Multi-Objective Design Optimization of Axial Flux Surface Mounted
Permanent Magnet Brushless DC Motor for Electric Vehicle
Application Based on Genetic Algorithm
Amit N. Patela* and Bhavik N. Sutharb
a Electrical Engineering Department, Institute of Technology, Nirma University,
Ahmedabad, India
b Electrical Engineering Department, Government Engineering College, Modasa, India
Correspondence details:
Prof. Amit N. Patel
Assistant Professor
Electrical Engineering Department
Institute of Technology
Nirma University, Ahmedabad, INDIA-382481
Email: [Link]@[Link]
.
Multi-Objective Design Optimization of Axial Flux Surface Mounted
Permanent Magnet Brushless DC Motor For Electric Vehicle
Application Based on Genetic Algorithm
Abstract: This paper presents the multi-objective design optimization of axial flux
permanent magnet brushless dc motor for electric vehicle application. Double rotor
sandwiched stator type axial flux permanent magnet motors are the most
compatible motor in electric vehicle applications. Motor rating of 250 W, 150 rpm
is calculated based on vehicular dynamics and application needs. The axial flux
permanent magnet brushless DC (PMBLDC) motor as per calculated rating has
been designed for direct drive application in electric two wheeler. It is always
expected that electric motors designed for electric vehicle application possess high
efficiency and less weight. Initially, optimization is carried out considering two
separate single objective functions efficiency and weight. Efficiency and weight of
motor are two opposite performance parameters hence it is challenging task to
improve both simultaneously. Finally, multi-objective optimization is performed
with an objective of simultaneous increase in efficiency and decrease in weight.
The main contribution of present work is to propose the best combination of design
variables obtained using genetic algorithm (GA) optimization technique and
design of motor based on optimized design variables. Optimization technique is
validated with help of 3-D finite element analysis (FEA). It is analyzed that results
obtained from multi-objective optimization are effective in efficiency
improvement and weight reduction. Proposed optimization technique can be
applicable for multi-objective performance enhancement of any nonlinear
engineering design comprising various design variables for specific application.
Keywords: Axial Flux PMBLDC motor, Computer Aided Design, FE Analysis,
Optimization, Genetic Algorithm
1. Introduction:
Electric Vehicles (EVs) are rapidly gaining popularity due to shortage of fossil
fuel and thrive to minimize carbon emission for an air quality improvement [1]. Each
major vehicle manufacturers are coming with new models of EV with futuristic
technologies to ensure efficient, reliable and environment friendly system [2]. Electric
motor powered by batteries act as a prime mover in EV. It is expected that electric motors
used in EV applications offer high operational efficiency with low weight. High
operational efficiency and low weight of motor will enhance battery life and vehicle drive
range. [3]. Permanent magnet (PM) motors using rare earth magnets are emerged as front
runners in EV applications. Application of PM material inherently increases efficie nc y
and power density of electric motor. Permanent magnet motors are classified as radial
flux motors or axial flux motors according to relative directions of magnetic flux and
exciting current. Flux travels radially and current travels axially in radial flux motor while
flux travels axially and current travels radially in axial flux motors [4]. The axial flux
permanent magnet motors offer futuristic advantages like high efficiency, high power
density, flat shape, better torque/current ratio and improved utilization of copper [5].
Axial flux permanent magnet motors are more compatible in direct drive application than
radial flux permanent magnet motors. Axial flux motors have various configuratio ns
based on number of stators & rotors and relative position of them [6]. Axial flux PM
motors are further classified as surface PM axial flux motor or interior PM axial flux
motor according to rotor construction. PMs are fixed on surface of rotor core in surface
mounted axial flux PM motor while PMs are buried inside rotor core in interior PM axial
flux motor. Electrical motors used in electric vehicles can be classified as direct driven
motors or indirect driven motors. Direct driven motors are more suitable in electric
vehicle applications. Axial flux permanent magnet motors are more advantageous
compared to radial flux permanent magnet motors in direct drive applications [7]. Low
speed axial flux machines are best suited in low speed applications such as electric
vehicle, gear less wind generator, ship propulsion and room less elevators [8]. Axial flux
permanent magnet motors are naturally suitable in wide range of applications having
unconventional geometric requirements with dimensional constraints. In this work double
rotor single stator surface mounted PM topology is selected for direct drive electric
vehicle application due to its best suitability [9].
