You are on page 1of 5

2019 8th International Conference on Modeling Simulation and Applied Optimization (ICMSAO)

A Design of an Axial Field Permanent Magnet


Generator for Small-Scale Water Turbines – Focus
on Stacking Factor Effects on Performance
Walid A. M. Ghoneim Mohamed Hussein Hamdy A. Ashour
Electrical Power and Control Dept. Electrical Power and Control Dept. Electrical Power and Control Dept.
Arab Academy of Science, Technology Arab Academy of Science, Technology Arab Academy of Science, Technology
and Maritime Transport and Maritime Transport and Maritime Transport
Alexandria, Egypt. Alexandria, Egypt. Alexandria, Egypt.
Walidghoneim1970@yahoo.co.uk mohamedabdohusien@gmail.com hashour@aast.edu

Abstract—This paper presents the modelling and design of losses is to decrease the conductivity of cores by using
a small scale Axial Field Permanent Magnet Synchronous ferromagnetic laminations covered by insulation such as
Generator driven by a low-head and low-speed water turbine. natural oxide surface, inorganic insulation, enamel, varnish
Firstly, the study suggests a water turbine that can be used to or chemically treated surface. To compensate for that in
extract small-scale hydraulic power, then means to calculate
analysis, the stacking factor was defined as the ratio of the
the produced mechanical power and performance. Secondly,
an overview of AFPM generators is presented. Afterwards, a steel lamination thickness to the total thickness [4]. Authors
generator is analytically designed through sizing equations, in [5] demonstrated analytical equations concerning AFPM
then its dimensions and performance is validated using FEM machines with overlapping and non-overlapping windings
software. This generator is redesigned after decreasing the and they validated them via FEM software. These equations
stacking factor to overcome disadvantages such as steel enabled them to hold comparisons between the windings
saturation and weight, which led to save steel consumption but types which concluded that the non-overlapping winding has
with lower performance. Then, FEM-based sensitivity analysis higher efficiency. The authors in [6] provided general sizing
is used to enhance the machine performance presenting a new equations used to obtain a primarily design for a 1.5 kW
hybrid machine comprising the advantages and disadvantages
torus machine. Consecutively, an embedded optimization
of cored and coreless ones.
method in the FEA software was defined and used to fulfil
Keywords— AFPM, FEM, TORUS-S, AFWT, CFWT objective functions such as efficiency, power to unit mass
ratio and cogging torque. The conclusion was an increase in
I. INTRODUCTION mass of 4.8% could lead to 27% increase in output power.
Hydropower generation systems take global attention as The authors in [7], established guideline tables for
it is environmentally friendly and renewable power source calculating the winding vital parameters, without assuming a
with highest power density. Most of hydropower is generated specific winding layout. A feasible range for the number of
by the large-scale hydroelectric plant that requires dam slots per pole per phase for machines with concentrated
constructions which lead to environmental drawbacks. As a windings was also proposed. Effects of slots on the winding
result, smaller hydropower systems attract much attention in space harmonics were discussed. The authors in [8] designed
recent years; mainly because of its simplicity and a swirl turbine for small hydropower plant and held a
environmental conservation [1]. Recently, axial flux comparison between different generators types selecting
machines have been more popular for applications which coreless AFPM as the best nominee. An analytical design
require light weight electric machines with high efficiency, was developed and an FEA model was built. Finally, a
torque and power density. The manufacturing of a compact prototype was manufactured to verify the design. The
multi-stage machine and the ability to apply a large number authors in [9] used sizing equations to design three
of poles make the AFPM generators highly suitable for use conventional radial PMSG with three different winding
for low-speed power generators [2]. In conventional radial layouts. Then, FEM software was used to simulate the
machines, air-gap flux density usually is radial; in AFPM different models and assessed their performance. They
machines, it is mainly axial. AFPM machines generally have concluded that the model with fractional slot concentrated
an axial stack length much shorter than that of conventional winding was the best selectee in terms of performance
radial machines of the same rating. So, selecting an AFPM characteristics. In this work, water turbines and AFPM
machine of compatible construction is considered in a variety machines performance are investigated. An AFPM generator,
of direct-drive applications, such as wheel-motor drives [3]. for a small hydro power plant is analytically studied and
Generally, electric machines whose yokes consist of validated by FEM software. Finally, a low stacking factor is
ferromagnetic materials are subjected to hysteresis and eddy proposed to eliminate saturation in teeth tips and a sensitivity
current losses if they are excited by time-varying magnetic analysis is made to define which design variable could
field. The common solution to decrease the eddy current effectively cope with low stacking factor drawbacks. That
illustrates the possibility of designing a new hybrid AFPM

