the results measured on the test
rig and the results from Advance with the conventional CVT with control strategy Sf = 1.3 (Variant A).
Both results are measured with a minimum secondary pressure of 6.6 [bar]. In Fig. 14 the torque loss and
secondary pressure for the test rig and for the Advance simulation are drawn. From this figure it appears
the results look sufficiently similar. The results show that the model of the torque converter in Advance
can be compared to the model of the torque converter on the test rig, because now there is only a minimal
difference in the secondary pressure between the measured cycle and the simulated cycle.

Efficiency and fuel consumption improvement can be reached when it becomes possible to decrease the
minimum line pressure (for example Variant C and Variant D from table 1). Decreasing the line pressure
will increase the risk of slip. As mentioned before, road irregularities may cause short torque peaks in the
drive train, causing a temporary macro-slip condition. The implemented slip controller should avoid these
problems. Interesting is the question in which way the control strategy Sf = 1.3 can handle with such
torque peaks.

Figure 14: Torque loss and secondary pressure

In Fig. 15 a simulation is shown with the reference CVT with a minimum line pressure of 6.6 [bar]
(Variant B) in comparison with a CVT with minimum line pressure of 1.35 [bar] (Variant D), both slip
controlled. These results show that the CVT with decreased minimum line pressure has lower torque
losses, while the amount of slip is within acceptable limits. So when the line pressure can be decreased,
the efficiency will increase and the fuel consumption will decrease. Also the fuel consumption for both
cases can be simulated. As expected, the fuel consumption for a CVT with minimum line pressure of 1.35
[bar] (Variant D) will be lower and gives a benefit of approximately 2.2 %. The same simulation can be
done for the FTP-72 cycle. For this cycle the CVT with decreased minimal pressure gives a profit of
around 1.5 %.

17
 Figure 15: Secondary pressure, torque loss and slip for the reference
CVT (B) and the CVT with decreased minimal pressure (D)
In Fig. 16 the comparison has been made between a CVT with decreased minimum line pressure psec_min =
1.35 [bar] (D) and the EMPAct system (E). The secondary clamping forces resulting from the
measurements are almost equal. The advantage of the EMPAct system is the removal of the oil pump, and
therefore the removal of these torque losses. The oil pump always rotates at engine speed and therefore
dissipates always an amount of power. The torque losses for the EMPAct system are significantly lower.

Because of the decrease in torque loss for the EMPAct system, the efficiency will increase and therefore
the fuel consumption will decrease. In comparison to a CVT with decreased minimum line pressure of
1.35 [bar] (Variant D), the new EMPAct system (Variant E) has an advantage of almost 4 percent for the
NEDC cycle.

Figure 16: Secondary clamping force, torque loss and slip for the CVT
with decreased minimal pressure (D) and the EMPAct system (E)
When the EMPAct system (E) is compared to the conventional CVT (Variant A) an advantage of 6
percent can be reached, see Fig. 17. For the FTP-72 cycle the same simulations can be made. Also here the
EMPAct system (E) results in lower torque losses and slip can be kept within acceptable limits. For the
FTP-72 cycle the EMPAct system (E) has a fuel consumption benefit of 4.2 percent compared to the CVT

18
with decreased line pressure (D). In comparison with the conventional CVT (A), the EMPAct system (E)
has a benefit of around 5.7 % with respect to fuel consumption.

Figure 17: Fuel consumption for Variant A and Variant E during the NEDC cycle.
The results for the two different speed cycles are summarized in the table 2.

