You are on page 1of 8

Design and Optimization of a Drivetrain with

Two-speed Transmission for Electric Delivery Step


Van

Alexei Morozov∗, Kieran Humphries† , Ting Zou∗, Sudarshan Martins† and Jorge Angeles∗
∗ Centrefor Intelligent Machines & Department of Mechanical Engineering
McGill University, 817 Sherbrooke Str. West, Montreal, H3A 0C3 Canada
Email: alexvit@cim.mcgill.ca, ting@cim.mcgill.ca, angeles@cim.mcgill.ca
† Department of Mechanical Engineering, McGill University

817 Sherbrooke Str. West, Montreal, H3A 0C3 Canada


Email: kieran.humphries@mail.mcgill.ca, sudarshan.martins@mcgill.ca

Abstract—The use of a two-speed transmission in electric this paper implemented a simulation on the conventional
delivery step vans to improve performance and efficiency is gasoline GM Workhorse chassis delivery truck model by using
discussed in this paper. Two important tasks relating to this ap- AVL Cruise, to test the energy consumption in the Orange
plication are considered. The first task includes the development County Bus Cycle (OCC). The simulation result shows a fuel
and optimization of a two-speed transmission, where the gear consumption of 39 L/100km, which fully correlates with the
ratios are used as design variables, while vehicle performance and
data obtained during field tests. Although it is not obvious how
energy consumption are design objectives. The procedure is based
on simulation and optimization. The simulation model and the to compare the energy consumption between the conventional
optimization process are described. The second task includes the and electric vehicles directly, due to the different physical
estimation of the efficiency and comparison of the performance nature of the energy sources, it is possible to convert the fuel
of an electric step van with a direct drive setup versus the vehicle consumption to its energy equivalent in electricity with units
with the two-speed transmission. of kWh/100km. The equivalent energy consumption obtained
for the conventional delivery truck is about 345 kWh/100km,
I. I NTRODUCTION which is much higher than the energy consumption of the elec-
Electric vehicles (EV) were first commercialized in the tric vehicles. Within the OCC cycle of an approximate distance
early twentieth century. Their popularity was eclipsed by of 10.53 km, the same parameter obtained in our research for
vehicles with internal combustion engine because of perfor- the electric delivery truck varies between 50.75 kWh/100km
mance limitations, but interest in electric vehicles has been and 58.89 kWh/100km, including a 30% electric energy loss
rekindled due to high fossil-fuel prices and the global warming during charging.
effects of combustion engines. Most of the larger players Due to the intrinsic merits of the electric vehicles, the
in today’s automotive market either currently produce or are authors of this paper decided to develop and optimize a vehicle
developing electric models. One potential sales category for with a transmission system. The vehicle efficiency is compared
emerging battery-powered vehicles is in commercial delivery to an electric version of the GM Workhorse chassis delivery
fleets. More companies are starting to include electric trucks truck without a transmission. Considering that in the current
into their fleets, such as FedEx, Staples, Purolator, and AT&T. market most research focuses on small passenger cars, the
Electric medium-duty delivery trucks are preferable to their work conducted on the electric delivery trucks in this paper
conventional counterparts for several reasons: no idling fuel attempts to broaden the scope of knowledge in this realm.
use; high efficiency due to electric motors; energy recovery The most widely used type of delivery truck is the step
during regenerative braking; fewer moving parts and low van, which is a medium-duty truck with a large storage space
maintenance cost. It is no surprise that such vehicles attract in the rear. Many companies prefer to acquire hybrid instead
the attention of modern businesses, even though the benefits of pure battery-electric vehicles because of range limitations
of these vehicles are highly dependent on driving conditions and the cost of installing charging infrastructure. For example,
and they may also have a limited driving range. Purolator (the Canadian package delivery operator) currently
In a recent study [1], researchers compared a 2011 Smith owns more than 550 hybrid electric trucks within their overall
Newton electric truck powered by a 120 kW electric motor fleet of more than 3000 vehicles. More original equipment
and a 2006 Freightliner truck powered by a Cummins diesel manufacturers (OEMs) are starting to offer low-cost battery
engine. The study demonstrated that the electric trucks used electric vehicles, such as AMP Electric Vehicles. They claim
only 48% to 82% of the energy that their diesel counterparts that their electric truck could provide an added initial cost
used, and created only 25% to 89% of the greenhouse gas payback in as short as three years [2].
emissions over the same cycles.
A possible next step in the development of pure EVs is the
For the purpose of comparing the efficiency of the con- use of multi-speed transmissions for EV powertrains. Discus-
ventional truck with its electric competitor, the authors of sions about the potential benefits of multi-speed transmission

