You are on page 1of 9

Applied Energy 115 (2014) 582590

Contents lists available at ScienceDirect

Applied Energy
journal homepage: www.elsevier.com/locate/apenergy

Electrical consumption of two-, three- and four-wheel light-duty electric


vehicles in India
Samveg Saxena , Anand Gopal, Amol Phadke
Environmental Energy Technologies Division, Lawrence Berkeley National Laboratory, United States

h i g h l i g h t s
 Model electrical consumption of 2-, 3- and 4-wheelers in India.
 Average city energy use is 33 Wh/km for scooters, 61 Wh/km for 3-wheelers.
 Average city energy use is 84 Wh/km and 123 Wh/km for low and high power 4-wheelers.
 The increased energy use from air conditioning is quantied.
 Energy use from variations in vehicle mass and motor efciency are quantied.

a r t i c l e

i n f o

Article history:
Received 9 August 2013
Received in revised form 18 October 2013
Accepted 27 October 2013
Available online 20 November 2013
Keywords:
Electric vehicles
Powertrain
Transportation
Vehicle to grid
India

a b s t r a c t
The Government of India has recently announced the National Electric Mobility Mission Plan, which sets
ambitious targets for electric vehicle deployment in India. One important barrier to substantial market
penetration of EVs in India is the impact that large numbers of EVs will have on an already strained electricity grid. Properly predicting the impact of EVs on the Indian grid will allow better planning of new
generation and distribution infrastructure as the EV mission is rolled out. Properly predicting the grid
impacts from EVs requires information about the electrical energy consumption of different types of
EVs in Indian driving conditions. This study uses detailed vehicle powertrain models to estimate per kilometer electrical consumption for electric scooters, 3-wheelers and different types of 4-wheelers in India.
The powertrain modeling methodology is validated against experimental measurements of electrical
consumption for a Nissan Leaf. The model is then used to predict electrical consumption for several types
of vehicles in different driving conditions. The results show that in city driving conditions, the average
electrical consumption is: 33 Wh/km for the scooter, 61 Wh/km for the 3-wheeler, 84 Wh/km for the
low power 4-wheeler, and 123 Wh/km for the high power 4-wheeler. For highway driving conditions,
the average electrical consumption is: 133 Wh/km for the low power 4-wheeler, and 165 Wh/km for
the high power 4-wheeler. The impact of variations in several parameters are modeled, including the
impact of different driving conditions, different levels of loading by air conditions and other ancillary
components, different total vehicle masses, and different levels of motor operating efciency.
2013 Elsevier Ltd. All rights reserved.

1. Introduction
India is one of the worlds most rapidly growing economies, and
is the third largest vehicle market in the world. Annual demand of
vehicles is rapidly growing in India, with 2020 annual projected
sales of 10 million passenger vehicles, 2.7 million commercial vehicles, and 34 million two-wheelers. India currently imports about
85% of its oil and is projected to reach 92% by 2020, creating a significant challenge for the balance of payments and the energy security
of the country [1]. Based on the pressing challenges with growth in
vehicle sales and energy security facing the country, the Central
Corresponding author.
E-mail address: samveg@berkeley.edu (S. Saxena).
0306-2619/$ - see front matter 2013 Elsevier Ltd. All rights reserved.
http://dx.doi.org/10.1016/j.apenergy.2013.10.043

Government of India has released the National Electric Mobility


Mission Plan (NEMMP) [1] which establishes a pathway for the
widespread deployment of hybrids, plug-in hybrids, and electric
vehicles in India. The NEMMP calls for the deployment of 57 million EVs (hybrids and full EVs) on the road by 2020, and the Government of India has committed Rs 22,500 cr (approximately $4.1
Billion USD) to this initiative. Similar goals for widespread electric
vehicle adoption have been set by other governments around the
world [24].
The rapid deployment of plug-in hybrid and fully electric vehicles (collectively called plug-in vehicles, PEVs, in this paper) called
for in the NEMMP places signicant demands on an already
strained electricity grid in India [5]. However, since range anxiety
is a signicant consumer perception barrier to EV deployment

