04CVT-57
Continuously Variable Transmission Modifications and Control
for a Diesel Hybrid Electric Powertrain
Copyright © 2004 SAE International
ABSTRACT
The Center for Transportation Research (CTR) Vehicle
Systems team modified a Nissan CK-2 Continuously
Variable Transmission (CVT) for a diesel hybrid
powertrain application. Mechanical and electrical
modifications were made to the CVT, both internal and
external to the transmission. The goal of this experiment
was to investigate and demonstrate the potential of CVT
for diesel engines hybrid electric vehicles (HEVs) in fuel
economy and emissions.
The test set-up consisted of a diesel engine coupled to
an electric motor driving a Continuously Variable
Transmission (CVT). This hybrid drive is connected to a
dynamometer and a DC electrical power source creating
a vehicle context by combining advanced computer
models and emulation techniques.
The experiment focuses on the impact particular
transmission control strategies have on measured fuel
economy and emissions specifically, nitrogen oxides
(NOx) and particulate matter (PM). The same hardware
and test procedure were used throughout the entire
experiment to assess the impact of different control
approaches.
INTRODUCTION
The CVT parallel hybrid configuration used in this
experiment provides tremendous flexibility in the choice
of both engine torque and speed operation. The electric
motor can replace, assist, or absorb the engine torque
independently from driver expectations. In addition, the
CVT allows decoupling between engine and wheel
speeds. This paper compares different transmission
control approaches and evaluates their impact on fuel
economy and emissions.
But first, in order to obtain a fuel economy and
emissions reference, we first operated the CVT in an
emulated manual transmission mode. Therefore, the
electric motor was disabled, and the CVT acted as a
manual transmission.
M. Pasquier
Argonne National Laboratory
Then, we controlled the hybrid CVT to keep the diesel
engine on its best efficiency curve. The best efficiency
curve describes the optimal engine operating point for
each power demand from an energy perspective.
Therefore, the engine torque and the CVT ratio were
both controlled to operate the engine at the most
efficient point while satisfying the power demand.
However, when the engine operates on its best
efficiency curve, it produces excessive NOx emissions.
We used simulation to design a trade-off between fuel
economy and NOx emissions.
For each engine power demand, we interpreted NOx
emissions and fuel consumption data to define the best
trade-off curve. The engine torque and the CVT ratio
were controlled to operate the engine on this curve while
satisfying engine power demand.
Fuel economy and emissions results obtained from the
experiments are compared and described in this paper.
The mechanical and electrical modifications made to the
CVT to improve efficiency and to control the
transmission are also discussed.
CVT MODIFICATIONS FOR HEV APPLICATION
The transmission was a modified Nissan CK-2 CVT,
which uses a Van Doorne push-type belt that is
commercially available in several Japanese production
vehicles. Mechanical and electrical modifications were
made to the CVT, both internal and external to the
transmission. In stock trim, an off-board transmission
control unit that controls torque converter lock-up, CVT
ratio, and hydraulic pressure accompanies the CVT. In
its original design, the mechanical hydraulic pump is
connected to the engine through the torque converter.
REMOVAL OF THE TORQUE CONVERTER
Hybridization requires disconnecting the engine from the
transmission by using a clutch, temporarily allowing the
electrical motor alone to propel the vehicle. The torque
converter was replaced by a clutch. The clutch benefits
system efficiency because clutch efficiency is better
than torque converter efficiency.
The reverse planetary gear was removed because HEVs
with electric-only capabilities can use the electric motor
for reverse.

CONTROL OF THE CVT RATIO

All CVT control is done with the ANL-developed PSAT-

PRO control software; additional hardware has been
added to support this approach [1]. The stock electronic
control unit for the transmission was not used. A new
driver for the stock ratio control stepper motor in the
hydraulic control circuit was installed to control CVT
ratio according to PSAT-PRO hybrid control computer.
Figure 3: Modified Nissan CVT

CONTROL OF THE CLAMPING PRESSURE

Supply clamping pressure is a key parameter for

efficient use of this type of transmission. Ideally, the
pressure should always be minimized to allow efficient
torque transmission while avoiding belt slipping (if
pressure is too low) or overheating and abnormal wear
(if pressure is too high).

Figure 1: Stepper motor for ratio control For this reason, a dynamic control algorithm has been
integrated into the powertrain controller to allow optimal
CVT operation. This algorithm uses instantaneous CVT
REMOVAL OF THE INTERNAL PUMP
input shaft torque and ratio measurement to calculate
optimum supply pressure and accordingly command
The main modification to the CVT was the removal of
hydraulic pump.
the internal high-pressure hydraulic pump. We replaced
it with an off-board pump. The pump is a key component
Figure 4 shows in parallel the CVT pressure, the input
of the CVT because it provides adequate belt/pulleys
torque used to dynamically control the hydraulic pump,
clamping. By using an external pump, much higher
and the vehicle wheel speed during a test.
power transmission efficiencies can be achieved
because an electrically driven pump allows the optimal
control of hydraulic clamping pressure by decoupling the
pump from the transmission input shaft (see Figures 2
and 3).

Figure 2: Removal of the CVT oil pump

Figure 4: CVT pressure control

The test data show a substantial improvement in the

efficiency of the transmission.