Objective function is specific feature of optimized machine. Electric motor design
is required to be optimized considering application requirements. In direct drive EV
applications important features of electric motors are high efficiency and less weight
hence it is necessary to perform multi-objective optimization. Efficiency and weight of
electric motor are two contradictory performance parameters. Two contradictory
objective functions are required to be tackled simultaneously in multi-objective
optimization [10]. Axial flux permanent magnet synchronous generator (AFPMSG) of
30 kW rating has been optimized for cost of active material using genetic algorithm and
validation is carried out with FEA [11]. Design optimization of stator structure of
brushless servo motor is discussed considering voltage, torque and power [12]. Optimal
design of electric motor is important design issue hence it is point of interest for many
researchers. Power density optimization of 1 kW, 70 V axial flux PM machine is
performed with GA technique [13]. Reza Ilka et al. discussed optimization of PM motor
with an objective to reduce loss and increase power density [14]. This paper is focused
on multi-objective design optimization of double rotor single stator axial flux PM motor.
Optimal decision is required to be taken considering trade-offs between two single
objective functions regarding efficiency and weight in direct drive applications. Electric
Vehicle requirements considered in present work are 150 kg. laden weight, maximum
speed 25 kmph. and acceleration requirement 25 kmph. in 9 second. Specifications of
electric motor for vehicular application are determined based on application needs and
vehicular dynamics [15]. Following rating of axial flux PMBLDC motor has been
determined for above mentioned application needs.
1. rated torque: 15.91 N.m.
2. rated speed: 150 rpm
3. rated power : 250 W
4. maximum power: 803.4 W
5. maximum torque: 58.32 N.m.
Double rotor single stator axial flux PM motor of 250 W and 150 rpm has been
designed considering appropriate design variables in this work. Computer Aided Design
(CAD) programming including decision making loops has been done. Multi-objective
design optimization has performed using Genetic Algorithm (GA) technique. The main
objective in the current work is to propose the best combination of design variables
obtained using genetic algorithm (GA) optimization technique and design of motor based
on optimized design variables. Validation of CAD based design and GA based multi-
objective optimized design is done with three dimensional Finite Element Analys is
(FEA). Subsequent sections of paper explain the work carried out and result analys is.
Section 2 explains important design considerations and design information. Derivation of
two objective functions for efficiency and weight are presented in section 3. Genetic
algorithm technique for optimization is discussed in section 4. Section 6 explains multi-
objective function and simulation details for optimization. Three dimensional FEA to
validate proposed multi-objective optimization has been presented in section 7.
2. Design of Axial Flux PMBLDC Motor
View of double rotor single stator axial flux PMBLDC motor considered in present work
is shown in Figure1.
Stator winding
Rotor
PM
Figure1 View of Axial Flux PMBLDC motor
Four main stages of electrical motor design are calculation of main dimensions, design of
stator, and design of rotor and performance prediction [16]. Size of motor is primarily
governed by electrical loading and magnetic loading. Type of PM material significa ntly
influence performance of motor. Axial flux PMBLDC motor is designed with high energy
Neodymium Iron Borone (NdFeB) 35 grade PM material. Salient properties of 35 grade
NdFeB are as under:
1. Residual flux density: 1.23 T
2. Maximum energy product: 35 MGOe
3. Coercive force: 890 KA/m
4. Relative permeability:1.0
Motor is designed with 16 rotor poles and 48 stator slots. Soft iron material is selected
for rotor core material. Stator core is made of tape wound laminated silicone steel. Three
phase winding incorporating 26 conductor per slot is connected in star pattern. Main
dimensions are determined and sizing of motor has been carried out based on performance
specifications and assumed design variables. Dimensions of magnetic sections are
determined based on flux and permissible flux density in respective magnetic materia l
[17]. Various design variables are assumed considering availability of materials and
manufacturability. Initially designed motor has been considered as reference motor in
present work. Following table shows design information of reference motor.