S. Saxena et al. / Applied Energy 115 (2014) 582590

[6], the absence of reliable charging points (which require a stable


electricity grid) in India will make it difcult to achieve the targeted levels of EV market penetration. Additionally, if the electricity grid is unable to accommodate PEV charging, it is possible that
diesel generators will be used to provide the unmet electricity demand. Although this local distributed generation solution may
accommodate PEV charging demand in the interim, it is not an
effective way to decouple the Indian transportation sector from
oil and can still lead to urban air quality problems.
The Government of India has recently joined the Electric Vehicle
Initiative (EVI) of the Clean Energy Ministerial, which seeks to facilitate the deployment of 20 million EVs by 2020. Under this initiative, Lawrence Berkeley National Laboratory is supporting the
NEMMP in assessing the real-world costs, benets and environmental impacts of EV uptake in India; this publication is the rst
in a series of studies in this effort.
To properly plan for the rapid deployment of PEVs in India,
there is a need for nely resolved temporal and spatial predictions
of PEV charging load on the electricity grid. The ability to properly
forecast PEV charging load is essential for utility grid operators to
ensure that adequate generation capacity is available at the correct
times, and ensure that distribution infrastructure can accommodate substantial PEV charging. Several studies [713] have developed methods to estimate PEV charging load for the US
electricity grid. The most rigorous of these studies [1416] follow
a three-step methodology (listed below) to predict temporally resolved PEV charging load proles. A modeling tool, called V2GSim, has been developed at Lawrence Berkeley National Laboratory
to streamline the simulation of vehicle-grid interactions and this
tool is available for use in potential research collaborations [17].
1. Estimating the time when vehicles are plugged in: Survey data is
used to provide information on how drivers use their vehicles,
including number of vehicle trips per day, time of departure
of each trip, trip travel length, arrival time of each trip, type
of vehicle, etc. In the United States, a common data source for
this information is the National Household Travel Survey
(NHTS) [18], however other data sources have also been used.
2. Estimating the amount of energy required to charge the vehicle
battery: Typically, a simplied vehicle model is used to estimate: (a) how much of the vehicle battery is depleted during
each trip, and (b) how much energy is required during charging.
A standard approach in prior studies [1416] is to assume a
constant value for electrical consumption (kWh/km or kWh/
mile) depending on the type of vehicle (i.e. car, van, SUV, truck,
etc.) that is being modeled. More accurate estimates of battery
depletion while driving can be obtained with detailed vehicle
physics models (such as models used in other papers), however
this approach may be prohibitively computationally expensive
when attempting to model hundreds, thousands, or millions
of PEVs on an electricity grid.
3. Estimating charging rates while a vehicle is plugged in: Using estimates of when different vehicles will plug in for charging from
step 1, how much charging is required from step 2, and information about the charging rate (i.e. level 1, level 2, or DC fast
charger), number of PEVs and any smart charging strategies,
aggregate charging load proles are estimated for a large number of vehicles within a given region (i.e. utility service territory,
state, or country).
Successful implementation of the NEMMP within the prescribed
timeline requires immediate planning and infrastructure deployment to ensure that the Indian electricity grid can cope with the
added charging load from large numbers of PEVs. Thus, the 3-step
analysis methodology described above must be applied to the Indian context, however much of the required data for India is not

583

available in published studies. For instance, for step 1 better data


is required to characterize vehicle usage patterns in India. For step
2, average electrical consumption (Wh/km) numbers are required
for vehicles specically in the Indian context (i.e. for vehicle sizes
representative of typical Indian vehicles driving in Indian trafc
conditions). The use of prior published electrical consumption values does not adequately account for typical Indian vehicles or for
the inuence of driving and usage factors (i.e. from dense trafc,
or the use of power-consuming devices like an air conditioner).
Electrical consumption data for scooters, 3-wheelers, and small
4-wheelers has previously been unavailable in the literature, particularly for the Indian context where driving conditions will be
different than in developed countries and air conditioning load will
be a signicant factor. For the Indian context in particular, it may
be inappropriate to use prior published Wh/km values because
two-wheelers and ultra-compact four-wheelers that are typical
in India are signicantly smaller and lighter than the US market,
and typical driving conditions are different in India with more frequent stopping, lower average speeds and potentially more sudden
acceleration and deceleration [19].
In support of the NEMMP and as a step towards predicting the
charging load of PEVs on the Indian electricity grid, the results presented in this study will enable better estimates of PEV charging
load on the Indian electricity grid. Specically, the results of this
study provide Wh/km values that are representative of typical
vehicles in India, driving in conditions representative of Indian
roads. The results of this study can then be used in Step 2 of the
3-step methodology above to estimate temporally resolved PEV
charging loads on the Indian electricity grid.
2. Specic objectives
In support of the India National Electric Mobility Mission Plan,
this study provides critical data to enable detailed predictions of
PEV temporal charging load proles for the Indian electricity grid.
Detailed vehicle powertrain modeling is used for:
1. Providing estimates of average electrical consumption
(Wh/km) for vehicles that are representative of typical
Indian two-, three- and four-wheel vehicles over drive
cycles that are representative of Indian driving conditions.
2. Providing correlations for the Wh/km results that account
for variations in vehicle use, such as variability in vehicle
mass, the use of air conditioners, and variations in powertrain component efciency.
3. Vehicle models
3.1. Vehicle powertrain models
A detailed vehicle powertrain model is used to estimate electrical consumption for four types of vehicles, with specications for
each vehicle listed in Table 1. The powertrain models are created
in the industry standard Autonomie powertrain modeling platform.
3.2. Drive cycles
Given that energy consumption of a vehicle depends signicantly on driving patterns [1925], several different drive cycles
are chosen. Five drive cycles are chosen based on Indian driving
conditions, including a New Delhi cycle [26], Pune cycle [27], the
modied Indian drive cycle (MIDC) [28], and an Indian urban and
Indian highway cycle. Additionally three US certication cycles
are also included for comparison purposes, the EPA UDDS, HWFET
and US06 cycles [29]. Figs. 14 compare the characteristics of each

584

S. Saxena et al. / Applied Energy 115 (2014) 582590

Table 1
Vehicle specications used in powertrain models.

Base vehicle mass (kg)


Motor max power output (kW)
Final drive ratio
Usable battery capacity (kWh)
Tire size
Drag coefcient
Frontal area (m2)
Baseline electrical accessory & AC load (W)
Estimated range in City (km)
Estimated range on Highway (km)
Top speed (km/h)

Scooter

3-Wheeler

Low power 4-wheeler

High power 4-wheeler

150
1.5
6.3805
2.16
1000  300
0.60
1.25
50
6471
N/A
50

500
5.46
6.3805
4.25
1000  4.500
0.35
2.40
100
6080
N/A
73

898
19
6.8737
6.54
P155/70R13
0.335
2.0
200
7095
3476
117

1493
80
7.9377
16.7
P205/55 R16
0.28
2.50
200
123138
73136
120

Fig. 1. Velocity characteristics of US and Indian drive cycles.