CVT INSTRUMENTATION

The bell-housing of the CVT transmission was modified

to mount a HBM T-10F flat torque sensor for measured
CVT input torque. Flanges that mate to the splined shaft
and the engine output were also machined.
 Figure 5: CVT torque sensor

The hydraulic pump was instrumented to evaluate its

energy use. This parameter needs to be taken into
account to validate the actual gain in efficiency of the
dynamic pressure control. Therefore, the power
consumed by the auxiliary pump was taken into
account. The electric power consumed by the pump was Figure 7: Modified CVT efficiency map
drawn from an emulated battery to reflect its impact on
fuel economy (see Figure 6). Figure 8 shows a comparison with a similar transmission
commercially available in the Toyota Opa CVT. This
Hi pressure
vehicle has been previously tested at the ANL’s
to CVT Advanced Powertrain Research Facility for a different
purpose, but the vehicle was instrumented with an
engine torque sensor and an axle torque sensor.
Therefore, data could be extracted from the tests for
Hydraulic CVT efficiency comparison [2].
pump Motor
Speed
controller

3-Phases motor
Low pressure
from sump

DC power source

Figure 6: Off-board hydraulic pump

We used pump specifications to evaluate its

instantaneous power. The manufacturer specifies pump
shaft torque and speed according to operating pressure
and flow. A constant efficiency is used to convert pump
drive power into electrical power. The power is then
drawn from the battery to discharge it and ultimately
impact fuel economy, which is corrected according to
battery state of charge.

EFFICIENCY COMPARISON Figure 8: Toyota Opa CVT efficiency data

The efficiency of the modified CVT was measured on
the complete speed, torque, and ratio range of the
transmission. The results of this mapping are shown in
Figure 7.
CVT CONTROL IMPACT ON A CONVENTIONAL
DIESEL POWERTRAIN
The use of a CVT for a conventional diesel powertrain
greatly increases the control possibilities because
transmission ratio can be changed continuously during
operation. In our case, the ratio ranges from 2.5 to 0.5
and is physically controlled by using a stepper motor.
There are several possibilities to operate the ratio during
a driving cycle. Ratio control is a key parameter of the
way the vehicle will perform on a given driving cycle.
For example, the ratio can be controlled to provide the
most efficient engine operation at all times. This study Results
will assess the impact of different CVT control
strategies. Test /11 /12 /13 Average Standard
103080/# Deviation
F.E. meas. 36.47 36.77 37.84 37.03 0.42
Because the ratio can vary continuously, the engine (mpg)
speed is decoupled from the vehicle speed (see Figure F.E. calc. 36.24 36.78 37.95 36.99 0.51
9). (mpg)
NOx 1.09 1.09 1.07 1.08 0.007
Engine (g/mi)
speed CO2 277.73 273.54 265.28 272.18 3.66
(g/mi)
Maximum speed
ar
CO 1.95 1.98 1.80 1.91 0.06
ar
ear

Ge Ge
ar

rd (g/mi)
Ge

th
1 st G

3 4 ar
Ge
2 nd

st
5 THC 3.03E-3 3.16E-3 2.67E-3 2.95E-3 1.47E-4
operational speed

O
Engine discrete

using gearbox

Engine operational (g/mi)

CVT

speed range using CVT

Test consisted of a Federal Urban Driving Schedule
O (duration, 1372 s; distance, 7.45 mi)
O
O
In the table, “F.E. meas.” denotes the fuel economy
Engine idle speed directly measured, while “F.E. calc.” denotes the fuel
economy calculated from emissions data. The average
Vehicle
speed of the three tests was used as a basis for comparison
during the entire study.
Specific vehicle speed

Figure 10 shows engine operating points from test data

Figure 9: Continuously variable ratio compared with during a FUDS cycle while applying a manual
discrete ratio transmission control with the CVT.
At each given vehicle speed, engine speed can be set to
any value allowed by the ratio range. For a manual or
automatic transmission, discrete engine speed must be
chosen among a limited number of gear sets.

As a consequence, engine operating point

(instantaneous engine torque and speed) can be set
freely (within range limits) while the vehicle is being
driven over a cycle. Setting the engine operating point
freely can greatly benefit fuel economy and emissions.
Adequate continuous control of the CVT ratio can keep
the engine running at a preferred efficiency operating
range, resulting in better fuel economy. This continuous
ratio change may also limit transient operation of the
engine, thus lowering pollutant emission.

DISCRETE RATIO CONTROL Figure 10: Engine operating point

The objective of this experiment was to obtain a The engine operated at low efficiency (<28%) most of
baseline for the study. The CVT was operated at five the time. Moreover, we noticed important speed and
discrete ratios; each ratio corresponded to one gear of a torque variations that had a negative impact on
manual transmission. An engine speed based shifting emissions. In the next experiment, we used the CVT to
schedule was used to determine adequate ratios during maintain the engine at a constant speed and assess the
vehicle operation. The clutch was fully disengaged impact of transients on emissions.
before and fully reengaged after a shifting event.
Although the durations of shifting events were higher IMPACT OF ENGINE SPEED TRANSIENTS
and CVT efficiency was lower than those characteristic
of an actual manual transmission (despite the torque To put a figure on the impact of engine speed transients
converter removal), using this shifting approach on fuel economy and emissions, the CVT ratio was
recreated engine operation as it would be in the context operated continuously to keep the engine at a constant
of a manual transmission vehicle. Therefore, this operating speed of 2000 rpm during this experiment.
experiment allowed assessing the impact of Therefore, CVT ratio operation was proportional to
transmission control only without considerations of vehicle speed during the cycle to minimize engine
transmission efficiency since the same hardware was speed transients.
used during the entire study.
Results torque converter was replaced by a clutch, and the CVT
gear pump was replaced by an auxiliary electrically
Test /17 /19 /20 Average Standard % driven pump. These changes benefited system
103080/# Deviation gain efficiency because clutch efficiency is better than torque
F.E. meas. 40.64 41.37 41.69 41.23 0.54 11% converter efficiency and also because an electrically
(mpg) driven pump allows optimal control of hydraulic
F.E. calc. 40.33 39.11 41.67 40.37 1.28 9% clamping pressure by decoupling the pump from the
(mpg) transmission input shaft. Moreover, compared with the
NOx 0.9 0.87 0.89 0.89 0.02 -18% existing pump, an electrically driven pump can be
(g/mile) operated when the engine is off and the vehicle is
CO2 249.99 257.54 241.88 249.8 7.83 -8% propelled electrically.
(g/mile)
CO 1.47 1.67 1.49 1.54 0.11 -19% Transmission input torque, speed, and ratio allowed
(g/mile)
calculation of the optimal clamping pressure. Closed
THC 2.44E-03 2.68E-03 2.46E-03 2.53E-03 1.30E-04 -14%
loop control of the pump provided adequate clamping
(g/mile)
pressure.