978-1-4799-6075-0/14/$31.00 ©2014 Crown

Authorized licensed use limited to: COLLEGE OF ENGINEERING - Pune. Downloaded on December 13,2021 at 10:31:17 UTC from IEEE Xplore. Restrictions apply.
in EV versus single-speed drives started as early as the mid- optimized and compared in this work are the single-speed
seventies and have become more active recently [3]–[6]. and the two-speed drivetrain battery electric vehicles (BEV).
Braking performance is outside of the scope of this study. In
As indicated by Jerew [7], the most common explanation general, the goal of electrification or hybridization is to lower
of why EVs do not need a transmission, at least from the the energy (fuel or electricity) consumption. In this case, the
viewpoint of electric vehicle manufacturers, is that an electric two-speed transmission option has been used to try to improve
motor has a much wider operating range than a standard the energy consumption of the electric delivery truck while
internal combustion engine (ICE). However, recent research maintaining or improving dynamic performance.
activities support the use of multi-speed transmissions with
electric vehicles, justified by simulation and testing results that Performance requirements are usually based on dynamic
show improvements in efficiency and performance. Recently testing criteria. The main performance requirements considered
the simulation of a two-speed EV drive, developed by Vocis in this study are top speed, full load acceleration time, and
Driveline Controls, was reported [8]. The authors stated that climbing grade. These items are considered in Section III. Note
multi-speed transmissions can reduce the energy consumption that each of these tests is carried out with the vehicle model
of the battery by 5%–10% during industry-standard testing. fully loaded to its Gross Vehicle Weight Rating (GVWR), to
The authors of another report [9] found that the two-speed simulate the top loading conditions.
transmission system operates mainly in second gear and that
the electric motor runs at higher torques and lower speeds than The task of the development of the two-speed transmission
with the single-speed transmission. This means that the motor for the electric delivery truck includes several phases. In the
ran in a higher efficiency region. The authors of yet another first phase, preliminary selection of the transmission gear
study [10] conducted a comparison of energy consumption ratios is undertaken following the procedure described by
in EVs with a single-speed system without a gearbox and Naunheimer [13]. Then, in the next step, these two ratios are
with a four-speed transmission. The results of this comparison verified and optimized using simulation in AVL Cruise and
show an overall efficiency improvement of up to 11% for the Matlab. For comparison purposes, the direct-drive BEV with
transmission-equipped vehicle. a low-speed motor and a two-speed BEV with a high-speed
motor will also be modeled and simulated. The simulation
In 2012, one of the world leaders in transmission technol- results for different models are discussed in Secs. IV-VI.
ogy, Oerlikon Graziano, in collaboration with Vocis Driveline
Control, reported the results of a case study on a two-speed 100

system for EVs [11]. These results clearly demonstrate the


advantages of the two-speed transmission when compared to a
speed (km/h)

single-speed model used in EVs via four main characteristics: 50


acceleration time; maximum climbing grade; maximum speed;
and energy consumption.
As a result of various research activities, current trends 0
0 200 400 600 800 1000 1200 1400 1600 1800 2000
show that more and more vehicle and transmission manufac-
turers are working on implementing multi-speed drivetrains in F IG . 1: Velocity profile of the OCC cycle
electric and hybrid electric cars. For example, the Furtive eGT
from Exagon Motors has an innovative four-speed transmis-
sion, and Antonov has proposed a three-speed transmission,
both designed to work with electric motors [12]. However, all III. S IMULATION AND O PTIMIZATION
previously mentioned studies explore the application of multi-
speed transmissions in the passenger and sports car segments. A. Two-speed transmission ratio selection
In this paper, the multi-speed electric delivery van option is
According to Naunheimer [13], the total powertrain ratio
studied.
iA of a vehicle is the product of three factors, namely,
II. TASK DEFINITION iA = iS iiE (1)
The work described here pertains to the development, where iS is the ratio of the coupling element (clutch or torque
simulation and optimization of the drivetrain of a battery converter), i the gearbox ratio, and iE the final drive ratio, as
electric step van with a two-speed transmission, the final aim illustrated in Fig. 2. In most cases, iS is assumed to be 1, so
being a comparison of its efficiency with that of a single-speed that eq. (1) is simplified:
version. For practical purposes, a 2004 GM Workhorse P32
chassis with a body made by ENOVA Systems was selected. iA = iiE (2)
When designing an electric or hybrid vehicle, goals and
performance requirements must be set. Instead of considering 1) Selection of the first gear ratio: The largest powertrain
battery or motor characteristics, it was decided to focus on the ratio iA1 is determined by the climbing grade requirements, as
definition and optimization of the parameters of the two-speed shown below [13]
transmission, namely its gear ratios. Energy consumption in rdyn mF g(fR cos αst + sin αst )
the OCC test run and vehicle performance were used as the i A1 ≥ (3)
TM,maxηtot
major objective functions. The velocity profile of the OCC
cycle run is included in Fig. 1. The two vehicles to be with all parameters listed in Tables I and II.