Table 2: Fuel consumption reduction for different clamping force control strategies
Clamping force control strategy Fuel consumption reduction [%]
Reference: Compared to: NEDC FTP-72
Safety strategy 1.3 (A) Slip controller (B) 0 0
Slip controller (B) Slip controller with 2.2 1.5
minimum pressure (D)
Slip controller with EMPAct (E) 4 4.2
minimum pressure (D)
Safety strategy 1.3 (A) EMPAct (E) 6 5.7

10 Conclusions and recommendations

Both model calculations and test rig measurements have been used to construct a detailed efficiency
model of a commercially available CVT. The model results are compared to the measurements results and
show similarity to these results. By subtracting pump loss and adding loss due to the EMPAct system and
by mapping variator loss as a function of over-clamping, the effect of the new actuation system and slip
control on the efficiency can be estimated and an indication is given for the consequences for fuel
consumption. By applying a servo-electromechanical actuation system and variator slip control,
significant reduction of power loss can be expected. This leads to a large improvement of efficiency,
especially at part load, which in turn leads to a significant fuel economy potential, especially on driving
cycles where part load conditions prevail (about 6% on the NEDC cycle). Simulations for fuel
consumption reduction are made with the simulation package Advance, developed by TNO Netherlands.

19
The proposed servo-electromechanical actuation system is currently in the realization phase. The slip
control technique will be optimized, but it has already been applied and demonstrated in a passenger car.

References
[1] K.G.O. van de Meerakker, P.C.J.N. Rosielle, B. Bonsen, T.W.G.L. Klaassen, N.J.J. Liebrand,
Mechanism proposed for Ratio and Clamping Force Control in a CVT, in Fisita 2004; Barcelona,
Spain, pp. 10, (2204).

[2] Faust, H., M. Homm, M. Reuschel, ”Efficiency-Optimized CVT Hydraulic and Clamping
System”, proceedings International CVT congress, VDI berichte 1709, pp.43-58, Munich,
Germany, 2002.

[3] Bradley, T.H. and A.A. Frank,”Servo-pump hydraulic control system performance and evaluation
for CVT pressure and ratio control”, proceedings International CVT congress, VDI berichte 1709,
pp.35-41, Munich, Germany, 2002.

[4] Abo, K., M. Kobayashi, M. Kurosawa, “Development of a metal belt-drive CVT incorporating a
torque converter for use with 2-liter class engines”, SAE Technical Paper 980823, 1998.

[5] Changenet, C., M. Pasquier, “Power losses and heat exchange in reduction gears: numerical and
experimental results”, International conference on gears, Munich, 2002. VDI berichte 1665.

[6] Bonsen, B., T.W.G.L. Klaassen, K.G.O. van de Meerakker, M. Steinbuch and P.A. Veenhuizen,
”Analysis of slip in a continuously variable transmission”, Proceedings of IMECE03 2003 ASME
International Mechanical Engineering Congress, Washington, D.C., November 1521, 2003,
IMECE 2003-41360.

[7] Ide, T., ”Effect of Power Losses of Metal V-belt CVT Components on the Fuel Economy”,
Proceedings of the International Congress on Continuously Variable Transmission, CVT ’99,
Eindhoven, the Netherlands, 1999.

[8] Bonsen, B., T.W.G.L. Klaassen, K.G.O. van de Meerakker, P.A. Veenhuizen and M. Steinbuch,
“Measurements and control of slip in a continuously variable transmission”, in Mechatronics
2004, Sydney, Australia 2004.

[9] Drogen, M. van, M. van der Laan, ”Determination of Variator Robustness under Macro Slip
Conditions for a Push Belt CVT”, SAE Technical paper, 2004-01-0480, SAE world Congress,
Detroit, 2003.

[10] Akehurst, S, “An investigation into the loss mechanisms associated with a pushing metal V-belt
continuously variable transmission”, PhD-thesis University of Bath 2001.

[11] Klaassen T.W.G.L., B.G. Vroemen, B. Bonsen, K.G.O. van de Meerakker, M. Steinbuch and P.A.
Veenhuizen, “Modeling and simulation of an electro-mechanically actuated push belt type
continuously variable transmission”, 3rd IFAC Symposium on Mechatronic Systems, Australia,
2004.

[12] Vink, W., J. Eelkema, J. Zuurbuur, “Modular vehicle simulation in Matlab/ Simulink”, Versuch,
Test und Simulation in Antriebsstrang und Fahrwerk Conference, Karlsruhe, Germany, 2002.