The test consisted of a Federal Urban Driving Schedule Results

(duration, 1372 s; distance, 7.45 mi); % Gain compared
with previous experiment Test /14 /15 /16 Average Standard %
103080/# Deviation gain
This experiment involved using a very simple control F.E. 43.3 43.56 43.04 43.3 0.26
algorithm that is not realistically applicable in a real- meas.
vehicle context. However, the results show the impact of (mpg) 17%
engine speed transients. We observed a gain of 10% in F.E. 42.54 44.28 41.88 42.9 1.24
fuel economy and a substantial reduction in emissions. calc.
Limiting engine transients improved both fuel efficiency (mpg) 16%
and emissions without a trade-off. This inference will be NOx 0.88 0.86 0.85 0.86 0.02

used in the development of a hybrid control strategy (g/mile) -20%

because the electric motor could compensate for a CO2 236.24 227.3 239.88 234.47 6.47
(g/mi)
driver’s high dynamics demands while limiting engine -14%
CO 1.89 1.6 1.6 1.7 0.17
transients.
(g/mi) -11%
THC 2.68E-03 2.49E-03 2.52E-03 2.56E-03 1.00E-04
Figure 11 shows engine operating points from test data
(g/mi) -13%
during a FUDS cycle while maintaining engine speed
constant. The test consisted of a Federal Urban Driving Schedule
(duration, 1372 s; distance, 7.45 mi); % Gain compared
with baseline experiment

This experiment was the same than the constant engine

speed operation, except for the control of the clamping
pressure. In the previous experiment, the clamping
pressure was set to a high value to avoid any belt
slippage. In this experiment, we wanted to evaluate the
benefit of operating the pump at a lower pressure by
using our pump control algorithm.

As expected, applying an optimal clamping pressure

Figure 11: Engine operating point – Constant speed
operation
IMPACT OF CLAMPING PRESSURE CONTROL
As described previously, the main modification of the
CVT was to replace the mechanical hydraulic pump,
which was connected to the engine through a torque
converter. The pump is a key component of the CVT
because it provides adequate belt/pulleys clamping. The
increased CVT efficiency and consequently improved
fuel economy by 5%. We also noticed a general trend
toward decreasing NOx and CO2 emissions. Concerning
CO and THC, the average results show higher level of
emissions. However, because of the uncertainties
associated with emissions data collection, the
comparison of averages is of limited accuracy. If we
look at the individual test data, the first test of the
second experiment generated 1.89 g/mi of CO. This
value is higher than that of the other tests and
contributed to a 10% increase in average CO emissions.
Concluding that, on the basis of a single test, pressure
control of the CVT causes an increase in CO emissions
would be erroneous. Because of its benefit on fuel
economy, the control algorithm was used for the between B and C, we can conclude that the difference in
remainder of the study. efficiency is not significant.

IMPACT OF ENGINE OPERATION

180

84

0. 37
0.290.287 49
75

0. 37
14

0.3 78
65

9
94

014
0.3309
0.283

184

0.3527
55

884
0. 317
9

449
0.30 0. 29 619

0.357
0.3222

0. 37
45

As we explained earlier, the CVT allows engine speed to

054

09
9
35 70

0.34 404 0.3396

60

24
0.313 0.309
B
160 25 50

0.339
be set to any value within the ratio range. Thus, we had

59
89
65

0.3 7449
0.304
40

69

0. 32664
55

19
an additional degree of freedom (compared with a

83
15 30 45 60

0.36 579

0. 3
140
50

0. 34839
35

4
66
55

manual or automatic transmission) to operate the

0.37 014

0.36 144
0.32

534

9
20 40

57
0. 33

0. 36
45

709
120 25 50

0. 35
engine. During this experiment, to improve fuel
4
4 27
39 30 14 35

9
48 36 0.

0.3222
0.3 0. 35

ENGINE Nm
40
04 45
44

economy, we applied continuous control of the CVT

0.3
100
10 A 15 20 25 30
0. 33
53435 40

ratio to keep the engine running near its most efficient 80

0. 35
709

25
0. 33
099
30
35

operating point for any given power demand.

20 489
0.35 274 0. 30 54
0.33969 794 0.309
24 0.300
0.34839 15 0. 31 30
59 25
0. 34404 0.313

C
10 0.32 664 20 0. 2961
9
60 0. 29184 0.28749 25
0.33534
9 15
32 22 20 0. 28314
0.33099 0. 0.27 879
0.27 444

The most efficient operating area for each engine’s

0.27 009
5 10 15 0.26 574 20
0. 31794 0.26139
0. 31359 0.25704
40 0.30 924 15 0.25269
0. 304890. 30 054 10 0.24 834
0.24399

power demand yields the best engine efficiency curve,

15
0.23964
5 29 619 0.28 749 10 0.23 094 0.22659
0. 87944
184 0.28 314 0.27 0. 2700 9
0.274 2489 10 0.20919
0.20 484
0. 29 0. 26574
0.222
0.21354
0.217 0.20 049
139 5 10

as shown in the Figure 12.