Authorized licensed use limited to: COLLEGE OF ENGINEERING - Pune. Downloaded on December 13,2021 at 10:31:17 UTC from IEEE Xplore. Restrictions apply.
iA
goals is to verify this. The main design parameters of both
iS i iE motors are listed in Table II.

TABLE II: TM4 MΦtive B and Sumo MD motors


powertrain
value
parameter symbol unit
clutch gearbox final drive MΦtive B Sumo MD
ICE
D motor voltage Vmotor V 320 320-450
maximum starting
TM,start Nm 315 2100
torque
maximum torque
TM,max Nm 343 2100
(peak)
maximum power
Ppeak kW 200 200
(peak)
F IG . 2: Hierarchical structure of the powertrain ratio of a maximum power
Pcont kW 150 100
vehicle (D denotes the differential) (continuous)
maximum speed nmax rpm 12000 3100
base speed nrate rpm 5500 1000

2) Selection of the second gear ratio: The vehicle top


speed is considered as an important criterion to determine the As a follow-up to eqs. (3) and (4), along with the parameter
smallest powertrain ratio. At the same time, characteristics of values listed in the foregoing tables, the initial ranges of the
the electric motor should also be taken into consideration: its geartrain ratios for the pure-electric truck are given below:

maximum angular velocity as well as angular velocity at rated iA1 ≥ 23.0563
power. Hence, the range of the second geartrain ratio is given (5)
7.5632 ≤ iA2 ≤ 16.5017
by [13], [14]
3.6πnrate rdyn 3.6πnmax rdyn 3) Final drive ratio: Typical values of the final drive ratio
≤ i A2 ≤ (4) iE follow [16]:
30vmax 30vmax

Definitions of all parameters in the above relations are listed for best fuel economy: iE = 2.73, 2.91
in Tables I and II. for best pulling performance: iE = 3.73, 4.10, 4.30, 5.13

TABLE I: Vehicle parameters The truck model analyzed in this paper requires high
pulling performance and comes fitted with a 5.13 final drive
parameter symbol unit value ratio. Because of this, iE = 5.13 is the final drive ratio used
vehicle mass (GVWR) mF kg 6373 in this paper.
dynamic wheel radius rdyn m 0.383
rolling resistance coefficient fR — 0.007 With reference to eqs. (2) and (5), the design range for the
coefficient of aerodynamic drag fD — 0.7
vehicle frontal area Sf m2 7.02
two-speed gearbox ratio i can be bounded as:
iE

final drive gear ratio — 5.13 i1 ≥ 4.4944
maximum vehicle speed vmax km/h 105 (6)
maximum climbing grade q — 30% 1.4743 ≤ i2 ≤ 3.2167
road incline angle at max grade αst deg 16.6992
overall powertrain efficiency ηtot — 96% B. Matlab-Cruise co-simulation: goals and strategy
The tasks of simulation and optimization are outlined next.
A permanent magnet electric motor, the TM4 Sumo MD,
was selected for the direct drive BEV model. This motor was AVL Cruise is a full-featured vehicle design and simulation
designed to run without any transmission and connect directly software package for conventional or alternative fuel vehicles.
to the final drive ratio of the vehicle. For comparison purposes, It is widely used in industry for component and vehicle design.
the TM4 MΦtive B electric motor was adopted for the two-
speed drivetrain design, with its torque and efficiency maps Yang et al. [17] reported on the parametric drive train
scaled up in order to have the same power output as the lower design, modeling and performance simulation of a parallel
speed Sumo. This was done in order to compare two motors diesel hybrid electric bus using AVL CRUISE and MAT-
with the same output power but different speed characteristics LAB/SIMULINK. Their performance criteria included fuel
for use with different types of transmission systems. The motor consumption, vehicle top speed and acceleration time. Priority
scaling in the torque dimension was considered a reasonable was given to the improvement of fuel consumption by em-
approximation based on previous experience and studies [15], ploying a downsized engine and operating the engine in high
where the torque was scaled as a design variable. This opera- efficiency regions. Gao et al. [18] used Matlab and AVL Cruise
tion directly scales the torque and power output of the motor; in combination with a hybrid optimization algorithm based on
the efficiency map was scaled along with the torque. the Multi-island Genetic Algorithm (MIGA) and Non-Linear
Programming by Quadratic Lagrangian (NLPQL), which was
With the addition of a two-speed transmission, extra losses built with the ISIGHT software package (Design Optimization
are introduced in the drivetrain. However, the two-speed trans- Solutions) of Dassault Systèmes. In the paper cited above, min-
mission may improve performance due to its operation in more imum fuel consumption and minimum fluctuation of battery
efficient motor-speed and torque ranges; one of our research state of charge (SOC) were chosen as the objective functions,