Fig. 3. Acceleration characteristics of US and Indian drive cycles.

Fig. 2. Stopping/idling characteristics of US and Indian drive cycles.

Fig. 4. Deceleration characteristics of US and Indian drive cycles.

drive cycle in terms of velocity, stopping/idling, acceleration and


deceleration characteristics. The values in these gures are normalized by the average values across all driving cycles to allow
easier comparisons.
Fig. 1 compares the velocity characteristics of the US and Indian
drive cycles. The plot shows that driving conditions on the Indian
cycles involve lower maximum speed, lower mean speed, and lower mean driving speed1. Even the speeds on the Indian highway cycle are considerably lower than the speeds on the US highway cycles.

Fig. 2 compares the stopping and idling characteristics of the US


and Indian drive cycles. As expected, the results show that stop frequency and fraction of total time stopped are much higher on the
city cycles as compared with the highway cycles. Of particular
importance, Fig. 2 shows that stop frequency is much higher in
the Indian city cycles than the US city cycle. The total fraction of
time stopped is highest in the Pune cycle, followed by the US city
cycle.
Fig. 3 compares the acceleration characteristics of the US and
Indian drive cycles. The highest acceleration values are seen in
the high speed US highway cycle (US06). Comparing the US and Indian city cycles, it is seen that greater maximum acceleration and

1
Mean speed is dened as the average of all velocities over the drive cycle. Mean
driving speed is dened as the average of all non-zero velocities.

585

S. Saxena et al. / Applied Energy 115 (2014) 582590

maximum acceleration from stop values are encountered in the Indian city cycles, however the average acceleration is higher in the
US city cycle.
Fig. 4 compares the deceleration characteristics of the US and
Indian drive cycles. The results show that maximum deceleration
and maximum deceleration to stop are higher in Indian city conditions than US city conditions, however higher levels of average
deceleration are seen in the US city cycle.
Summarizing the results in this section, Figs. 14 compared the
drive cycle characteristics for the US and Indian drive cycles. It was
generally observed that the Indian drive cycles involve lower driving speeds, greater frequency of stopping, and higher levels of
maximum acceleration and deceleration. These results suggest that
driving in India may involve more severe stop-and-go conditions,
and previous studies [19,25] have found that these types of driving
conditions create unique opportunities for achieving greater levels
of fuel savings with vehicle electrication.
Fig. 5. Model validation: comparison of modeled and measured electrical consumption for a Nissan Leaf.

3.3. Parametric variations


In addition to the vehicle speed proles while driving, other
parameters will signicantly inuence vehicle energy consumption as well. The vehicle modeling that is discussed in this paper
captures the impact on vehicle energy usage from several parameters that will change with different vehicle designs, usage patterns,
and driving conditions. For warm climates like India, ancillary
components such as vehicle air conditioning load will have a
signicant impact on energy consumption [3031]. Loading the
vehicle with more passengers or cargo will also impact energy
consumption. Additionally, variations in powertrain component
efciency will also impact energy consumption. Table 2 lists the
range of parameter variations that were explored using the vehicle
powertrain models for their impact on vehicle energy
consumption.
3.4. Model validation
To ensure that the electrical consumption estimates presented
in the results section of this paper are reasonable, the same modeling methodology is followed to create a powertrain model for a
Nissan Leaf electric vehicle, for which there are well documented
values of electrical consumption under various driving conditions.
A vehicle powertrain model was constructed with specications
resembling a Nissan Leaf, and electrical consumption model estimates were compared against published measurement data [32]
for the EPA UDDS, Highway, and US06 drive cycles over a range
of total vehicle mass. Fig. 5 shows a comparison of the modeled
and measured electrical consumption values for a Nissan Leaf.
The modeled and experimentally measured electrical consumption values plotted in Fig. 5 show that the vehicle powertrain model reasonably predicts both the trends and absolute values of
electrical consumption for a range of different vehicle masses for
all three drive cycles. The largest difference in absolute values
between the model and the experimental measurements is
11.50%, which occurs for the lowest vehicle mass on the highway
cycle. It is typically the case that increased vehicle mass leads to
increased energy consumption, however the experimental

measurements on the highway cycle do not display this expected


trend. This may be due to experimental error because obviously
the vehicle mass will have a signicant inuence on the electricity
consumption of a vehicle. This expected trend is indeed seen for
the UDDS and US06 experimental measurements, thus the data
points at the lowest mass values for the highway cycle seem higher
than expected. As a result of the overall agreement of trends and
absolute values shown in Fig. 5, the modeling methodology is considered accurate enough for the purposes of this study.
4. Results
4.1. Baseline electrical consumption estimates
Table 1 lists the vehicle specications that were used in the
powertrain models for an electric scooter, electric 3-wheeler, low
power EV 4-wheeler and high power EV 4-wheeler. These powertrain models provide the electrical consumption per kilometer estimates over several different drive cycles in Fig. 6 and Table 3. There
are several numerical values in Fig. 6 which are crossed out (particularly for the electric scooter and 3-wheeler). These crossed out
values denote that the vehicle was unable to perform on the drive
cycle, either because the drive cycle requests speeds which are
higher than the maximum speed capability of the vehicle, or
because acceleration proles are demanded which exceed the
capabilities of the powertrain components. Thus, these crossed

Table 3
Electrical consumption range of each vehicle.
Electrical consumption (Wh/km)

Scooter
3-Wheeler
Low power EV
High power EV

Avg city

Avg hwy

Range

33
61
84
123

38
85
133
164

3140
5397
70192
101224

Table 2
Range of parameter variations explored for their impact on vehicle energy consumption.