0.26
0.25 704 399 29
48340.234964 0. 19179 0.18 744 0.18 309
20 0.25 269 0.2 0.2 0.235 5 0. 19614 0.17874
0.17 439
0.17004
1000 1500 2000 2500 3000 3500 4000

ENGINE RPM

Figure 13: Selection of lower operating speed

To complete the analysis, we can compare the

efficiency in the low power area at 1500 rpm with the
low power area at 1900 rpm (blue arrows in Figure 13).
Similarly, the difference in efficiency is not significant
and, therefore, according to the results of the simulation,
we should only expect a modest improvement in fuel
economy by setting the low engine speed limit at
1500 rpm instead of 1900 rpm.

However, if we impose a low engine speed limit at

Figure 12: Engine best efficiency operating curve
CVT offers additional control opportunities to maintain
the engine on its best efficiency curve. However, for a
low power-level requirement (below 15 kW), the engine
would have to operate at a low speed (below 1500 rpm)
to stay on the curve (point A). Thus, we needed to
impose a low engine speed limit so that engine torque
could leave the curve to satisfy driver’s demand. This
engine speed limit was selected by using simulation,
and an example of the simulation analysis is discussed
below for two different engine speed limits (1500 rpm
and 1900 rpm) [3].
If the minimum engine speed is set at 1500 rpm, for an
engine power (1) below the intersection between
1900 rpm and the best efficiency curve (point B) and (2)
above the intersection between 1500 rpm and the best
efficiency curve (point A), the engine is kept on the best
efficiency curve (see Figure 12).
1500 rpm, for low vehicle speed and low acceleration
demand conditions, we will generate more engine torque
transient than if the limit is set at 1900 rpm. Figure 13
shows that it takes more torque variation to go from
10 kW to 15 kW if the engine is at 1500 rpm than if the
engine is at 1900 rpm.
Even if the engine stays longer on its best efficiency
curve and the low engine speed limit is 1500 rpm, the
fuel economy impact is not significant. However,
imposing a low engine speed limit of 1900 rpm reduces
engine torque transient under conditions of low vehicle
power. Therefore, for subsequent research, we operated
the engine above 1900 rpm.
Figure 14 shows engine operating points from test data
during a FUDS cycle as the engine is maintained on its
best efficiency curve.

If the low engine speed limit is set at 1900 rpm, for the
same engine power requirement, the engine stays at a
speed of 1900 rpm and the torque decreases to match
power demand. The engine operates between points B
and C, as shown in Figure 13. However, when
comparing the area between A and B with the area
Figure 14: Engine operating point – Best efficiency
curve operation

Figure 14 shows that the control does not maintain the

engine at its best efficiency all of the time. This test was
performed at a low engine speed of 1500 rpm to confirm
the impact of torque transient on emissions predicted by
simulation. Therefore, when the power level required
was below 15 kW, the control reduces the engine torque
and leaves the curve. Moreover, at high vehicle speed,
the CVT ratio range becomes a constraint. At high
vehicle speed, once the ratio reaches its minimum, the
engine speed keeps accelerating with vehicle speed. Figure 15: Comparison between the different
Acceleration becomes particularly problematic when the conventional diesel powertrain control approaches
power demand decreases because we have to operate
the engine at low efficiency. Moreover, if we are not Those experiments introduced the notion of a trade-off
limited by the CVT ratio range, we still have difficulties between fuel economy and emissions. They also
in controlling the ratio dynamically enough to follow the demonstrated two limitations of the CVT in a
power variations. conventional diesel vehicle: ratio range and CVT
dynamics.
Results
Those issues can be addressed in a hybrid environment
Test /01 % gain
where the electric motor can compensate for the slow
103090/#
CVT dynamic and charge the battery to store energy
F.E. when it is more efficient to operate the engine at a
meas. higher load.
(mpg) 42.22 14%
F.E. CVT CONTROL IMPACT ON A HYBRID DIESEL
calc. POWERTRAIN
(mpg) 44.45 20%
NOx Hybridization brings a second degree of freedom
(g/mi) 1.05 -3% compare to a conventional powertrain with a CVT. The
CO2 electric motor can compensate for the slow dynamic of
(g/mi) 226.55 the CVT ratio change but also allows the engine to
-17%
operate at a higher power than that demanded by the
CO
driver. The excessive power can be used to recharge
(g/mi) 1.49 -22% the battery by using the motor as a generator. If the
THC battery state of charge (SOC) becomes too high or the
(g/mi) 2.51E-03 -15% driver’s power demand too low, an option is to turn off
the engine and propel the vehicle in electric-only mode.
The test consisted of a Federal Urban Driving Schedule Consequently, the control of the engine is more
(duration, 1372 s; distance, 7.45 mi); % Gain compared accurate than with a conventional powertrain with CVT.
with baseline experiment
IMPACT OF ENGINE OPERATION
This experiment demonstrates that we can achieve the
same fuel economy by operating the engine on its best Engine Best Efficiency Operation
efficiency curve as by operating the engine at 2000 rpm.
We can assume that the higher-efficiency operating Figure 16 displays typical engine operation during a test
point compensates for the transient effect on fuel that was performed with the best efficiency strategy.
economy. However, NOx emissions increase. When Engine power demand is calculated to match driver
performing the steady-state emissions map of the expectations and SOC requirement. The engine power
engine, we noticed that the best efficiency curve is at a required is then interpreted by the control strategy in
very high level of NOx. Nevertheless, looking at the order to set engine torque command and CVT ratio
measured engine operating point in Figure 14, it is command. As a result, the engine operates near the
difficult to determine if the 22% increase in NOx best efficiency curve.
(compared to previous experiment) was due to engine
operating area or engine transients.