Authorized licensed use limited to: COLLEGE OF ENGINEERING - Pune. Downloaded on December 13,2021 at 10:31:17 UTC from IEEE Xplore. Restrictions apply.
to the vehicle driveline, whereas the cost also depends on
many other components of the car. Moreover, since the object
of the research is a delivery step van working in the urban
environment, it is assumed that top speed and gradability are
less critical criteria than acceleration and energy-saving capa-
bilities. Thus, the main priorities for the design optimization
are used as the objective functions, namely,
• economy: to minimize the overall energy consumption
in the OCC test run;
• vehicle dynamic performance: to minimize the accel-
eration time from standstill to 100 km/h.
However, maximum gradability and top speed are considered
as well in that they limit the range of possible gear ratios, as
explained in section III-A.
2) Design variables: The design variables in this optimiza-
tion problem are the gear ratios x and y, stored in w = [x, y]T .
F IG . 3: Block diagram of the AVL Cruise simulation
model for a two-speed BEV: transmission was added 3) Optimization procedure: According to Sec. III-B1, in
between motor and final drive order to reach the optimization objectives of the minimum
acceleration time and minimum overall energy consumption,
single optimization of only one objective is not enough to rep-
resent the system optimization. In order to meet the optimiza-
whereas the SOC limit coefficients and the engine torque tion objectives simultaneously, multi-objective optimization is
coefficients were adopted as the optimization parameters. used. The optimization procedure is summarized below:
Sangtarash et al. [19] used AVL Cruise in their analysis of
braking performance and fuel economy for different braking a) Approximate the two objective functions—the overall
strategies in different drive cycles. Xiong et al. [20] optimized energy consumption in OCC test run, ue , and the acceleration
economical gear-shifting schedules for a purely electric bus time from standstill to 100 km/h, ta —by means of surrogate
with a three-speed transmission. This comparison was carried models1 . Each objective function is represented as a second-
out using AVL Cruise as a simulation tool with gear-shifting order bivariate polynomial, which is the response surface
schedules designed using different shifting methods. model in two variables, x and y [22], namely
Vehicle powertrain models can be readily built in the Cruise zi = β0 + β1 xi + β2 yi + β3 x2i + β4 yi2 + β5 xi yi , i = 1, · · · , 6
GUI and then component specifications entered for each item. (7)
The electric truck simulation is based on a simple rear-wheel where {βj }5j=0 are the unknown polynomial coefficients, while
drive electric vehicle platform. The two-speed transmission xi , yi and zi are values of the two gear ratios at the ith sample
model has an added transmission block and gear-shifting point and of the objective function, calculated for this pair of
program. gear ratios.
Optimization in AVL Cruise is limited to a matrix calcu- Upon assembling all six sub-equations (7), the relation
lation or component variation, i.e., varying certain parameters below follows [23]:
and running the calculations again to obtain many different z = Wβ (8)
results for different sets of parameter values or different in which
components. This method does not directly lead to an optimal ⎡ ⎤
⎡ ⎤ ⎡ ⎤ 1 x1 y1 x21 y12 x1 y1
solution, but only shows the results of many possible solutions z1 β0
within the design space. ⎢ ⎥ ⎢ ⎥ ⎢ 1 x2 y2 x22 y22 x2 y2 ⎥
z = ⎣ ... ⎦ , β = ⎣ ... ⎦ , W = ⎢
⎣ .. .. .. ⎥⎦
1) Design objectives: In general, for a vehicle design, some . . .
z6 β5
of the important performance evaluation criteria are: 1 x6 y6 x26 y62 x6 y6
(9)
• top speed;
Since W is a 6 × 6 square matrix, the linear system shown
• acceleration time from 0 to 60 mph (100 km/h); in eq. (8) has a unique solution, as long as W is not singular,
for the coefficient vector, namely
• maximum gradability;
β = W−1 z (10)
• system reliability/longevity;
• maintenance cost; Then, a set of simulation tests in AVL Cruise were con-
ducted, with the simulation results ported into Matlab.
• fuel or energy consumption.
1 A surrogate model is defined as an approximation of the input/output
The maintenance cost and the life time of the vehicle are function [21], which is built to mimic the behavior of the simulation model
out of the scope of this paper, as the research interest is limited with cheaper computational cost.