20
[13] Veenhuizen, P.A., B. Bonsen, T.W.G.L. Klaassen, K.G.O. van de Meerakker, M.
Steinbuch and F.E. Feldpaus, “Simulated behavior of a vehicle with V-belt type geared
neutral transmission with variator slip control, the Netherlands.

[14] Pulles R.J., B. Bonsen, M. Steinbuch and P.A. Veenhuizen, “Slip controller design and
implementation in a Continuously Variable Transmisión”, American Control Conference,
Portland, Oregon, United States, 2005.

21
Attachements
This chapter gives recommendations which are not important enough to mention in the paper, but can be
used for future work. Also an explanation of the next appendices is given.

A Recommendations and Appendices

The modeling of the CK2 CVT in Matlab (chapter 3) seems be done in a correct way. For all components
of the CVT models of loss contributions are implemented, only for the torque converter a model is
missing. When the model will be used for situations with an open torque converter, also a model of the
torque converter should be implemented.

In Advance the simulations can be made which can be compared to the measurements. The simulation in
Advance runs at low speed when the wrong solver is chosen (chapter 9). It is therefore important to
choose the correct solver (ode23tb(stiff/TR-BDF2)) and the correct initial step size and relative tolerance.
Interesting is to find out if it is possible to use other solvers by adapting the model in the simulation.

To be sure of the results simulated with Advance for the EMPAct system (chapter 8), the results should be
compared to the measurements when the EMPAct system is operational on the test rig.

The information given in Appendix B – Appendix G is included in this report to give some more details
about subjects treated in the paper. In Appendix B a cross-section of the Jatco CK2 CVT is shown, while
in Appendix C the torque loss definition is given which will also be used in Appendix D, the Matlab
model. This Matlab model is compared to simulations resulting from the simulation package Advance,
which is described in Appendix E. In Appendix F the structural representation of the CK2 model is shown
which is used in Advance. Finally, in Appendix G the modeling of the EMPAct system is mentioned.

22
B Cross-section of the Jatco CK2 CVT
Fig. B.1. shows the cross-section and important part names of the Jatco CK2 CVT, which is used in this
research program. For a complete and detailed description of the CK2 and its components the reader is
referred to [7].

Figure B.1: Cross-section of the Jatco CK2

23
C Torque loss definition
Before making a Matlab model of the Jatco CK2 CVT (Chapter 3 and Appendix D) first the definition of
torque losses will be explained. The torque losses can be calculated with the Eq. C.1 [13].

Tloss = T pri − rT 0Tsec

(C.1)

where:
Tpri = the primary torque of the variator
rT0 = the torque ratio at zero load
Tsec = the secondary torque of the variator

From earlier experiments [13] it becomes clear that the torque ratio at zero load rT0 is equal to the
geometric ratio rG and to the speed ratio at zero load rω0. So this means:

rT 0 = rG = rω 0 (C.2)

Due to the slip also a speed loss will be present. This speed loss can be calculated with Eq. C.3.

ω sec
ω loss = ω pri −
rω 0 (C.3)

Slip sω can be defined with the following equation:

rω
sω = 1 −
rω 0 (C.4)

where the speed ratio rω can be defined as:

ω sec
rω =
ω pri
(C.5)

With Eq. C.2 and Eq. C.4 it is possible to transform Eq. C.1 in Eq. C.6.

rω
Tloss = T pri − Tsec
1 − sω (C.6)

The same transformation can be made for Eq. C.3. Combining this equation with Eq. C.4 gives a new
expression for the slip losses. These slip losses can now be calculated with:
ω sec
ω loss = ω pri − (1 − sω )
rω (C.7)

24
To translate both losses to power losses first Eq. C.1 and Eq. C.3 have to be transformed. For Tsec and ωsec
the following equations can be found:

T pri − Tsec
Tsec =
rT 0 (C.8)
and

ω sec = (ω pri − ω loss )rω 0

(C.9)

Now the power on the output shaft of the CVT variator can be calculated. In this equation rT0 and rω0 have
the same value and so dividing these parameters gives value 1.

