See discussions, stats, and author profiles for this publication at: https://www.researchgate.

net/publication/254806897

Analysis of Slip in a Continuously Variable Transmission

Article · January 2003

DOI: 10.1115/IMECE2003-41360

CITATIONS READS
47 789

5 authors, including:

B. Bonsen M. Steinbuch
Eindhoven University of Technology Eindhoven University of Technology
24 PUBLICATIONS   320 CITATIONS    634 PUBLICATIONS   11,900 CITATIONS   

SEE PROFILE SEE PROFILE

Some of the authors of this publication are also working on these related projects:

Computational design synthesis for discrete topology design View project

Variator efficiency optimization View project

All content following this page was uploaded by M. Steinbuch on 25 December 2014.

The user has requested enhancement of the downloaded file.

DSC TOC

Proceedings of IMECE’03
2003 ASME International Mechanical Engineering Congress
Washington, D.C., Novemberof15–21,
Proceedings 2003
IMECE2003
2003 ASME International Mechanical Engineering Congress and R&D Expo
November 15-21, 2003, Washington, DC USA

IMECE2003-41360
IMECE2003-41360

ANALYSIS OF SLIP IN A CONTINUOUSLY VARIABLE TRANSMISSION

B. Bonsen, T.W.G.L. Klaassen, K.G.O. van de Meerakker,

M. Steinbuch and P.A. Veenhuizen
Eindhoven University of Technology
Mechanical Engineering
Control Systems Technology
PO Box 513, WH -1.126
5600 MB Eindhoven, The Netherlands
Email: [email protected]

ABSTRACT pulleys which are wedge shaped. By changing the position of the
High clamping force levels reduce the efficiency of the Con- pulleysheaves the ratio of the CVT can be adjusted. The V-belt
tinuously Variable Transmission (CVT). However, high clamp- consists of blocks which are held together by two rings that in
ing force levels are necessary to prevent slip between the belt turn exist of a set of bands. To achieve torque transmission suf-
and the pulleys. If a small amount of slip is allowed, the clamp- ficiently high clamping force levels are needed to prevent slip in
ing force level can be reduced. To achieve this, slip in a CVT the variator. Because the torque level is not exactly known at all
is investigated. From measurements on an experimental setup, times, since no torque sensor is used due to cost considerations, a
Traction curve data and efficiency measurements are derived. A safe clamping force level based on the maximum possible load is
model describing slip in a CVT is verified using measurements maintained at all times. This safety level is based upon assumed
with a belt with increased play. It is found that small amounts maximum shockload levels from the road, like bumps, and the
of slip can be controlled in a stable way on the setup. The trac- engine torque.
tion curve was mostly dependent on the CVT ratio. Efficiency is In order to maintain these safety levels higher clamping force
found to be highest for 1 to 2% slip depending on the ratio. The levels are maintained then needed. Higher clamping force levels
model is in reasonable agreement with the measurements. cause more losses in the CVT. These losses are caused by in-
creases in power consumed by the hydraulic pump, by increases
in the losses due to slip in the belt if a pushbelt is used, and by
increases in deformation in the belt and in the pulleys. Further-
1 Introduction
more wear is increased and fatigue life is reduced.
Applying a Continuously Variable Transmission (CVT) in
In order to reduce these clamping force levels a method is needed
an automotive driveline has several advantages. A CVT can op-
to detect slip in the variator fast enough to prevent slip from
erate at a wider range of transmission ratios, therefore the engine
reaching destructive levels. A method to detect and control slip
can be operated more efficiently than with a stepped transmis-
is therefore needed.
sion. Also, a CVT does not interrupt the torque transmission
In this paper measurements are presented of the traction curve
when shifting. This gives a more smooth ride than a stepped
in a V-belt CVT. Also a quantitative evaluation is made of the
transmission does. A V-belt based Continuously Variable Trans-
model proposed by [1].
mission 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 figure 1. The variator consists of two