Authorized licensed use limited to: COLLEGE OF ENGINEERING - Pune. Downloaded on December 13,2021 at 10:31:17 UTC from IEEE Xplore. Restrictions apply.
b) Optimize both objective functions simultaneously using V. S IMULATION OF THE ELECTRIC DELIVERY TRUCK
the multi-objective genetic algorithm function included in the WITH THE TWO - SPEED TRANSMISSION
Matlab Global Optimization Toolbox, to obtain a Pareto front
of optimal solutions [11]. A. Model review
The two-speed BEV model uses the same base parameters
c) Input the optimal sets of transmission ratios returned as the direct drive model, shown in Table I, with one difference:
by Matlab into AVL Cruise, and calculate the two design the motor and transmission system. The motor is based on the
objectives. TM4 MΦtive B design, but has been scaled to match the peak
power output of the TM4 Sumo, for comparison purposes. The
The results of gear ratio optimization are discussed in motor is designed to run at much higher speed, but has a lower
Sec. V. peak torque rating. The transmission gear is intended to make
up for this lack of starting torque while allowing the vehicle
to reach the required top speed. Table II shows the modified
IV. S IMULATIONOF THE ELECTRIC DELIVERY TRUCK TM4 MΦtive B motor specifications.
WITH THE DIRECT DRIVE MOTOR
The gear-shifting program used for these tests is based on
A. Model review the standard gear shift program block in Cruise. An upshift
line and downshift line are defined on the motor torque-speed
For the vehicle used for the simulation of the Class- map in order to keep the motor operation in the high-efficiency
4 delivery truck based on a GM Workhorse platform, the range. The ideal shifting strategy would be to determine the
main parameters used to build the model of this vehicle are efficiency of the motor in both gears continuously and to
included in Table I. ”Direct drive” in this case means that choose the one with the higher efficiency. However, this option
the motor is connected directly to the final drive ratio of the was not available in AVL Cruise and is not realistic, since it
differential; no reduction gearbox is included in this model may cause rapid gear switching. It has been shown [24] that
(most current BEVs use a single-gear-reduction gearbox and the efficiency difference between running an optimal and sub-
the final drive). This removes the losses associated with a optimal shifting method is a small percentage. Also, since the
single-speed gearbox. The motor used for the direct-drive shifting strategy was defined based on motor speed and not
simulation has been specifically designed for this type of vehicle speed, the shifting points do not need to be modified
installation: it runs at lower RPM and has higher torque output for different transmission ratios.
than most BEV motors. The basic motor specifications of the
TM4 Sumo MD motor are listed in Table II. B. Optimization and verification of computed transmission
ratios
Compared with the two-speed BEV model illustrated in
Fig. 3, the direct drive BEV has a similar layout, but does not For the two-speed transmission, the objective functions
include the two-speed transmission module, which leaves only of minimum overall energy consumption and minimum ac-
the final drive ratio accessible to optimize. Hence, it is difficult celeration time from standstill to 100 km/h are calculated
to satisfy the basic performance requirements simultaneously, in AVL Cruise using many test runs. For each combination
i.e., the climbing grade of 30% and the maximum vehicle of transmission ratios in the design space, the corresponding
velocity of 105 km/h, since the designer has control of only overall acceleration time and energy consumption are calcu-
the final drive ratio. Simulation tests are conducted for two dif- lated as displayed in Figs 4 and 5. The first and second
ferent values of iE : 3.15 and 5.13, respectively. The simulation powertrain ratios, i1 , i2 , must satisfy the relations displayed
results are listed in Table III. It can be observed that although in eqns. (3 & 4) in order to meet the basic requirements of
low energy consumption can be achieved in both cases, the 30% climbing grade and 105 km/h maximum velocity.
direct drive BEV fails to meet some dynamic performance re- Figures 4 and 5 illustrate the acceleration time results
quirements. For iE = 3.15, the maximum velocity of 105 km/h from standstill to 100 km/h and overall energy consumption
can be reached; however, the resultant climbing grade is only with respect to i1 and i2 , respectively. A nonlinear negative
26.88%, which is well below 30%. When iE = 5.13, the relationship of the acceleration time with both i1 and i2 is
climbing grade is 46.92%, while the maximum velocity yields observed, i.e., the larger the two gear ratios, the lower the
only 65 km/h, failing to reach 105 km/h. acceleration time it can reach. Similar behavior of the overall
energy consumption is observed with i1 and i2 , as illustrated
in Fig. 5.
TABLE III: Simulation results for the electric delivery
truck with the direct drive motor According to Knowles at al. [11], the Pareto optimization
method based on a genetic algorithm is a robust tool to
maxi accel energy
iE
climbing
velocity time consumption realize multi-objective optimization. Details of this method are
grade included in [11], [25]. Figure 6 illustrates the two optimization
(km/h) (s) (kWh)
3.15 26.88% 105 18.26 4.111 objective functions of the system in terms of a Pareto front for
5.13 46.92% 65 –2 3.712 all optimum gear-ratio values, i.e. the minimum overall energy
consumption and acceleration time from standstill to 100 km/h.
It is noteworthy that more than one optimum gear-ratio
2 This result is not available because the maximum velocity the model can pair is obtained. Some optimum gear ratios yield a lower
reach is below 100 km/h. acceleration time but higher energy consumption, while other