Table [Link] information of reference motor
Parameter Value
Torque 15.91 N.m.
Stator outer diameter 0.182 m.
Stator inner diameter 0.104 m.
Axial length 0.090 m.
Initial efficiency 88.15 %
Initial weight 9.9 kg.
3. Objective Functions
Efficiency and weight are two main performance parameters of axial flux PM motors.
Various losses can be calculated as under based on geometrical dimensions and materia l
properties for efficiency estimation. Losses are classified as copper loss, iron loss and
frictional loss. Copper loss (Pcu) due to winding resistance can be obtained from following
equations [18].
2
Pcu = 2* I ph * R ph
(1)
ρ * ns2 * ( Ro − Ri )
Rs =
K cp * As
(2)
ρ * ns2 * π * ( wbi + 2d s )
Re =
4* K cp * As
(3)
N s * ( Rs + Re )
R ph =
N ph
(4)
Where Rs slot resistance, Re end turn resistance, Iph current per phase, Rph phase
resistance, ns number of conductors per slot, Ro outer diameter, Ri inner diameter, kcp
packing factor and As slot area.
Iron loss in motor is dependent on hysteresis loss and eddy current loss. Hysteresis loss
and eddy current loss are governed by flux density and frequency of flux reversal.
Estimation of iron loss is done with following equation
=Pi K h Bm1.6 f + K e Bm2 f 2 (5)
Where Kh hysteresis coefficient, Bm maximum flux density in core, f frequency of flux
reversal and ke coefficient of eddy current loss.
1
=Pfr C f ρ r (π n 3 )(D5o − D5i )
2 (6)
Frictional loss (Pfr) is governed by friction coefficient, air density, main dimensions and
speed. Generally frictional loss is usually assumed to be 20% of iron loss in low speed
machine. Motor efficiency is expressed as under,
Pout
η=
Pout + Losses
(7)
Pout
ηopt =
Pout + Pcu + Pi + P
fr (8)
The abovementioned equation is chosen as an objective function for efficie nc y
optimization.
Weight of particular motor section can be calculated by product of volume and materia l
density respectively. Following are equations derived to determine volume and weight of
various motor sections.
=Vsc 0.9* π ( Ro2 - Ri2 ) * 2(W + d s ) - N s (d s * w )( Ro - Ri ) (9)
sbi sb
Wsc = Vsc * ρi
(10)
=Vcu [ Ass * S * N s ( Ro − Ri )]
f
(11)
Wcu = Vcu * ρcu (12)
=Vrbi 2* Lcr * π ( Ro2 − Ri2 ) (13)
Wrbi = Vrbi * ρi (14)
2 2 (15)
= [(2π ( Ro − Ri ) * Lm ) − (( Ro − Ri ) * Lm * τ f * N m * 2)]
V pm
W pm = V pm * ρ pm
(16)
Weight, W = Wsc + Wcu + W + W pm (17)
rbi
Following equation is chosen as objective function for weight optimization.
Wopt
= k π ( R 2 − R 2 ) * (2W + 2d ) − N (d * w )( R − R ) ρ + [ A * S * N ( R − R )]* ρ +
i o i sbi s s s sb o i i ss f s o i cu
(18)
[(2π ( Ro2 − Ri2 ) * Lm ) − (( Ro − Ri ) * Lm * τ * N m * 2)]* ρ pm + 2 * Lcr * π ( Ro2 − Ri2 ) * ρi
f
where Vsc volume of stator core, Wsbi width of stator back iron, ds depth of stator slot,
Ns number of slots, Wsb width of slot base, Wsc weight of stator core, Vcu volume of
copper, Wcu weight of copper, Ass area of stator slot, Sf space factor, ρi density of iron,
ρcu density of copper, Lm thickness of PM, τf magnet fraction, Nm number of pole, Vpm
volume of PM, Wpm weight of PM and ρpm density of PM.