Figure 15 shows a summary of the study of the

conventional diesel powertrain.
 Best efficiency strategy Fuel Economy (FE)

80
75
70
Corrected FE : 61.1mpg 65
60
55
50
Corrected FE : 58.2 mpg 45

FE (mpg)
40
35
30
25
20
Calculated FE test value
15
Measured FE test value 10
5
0
-25 -20 -15 -10 -5 0 5 10 15 20 25 30 35
Delta SOC (%)

Figure 16: Engine operation point using the best Figure 17: Regression of fuel economy versus delta
efficiency strategy SOC – Best efficiency strategy

The color density on this plot is proportional to the time Test

103110/#
/01 /02 /03 /06 /12 /15 Corrected
FE
Average Standard
Deviation
%
gain
spent by the engine on the operating area. By F.E. 45.23 58.33 67.23 52.69 48.38 69.95 58.2 57 1.4

comparing Figure 14 with Figure 16, we notice that the meas.

(mpg) 57%
engine stays closer to the best efficiency curve. At high F.E. 44.5 61.19 69.39 55.63 50.69 76.08 61.1 59.6 2.3

load and an engine speed above 2300 rpm, the control calc.
(mpg) 65%
reduces the engine torque because the motor cannot Delta 29.8 -1.91 -19.33 9.71 18.48 -21.06 0 2.62 NA
SOC
absorb all of the excessive power generated by the (%) NA

engine. Because the motor is above its base speed, it is NOx

(g/mi)
1.77 1.29 1.12 1.59 1.84 1.13 NA 1.46 0.32
35%
limited in maximum negative torque capacity. The PM NA NA NA 0.086 0.087 0.061 NA 0.078 0.015

driver’s power demand takes priority on the best (g/mi)

CO2 228.7 166.3 146.6 182.8 200.4 133.8 NA 176.4 35.1
NA

efficiency curve, and the control has to leave the (g/mi) -35%
CO 4.30E-02 2.20E-02 2.90E-02 8.20E-02 #NV# #NV# NA 0.044 0.027
optimal (in terms of fuel consumption) mode of (g/mi) #NV#
operation. We also notice that we can operate the THC #NV# 3.30E-03 3.80E-03 9.40E-03 4.90E-03 9.40E-03 NA 0.0062 0.003
(g/mi) #NV#
engine at a lower speed than in conventional
configuration and stay on the curve.
The test consisted of a Federal Urban Driving Schedule
(duration, 1372 s; distance, 7.45 mi); % Gain compared
Results
with baseline experiment
Both fuel usage and battery SOC must be taken into
With the assistance of the electric motor in hybrid mode,
account when calculating or measuring the fuel
the engine operates closer to its best efficiency curve.
economy (FE). Each test consists of the Federal Urban
As a consequence, fuel economy increases by 38%, in
Driving Schedule. The vehicle’s performance is
comparison with the same control principles used in a
evaluated over several tests to check test repeatability.
conventional mode with CVT and by 57% in comparison
Initial and target battery SOC is changed before each
with the CVT being controlled as a 5-speed manual.
test. As a result, we have a range of SOC variation that
This experiment is critical to demonstrating the potential
allows a corrected FE calculation.
of CVT control. The 57% improvement in fuel economy
denotes the impact of control, but the same components
Figure 17 shows how to estimate the corrected FE (i.e.,
were used in both experiments (baseline and hybrid).
what the fuel economy would be in case of SOC
The baseline powertrain uses all of the components of
equalization over the test cycle).
the complete hybrid powertrain. To consider only the
control aspect, the baseline is penalized by the CVT
transmission operating as a manual transmission (which
would be more efficient) and by the electric motor inertia
and bearing losses.

Despite the improvement in fuel economy, we also

noticed a 39% increase in NOx emissions. We already
observed this tendency in conventional operation, but
now that the engine is always operating on the best
efficiency curve, we can correlate this increase in NOx
emissions to the engine operating point. Figure 18
shows the engine operating points on the NOx emissions strategy in order to set engine torque command and
map of the engine. CVT ratio command. As a result, the engine operates
near the trade-off curve.

180 3.06 67e-005

-005
7 12

0.0001
0.000 0. 000 0.000
0.00 0120.00 0125 6 7
7. 36 33e-
010
9.2e-006

10 42

0. 00011 04

0.0001 88
7. 05

0.00 010733
8. 8.5

0.00011347
0.000

9.2e-0 05
1.5333e -005

0. 00

7.
5.82 67e
28 8 9.50

9.5067e
5.52 e- 005 33e -005 0 67e

e- 00 005
66
e-0 67e

4.29 33e-005

0.0001 1653

14413 141 13 8
2. 76e-00 5

4.6e-005
3.98 67e-005
2.4533e-005

67

472
67
0 5 -00

318 7
e -0

0.00 013
1.84 e-005

6.7

e-
9.8133e-005
3.68 e-005

6. 44 6. 13

0.00 0122
3.3733e-005

5 05

00
46

5.21

-005

5
7e 8.8933 9.2e -005

e- 00 33e
1.2267e- 005

-0 e-005
160 05 7. 97

493

07
73
33e-

5 -005
005
4 .9

2
01
01 2 7
00 0 4
-0

8. 28

8.5

0.00 01196
0. 001
05

e-00

86
140 7. 057 .3 0
5 0.