Authorized licensed use limited to: COLLEGE OF ENGINEERING - Pune. Downloaded on December 13,2021 at 10:31:17 UTC from IEEE Xplore. Restrictions apply.
higher acceleration time. The optimization method does not
limit the designer to one strict point to reach the optimization
Acceleration time from standstill to 100 km/h (s)

objectives, but provides the possibility of a choice, depending


29 on whether the overall energy consumption, or the acceleration
time, is more important to the customer. For the objective of
27
minimum acceleration time from standstill to 100 km/h and
25 minimum overall energy consumption, gear ratios of i1 higher
23
than 5.5 and i2 higher than 3.0 are recommended.
21 As illustrated in Fig. 6, all the optimum points of the objec-
19
tive functions are normalized to their maximum values. Based
4 on the normalized points, a linear numerical interpolation is
17
0 3
used to obtain the Pareto frontier of the system. Denoting the
2 optimum objective function values for gear ratios i1 and i2 by
4
6
2 μ1 and μ2 , respectively, the distance to the origin from the
8 Pareto frontier μ is
1 Gear ratio i2
Gear ratio i1
10

μ = μ21 + μ22 (11)
F IG . 4: Acceleration time from standstill to 100 km/h
against i1 and i2 generated by AVL Cruise
The bi-objective minimum is thus the point of the Pareto
frontier lying closest to the origin (minimum μ), which yields
ta = 16.861 s and ue = 4.763 kWh, i.e., i1 = 5.90 and
i2 = 3.11.
Overall energy consumption in OCC cycle run (kWh)

Afterwards, the candidate sets of gear ratios thus obtained


4.90
are exported into AVL Cruise, to calculate the overall en-
ergy consumption and acceleration time, for the purpose of
4.85 validating whether the optimization objectives are reached.
Figures 7 and 8 illustrate the acceleration time and overall
4.95 energy consumption for the optimum first and second gear
ratios, respectively. Good agreement is observed between AVL
4.80
Cruise and the Pareto optimization methodology. Therefore,
4.75 the proposed optimization method is deemed validated.
4

3 Based on the verification via AVL Cruise, the minimum


0 acceleration time of 17.05 s and minimum energy consumption
2
4 2 of 4.77 kWh are reached at i1 = 5.9 and i2 = 3.11,
6
8 which validates the optimization results by Pareto frontier. The
1 Gear ratio i2
10 corresponding dynamic performance is shown in Table IV. It
Gear ratio i1
can be seen that the climbing grade and maximum velocity
F IG . 5: Overall energy consumption against i1 and i2 both meet the design requirements.
generated by AVL Cruise
Normally, for small and medium cars, the difference be-
tween two gearbox ratios can be up to a ratio of 1.6–1.8.
Objective 2, ue (overall energy consumption, kWh)

However, for heavy vehicles, with high operating torques, the


1
step between the two speed ratios can reach 1.9–2.0. Moreover,
0.995 in EVs and HEVs, some flexibility can be allowed for the
0.99 transmission ratios. The step between the two ratios in EVs and
0.985 HEVs can be higher than their conventional counterparts due
0.98 to the superior control over speed and torque of the traction
0.975 motor, while the speed/torque range is more limited for the
0.97 internal combustion engine. In our research, the ratio between
Optimum point
0.965 the gears is 1.9, which falls within the suggested EV ratio
0.96 range.

0.0 0.96 0.97 0.98 0.99 1


Objective 1, ta (acceleration time, s)
TABLE IV: Optimization result for the two-speed BEV
F IG . 6: The Pareto front plot for the two optimization with MΦtive B motor
objectives energy
maxi accel
climbing consump-
i1 i2 velocity time
grade tion
(km/h) (s)
(kWh)
5.9 3.11 39.69% 120 17.05 4.77
optimum gear ratios show lower energy consumption with

Authorized licensed use limited to: COLLEGE OF ENGINEERING - Pune. Downloaded on December 13,2021 at 10:31:17 UTC from IEEE Xplore. Restrictions apply.
expensive. These factors should be taken into account when
selecting a combination of motor and transmission.
Acceleration time from standstill to 100 km/h (s)