T pri − Tloss
Psec = Tsecω sec = (ω pri − ω loss )rω 0 = (T pri − Tloss )(ω pri − ω loss )
rT 0 (C.10)

When the last part of Eq. C.10 was multiplied the following equation for Psec can be found:

Psec = T pri ω pri − Tloss ω pri − ω loss T pri + ω loss Tloss

(C.11)

Rewriting Eq. C.11 gives the following expression for the power losses due to torque loss and speed loss:

Ppri − Psec = Ptorque −loss + Pslip −loss − Premaining

(C.12)

where the last term Premaining can be neglected because of its small size.

25
D Matlab model
This Matlab model is validated with measurements from the test rig and the data mapped in lookup tables
is implemented in Advance. Before running the model first an engine speed ωeng should be defined. This
speed can vary between 500 and 2500 [rpm]. Then a column of torques is generated where the maximum
torque is 200 Nm. This 200 Nm is the maximum torque the CK2 CVT can handle. Also a column of
geometric ratios is generated where the ratio varies between 0.43 and 2.2. These two borders are the low
and OD ratio of the CK2 CVT. Now the different functions will be given which are used in the CVT
model.

D.1 Function r2Rp

The function r2Rp calculates the running radius of the primary pulley Rpri. This radius can be obtained
using Eq. D.1:

L = 2CD cos(λ ) + R pri (π + 2λ ) + Rsec (π − 2λ )

(D.1)

where:
L = Circumferential length of the belt
CD = Pulley center distance
λ = Pulley wedge angle
Rpri = Running radius of the belt on the primary pulley
Rsec = Running radius of the belt on the secondary pulley

Together with the equation for the geometric ratio rG it is possible to transfer Eq. D.1 to an ABC-formula
with only an unknown parameter Rpri. For this equation the circumferential length of the belt and the
pulley center distance should be known.

D.2 Function Fclamp

Function file Fclamp calculates the minimum clamping force on the secondary pulley. In Eq. D.2 the really
desired clamping force Fsec is given:

Fsec = F pri + Fs + Fc (D.2)

where:
Fpri Clamping force on the primary pulley
Fs Spring force
Fc Centrifugal force

Next a safety strategy Sf has to be chosen so the clamping force on the secondary pulley can be
determined. Sf determines the clamping force on the secondary pulley.

D.3 Function sω
In the function file sω the slip acting on the belt can be calculated. In figure D.1 a traction curve is shown
of a measurement of slip versus the friction coefficient µ. The derivative of the linear part near to zero is

26
called α. From many experiments all different values for α are found which depends on clamping force
and geometric ratio. For three different ratios (low, medium and OD ratio) and variable secondary
clamping forces (Fsec) the measurement data has been plotted in Fig. D.2.

Figure D.1: Slip percentage versus µ Figure D.2: Fit through data of measurements

Through this measured data a first order polynomial fit has been plotted. This polynomial fit gives a
relation between the secondary clamping force Fsec and the value of α given by Eq. D.3.

α = −0.001(Fsec − Fc ) + α int (D.3)

where:
Fsec = Clamping force on the secondary pulley
Fc = Constant clamping force on the secondary pulley
αint = Interpolated value of α for different ratios

At the level of Fc (about 7.8 kN) three nearly constant values for α were measured. For high axial forces
three different constant values for α were found. The minimum value of α belongs to the constant α for
high axial forces. This protects α to become never too small or even negative. By interpolating the real
geometric ratio over the measured values for low, medium and OD you can find an estimated value for the
acting αint. Together with α it is possible to determine an interpolated value for the slip sω. In Eq. D.4 the
definition of slip is given.

µ eff
sω =
α (D.4)

Normally the slip value sω will be around the 1 %.

D.4 Function Tpump, Tvar

To calculate the oil pump and variator torque losses the lookup tables described in chapter 3 has been
used. The pump losses depend on the engine speed, secondary pressure and oil temperature, while the
variator losses depend on the primary pulley speed, geometric ratio and secondary pressure.