7e
-005

5.52 5.82 67 33e6 e- 7.6667e

9.506
-
6.44 6. -00 0 5

00
4.6e-0 -005
5.e- e -0

5
2.14 67e

2100

9. 81
05 e- 007405
4.2933e-005 335e- 05 6. 13 33e567e -005

7e -00 89 3
4.00

33
0. 0
3. 9867e -005 90567 -005

e -0 9.2 05
0 01

8.
e-00 0 .0

5 3e
05 5 0 0 1 04
3.68 e-0

05 e
7.9
3.066 10
120

73
7e 73

-0
05 2.76e-005 -005

3e
e -0 3

-0
3 .3

-0
-005 7 33

05
2.45 33e
ENGINE Nm

05
73
67e 3. 3 -005 3e 3.6

0.0
3. 06 2.1467e-005 -00 8
5 e- 0 .98
3

00
0 5 67

11
1.84e-005

4.
e- 4. 6

29

34
5.5 5.

7.0 6
7. 5 .446. 13
5 00 e -0
100

33

36 33 6 e-0 3

7
3.06 67e-00 1 .5 5 2 82

7.6-00-0045 6 0 5 e -0

0.
5.e2-0 0 67e

e-0
05

e e .7

00
3 33

0 .0 12 9.
6 6 5 7e -0 05
05
1 335 -0 0

8.2 5

01
8.58 e-0
7e
e -0

0 01
e -0 5

0
2.76e-005

-0

86 05
1. 22 05

04
4.9 05

7e
67e-

2 7 9. 81 33
0 67

-0
2. 4533e-0 05 005

05

05
e -0

50
2.
76
05

67e 8. 89
-006

80 2.14 67e-

e-0
005 3. 06

-005 33e
2.4

05
67e
9. 2e

e -0 9. 2e05
1.84 e-0 53

2 .1
-005

7. 97
1.533 05 3e 3.37

05
46
-0

1.
1.22 3e -00 05 33e

33e
7e
84 3e -0
9.2e
1.
-003 .68

-0
67e 5

-0
e-0
53
-005

-005
-006

05

-0 05
1.2 5 e-0
3

3.9
05 5
26 05

86
4.29
60 7e
0

7e
-0

4.6e-005
9.2e-0

-0
05

303e5
06

-005

5.52e
5.826 05 3e -00
6
-00

7.0533
7.36e e-
6.133
33e

5.2
2. 76

-0
3e- 006

7e -00 5067e- 00
3. 0 6.13

7.6-00
13
6 67

6.44e-
40

e- 002.4

6 67e
e- 00

6.7
5
6.1333
005
6

2. 14

4.9

5
5 5367e
1.843 33 e

467
1. 5

3.0667-005
1.2

-005
005e-005
9.
6.13 2e-0 2 67

3e -005

e-005
e- 00-0 0
33e 06 e -0

e- 005
006

55
-006 05
67e-

5
20 3.0667e- 006 3. 06

1000 1500 2000 2500 3000 3500 4000

ENGINE RPM

Figure 18: NOx emissions map based on steady-state

data

This map has been established by using ANL test data

Figure 20: Engine operation point using the trade-off
of steady-state engine operation [4]. The best efficiency
strategy
curve is located in high NOx-emissions areas.
The color density on this plot is proportional to the time
This experiment demonstrates the need for a trade-off
spent by the engine in the operating area. By comparing
between fuel economy and harmful NOx emissions.
Figure 16 with Figure 20, we see that the engine is
operating at lower load and higher speed. Consequently,
Engine Trade-Off Operation
the motor is able to absorb the excess power generated
by the engine and is not limited by its negative torque.
Nitrogen Oxide (NOx) emission data allow definition of a
minimal NOx curve that can be used to determine the
Results
lowest NOx production at each power. For each power, if
several NOx minima exist, the controller selects the
most efficient operating point. The engine torque and Reduced NOX strategy Fuel Economy (FE)
the CVT ratio are controlled to operate the engine on 70
this NOx/FE trade-off curve while satisfying engine
power demand. Figure 19 shows the trade-off curve that 65
has been calculated in simulation [5].
60
FE (mpg)

180
55
3.06 67e-005

-005

7 12
0.0001
0.000 0. 000 0.000
0.00 0120.00 0125 6 7
7. 36 33e -0

010
9.2e-006

10 42
0. 00011 04

0.0001 88
7. 05

0.00 010733

8. 8.5
0.00011347

0.000
9.2e-0 05
1.5333e -005

0. 00
7.
5.82 67e

28 8 9.50
9.5067e
5.52 e- 005 33e -005 0 67e

e- 00 05
66

e-0 67e
4.29 33e-005

0.0001 1653

14413 141 13 8
2. 76e-00 5

4.6e-005
3.98 67e-005

Corrected FE : 53.7 mpg

2.4533e-005

67
472
67

0 5 -00
318 7

e -0
0.00 013
1.84 e-005

6.7
e-0

9.8133e-005
3.68 e-005

6. 44 6. 13

0.00 0122
3.3733e-005

46 5 05
5.21

05
-005

7e 8.8933 9.2e -005

e- 00 33e
1.2267e- 005

Corrected FE : 52.3 mpg

-0 e-005
160 05 7. 97
493

07
73

33e-
5 -005

005
4 .9

2
01
01 2 7
00 0 4 50
-0

8. 28
8.5

0.00 01196

0. 001
05

e-00
86

140 7. 057 .3 0
5 0.
7e
-005

5.52 5.82 67 33e6 e- 7.6667e

9.506
-0

4.6e-0 6.44 6. -00 0 5 -005

05

5.e- e -0
2.14 67e

2100
9. 81

05 7405
4.2933e-005 335e- 05 6. 13 e- 00 67
7e -00 89 3

4.00 5 e- 005
33e-
Calculated FE test value
33

0. 0
3. 9867e -005 90567 005
e -0 9.2 05

0 01
8.

e-00 0 .0
5 3e

05 5 0 0 1 04
3.68 e-0
05 e
7.9

3.066 10
120
73

7e -00

45
73
-0

05 2.76e-005
3e

e -0 5 3
-0

3 .3
-0

-005 33
05

2.45 33e
ENGINE Nm

05

7 73
67e 3. 3 -005 3e 3.6
0.0

3. 06 2.1467e-005 -00 8 3
5 e- 0 .98
Measured FE test value
00

0 5 67
11

1.84e-005
4.

e- 4.6
29

34

5.5 5.
7.0 6
7. 5 .446. 13

5 00 e -0
100
33

36 33 6 e-0 3

3.06 67e-00 1 .5 5 2 82
7.6-00-0045 6 0 5 e -0

0.