Ancillary loading (i.e. A/C) (kW)


Vehicle mass (kg)
Motor efciency (%)

Scooter

3-Wheeler

Low power 4-wheeler

High power 4-wheeler

0.00.30
150300
5590

0.00.50
500800
5590

0.203.0
8981200
5590

0.204.0
14931800
5590

586

S. Saxena et al. / Applied Energy 115 (2014) 582590

Fig. 6. Electrical energy consumption rate for different types of EVs on different
drive cycles.

Fig. 8. Variation of vehicle electrical consumption with different total vehicle


masses.

4.2. Impact of parameter variations on vehicle electricity consumption


The electrical consumption was calculated for several vehicles
on several US and Indian drive cycles in Section 4.1, with specications dened in Table 1. Vehicles on the road, however, will rarely
have exactly the same specications as those dened in Table 1,
thus this section explores how different parameters will impact
electrical consumption of each vehicle.

Fig. 7. Variation of vehicle electrical consumption with different ancillary component loading.

out values should not be given much weight but instead simply
considered for reference.
The results in Fig. 6 show that electrical consumption per kilometer is highest for the 4-wheelers and lowest for the electrical
scooter, which comes as no surprise given the differences in vehicle mass. For the vehicles which are capable of sustaining highway
speeds (i.e. only the 4-wheelers), electrical consumption is signicantly higher for high speed highway driving.

4.2.1. Ancillary component and air conditioning loads


For hot climates like India, energy use by air conditioners will
have a signicant impact on the electricity consumption of a vehicle. Additionally, other ancillary components (like vehicle control
electronics, radio, and lights) will consume energy. Fig. 7 presents
the impact on vehicle electricity consumption from different levels
of loading by ancillary components. As two- and three-wheelers
typically do not have an enclosed cabin they will not have air conditioners, and thus their maximum loading from ancillary components will be lower. Thus, in Fig. 7 the modeled range of energy
consumption from ancillary components for the two- and threewheel vehicles is much lower than for the four-wheelers.
Fig. 7 shows that for each vehicle on all the different drive cycles, vehicle electricity consumption (Wh/km) increases linearly
with increasing loading from ancillary components. It is particularly important to note that the slope of this linear increase is different across the different drive cycles. The equation of t for the
relationship between ancillary component loading and vehicle
electricity consumption follows the form of Eq. (1), where x is

Table 4
Coefcients for equation of t for impact of ancillary component loading (kW) on vehicle electricity consumption (Wh/km).
UDDS

HWFET

US06

India urban

Delhi

Pune

MIDC

61.62
30.40
1.00

57.59
28.84
1.00

40.59
33.97
1.00

22.75
67.15
1.00

59.28
46.94
1.00

55.40
50.46
1.00

34.42
69.80
1.00

47.11
67.81
1.00

23.72
80.78
1.00

60.64
57.03
1.00

56.86
69.33
1.00

34.49
89.80
1.00

46.14
112.28
1.00

22.88
118.40
1.00

59.57
88.84
1.00

55.73
113.99
1.00

33.38
124.40
1.00

2 Wheeler

m
b
R2

42.35
32.83
1.00

48.13
28.85
1.00

3 Wheeler

m
b
R2

36.47
65.70
1.00

45.85
50.62
1.00

4 Wheeler low power

m
b
R2

35.30
87.86
1.00

15.43
117.48
1.00

15.73
189.49
1.00

4 Wheeler high power

m
b
R2

34.22
128.27
1.00

14.21
142.42
1.00

14.70
220.64
1.00

India highway

587

S. Saxena et al. / Applied Energy 115 (2014) 582590


Table 5
Coefcients for equation of t for impact of vehicle mass (kg) on vehicle electricity consumption (Wh/km).
UDDS

HWFET

US06

India urban

India highway

Delhi

Pune

MIDC

0.02
30.88
0.95

0.03
27.64
0.98

0.02
32.73
0.96

2 Wheeler

m
b
R2

0.03
29.94
1.00

0.02
28.12
0.95

3 Wheeler

m
b
R2

0.07
34.43
1.00

0.07
22.27
1.00

0.05
45.54
1.00

0.04
32.00
1.00

0.07
23.64
1.00

0.04
51.03
1.00

4 Wheeler low power

m
b
R2

0.07
30.87
1.00

0.05
77.49
1.00

0.07
127.74
1.00

0.07
18.62
1.00

0.06
31.96
1.00

0.04
32.06
1.00

0.07
22.60
1.00

0.05
48.43
1.00

4 Wheeler high power

m
b
R2

0.06
44.05
1.00

0.04
83.86
1.00

0.07
114.44
1.00

0.06
31.90
1.00

0.05
42.89
1.00

0.04
39.43
1.00

0.06
37.38
1.00

0.05
60.27
1.00

Fig. 10. Powertrain architecture for electric vehicle models.

loading has a greater impact on vehicle electricity consumption


at lower speed driving conditions (i.e. city driving), and is not sensitive to vehicle type.