VII. C ONCLUSIONS
24.5
Based on the two-speed electric delivery step van, an in-
23.0 novative approach to the optimization of the gear transmission
ratios is proposed in this paper. The objective functions are
21.5
chosen, according to the requirements of electric vehicles,
20.0 as the minimum overall energy consumption and minimum
acceleration time from standstill to 100 km/h. An electric
18.5 4.5 step van was simulated in the AVL Cruise software package,
17.0 5 which was used to obtain the relationship of the overall energy
1.5
consumption and acceleration time with respect to both gear
2 5.5
2.5
ratios. As these two objective functions are to be optimized
3
6 simultaneously, the multi-objective optimization methodology
Gear ratio i1
is used. The authors applied the Pareto optimization method
Gear ratio i2
based on a genetic algorithm and validated the optimum gear
F IG . 7: Acceleration time from standstill to 100 km/h ratio values using AVL Cruise.
against i1 and i2 after optimization
Using the two-speed transmission with the high-speed mo-
tor, the minimum acceleration time from standstill to 100 km/h
of 17.05 s and minimum energy consumption of 4.77 kWh
is reached with i1 = 5.9 and i2 = 3.11. The direct drive
Overall energy consumption in OCC cycle run (kWh)

BEV with a low-speed motor, on the other hand, is not able to


reach both basic requirements of climbing grade and maximum
velocity at the same time, as per the results listed in Table III.
In addition, based on the layout of this direct drive BEV model
5.05 with a low-speed motor, its two-speed version was developed
and simulated, whose energy efficiency is better than its high-
4.95 speed counterpart, with a compromise of higher overall cost.
The approach reported in this paper provides a straight-
4.85
4.5 forward and robust methodology to optimize powertrain gear
5
ratios for electric vehicles that can be readily applied to the
4.75
optimum design of other systems in electric vehicles.
1.5 5.5
2 The research work conducted in this paper also shows a
2.5
6 significant difference in a two-speed transmission with the
3 Gear ratio i1
electric trucks versus electric passenger cars, where an effi-
Gear ratio i2 ciency increase was much more significant.
F IG . 8: Overall energy consumption against i1 and i2 after
optimization ACKNOWLEDGEMENTS
The research work reported here was supported by a
grant from Automotive Partnerships Canada, APCPJ418901-
VI. A LTERNATIVE M OTOR 11. Technical support was also provided by industrial partners
Infolytica, Linamar, and TM4.
As an alternative to the two-speed vehicle with a high-
speed motor, the TM4 Sumo MD low speed motor was also R EFERENCES
tested with a two-speed transmission. A matrix calculation with
various gear ratios was carried out in AVL Cruise; the results [1] D. Y. Lee, V. M. Thomas, and M. Brown, “Electric urban delivery
show a similar trend to those illustrated in Figs. 4 and 5. Both trucks: Energy use, greenhouse gas emissions, and cost-effectiveness,”
American Chemical Society Publications, Environ.Sci.Technol., vol. 47,
efficiency and acceleration time improved using a lower-ratio pp. 8022–8030, 2013.
first gear. Using a lower-ratio second gear improved efficiency [2] M. Kent, “From car converter to truck OEM,” Charged EV Magazine,
and decreased acceleration time slightly in the test range. One vol. 10, pp. 54–59, 2013.
candidate set of gear ratios is i1 = 1.12, i2 = 0.62. This leads [3] G. W. Osborne, “Electric motor driven vehicle having driver controlled
to a ratio of 1.8 between the gears with an energy use of variable speed drive,” US Patent, no. 3202234, Aug 24, 1965.
4.46 kWh during the OCC cycle and an acceleration time of [4] G. Osborne, “Automatic or manual multi-speed transmission for elec-
16.1 s, which is lower than its high-speed counterpart. trically propelled vehicle,” US Patent, no. 3888326, June 10, 1975.
[5] F. Pollock and D. Trombley, “Output torque matching in a multi-speed
It is noteworthy that, in general, the low-speed motor with electric vehicle,” US Patent, no. 5629593, May 13, 1997.
two-speed transmission appears to be more efficient than its [6] D. Deng, J. Xu, and K. Farkas, “Backlash elimination in the drivetrain
high-speed counterpart. However, it is heavier, larger and more of an electric vehicle,” US Patent, no. 5821720, Oct 13, 1998.