5.e2-0 0 67e
e-0

05
e e .7

00

3 33
0 .0 12 9.
6 6 5 7e -0 05
05

1 335 -0 0
8.2 5

01
8.58 e-0
7e

e -0
0 01

e -0 5
0

2.76e-005
-0

86 05

1. 22 05
0

04

4.9 05
7e

67e-
2 7 9. 81 33

40
0 67
3

-0

2. 4533e-0 05 005
05

05

e -0
50
2.
76

05
67e 8. 89
-006

80 2.14 67e-
e-0

005 3. 06
-005 33e

2.4
05

67e
9. 2e

e -0 9. 2e05

-25 -20 -15 -10 -5 0 5 10 15 20 25

1.84 e-0 53
2 .1

-0 05
7. 97

1.533 05 3e 3.37
05
46

-0
1.

1.22 3e -00 05 33e

33e
7e
84 3e -0

9.2e
1.

-00 3 .68
-0

67e- 5
-0
e-0
53

Delta SOC (% )
-005

-0 06 005
05

-0 05

1.2 5 e-0
3

3.9
05 5

26 05
86
4.29

60 7e
0

7e

-0
4.6e-005

9.2e-0
-0

05
33e
05

06
-0 05

5.52e
5.826 05 3e -00

6
-00
7.0533
7.36e e-

6.133
33e
5.2
2. 76

-0

3e- 006
7e -00 5067e- 00

3. 0 6.13
7.6-00
13

6 67
6.44e-

40
e- 002.4

6 67e

e- 00
6.7
5
6.1333
005

6
2.184

4.9

5
5 533e
1. 3 33 e e -0

467

Figure 21: Regression of fuel economy versus delta

1. 5 2 6

3.0667-005

-005
005e-005
46e-

9.
1.2

6.13 2e -0
e-005
7e00

33e 0 6
7

-0505
-0 0 05

e- 005

006 -006
67e-
5

20 3.0667e- 006 3. 06

1000 1500 2000 2500

ENGINE RPM
3000 3500 4000
SOC – trade-off strategy

Figure 19: NOx/FE trade-off curve

Figure 20 displays typical engine operation during a test

that was performed with the trade-off strategy. Engine
power demand is calculated to match driver
expectations and the battery SOC requirement. The
engine power required is then interpreted by the control
Test /08 /10 /11 /16 /17 Corrected Average Standard %
The NOx/fuel economy trade-off approach yields a 50%
103110/# FE Deviation gain reduction in NOx and yet improves fuel economy by
F.E. 51.97 50.36 42.97 59.98 60.25 52.3 53.1 0.5
meas.
41%. However, it also results in an increase in PM
(mpg) 41% emissions.
F.E. 54.11 52.48 44.64 59.55 61.09 53.7 54.4 0.9
calc.
(mpg) 45% Those results refer to an emulated conventional vehicle,
Delta
SOC
1.26 2.7 21.53 -16.86 -17.33 0 -1.74 NA
which is actually penalized by the inefficiency of a CVT
(%) NA
acting as a manual transmission, as well as the losses
NOx 0.521 0.559 0.819 0.407 0.409 NA 0.543 0.17 and inertia of a disabled electric motor. The NOx and
(g/mi)
PM 0.119 0.125 0.211 0.112 0.112 NA 0.136 0.042
-50%
PM emissions results are averages of the tests
(g/mi) NA performed and are not SOC corrected. Consequently,
CO2
(g/mi)
187.2 193.4 227.4 170.3 165.6 NA 188.8 24.5
the average delta SOC should be considered in order to
-31%
CO 0.372 0.26 0.272 0.293 0.456 NA 0.331 0.083 interpret the analysis correctly.
(g/mi) NA
THC 8.07E-02 2.22E-02 2.43E-02 1.97E-02 7.60E-02 NA 0.0446 0.0309
(g/mi) NA CONCLUSION

The test consisted of a Federal Urban Driving Schedule In this study, we presented different control approaches
(duration, 1372 s; distance, 7.45 mi); % Gain compared for the CVT from conventional 5-speed manual mode to
with baseline experiment hybrid trade-off strategy. The objective of the study was
not to optimize the control strategy but to assess the
Compared with the previous experiment, we have a fuel impact of CVT control and to demonstrate its potential
economy penalty, but NOx emissions are reduced. for helping diesel hybrid powertrain to minimize the fuel
Figure 22 and 23 show a comparative analysis of the economy penalty and yet meet emissions regulations.
baseline experiment and the two hybrid modes of Figure 24 shows a comparison of our results for the
operation. experiments in conventional mode, and Figure 25 shows
the results for the experiments in hybrid mode.
180
Fuel Economy result Nitrogen Oxydes result
Fuel
160
Economy Conventional 5-speed emulated based on FUDS
1 Conventional best efficinecy strategy based on FUDS
42 mpg
Tier1 based on FTP
140 0.9
37 mpg LEV based on FTP
comparison to reference (%)