Fig. 9. Variation of vehicle electrical consumption with different average motor


operating efciency.

the electrical loading in kW, m is the slope, and b is the y-intercept


(in this case, the electrical consumption if there was no loading
from ancillary components). Values for m and b for each vehicle
on each drive cycle are listed in Table 4.

y mx b

The tting equation parameters in Table 4 shows that the slope


of each tting equation for a given drive cycle is not sensitive to
vehicle type. The slopes are generally higher for lower speed driving conditions. These results suggest that ancillary component

4.2.2. Variations in vehicle, passenger and cargo mass


Individual vehicles are bound to be loaded with different mass
due to variations in the number of passengers or cargo being carried. Fig. 8 shows the impact of variations in total vehicle mass
for each vehicle on each drive cycle. The 3- and 4-wheelers will
have greater carrying capacity and thus the range of vehicle masses
modeled is larger for these vehicles.
Fig. 8 shows that vehicle electricity consumption is also linearly
dependent on vehicle mass, with increased electricity consumption
for greater total vehicle mass. The equation of t relating changes
in vehicle electrical consumption with changes in vehicle mass follows the form of Eq. (1) as well, with x being vehicle mass in kg.
Table 5 lists the coefcients for the equations of t for variations
in vehicle mass. The tting coefcients in Table 5 suggest that
the impact of vehicle mass on vehicle electricity consumption is

Table 6
Coefcients for equation of t for impact of motor efciency (%) on vehicle electricity consumption (Wh/km).
UDDS

HWFET

US06

India urban

Delhi

Pune

MIDC

49.53
70.89
0.99

51.55
69.97
0.99

54.34
77.54
0.99

118.2
161.4
0.99

74.5
110.2
0.99

99.9
131.3
0.99

118.0
164.6
0.99

286.7
328.7
0.99

266.0
322.9
0.99

167.1
219.4
0.99

274.9
321.1
0.99

251.1
318.9
0.99

268.7
328.7
0.99

266.0
322.9
0.99

167.1
219.4
0.99

274.9
321.1
0.99

251.1
318.9
0.99

2 Wheeler

m
b
R2

54.11
75.97
0.99

50.93
69.54
0.99

3 Wheeler

m
b
R2

125.9
165.3
0.99

106.2
135.6
0.99

4 Wheeler low power

m
b
R2

198.8
248.5
0.99

189.0
271.7
0.99

319.0
441.8
1.00

4 Wheeler high power

m
b
R2

304.2
361.7
0.99

231.1
326.1
0.99

438.0
568.6
0.99

India highway

588

S. Saxena et al. / Applied Energy 115 (2014) 582590

fairly consistent across the different drive cycles and across the different vehicles (especially the 3-wheeler and both 4-wheelers).
4.2.3. Variations in average motor efciency
The values chosen for the motor efciency maps used for the
baseline vehicle simulations (in Section 4.1) were established to
t the Nissan Leaf model validation results in Section 3.4. Electric
vehicles released in the Indian market, however, may use different
types of motors with different efciency operating proles, thus
this section explores the impact of changes is motor operating efciency. Fig. 9 shows the variation of vehicle electricity consumption with average motor operating efciency for the different
vehicles driving on the different drive cycles.
As expected, the results in Fig. 9 show that vehicle electricity
consumption decreases as a more efcient motor is used. A particularly interesting result, however, is that changes in motor efciency have very little impact on electricity consumption for the
smaller vehicles, especially the two-wheeler. This result is of signicant importance as it suggests that the use of less expensive
motors, which may be less efcient, can be used to lower the cost
of electric scooters while having minimal impact on vehicle electricity consumption (and thus vehicle range). For the larger vehicles and for higher speed driving conditions (i.e. on highways),
however, motor efciency impacts electrical consumption signicantly and thus better motors must be used. The results for the larger vehicles in Fig. 9 show that the relationship between vehicle
electricity consumption and motor efciency is not perfectly linear
(i.e. a slight curvature can be seen on the plots), however the R2 tting parameters in Table 6 show that a linear equation of the form
of Eq. (1), with x being the average motor efciency (%), produces a
good t.

5. Conclusions
Given the ambitious targets for electric vehicle deployment in
India under the National Electric Mobility Mission Plan which
was announced by the Government of India, there are signicant
concerns with the impact that EV charging will have on an already
strained Indian electricity grid. This study is part of a larger effort
towards estimating the impact on the Indian electricity grid from
substantial deployment of EVs on the Indian grid to subsequently
plan the deployment of new generation and distribution
infrastructure.
This study used detailed vehicle powertrain models to estimate
the per kilometer electrical consumption of several types of EVs,
including a scooter, a 3-wheeler, a low power 4-wheeler, and a
high power 4-wheeler. Electrical consumption data for scooters,
3-wheelers, and small 4-wheelers has previously been unavailable
in the literature, particularly for the Indian context where driving
conditions will be different than in developed countries and air
conditioning load will be a signicant factor. The powertrain model
methodology was validated against experimental measurements
for a Nissan Leaf. The main conclusions from this study are as
follows:
1. Average electrical consumption: Vehicle size has the greatest
impact on per km electrical consumption, followed by the
driving characteristics (i.e. city vs. highway driving). In city
driving conditions average electrical consumption results were:
33 Wh/km for the scooter, 61 Wh/km for the 3-wheeler, 84 Wh/
km for the low power 4-wheeler, and 123 Wh/km for the high
power 4-wheeler. For highway driving conditions average electrical consumption results were: 133 Wh/km for the low power
4-wheeler, and 165 Wh/km for the high power 4-wheeler. The