1 Copyright
c 2003 by ASME
 Active arc T Q
Secondary side
β
α
Primary side
Idle arc
Figure 2. CVT torque transmission scheme

The second variable in the traction curve is the slip in the variator.
Slip is defined as:

ωs
ν=| − 1| (2)
ω p r0

Where ωs is the angular speed of the secondary axle, ω p is

the angular speed of the primary axle and r0 is the geometrical
ratio, which is defined by:

Rp
r0 = (3)
Rs

R p is the running radius on the primary pulley. A model of

Figure 1. Layout of a CVT and a metal pushbelt the torque transmitting mechanism of the V-belt is given by [2].
This model gives insight on the tensile and compressive forces
acting within the bands and between the blocks of the belt (see
2 Traction curve figure 2).
The V-belt type CVT utilizes friction to transmit power
from the primary pulley to the secondary pulley. The traction
curve is the dimensionless relationship between transmit- 2.1 Tangential slip
ted torque and the slip. The maximum input torque that Slip is defined in equation 2. When the CVT transmits
can be transmitted by the CVT is dependent on the applied power a certain amount of slip can be measured almost linear
clamping force. The traction coefficient is therefore chosen to with the applied torque. This is called the microslip regime of
be a dimensionless value. The traction coefficient µ is defined as: the CVT, because traction is still increasing in this regime with
increasing slip. The microslip is caused by gaps between the
blocks on the idle part of the driving pulley as shown in figure
3 [1]. On the driving pulley an idle arc exists where no slip oc-
curs. Also an active arc exists (see figure 2), where slip occurs
Tq cos φ
µ= (1) relative to the total play in the belt and the active arc length.
2Fs Rs However, when the maximum torque capacity of the CVT is
reached slip will increase dramatically. This situation, macroslip,
In which Tq represents the input torque, Rs represents the is not stable during normal operation of the CVT, because the
secondary running radius of the belt on the pulley, Fs represents traction coefficient decreases with increased slipspeed.
the secondary clamping force and φ is the pulley wedge angle. It is assumed that the total gap δt is evenly distributed along

2 Copyright
c 2003 by ASME
 Equation 7 gives a value for the friction caused by viscous
friction component. Equation 8 gives a value for the coulomb
friction component. a0,1 , c0 and v1 are coefficients which can be
chosen to match the measured values.
With these equations we can derive slip and traction from mea-
sured data as shown in section 4. With Asayama [1995] we can
d
dm obtain the tension and compression force distribution needed to
calculate the lengthening of the belt. Also, we can calculate the
idle arc from this model. From the idle arc, the length of the belt
and the initial gap we can calculate an estimate for slip in the belt
for a given load.