978-1-5386-7684-4/19/$31.00 ©2019 IEEE


2019 8th International Conference on Modeling Simulation and Applied Optimization (ICMSAO)

machine with major advantageous of cored and coreless armature yokes, with surface-mounted or internal permanent-
machines. magnet, and as single-stage or multistage [11]. In order to get
the main dimensions of electric machines, power
II. WATER TURBINE OVERIEW formulations that relate machine sizing to output power at a
Water turbines turn the kinetic energy of water into constant frequency are necessary [6]. For an AFPM machine
mechanical energy. Then electrical generators convert the this relation could be expressed as:
mechanical energy from the turbines into electrical energy.
All hydro turbines can be classified into two types according 1 f
Pout = Kp Ki Ke Kl η A Bg 1 λ D o Le (3)
to the rotational axis orientation the axial flow water turbine 1 + K∅ m1 p

AFWT and cross flow water turbine CFWT. AFWT has an


axis of rotation parallel to the current direction. On the other where, Kφ is the electrical loading percentage on rotor and
hand, CFWT has a rotational axis perpendicular to the stator, and Kl is the aspect ratio coefficient connecting to a
current. Therefore, the turbine operates whatever the flow specific machine structure, m is the number of machine’s
direction. Many turbines which have been mainly used in phases, m1 is the number of each stator’s phases (if there are
wind power plants for years are adapted to be placed in a free multiple stators, each stator contains the same m1), Kp and Ki
stream such as Savonius, Gorlov and Darrieus turbines. are constants relating to the shapes of electrical power and
Savonius rotor will be considered in this paper because it is current waveform respectively, their values changes
simple in design and construction. It is made by cutting a according to the different shapes of waveforms. Ke is the
cylinder into two halves along the central plane and then EMF coefficient, which includes the winding factor and the
arranging them into ‘S’ shape as shown in Fig. 1 [10]. per unit air gap area occupied by salient poles (if they exist),
η is the machine efficiency, f is the machine’s frequency, p
is the pole pairs number, λ is the AFPM machine’s diameter
ratio Di /Do, A is the electric loading, Bg is the magnetic
loading, Do is the machine’s outer diameter and Le is the
machine's effective axial length. The main difference
between cored and coreless AFPM is the presence of
ferromagnetic materials in the stator cores. As a result, iron-
cored machines suffer from core losses while coreless
machines do not and the values of the synchronous
reactance of iron-cored machines are greater than those of
coreless machines [12]. If electric steel is used to
manufacture cored machines, it will be theoretically either
Fig. 1. Savonius rotor laminated spirally or axially as shown in fig. 2. The spiral
lamination is the best choice because it is more popular
The output power of a Savonius turbine (Ptur) and tip while the axial laminations of the stator introduce a
speed ratio (TSR) are expressed as follows [10]: difficulty in keeping the stator mechanically integral [13].

1
ptur = Cp H D U 3
∞ (1)
2

V
TSR = U (2)

where Cp is the power coefficient, ρwater is the water density,


H is the turbine rotor height, D is the turbine rotor diameter,
v is the turbine speed and U∞ is the water current speed. A
suggested Savonius turbine will be theoretically installed Fig. 2. Spiral and axial laminations of cored AFPM [13]
(a) Spiral (b) axial
horizontally in a water fall of a reservoir in a waste water
treatment station. The mass flow rate is 500 kg/s and the
There are promising alternatives to laminated steel such
turbine will be located at 1.5 m below the reservoir. Turbine as amorphous ferromagnetic alloys and soft magnetic
parameters such as dimensions and output power are powder composites but they have hardships concerning the
calculated neglecting turbine losses according to Bernoulli manufacturing processes [12]. The efficiency of any electric
and aforementioned equations and the results are shown in machines can be expressed by equation (4):
Table 1.

TABLE I. TURBINE PARAMETERS Pout


η= (4)
P [Watt] H [mm] D [mm] Pout + PCor + Pcu + Pm
7500 367 248

III. AFPM MACHINES OVERVIEW where Pcor is the core loss, Pcu is the Joule loss and Pm is the
mechanical loss. The Joule losses are calculated using the
Brushless AFPM machines could be classified as single-
sided or double-sided, slotted or slot-less, with or without
2019 8th International Conference on Modeling Simulation and Applied Optimization (ICMSAO)

value of electric resistance per phase Rph and the phase Relative permeability μr 1.05
current Irms as shown [12]: Maximum flux density in core Bcs, Bcr [T] 1.7
Dimensional results
Outer diameter Do [mm] 660
pcu = m Irms 2 R ph (5) Inner to outer diameter ratio λ 0.577
Permanent Magnet Thickness lm [mm] 4
Number of Turns Nc 90
Air gap length lg [mm] 1
ρ Lph (6)
Rph = Steel weight Mst [Kg] 123.31
ap aw sa Total weight M [Kg] 146.74
The output dimensions of the linear design are used to
where ρ is conductor resistivity, Nph is the number of series create a model using Maxwell RMxprt software. The phase
turns per phase, Lph is the conductor total length per phase, induced voltage and flux density magnitude spectrum are
ap is the number of Parallel paths, aw is the number of parallel illustrated in Fig. 3 and Fig. 4 respectively.
conductors and sa is the conductor cross section area. The
core losses are classified into two different types: eddy
current and hysteresis. These losses are difficult to be
measured experimentally and are analytically obtained from
[4].