scooter and 3-wheeler were incapable of sustaining highway


speeds. Readers are referred to Section 4.1 for a detailed breakdown of electrical consumption for different driving conditions.
2. Impact of air conditioners and ancillary component loads on electrical consumption: Ancillary components have a signicant
impact on electrical consumption, with per km electrical consumption increasing linearly with greater ancillary component
loads. The slope of increasing electrical consumption is larger
for lower speed driving conditions (i.e. in cities), but is not sensitive to vehicle type.
3. Impact of variations in vehicle mass on electrical consumption: Per
km electrical consumption also increases linearly with increasing vehicle mass (i.e. for more passengers or cargo). The slope of
increase is fairly consistent across different driving conditions
and vehicle types.
4. Impact on variations in motor efciency on electrical consumption:
Per km electrical consumption decreases linearly with greater
motor operating efciency, however the slope of this decrease
is highly sensitive to vehicle size. An important nding is that
for smaller vehicles, like scooters, increasing motor efciency
has little impact on electrical consumption. As a result, the
use of inexpensive and less efcient motors to minimize the
cost of electrical scooters will only have minimal impact on
electrical consumption and thus on EV range. For larger vehicles, however, motor efciency has a signicant impact, with
more efcient motors allowing signicantly reduced electrical
consumption. For larger vehicles there is also an impact of driving characteristics, with higher speed driving conditions showing greater variation of electrical consumption with changes in
motor operating efciency.

Acknowledgements
This work was supported by the Assistant Secretary of Policy
and International Affairs, Ofce of Policy and International Affairs,
of the US Department of Energy and the Regulatory Assistance
Project through the US Department of Energy under Contract No.
DE-AC02-05CH11231.
Appendix A.
This Appendix presents a brief description of the powertrain
and component models that are used to model the four types of
electric vehicles considered in this study. For a detailed description
of each model, readers are referred to the documentation associated with the commercially available powertrain modeling software Autonomie, which was used in this study.
A.1. Overall powertrain architecture
The electric vehicle models in Autonomie include the component models shown in Fig. 10, as well as an overarching propulsion
and brake control model.
The propulsion control model translates driver acceleration
commands, which are governed by the specied drive cycle, into
motor torque demands while simultaneously considering vehicle
and motor speed, battery state of charge, maximum torque output
before wheel slip at a given speed, and loading from ancillary
components.
The braking control model performs a similar function of translating driver braking commands, which are governed by the specied drive cycle, into braking torque demands while considering
several factors and constraints. One further function of the braking

589

S. Saxena et al. / Applied Energy 115 (2014) 582590

model is to specify the braking torque provided by the traction


motor and the mechanical brakes. In general, braking torque is
provided entirely by the traction motor until the motor or
battery power limits are encountered. Beyond these limits,
mechanical braking is used to absorb the remaining required braking torque.

T module air T air  1=2

SOC

I
dt
3600

Ahinit
Ahmax

In Eq. (1), I is the charging or discharging current requested from


the battery, Ahinit is the amount of energy stored in the battery
when the model is initialized, and Ahmax is the maximum energy
storage capacity of the battery as a function of cell operating temperature, as shown in Eq. (2):

Ahmax f T cell

The values for Ahmax are specied in an initialization le using measurement data for the maximum capacity of a cell that is discharged
at a C/5 rate.
The open circuit voltage and the internal resistances of the cell
on charging or discharging are determined as a function of SOC and
cell temperature, as shown in Eq. (3) through Eq. (5) respectively:

V OC f SOC;T cell

Rint;chg f SOC;T cell

Rint;dis f SOC;T cell

Open circuit voltage and internal resistance data on charging and


discharging is specied in an initialization le based on experimentally measured data. For lithium ion batteries, open circuit cell voltage typically spans a range from 3.5 to 4.2 V.
The cell output voltage at any given operating condition (i.e. at
the battery terminals) is calculated using Eq. (6) for charging and
Eq. (7) for discharging:

11

Finally, the module temperature is calculated through the balance of heat generation and heat dissipation rates in Eq. (12),
and it is assumed that each cell within the module will have the
same temperature.

A.2. Battery model


The battery model calculates the state of an individual cell and
assumes that all cells operate identically. Cell state of charge (SOC)
is calculated according to the coulomb counting approach in Eq.
(1):

Q_ cooling
_ cooling air C p;module
m

T cell T module


R_
Q gen  Q_ cooling dt
mmodule C p;module

12

A.3. Motor model


The motor model provides the torque demanded by the propulsion controller, while taking into account the effects of losses and
rotor inertia. Motor temperature is used to determine the time that
the motor can spend above the maximum continuous rated torque
levels.
The maximum continuous torque is specied in an initialization
le for a full range of motor speeds according to experimental data.
The absolute maximum torque output is specied according to
a predened value for continuous to peak torque ratio. The
motor efciency map is specied for a full range of torque and
speed points in the initialization le using experimentally
measured data. The motor model inputs are the command to the
motor (i.e. required propulsion torque), the input voltage, and
the motor speed.
The maximum propulsion and regenerative torque capabilities
of the motor are determined as a function of motor speed. The
maximum torque map (as a function of speed) is specied in the
vehicle initialization le. The specied torque map enables maximum torque at low speeds (up to roughly 2000 RPM) and subsequently decaying maximum torque up to the high speed limits of
the motor.
Section 4.2.3 of this paper examines the impacts of different
levels of motor efciency for the vehicles that are modeled. Motor
efciency is scaled by multiplying the efciency map specied in
the initialization le by a scaling factor. The scaling factor is dened as the ratio of desired maximum motor efciency over the
maximum efciency specied in the map dened in the initialization le.