2.2 Radial slip

Figure 3. Gaps in the belt Not only slip in tangential direction occurs, but also slip in
radial direction. The first reason for radial slip is spiral running.
When the belt runs along the arc of contact the radius at which it
the idle arc of the driving pulley. The traction curve (figure 5)
runs is not constant. This effect is caused by pulley deformation.
shows that torque transmission increases almost linearly with an
One type of deformation is the bending of the axle between both
increase in slip, until a certain maximum torque is reached. δt
pulley sheaves. The belt is not fully wrapped around the pulley,
can be estimated by adding an initial gap δo to the increase in
therefore the resulting normal force of the blocks on the pulley is
belt length due to the internal stresses in the bands and a decrease
not axial. This causes a bending moment in the axle [4].
in length of the blocks due to the compressive forces.
A second effect is the bending of the pulley itself. This effect is
mostly dependent on the local normal force exerted on the pulley
by the blocks. This effect is small when the belt is running on a
δt = δo + dL (4) small running radius, but on a large running radius this effect is
significant. The second reason for slip in radial direction is due
To calculate the slip caused by these gaps we can use the follow- to shifting. When the CVT is shifted to a different transmission
ing equations: ratio, radial slip is forced. This is done by changing the clamping
force ratio. The amount of radial slip that is forced depends on
the shifting speed and the (primary) angular speed.
d · δt
δm = (5)
αR p,s + δt + d
3 Experimental setup
δm
ν= (6) In the experiments the geometric cvt ratio is fixed and the
d + δm clamping forces are constant, the traction coefficient then de-
pends only on the slip in the system. The traction curve can
In equation 5, α is the idle arc, d is the width of a belt ele- be constructed from output torque and slip measurements. The
ment and δt is the total gap between the elements in the belt. To test rig motors deliver a maximum torque of 298 Nm with a
calculate the amount of slip the total gap δt has to be known. Fur- maximum speed of 525 rad/s. Both motors are equipped with
thermore the decrease of the friction coefficient with increasing a Heidenhain ERN1381 incremental rotary encoder with 2048
slip speed has to be taken into account as shown by Kobayashi pulses/rev. The torque at both sides is measured using a HBM
[1998]. For this purpose a Stribeck curve is used [3]. This ef- T20WN torque sensor. The maximum allowable torque is 200
fect has an influence on the traction coefficient in the macroslip Nm with speeds up to 1050 rad/s. A separate hydraulic unit is
regime. When macroslip occurs the traction will decrease with used to provide the required flow and pressure for the clamping
increasing slip. The Stribeck effect is modelled using equation 9. forces. Figure 4 gives a schematic overview of the experimental
setup.

f = c0 v (7)
− vv 2
4 Experimental results
g = a0 + a1 e 1 (8) The geometric ratio of the CVT was fixed during the exper-
µv = sign(v) · g + f (9) iments using a so-called ratio ring and the limits of the primary

3 Copyright
c 2003 by ASME
 0.12

M o to r 1 E n c o d e r
0.1
T o rq u e s e n s o r
M o to r 2 0.08

µ [−]
H y d r a u lic u n it
0.06

0.04

0.02
Figure 4. Experimental setup
0
0 1 2 3 4 5 6
Slip [%]
0.12
Figure 6. Traction coefficient in overdrive, ωs = 150, 225, 300
0.1

0.08
0.12
Overdrive
µ [−]

0.06 Medium 0.1

0.04 Low
0.08
0.02

µ [−]
0.06
0
0 1 2 3 4 5 6 0.04
Slip [%]

0.02
Figure 5. Traction coefficient at 300rad/s, ratio low(0.4), Medium (1.1)
and overdrive (2.26) 0
0 1 2 3 4 5 6
Slip [%]

pulley. This ratio ring limit the movement of the pulley. Primary Figure 7. Traction coefficient in low, ω p = 150, 225, 300
and secondary pressure was held constant (i.e. clamping forces
were held constant) during the experiments.
Traction curve
0.12

4.1 Traction coefficient 0.1

The traction coefficient was measured at different ratios, at
different primary speeds and at different pressures. In figure 6 0.08

and 7 can be seen that the traction coefficient depends little on 8 bar
µ [−]

0.06
primary speed or secondary clamping pressure, but mostly on 5 bar
the transmission ratio, as can be seen in figure 5. An increase in 0.04
clamping force causes more slip (see figure 8). This is caused by
0.02
an increase in tension in the bands and therefore in an increase in
length of the belt. This causes the play to increase. 0
0 1 2 3 4 5 6
Slip [%]

4.2 Efficiency Figure 8. Traction coefficient for resp. 5bar and 8bar secondary clamp-
The efficiency depends on pressure and on ratio. From figure ing pressure
12 can be seen that an increase in pressure causes a decrease in
efficiency. This effect is caused by the internal friction in the
belt. Slip between the blocks and the bands also causes a strong From figure 10 and 11 can be seen that input speed also has an
dependency on ratio (see figure 9). Efficiency is clearly higher influence on efficiency.
in medium than in overdrive or low. In medium no slip occurs
between the blocks and the bands, but in overdrive or low the
bands slip over the blocks. At high clamping levels this effect is 4.3 Play
greater, because the normal forces acting between the blocks and The microslip region is dependent on play in the belt. An
the bands increase linearly with an increase in clamping level. experiment has been carried out with a belt with increased play.