4. Genetic Algorithm an Optimization Technique
Genetic Algorithm (GA) is considered the most appropriate optimization technique for
electrical motor design because it is nonlinear process involving many design variables
having specified upper band and lower band [19]. Usually design variables affect each
other and vary simultaneously. The nonlinear nature of the GA makes it the most suitable
for motor design optimization. GA starts with number of design variables used for optimal
design, the function which is to be optimized (fitness function), number of populatio ns
and generations, upper and lower limits of design variables. Figure 2 illustrates block
diagram of GA based optimization technique. To optimize motor design there are mainly
four operators: (i) Generate population (ii) Selection (iii) Crossover (iv) Mutation.
Start
Input: No. of variables,
generations and population
Initial population
generation
New population Process of selection
Process of crossover Process of crossover
and mutation
Process of mutation
Is
criteria
No satisfied ?
Yes
Stop
Figure 2. Block diagram for genetic algorithm
The population is arbitrarily produced from the given ranges of various design variables.
Population is a set of different chromosomes which have different genes and different
values for the objective function. Each of the chromosomes are arbitrarily generated and
only one of the chromosomes from the whole population is considered.
In selection procedure, whole population is initially sorted according to the fitness value.
The selection process retains the chromosome with high efficiency and discards the
chromosomes with lowest efficiency.
The process of crossover ensures that sufficient diversity is maintained during the entire
process of genesis. The mating pool has that population which is selected based on its
fitness. Genes of the two different chromosomes are exchanged arbitrarily and it produces
to two completely new and different chromosomes.
The mutation process presents abrupt and random changes in the original chromosomes.
The mutation is done by either increasing or decreasing the original genes by percentage
indicated by the randomly generated number.
5. Details of Simulation
Parametric analysis is carried out using CAD programming to select influential design
variables. Five influential design variables with appropriate range for optimization are as
under. Materials availability, material properties and manufacturability govern range of
design variables.
Air gap flux density: 0.4 T < Bg < 0.9 T
Slot electrical loading: 100 A <Is < 400 A
Stator diametric ratio: 1.3 <Kr< 2.5
Current density: 4 A/mm2 < Jmax < 10 A/mm2
Air gap length: 0.25 mm < lg <1.0 mm
Range of air gap flux density depends on type of PM material and ferromagnetic materia ls
used. Range of slot electric loading and current density depend permissible temperature
rise, cooling arrangement and thermal stability. Range of air gap length is governed by
manufacturability, acoustic noise and cogging torque [17]. Table 2 illustrates each
chromosome’s 1x5 array for proposed optimization.
Table 2. Chromosome representation
Bg Is Kr Jmax lg
Single objective optimization:
Optimization of electrical machines is making trade-off between different objectives and
is usually done for any one of the performance parameters in single onjective
optimization. In this section, efficiency and weight are considered as two single objectives
for optimization. Optimization using GA technique is carried out considering these two
objective functions separately and analysis is done. Influence of number of optimized
design variables on fitness function is shown in Table 3. It is analyzed fitness functio ns
improve i.e. efficiency increases and weight reduces as number of design variables are
increased.
Table 3. Influence of number of variables on fitness functions
Variables Efficiency Weight
Bg , K r 90.72 % 9.25 kg.
Bg , K r , J max 90.83 % 8.55 kg.
Bg , K r , J max , I s 91.30 % 7.95 kg.
Bg , K r , J max , I s , lg 91.50 % 7.56 kg.