V out;chg V OC  gcoul Iout Rint;chg

A.4. Torque coupling, nal drive and wheel model

V out;dis V OC  Iout Rint;dis

The nal drive and torque coupling models are functionally


similar, and serve to apply a xed gear reduction ratio to both torque and speed by taking into account the losses. The torque coupling and nal drive are assumed to be 97% efcient across the
entire torque/speed range.
The wheel model serves to transform rotational energy into linear. Losses from mechanical braking and tire friction are calculated
within this model. Linear force exerted or absorbed by the tires is
calculated using Eq. (13):

gcoul is the coulombic efciency, which for these models is simply


set to 1.0.
A simple thermal model is included as part of the battery model
to estimate the cell operating temperature. The rate of heat generation in a cell is calculated according to Eq. (8) while charging and
Eq. (9) while discharging:

Q_ gen;chg I2out Rint;chg  V out Iout 1  gcoul

Q_ gen;dis I2out Rint;dis

Heat dissipation is calculated by assuming a fan ows cooling air


across the cells within a pack. Eq. (10) is activated when the cell
temperature rises above a specied threshold to cause the battery
management system to turn the cooling fan on.

T module air  T module


Q_ cooling
Thermal resistance

10

In situations where the cooling fan remains off, Eq. (10) is simply set
to zero. The module air temperature is calculated using Eq. (11):

F T=rwheels

13

In Eq. (13), T is the total input or output torque to the tires, and
rwheels is the wheel radius. Torque input or output is calculated
using Eq. (14):

T T in  T braking  T res

14

In Eq. (14), Tin is the torque input from the vehicle powertrain,
Tbraking is the braking torque exerted by the mechanical brakes,
and Tres is the resistive torque from tire rolling resistance which is
calculated using a third-order polynomial function of speed. The

590

S. Saxena et al. / Applied Energy 115 (2014) 582590

coefcients for the polynomial are specied in an initialization le


based on experimentally measured data.
A.5. Chassis model
By balancing the total powertrain output against the total
opposing forces, the linear acceleration and vehicle speed is nally
calculated at the chassis model. Powertrain output (or regenerative
input) is calculated at earlier sub-models based on the specied
drive cycle and the specied component parameters. Opposing
forces include factors such as hill climbing, aerodynamic losses,
and tire rolling resistance.
Aerodynamic losses are calculated within the chassis model
using Eq. (15):

F loss;aero 1=2qC d AV 2

15

In Eq. (15), q is the density of air, Cd is the vehicle drag coefcient, A


is the frontal area of the vehicle, and V is the vehicle speed.
Opposing force from hill climbing is calculated using Eq. (16):

F loss;hill mg sinh

16

In Eq. (16), m is the vehicle mass, g is the acceleration from gravity,


and h is the hill grade.
The acceleration of the vehicle is subsequently calculated using
Eq. (17):

F in  F loss
mstatic mdynamic

17

In Eq. (17), Fin is the input from the vehicle powertrain, Floss is the
sum of all opposing forces, mstatic is the static mass of the vehicle
and mdynamic is the dynamic mass of the vehicle from rotating components. Vehicle speed is calculated by integrating Eq. (17) over
time.
A.6. Ancillary components models
Power losses from ancillary components (such as air conditioning and electronic in-vehicle equipment) are calculated as a specied continuous power draw. The power that is owed to
ancillary components is assumed to travel through a power converter which maintains its output voltage at the required voltage
input for ancillary components (i.e. 12 V). The power converter is
assumed to have 95% conversion efciency.
References
[1] Department of Heavy Industry, Government of India, National Electric Mobility
Mission
Plan
2020,
<http://dhi.nic.in/NEMMP2020.pdf>,
[accessed
19.03.2013].
[2] Foley A, Tyther B, Calnan P, Gallachoir BO. Impacts of Electric Vehicle charging
under electricity market operations. Appl Energy 2013;101:93102. http://
dx.doi.org/10.1016/j.apenergy.2012.06.052.
[3] U.S. Department of Energy, EV Everywhere grand challenge blueprint, 2013,
<http://www1.eere.energy.gov/vehiclesandfuels/electric_vehicles/pdfs/
eveverywhere_blueprint.pdf>.
[4] The Central Peoples Government of the Peoples Republic of China, Energysaving and new energy automotive industry development plan, 2012, <http://
www.gov.cn/zwgk/2012-07/09/content_2179032.htm>.
[5] Romero JJ. Blackouts illuminate Indias power problems. IEEE Spectrum
2012;49(10):112. http://dx.doi.org/10.1109/MSPEC.2012.6309237.
[6] Hidru MK, Parsons GR, Kempton W, Gardner MP. Willingness to pay for electric
vehicles and their attributes. Resour Energy Econ 2011;33(3):686705. http://
dx.doi.org/10.1016/j.reseneeco.2011.02.002.
[7] Collins MM, Mader GH. The timing of EV recharging and its effects on utilities.
IEEE Trans Vehicular Technol 1983;VT-32(1):907. http://dx.doi.org/10.1109/
T-VT.1983.2394.