0.8

0.7
120 + 57% +41% + 35% 0.6

0.5
100
0.4
CO2 THC
0.3
272 g/miles 226 g/miles
80 0.2
- 50% 0.1
60 0 0.03 g/miles

40
1.49 g/miles

20 37 58.2 52.3 1.08 1.47 0.61 1.91 g/miles

mpg mpg mpg g/mile g/mile g/mile 1.05 g/miles

0
FE mpg
Conv reference HEV best efficiency nox
HEVg/mile
optimal nox
1.08 g/miles

NOx CO

Figure 22: FE and NOx result summary chart

1.6
Figure 24: Comparison analysis for the conventional
PM emission NOx emission
mode
1.4
1.46
Figure 24 shows that CVT control has a limited impact
PM and NOx emission (g/mile)

1.2 g/mile

1 on emissions for a conventional powertrain. In hybrid

0.8 mode, we increased the degree of freedom and the
0.6
potential for control impact on both emissions and fuel
0.54
economy.
0.4
g/mile
0.078 0.136
0.2 g/mile
g/mile
0
pm
HEVg/mile
best efficiency nox
HEV optimal g/mile
NOx

Figure 23: Hybrid operation PM and NOx summary

Operating the engine with the best efficiency approach

increases fuel economy by 57%. This efficient engine
utilization results in a 35% increase in NOx emissions.
 Fuel
Studies Using Advanced Control and Testing
1
Economy
58 mpg
Hybrid trade-off strategy based on FUDS
Procedures, 18th International Electric Vehicle
Symposium (EVS18), 2001, Berlin, Germany.
Hybrid best efficiency strategy based on FUDS
0.9
52 mpg Tier1 based on FTP
0.8 LEV based on FTP

0.7

4. Cole, R.; Hillman, G.; and Sekar, R., Baseline

0.6
CO2 THC
0.5
176 g/miles
189 g/miles
0.4

0.3
Performance and Emissions of 1.7-Liter Mercedes
0.2

0.1
Benz Engine, January 2002, DOE Report.
0.04 g/miles
0
0.33 g/miles

5. Pasquier, M.; Duoba, M.; Hardy, K.; Rousseau, A.;

0.08 g/miles

0.54 g/miles
and Shimcoski, D., Evaluation of a CIDI Pre-
PM 0.14 g/miles CO
transmission Parallel Hybrid Drivetrain with
Continuously Variable transmission, 19th
International Battery, Hybrid and Fuel Cell Electric
NOx
1.46 g/miles
Vehicle Symposium & Exhibition (EVS19), 2002,
Busan, Korea.
Figure 25: Comparison analysis for the hybrid mode
CONTACT
Figure 25 shows the trade-off between fuel economy
and NOx emissions. It demonstrates and quantifies CVT Maxime Pasquier
control impact on fuel economy and emissions for diesel Argonne National Laboratory
hybrid vehicles. To complete the evaluation of diesel 9700 South Cass Ave.
hybrid technology, after-treatment devices should also ES/362-H220, Argonne, IL 60439-4815
be considered. Optimal control of the system would Tel: 630-252-9717 Fax 630-252-3443
include engine calibration and advanced after-treatment E-mail: mpasquier@anl.gov
devices, such as particulate filters or an NOx absorber.
At this time, particulate filter technology is more mature Maxime Pasquier is head of the Hardware-In-the-Loop
than the NOx absorber. However, the development of an Department at Argonne National Laboratory. He
after-treatment control integrated to the vehicle control received his engineering diploma at the Industrial
strategy would complete the demonstration of a diesel System Engineering School in La Rochelle, France. He
hybrid as a short-term bridge to a hydrogen economy. then received Technology Research Diploma from La
Rochelle University based on Hybrid Electric Vehicles
modeling. After working for PSA Peugeot Citroen for 3
ACKNOWLEDGMENTS years in the Hybrid Electric Vehicle research
department, he joined Argonne National Laboratory
This work was supported by the U.S. Department of where he is now responsible of the advanced vehicle
Energy, Office of Energy Efficiency and Renewable technology validation in a vehicle system
Energy, under contract W-31-109-Eng-38. The authors
are grateful for the support of and guidance by the
FreedomCAR and Vehicle Technologies Program.
DISCLAIMER

This report was prepared as an account of work

REFERENCES sponsored by an agency of the United States
Government. Neither the United States Government nor
1. Rousseau, A., and Pasquier, M., Validation of a any agency thereof, nor The University of Chicago, nor
Hybrid Modeling Software (PSAT) Using its any of their employees or officers, makes any warranty,
Extension for Prototyping (PSAT-PRO), express or implied, or assumes any legal liability or
Proceedings of the 2001 Global Powertrain responsibility for the accuracy, completeness, or
Congress, Detroit, MI, USA. usefulness of any information, apparatus, product, or
process disclosed, or represents that its use would not
2. Min, B.; Matthews, R.; Duoba, M.; Ng, H.; and infringe privately owned rights. Reference herein to any
Larsen, R., Direct Measurement of Powertrain specific commercial product, process, or service by
Component Efficiencies for a Light-Duty Vehicle with trade name, trademark, manufacturer, or otherwise,
a CVT Operating over a Driving Cycle, SAE Paper does not necessarily constitute or imply its
2003-01-3202, SAE Powertrain & Fluid Systems endorsement, recommendation, or favoring by the
Conference & Exhibition, October 2003, Pittsburg, United States Government or any agency thereof. The
PA, USA. views and opinions of document authors expressed
herein do not necessarily state or reflect those of the
3. Pasquier, M.; Duoba, M.; and Rousseau, A., United States Government or any agency thereof.
Validating Simulation Tools for Vehicle System