27
D.5 Function FRloss
Function file FRloss determines the final reduction torque losses. The final reduction consists of four gears
which will decrease the speed and increase the torque. The final reduction losses depend on the ingoing
speed, the ingoing torque and the oil temperature. The power losses can be divided into tooth friction
losses, bearing losses and oil churning losses.

In this function file first the parameters of the gears should be noted. When all the parameters are given
the function file for the oil churning losses can be loaded. These oil churning losses depends on the
ingoing torque of the gear, gear parameters and oil properties. Next the tooth friction losses can be
determined. These losses due to tooth friction depends on the gear parameters, ingoing torque of the
driven gear and the oil properties. The results of this function file are the axial and radial force acting in
the tooth contact and of course the tooth friction power losses. For the second set of gears the same can be
done. The third losses in the final reduction are the losses of the bearings. Also for these losses a function
file has been written. This function file depends on the different axial and radial forces acting in the tooth
contact points and on the different speed of gears. When the bearing losses are add to the oil churning
losses and the tooth friction the total final reduction power losses are known and therefore the efficiency
of the whole final reduction.

PFR −out = PFR −in − PFR −loss (D.5)

28
E Advance
In Advance the user is able to perform a wide range of different vehicle simulations. The modular setup of
the toolbox opens the possibility for the user to combine Advance with models of vehicle subsystems
developed outside Advance in user-defined libraries. In Advance different modules have been created that
represent specific areas of the simulation of vehicle dynamics. The user sees these different areas
represented by subsystems in the Simulink basic vehicle model layout. When working with Advance the
user can create efficient models that are tailor-made for the behavior that is investigated. Advance has
been built in MATLAB/Simulink. The calculations that are performed in each standard module have been
transformed into efficient s-functions to improve the speed of the simulations. Advance is developed so
that it can be used for real-time applications.

Advance models are divided into the following levels:

Top level
This is the uppermost level of the model, representing the complete vehicle. This level appears when a
new model is created.

Areas
The areas, located in the top level of the model, represent the various vehicle subsystems, such as power
train and chassis. While they have no built-in functionality, they are used to define the model structure.
Areas can contain Advance modules, standard Simulink sub-systems and blocks.

Modules
Modules simulate the behavior of each of the vehicle components. These include e.g. the body, engine and
wheel kinematics modules. The available modules are present in the Advance library in the Simulink
Library Browser. Modules need to be placed in areas or lower in the vehicle structure.

Top level of an Advance model

The top level of the model structure shows six areas that contain all the vehicle modules that are used (see
Fig. E.1). The model handling, the setting of the parameter files, viewing the time traces and selecting the
output functions are all placed on the top level of the model layout in the green areas. The collection of the
data that comes out of the model areas is part of the data bus that is used in Advance. The data bus is used
to enable data access in all model levels.

The six main vehicle areas are gray. They represent six important aspects of the modeling of the vehicle
dynamics. The areas don't have any direct functionality. However, by using them the Simulink model will
always have a similar organization and the fixed structure will support the intuitive use of the model.
Specialized graphical user interfaces support the user when performing the simulations.

The following vehicle areas have been defined:

test
In this area a predefined vehicle test can be placed. Such a test can be e.g. a step steer test or a power train
load cycle test (for example the NEDC or FTP-72 cycle).

29
driver
In this area the used driver model can be placed. In the first release of Advance driver models will be part
of the library.

control
This area contains the vehicle controllers that are used to control vehicle dynamics and power trains.
These units are i.e. gear and throttle control for automatic vehicles and traction control systems.

power train
The power train area is created for all power train components.

chassis
In Advance the chassis is defined as the part of the vehicle containing the tires, wheels and suspension.

body
Different models for the main body of the vehicle.

Figure E.1: The top level of the model structure

30
F Modeling the CK2 in Advance
For modeling the CK2 in Advance an adapted model of R. Pulles has been used [14]. The model is created
in Advance using existing variator models that describe shifting behavior, torque transmission, and belt
slip. Additionally, the measured torque losses caused by the oil pump and the variator are incorporated in
the model using look-up tables. And finally, the line pressure circuit and the ratio control circuit were
modeled to complete the model. The structure of the total model is shown in Fig. F.1.