[8] Rahman S, Shrestha GB. An investigation into the impact of electric vehicle
load on the electric utility distribution system. IEEE Trans Power Deliv
1993;8(2):5917. http://dx.doi.org/10.1109/61.216865.
[9] Farmer C, Hines P, Dowds J, Blumsack S. Modeling the impact of increasing
PHEV loads on the distribution infrastructure, In: 43rd Hawaii international
conference on system sciences; 2010, doi:10.1109/HICSS.2010.277.
[10] Denholm P, Kuss M, Margolis RM. Co-benets of large scale plug-in hybrid
electric vehicle and solar PV deployment. J Power Sources 2013;236:3506.
http://dx.doi.org/10.1016/j.jpowsour.2012.10.007.
[11] Qian K, Zhou C, Allen M, Yuan Y. Modeling of load demand to EV battery
charging
in
distribution
systems.
IEEE
Trans
Power
Systems
2011;26(2):80210. http://dx.doi.org/10.1109/TPWRS.2010.2057456.
[12] Wang J, Liu C, Ton D, Zhou Y, Kim J, Vyas A. Impact of plug-in hybrid electric
vehicles on power systems with demand response and wind power. Energy
Policy 2011;39(7):401621. http://dx.doi.org/10.1016/j.enpol.2011.01.042.
[13] Wu D, Aliprantis DC, Gkritza K. Electric energy and power consumption by
light-duty plug-in electric vehicles. IEEE Trans Power Systems
2011;26(2):73846. http://dx.doi.org/10.1109/TPWRS.2010.2052375.
[14] Kelly JC, MacDonald JS, Keoleian GA. Time-dependent plug-in hybrid electric
vehicle charging based on national driving patterns and demographics. Appl
Energy 2012;94:395405. http://dx.doi.org/10.1016/j.apenergy.2012.02.001.
[15] Weiller C. Plug-in hybrid electric vehicle impacts on hourly electricity demand
in the United States. Energy Policy 2011;39(6):376678. http://dx.doi.org/
10.1016/j.enpol.2011.04.005.
[16] Peterson SB, Whitacre JF, Apt J. Net air emissions from electric vehicles: the
effect of carbon price and charging strategies. Environ Sci Technol
2011;45(5):17927. http://dx.doi.org/10.1021/es102464y.
[17] Lawrence Berkeley National Laboratory, V2G-Sim - A platform for
understanding vehicle-to-grid integration, <http://powertrains.lbl.gov/
research-projects/v2g-sim>.
[18] U.S. Department of Transportation, National Household Travel Survey, <http://
nhts.ornl.gov/download.shtml>, [accessed 19.03.2013].
[19] Saxena S, Phadke A, Gopal A, Srinivasan V. Understanding fuel savings
mechanisms from hybrid vehicles to inform optimal battery sizing for India.
Int J Powertrains IJPT-44428 2013;1:23.
[20] Ericsson E. Independent driving pattern factors and their inuence on fuel-use
and exhaust emission factors. Transport Res Part D: Transport Environ
2001;6(5):32545. http://dx.doi.org/10.1016/S1361-9209(01)00003-7.
[21] Kudoh Y, Ishitani H, Matsuhashi R, Yoshida Y, Morita K, Katsuki S, et al.
Environmental evaluation of introducing electric vehicles using a dynamic
trafc-ow model. Appl Energy 2001;69(2):14559. http://dx.doi.org/
10.1016/S0306-2619(01)00005-8.
[22] Joumard R, Andr M, Vidon R, Tassel P, Pruvost C. Inuence of driving cycles on
unit emissions from passenger cars. Atmos Environ 2000;34(27):46218.
http://dx.doi.org/10.1016/S1352-2310(00)00118-7.
[23] Hansen JQ, Winther M, Sorenson SC. The inuence of driving patterns on petrol
passenger car emissions. Sci Total Environ 1995;169(13):12939. http://
dx.doi.org/10.1016/0048-9697(95)04641-D.
[24] Wang H, Fu L, Zhou Y, Li H. Modelling of the fuel consumption for passenger
cars regarding driving characteristics. Transport Res Part D: Transport Environ
2008;13(7):47982. http://dx.doi.org/10.1016/j.trd.2008.09.002.
[25] Saxena S, Gopal A, Phadke A. Understanding the Fuel Savings Potential from
deploying hybrid cars in China. Appl Energy 2014;113:112733. http://
dx.doi.org/10.1016/j.apenergy.2013.08.057.
[26] Chugh S, Kumar P, Muralidharan M. et al. Development of Delhi driving cycle:
a tool for realistic assessment of exhaust emissions from passenger cars in
Delhi. SAE technical paper 201201-0877; 2012. doi:10.4271/2012-01-0877.
[27] Kamble SH, Mathew TV, Sharma GK. Development of real-world drive cycle:
case study of Pune, India. Transport Res Part D: Transport Environ
2009;14(2):13240. http://dx.doi.org/10.1016/j.trd.2008.11.008.
[28] Automotive Research Association of India, Testing type approval and
conformity of production (COP) of vehicles for emission as per CMV rules
115, 116 and 126, Chapter 3, Doc. No. MoRTH/CMVR/TAP-115/116, Issue 4, pp.
914,
<http://www.araiindia.com/CMVR_TAP_Documents/Part-14/Part14_Chapter03.pdf>.
[29] Environmental Protection Agency, Detailed Test Information, <http://
www.fueleconomy.gov/feg/fe_test_schedules.shtml>, [accessed 18.06.2013].
[30] Johnson VH. Fuel used for vehicle air conditioning: a state-by-state thermal
comfort-based approach, SAE technical paper 200201-1957; 2002.
doi:10.4271/2002-01-1957.
[31] Lukic SM, Emadi A. Performance analysis of automotive power systems: effects
of power electronic intensive loads and electrically-assisted propulsion
systems. IEEE Vehicular Technol Conf 2002;56(3):18359. http://dx.doi.org/
10.1109/VETECF.2002.1040534.
[32] Carlson RB, Lohse-Busch H, Diez J, Gibbs J. The measured impact of vehicle
mass on road load forces and energy consumption for a BEV, HEV, and ICE
Vehicle. SAE Int J Alternat Powertrains 2013;2(1):10514. http://dx.doi.org/
10.4271/2013-01-1457.

You might also like