Note that in the present study the population size selected was 100 with cross over
probability as 75% and 1% mutation rate. The simulations were always carried out for
300 generations, however in few cases it is observed that the optimized result is achieved
within 53 generations. The execution time for the simulation was found to be 12.34 sec.
for 300 generations. The simulations were carried out on Intel CPU core 3, I3-4150 @
3.50 GHz with 4 G RAM. Fitness functions improve as number of generations are
increased. Optimum efficiency is converged after 53 generations and optimum weight is
converged after 68 generations. Table 4 illustrates optimum efficiency and weight for
constraint as well as unconstraint optimization based on GA technique and its comparison
with initial CAD based design. Present study incorporates constraints related to
geometrical dimensions and materials properties.
Table 4. Comparison of CAD and GA based design
Initial Optimized design
Parameters
Design Unconstraint Constraint
Efficiency (%) 88.15 91.50 90.85
Weight(kg) 9.90 7.56 8.85
Optimum efficiency of 91.5% converged when efficiency is considered as single
objective of optimization. Optimum weight of 7.56 Kg. converged when weight is
considered as single objective for optimization.
Multi-objective Optimization:
This section presents multi-objective optimization of axial flux PM motor considering
both efficiency and weight concurrently. Multi-objective fitness function formulated in
this work is combinational function for optimization of efficiency and weight. Reduction
of motor weight results in to reduction of dimensions. Due to reduction of motor
dimensions copper conductor area also reduces henceforth winding resistance and copper
losses increase. In similar line of that, reduction of dimensions result in to reduced
sectional area of iron. Reduced sectional area of iron increase flux density and iron losses.
Increase in copper and iron losses reduce motor efficiency. Efficiency and weight of
electric motor are to contradictory performance parameters. Following is fitness functio n
for GA based multi-objective optimization of two contradictory performance parameters.
W ( x) η ( x)
f ( x)
= −
Wmin ηmax (19)
Multi-objective problem is treated as single objective problem with above mentio ned
function. Optimum efficiency and weight are obtained from minimization of above-
mentioned multi-objective fitness function using GA technique. Optimized weight (Wmin)
and efficiency (ηmax ) are determined from single objective optimization. Multi-objective
function is minimized with GA optimization technique. As a result of multi-objective
optimization optimum efficiency of 89.78 % and optimum weight of 8.43 kg. converged
after 92 iterations. Optimized value of design variables are: air gap flux density (Bg )
0.7946 T, diametric ratio (Kr) 1.99, slot electric loading (Is) 166 A, current density (Jmax)
4 A/mm2 and 0.7 slot packing factor (Kcp ).
Figure 3. Variation of f(x) with number of generations
The convergence of multi-objective function is shown in Figure 3. Multi-objective
function is converged after 92 generations with an execution time of 21.45 sec.
Comparison between initial design and optimized design considering multi-objective
fitness function is shown in Table 5. It is evident that two performance parameters are
improved as a result of multi-objective optimization.
Table 5. Comparison of Initial and multi-objective optimized design
Multi-objective optimize d
Initial design
Parameters
Design
Unconstraint Constraint
Efficiency (%) 88.15 89.78 88.7
Weight (kg ) 9.9 8.43 8.9
Outer diameter(mm) 178 162 162.4
Inner diameter(mm) 102.8 81.2 81.4
Thickness of PM (mm) 2.7 3.1 3.4
Axial length(mm) 90.0 84.8 87.9
Efficiency obtained from constraint optimization is more than initial design but less than
unconstraint optimized design similarly weight obtained from constraint optimization is
less than initial design but more than unconstraint optimized design. Constraints
associated in this work are related geometrical dimension and material properties.
Relative comparison between initial design and optimized design are shown in Figure 4.
This performance comparison chart is prepared on per unit bases considering initia l
design as reference.
Figure 4. Comparison of Initial and GA based multi-objective optimized design
Enhancement of GA based optimized model has been authenticated by data analys is.
Reduction of motor weight significantly enhances performance of direct drive
application. Efficiency improvement leads to enhanced battery life and drive range.