Authorized licensed use limited to: COLLEGE OF ENGINEERING - Pune. Downloaded on December 13,2021 at 10:31:17 UTC from IEEE Xplore. Restrictions apply.
[7] B. Jerew. (2014) Future electric vehicles may have multi-
speed gearboxes. The Green Optimistic Newsletter. [Online].
Available: http://www.greenoptimistic.com/2014/01/08/future-electric-
vehicles-may-multi-speed-gearboxes/U2KZUSjRbCm
[8] M. Hanlon. (2009) New multi-speed electric vehicle transmission
improves ev performance and range. Gizmag Blog. [Online].
Available: http://www.gizmag.com/new-multi-speed-electric-vehicle-
transmission-improves-ev//-performance-and-range/11670
[9] A. Sorniotti, T. Holdstock, G. Pilone, F. Viotto, S. Bertolotto, M. Everitt,
R. Barnes, B. Stubbs, and M. Westby, “Analysis and simulation of the
gearshift methodology for a novel two-speed transmission system for
electric powertrains with central motor,” Proceedings of the Institution
of Mechanical Engineers, Part D: Journal of Automobile Engineering,
no. 226, pp. 915–929, 2012.
[10] Q. Ren, D. Crolla, and A. Morris, “Effect of transmission design
on electric vehicle performance,” in Proc IEEE Vehicle Power and
Propulsion Conference (VPPC), Dearborn, USA, 2009.
[11] J. Knowles, D. Corne, and K. Deb, Multiobjective Problem Solving
from Nature: From Concepts to Applications. Springer-Verlag, Berlin
Heidelberg, 2008.
[12] L. Masson. (2011) Multi-speed transmissions mean better
efficiency for electric cars. Plugin cars Forum. [Online]. Avail-
able: http://www.plugincars.com/efficiency-multi-speed-transmissions-
electric-cars-107656.html
[13] H. Naunheimer, Automotive Transmissions. Fundametals, Selection, De-
sign and Application, 2nd edition. Springer-Verlag, Berlin-Heidelberg,
2011.
[14] P. D. Walker, S. A. Rahman, B. Zhu, and N. Zhang, “Modelling, simu-
lations, and optimisation of electric vehicles for analysis of transmission
ratio selection,” Advances in Mechanical Engineering, vol. 2013, 2013.
[15] B. Egardt, N. Murgovski, M. Pourabdollah, and L. Johannesson Mardh,
“Electromobility studies based on convex optimization: Design and
control issues regarding vehicle electrification,” Control Systems, IEEE,
vol. 34, no. 2, pp. 32–49, 2014.
[16] B. Dahl. (2014) Differential gear ratio. [Online]. Avail-
able: http://www.4-wheeling-in-western-australia.com/differential-gear-
ratio.html
[17] Y. Yang, H. Zhao, and H. Jiang, “Drive train design and modeling
of a parallel diesel hybrid electric bus based on AVL/Cruise,” EVS25
World Battery, Hybrid and Fuel Cell Electric Vehicle Symposium, World
Electric Vehicle Journal, vol. 4, 2010.
[18] J. Gao, Z. Liu, Z. Guo, and Y. Wei, “Optimization of hybrid electric bus
control strategy with hybrid optimization algorithm,” in Proceedings
of the 2nd International Conference On Systems Engineering and
Modeling (ICSEM-13), 2013.
[19] F. Sangtarash, H. Nehzati, S. Haddadi, and B. Haghpanah, “Effect of
different regenerative braking strategies on braking performance and
fuel economy in a hybrid electric bus employing CRUISE vehicle
simulation,” SAE Int. J. Fuels and Lubricants, vol. 1, no. 1, pp. 828–
837, 2009.
[20] G. Xiong, J. Xi, Y. Zhang, Y. Jin, and H. Chen, “Optimal design and
simulation evaluation of economical gear-shifting schedule for amt in
pure electronic bus,” in IEEE International Conference on Industrial
Technology (ICIT), 2010, pp. 1131–1135.
[21] J. Kleijnen, “Kriging metamodelling in simulation: a review,” European
Journal of Operational Research, vol. 192, no. 3, pp. 707–716, 2007.
[22] A. Khuri and J. Cornell, Response Surfaces: Designs and Analyses:
Second Edition. Marcel Dekker, Inc., 1996.
[23] D. Montgomery, Design and analysis of experiments: response surface
method and designs. New Jersey: John Wiley & Sons, Inc., 2005.
[24] A. Pakniyat and P. Caines, “The gear selection problem for electric
vehicles: an optimal control formulation,” in to be presented at the
13th International Conference on Control, Automation, Robotics and
Vision (ICARCV 2014), Marina Bay Sands, Singapore, 2014.
[25] K. Deb and H. Gupta, Evolutionary Multi-Criterion Optimization:
Searching for Robust Pareto-Optimal Solutions in Multi-objective Op-
timization, ser. Lecture Notes in Computer Science. Springer Berlin
Heidelberg, 2005, vol. 3410.

Authorized licensed use limited to: COLLEGE OF ENGINEERING - Pune. Downloaded on December 13,2021 at 10:31:17 UTC from IEEE Xplore. Restrictions apply.

You might also like