Stepper motor Duty cycle line

position pressure solenoid

Jatco CK2
model
Ratio control circuit Line pressure circuit
model model

pp ps

ωs ωp

xpp, vpp

Shifting model

Fs
ωs ωp

Var losses
Belt model
Pump losses

Tp,cvt Tvar,loss Tp,loss

(Tp,cvt-Tvar,loss)/rg

Tp,cvt Ts,cvt Tp,loss

Figure F.1: Structural representation of the CK2 model in Advance

In this figure the stepper motor position results from the ratio controller. Here the speed ratio is controlled
to the reference ratio resulting from the variogram. For safety strategy 1.3 the duty cycle line pressure
solenoid is calculated with Eq. 3 in chapter 3. Here the secondary clamping force is multiplied with a
factor 1.3. For the slip controller the duty cycle line pressure solenoid is calculated with the same equation
only with safety factor 1. This part is calculated as feed forward for the duty cycle. The total duty cycle is
completed with a value calculated from the difference between slip and slip set point.

31
G Modeling the EMPAct system in Advance
For the EMPAct system a little model has been written which predicts the torque losses due to tooth
friction losses in the planetary gear sets. This is because the simulations described in chapter 8 were not
done in time for the deadline of the paper.

The model calculates from the different torques on the primary and secondary shaft the primary and
secondary clamping force respectively. Together with the speed belonging to both shafts the sliding speed
can be calculated. Multiplying the clamping force with the sliding speed the power of tooth contact is
calculated. For each shaft there are four contacts and for each contact an efficiency of 99% has been taken.
Adding all the power losses gives an amount for the power loss due to the tooth friction contact in the
planetary gear sets.

Also for the oil pump for auxiliaries and for the electric motors in Fig. 7 and amount of power is absorbed
too. All this power is all subtracted from the input power on the primary shaft and the remaining power is
available on the secondary shaft.

32
Nomenclature & Acronyms
ACRONYMS

Symbol Description

CVT Continuously Variable Transmission

NEDC New European Driving Cycle
FTP-72 Federal Test Procedure
EMPAct Servo-electromechanical actuation system

SYMBOLS

Symbols Description [Unit]

CD Pulley center distance [m]

Asec Secondary cylinder piston area [m2]
Fpri Primary pulley clamping force [N]
Fsec Secondary pulley clamping force [N]
Fc Centrifugal force [N]
Fs Spring force [N]
L Length of V-type belt [m]
psec Secondary pressure of hydraulic oil [Pa]
psec-min Minimum secondary pressure of hydraulic oil [Pa]
Ppri Power on the primary shaft [W]
Psec Power on the secondary pulley [W]
Pslip-loss Power loss due to slip [W]
Ptorque loss Power loss due to transferring torque [W]
Premaining Remaining term for power loss [W]
Rp Primary pulley running radius [m]
rG Geometrical ratio [-]
rω Rotational speed ratio [-]
rω0 Zero load speed ratio [-]
rTo Torque ratio at zero load [-]
Sf Safety factor [-]
sω Relative slip number [-]
Tpri Primary shaft torque [Nm]
Tsec Secondary shaft torque [Nm]
Tloss Torque loss measured at primary side [Nm]
α Derivative of linear part of a traction curve [-]
λ Pulley wedge angle [rad]
µ Friction coefficient of V-belt type belt and pulley sheave contact [-]
µeff Effective friction coefficient [-]
ωloss Variator slip loss [rad/s]
ωpri Rotational speed primary intermediate shaft [rad/s]
ωsec Rotational speed secondary intermediate shaft [rad/s]