6. Finite Element Analysis
Three dimensional finite element analysis (FEA) is carried out to validate CAD and GA
based constraint design. Data obtained from CAD and GA based design is used as input
for FEA. Appropriate meshing is done after modeling. Magnetic field calculations are
carried out to obtain torque profile and flux density spectrum. Flux density distributio n
of CAD based designed motor and GA based constraint designed motor is obtained from
FEA and shown in Figure 5 and Figure 7 respectively. Flux densities obtained in various
sections of magnetic circuit are in line with the assumed flux densities respectively.
Figure 5. Flux density spectrum of initially designed Axial Flux PM motor
Torque profile of CAD based designed motor and GA based optimized (constraint) motor
design are shown in Figure 6 and Figure 8 respectively.
Figure 6. Torque profile for CAD based motor design
Figure 7. Flux density spectrum of GA based constraint optimized Axial Flux PM motor
Figure 8. Torque profile for GA based optimized (constraint) motor design
Results obtained from the FEA validate CAD and GA based designs and comparative
analysis is shown in Table 6. Average torque obtained from FEA is fairly matching with
CAD based designed motor and GA based optimized motor. Torque developed in FEA is
marginally less by 2.45 % and 3.64 % with reference to CAD and GA based optimizatio n
respectively. This marginal difference is ascribed due to empirical formulas and nonline ar
characteristic of magnetic materials.
Table 6. Validation of Designed Axial flux PMBLDC motor
Optimized Design
Initial Design
Motor Parameters (constraint)
CAD FE GA FE
Average Torque (Nm) 15.91 15.52 15.91 15.49
Air gap flux density(T) 0.75 0.76 0.82 0.79
Stator core flux density(T) 1.5 1.60 1.5 1.62
Stator teeth flux density (T) 1.7 1.75 1.7 1.78
Rotor core flux density(T) 1.5 1.62 1.5 1.58
Phase inductance ( mH) 17.4 17.9 13.8 14.1
7. Conclusion
Axial flux surface mounted dual rotor sandwiched stator PMBLDC motor is the most
compatible motor in electric vehicle applications. It offers various attractive features viz.
high efficiency and flat shape compared to radial flux PM motors. Electric vehicle
application necessitates motor with high efficiency and low weight. GA based
unconstraint as well as constraint design optimization has been performed to enhance two
contradictory performance parameters of 250 W, 48 V axial flux PMBLDC motor.
Efficiency of motor is increased from 88.15 % to 89.78 % and weight is reduced from 9.9
kg. to 8.43 kg. using unconstraint design optimization. From constraint design
optimization, efficiency is increased from 88.15 % to 88.70 % and weight is reduced from
9.9 kg. to 8.8 kg. Efficiency increment along with weight reduction is noteworthy
enhancement in performance of motor with GA based multi-objective design
optimization. It is observed that multi-objective optimized motor is more realistic and
practically implementable. Electromagnetic analysis with finite element (FE) technique
is performed to authenticate the design optimization. Fair agreement between results from
finite element analysis and constraint as well as unconstraint multi-objective design
optimization validate correctness of GA based optimization technique. It is concluded
that performance of non-linear engineering system having contradictory performance
parameters can be enhanced with application of GA based optimization technique.
References:
1. Arumugam Palanivel, Sutha Padmanabhan.2018. “Software-based performance estimation
and real-time thermal analysis of brushless direct current motor with corroded permanent
magnets.” Computers and Electrical Engineering, Elsevier, Vol.71, pp.938-952.
2. I Fraser, J W Twidell and R Rue.2011. “Instrumenting, Testing and analysing an Electric
Vehicle.” International Journal of Ambient Energy, Vol.17, No.3, pp.162-168.
3. X. D. Xue, K.W. E. Cheng, T.W. Ng, and N. C. Cheung. 2010. “Multi-Objective
Optimization Design of In-Wheel Switched Reluctance Motors in Electric Vehicles.” IEEE
Transaction on Industrial Electronics, Vol. 57, No. 9.