4 Copyright
c 2003 by ASME
 1
1
0.98
0.98 Medium 5bar
0.96
0.96
0.94
0.94
0.92
0.92

η [−]
0.9
η [−]

0.9 LOW
0.88
0.88 OD
0.86 8bar
0.86
0.84
0.84
0.82
0.82
0.8
0.8 0 1 2 3 4 5 6
0 1 2 3 4 5 6 Slip [%]
Slip [%]

Figure 12. Traction coefficient for resp. 5bar and 8bar secondary
Figure 9. Efficiency at 300rad/s, ratio low(0.4), Medium (1.1) and
clamping pressure
overdrive (2.26)

1
0.09
0.98 150rad/s
0.08
0.96 225rad/s
0.07
0.94
0.92
0.06

µ [−]
η [−]

0.9 0.05

0.88 300rad/s 0.04

0.86 0.03

0.84 0.02
0.82 0.01
0.8 0
0 1 2 3 4 5 6 0 1 2 3 4 5 6
Slip [%] Slip [%]

Figure 10. Efficiency in overdrive, ωs = 150, 225, 300 Figure 13. Effect of play in the belt, ω p = 30rad/s, in low, with in-
creased gap (1.8mm)

1
wp=300[rad/s] LOW
0.98
w =225[rad/s] LOW
p
0.96 wp=150[rad/s] LOW
0.94 0.1
0.92
η [−]

0.9 0.08
0.88
µ [−]

0.06
0.86
0.84
0.04
0.82
0.8 0.02
0 1 2 3 4 5 6
Slip [%]
0
0 1 2 3 4 5 6
Figure 11. Efficiency in low, ω p = 150, 225, 300 Slip [%]

Figure 14. Effect of play in the belt, ω p = 30rad/s, in overdrive, with

increased gap (1.8mm)
One block was taken out of the belt. The performance of the belt
was measured with a total gap of 1.8mm. The cumulative gap
in the belt was 0.3mm in the other experiments. A significant
difference is measured in the LOW ratio of the CVT. In figure
4.3 the traction curve is shown for the low ratio of the CVT for in overdrive there is no significant change in the traction curve,
the belt with increased play. Also the result of the numerical see figure 4.3. However, the model is less consistent with the
model is shown in figure 4.3. The results for overdrive show that tractioncurve in overdrive than in low.

5 Copyright
c 2003 by ASME
5 Conclusion
The traction curve is mostly ratio dependent. This can be
explained with the shown model as explained in section 4.
Transmission efficiency is dependent on applied pressure, input
speed and the CVT ratio.
Gaps between the blocks of the belt cause at least part of the
tangential slip of the belt. This was confirmed by the experiment
with increased play in the belt. The consistency of the model is
better in low than in overdrive.
Future research will be directed at controlling slip in the CVT.
This can enhance the efficiency of the CVT.

REFERENCES
[1] Kobayashi, D., Mabuchi, Y., and Katoh, Y., 1998. “A study
on the torque capacity of a metal pushing v-belt for cvt’s”.
SAE Technical papers [].
[2] H. Asayama, J. Kawai, A. T. M. A., 1995. “Mechanism of
metal pushing belt”. JSAE Review 16 [], pp. 137–143.
[3] H. Olsson, K.J. Astrom, C. C. d. W. M. G. P. L., 1997. Fric-
tion models and frictino compensation. Tech. rep., Lund In-
stitute of Technology.
[4] Sorge, F., 1996. “Influence of pulley bending on metal v-belt
mechanics”. Proceedings of the International Conference on
Continuously Variable Power Transmissions [].

6 Copyright
c 2003 by ASME