33
Table of figures
Figure 1: Low ratio and Overdrive...................................................................................................7
Figure 2: V-belt with steel bands .....................................................................................................7
Figure 3: Traction coefficient µeff as a function of the slip sω measured with an input speed of 225
rad/s for ratios low (0.43), medium (1) and overdrive (2.25) ..................................................9
Figure 4: Variator input torque versus variator output torque at fixed secondary clamping force at
low and OD ratio. The line represents a fit of Eq. 4 to the data.............................................10
Figure 5: Variator efficiency versus safety factor Sf (over-drive, 1500 [rpm]) .............................10
Figure 6: Loss breakdown of the reference transmission (over-drive, 1500 [rpm]) ......................11
Figure 7: Schematic representation of the EMPAct system...........................................................12
Figure 8: Loss breakdown of a transmission with EMPAct system and variator slip control (over-
drive, 1500 [rpm]) ..................................................................................................................13
Figure 9: Loss reduction by various measures (Variants B through E) compared to the reference
transmission (Variant A). .......................................................................................................14
Figure 10: Overview of the test rig used for loss measurements ...................................................14
Figure 11: Speed cycles: NEDC and FTP-72.................................................................................15
Figure 12: Secondary pressures for the NEDC and FTP-72 cycle.................................................16
Figure 13: Torque loss for the NEDC and FTP-72 cycle...............................................................16
Figure 14: Torque loss and secondary pressure .............................................................................17
Figure 15: Secondary pressure, torque loss and slip for the reference CVT (B) and the CVT with
decreased minimal pressure (D).............................................................................................18
Figure 16: Secondary clamping force, torque loss and slip for the CVT with decreased minimal
pressure (D) and the EMPAct system (E) ..............................................................................18
Figure 17: Fuel consumption for Variant A and Variant E during the NEDC cycle. ....................19
Figure B.1: Cross-section of the Jatco CK2...................................................................................23
Figure D.1: Slip percentage versus µ .............................................................................................27
Figure D.2: Fit through data of measurements...............................................................................27
Figure E.1: The top level of the model structure............................................................................30
Figure F.1: Structural representation of the CK2 model in Advance.............................................31

34
Table of tables
Table 1: Improvement measures ....................................................................................................13
Table 2: Fuel consumption reduction for different clamping force control strategies...................19

35
Samenvatting
Moderne controle technieken maken het mogelijk om aandrijfcomponenten van voertuigen onder meer
optimale condities te laten opereren dan voorheen. Krachtige electrische servo-motoren en moderne
controllers zijn noodzakelijk voor de ontwikkeling en verfijning van automatische transmissies zoals de
Double Clutch Transmissie en de Continu Variabele Transmissie. Dit alles om het brandstof verbruik,
comfort en performance niveau te bereiken wat verwacht kan worden van huidige voertuigaandrijvingen.

Om aan te tonen wat het effect is van de verschillende verbeteringen met betrekking tot het
brandstofverbruik kunnen in het begin van het ontwikkelings proces gedetailleerde simulaties gemaakt
worden welke gebasseerd zijn op de bovenstaande technologie. Dit paper beschrijft het resultaat van een
simulatie studie naar de effecten op gebied van efficientie en het brandstofverbruik in een voertuig
uitgerust met een CVT, wat resulteert uit de introductie van het volgende nieuwe actuatie systeem en
controle technieken [1]:

• Vervanging van de hydraulische hoge druk pomp door een servo-electromechanisch actuatie
systeem (EMPAct) en
• Introductie van variator slip controle

Voor deze studie is het simulatie pakket Advance van TNO gebruikt. In dit pakket zijn gedetailleerde
modellen van alle verliesbijdragende componenten van de CVT geimplementeerd. De winst in
brandstofverbruik is substantieel, ongeveer 6% op de NEDC cyclus.

36
Acknowledgment
I would like to thank my coach Bram Veenhuizen for his support and guidance during the project. Also I
want to thank Tim Klaassen for his support during my final project, especially for his help on the test rig.
Special thanks go out to all students in the student’s room in general and to Rob Pulles in special for all
his help on the test rig and modeling.

Finally I want to thank my family and my girlfriend in special for all their support during my final study.

37