4. S. Leung and [Link].1980. “A New Design Approach for Small Axial Field Electrical
Machines.” IEEE Transactions on Power Apparatus and Systems, [Link]-99, No.4,
pp.1679-1685.
5. [Link]. 1987. “Axial Field Electrical Machines-Design and Applications”, IEEE
Transaction on Energy Conversion, Vol. EC-2, No.2, pp 294-300.
6. M. Aydin, S. Huang and T.A. Lipo. 2004. “Axial Flux Permanent Magnet Disc Machines:
A Review.” in Conf. Record of SPEEDAM, pp. 61-71.
7. N A Rahim, Hew Wooi Ping and M Tadjuddin. 2007. “Design of Axial Flux Permanent
Magnet Brushless DC Motor for Direct Drive of Electric Vehicle.” IEEE Power Engineering
Society General Meeting, pp.1-6.
8. René Louis Ficheux, Federico Caricchi, Fabio Crescimbini and Onorato Honorati.2001.
“Axial Flux Permanent Magnet Motor for Direct Drive Elevator System without Machine
Room.” IEEE Transactions on Industrial Applications, Vol. 37, No. 6.
9. P.R. Upadhyay, K.R. Rajagopal and B.P. Singh.2003. “Computer aided design of an
axial-field permanent magnet brushless dc motor for electrical vehicle.” Journal of Applied
Physics, Vol. 93, pp 8689-8693.
10. Goga Cvetkovski and Lidija Petkovska.2016. “Genetic algorithm as a tool for multi-
objective optimization of permanent magnet disc motor.” Archives of Electrical Engineering,
Vol.65, No.2.
11. N. Rosatami, M.R. Feyzi, J. Pyrhonen, A. Parviainen and V. Behjat. 2012. “Genetic
Algorithm Approach for Improved Design of a Variable Speed Axial-Flux Permanent-
Magnet Synchronous Generator.” IEEE Transactions on Magnetics, vol.48, no.12, pp. 4860-
4865.
12. Bogdam Dumitru VARATICEANU, Paul MINCIUNESCU, Silviu Stefan MATEL and
Constantin NICOLESCU. 2017. “Optimization and Validation of a 10 kW Permanent
Magnet Brushless Motor for Small Electric Vehicles.” Electric Vehicles International
Conference & Show, pp.1-5.
13. Amin Mahmoudi, Solmaz Kahourzade, Nasrudin Abd Rahim, and Wooi Ping Hew.2012.
“Design, Analysis, and Prototyping of an Axial-Flux Permanent Magnet Motor Based on
Genetic Algorithm and Finite-Element Analysis.” IEEE Transactions on Magnetics, vol.49,
no.4,pp. 1479-1492.
14. Reza Ilka, Ali Roustaei Tilaki, Hossein Asgharpour-Alamdari and Reza Baghipour. 2014.“
Design Optimization of Permanent Magnet-Brushless DC Motor using Elitist Genetic
Algorithm with Minimum loss and Maximum Power Density. “International Journal of
Mechatronics Electrical and Computer Technology, Vol.4, No.10, pp 1169-1185.
15. Ajay Nair and K.R. Rajagopal.2010. “Generic Model of an electrical vehicle for dynamic
Simulation and performance prediction.” International Conference on Electrical Machines
and Systems, pp 753-757.
16. D.C Hanselman. 1994. Brushless Permanent Magnet Motor Design, New York: McGraw-
Hill.
17. Jackek F Gieras, Rong-Jie Wang and Maarten J. Kamper.2004. Axial Flux Permanent
Magnet Brushless Machines, Kluwer academic publishers.
18. J. R. Handershot and T. J. E. Miller. 1994. Design of Brushless Permanent Magnet Motors,
Oxford Univ. Press, UK.
19. M. Łukaniszyn, M. JagieŁa, and R. Wróbel. 2004. “Optimization of Permanent Magnet
Shape for Minimum Cogging Torque Using a Genetic Algorithm.” IEEE Transactions on
Magnetics, vol.40, No.2, pp. 1228-1231.