= α
+ h (7)

where B is the teeth flux density value, h is the material


thickness in a perpendicular plane to the magnetic field Fig. 3. phase induced voltages
direction, Kh is the hysteresis factor, Ke is the eddy current
factor and α is a factor connecting to the material
characteristics and dimensions. The core losses of the
0.5mm electric steel sheet are clarified as follows [13]:

.
= 0.014492 + 0.00004219 (8)

Mechanical losses are divided into friction and


ventilation. Friction losses occur in the bearings and could
be described as follows [8]:

= 0.06 (9)
where Kfp is the bearing constant, mr is the total weight of
moving parts and n is the speed in rpm. The loss caused by
ventilation was neglected if the considered machine is
naturally cooled without forcing air [12].
IV. CASE STUDY VALIDATION Fig. 4. Flux density magnitude

A 100 rpm, three phase double-sided internal stator Fig. 4 shows that the flux density magnitude in the
(TORUS-S) cored AFPM generator was designed for a machine obtained from Maxwell RMxprt software does not
hydro-power plant using the sizing equations while exceed 1.4 T in almost all the surface of the machine except
considering the requirements of the application and design in small areas around teeth tips. For further validation, a
limitations as detailed in Table 2. comparison is held between the analytical results and the
ones obtained from the FEM software package. The
TABLE II. ANALYTICAL DESIGN REQUIREMENTS comparison is shown in Table 3.
AND RESULTS
TABLE III. ANALYTICAL AND SIMULATED RESULTS COMPARISON
Generator specifications and restrictions
Output power P [Watt] 7500 parameters Proposed Maxwell Deviation%
Mechanical speed nm [rpm] 100 Linear RMxprt Linear vs
Phase voltage V [V] 220 Design RMxprt
slots/ poles combination Ns/2P 28/24 Induced Phase Volt. [V] 220 205 7.317073
Shear stress σ [K.pa] 27.579 Current [A] 9.96 10.20 -2.35294
Slot fill factor Kcu 0.79 Pcu losses [Watt] 809.35 863.4 -6.26013
Current density J [A/mm2] 6 Phase inductance [mH] 102.17 98.03 4.223197
Stacking factor 0.8 Efficiency [%] 89.23 87.90 1.763367
yokes and PM materials constraints (Nd Fe B ) Teeth maximum flux
1.7 2 -15
Residual flux density Br [T] 1.1 density [T]
2019 8th International Conference on Modeling Simulation and Applied Optimization (ICMSAO)

The deviation in phase induced voltage between the linear complete hydro power generation set. A 7.50 kW AFPM
analysis and simulation occurred since the linear analysis generator was designed for an existing water treatment site
neglects the leakage flux, which travels between magnets using analytical analysis. FEM software was used to validate
without passing through the stator windings, also neglected the analytical design proving that linear analysis is congruent
the presence of steel reluctance. Another major deviation and efficient enough in the machine design, but FEM was
related to the teeth maximum flux density, which occurs in essential to account for saturation in teeth tips and to
any slotted structure as the magnet edges rotate and overlap decrease ferromagnetic materials consumption in the
teeth tips, hence causing saturation. Other FEM results machine.
validated the proposed linear design procedures and the
.
maximum deviation percentages did not exceed 7% for the
remaining parameters. To overcome the disadvantages
relevant to steel laminations weight of the proposed machine
and the saturation occurring in teeth tips, the machine is
redesigned after decreasing the value of the stacking factor
from 0.8 to 0.55. As a result, the steel weight sharply
plummets but at the cost of reducing the induced voltage.
V. SENISTIVITY ANALYSIS
To compensate the reduction in the induced voltage, an
optimization with sensitivity analysis was carried out. The
magnet length lm, the inner to outer diameter ratio λ and the
number of turns per coil Nc are selected as input design
variables, while the induced voltage is chosen as an objective
function. Sensitivity analysis tool – built in Maxwell FEM
(a)
software – is used to identify which of these design variables
is more influential in maximizing the objective function. The
modified design variables, which approximately achieved the
required induced voltage with a reduction in total weight
between 24.75 % and 26.26 % are shown in Table 4.

TABLE IV. SENSITIVITY ANALYSIS SUPERIOR RESULTS


Stacking lm Nc Vrms M
variation factor λ
[mm] [turns] [volt] [kg]
1 4 0.579 90 210.30 108.20
2 0.55 4 0.577 94 219.71 110.42
3 4.5 0.577 90 220.62 109.31
Fig. 5 shows the variations of the amplitude of phase
voltage versus the design variables. First of all, the inner to
(b)
outer diameter ratio achieved the maximum induced voltage
at (λ =0.579), which is approximately the same value to get
the maximum output power in the linear analysis. Hence,
varying it from 0.577 to 0.579 will not have a drastic
contribution on the induced voltage. Secondly, the linear
relation between the induced voltage and the number of
turns per coil shows that a 4.4% increase in the number of
turns per coil (four turns) increased the phase voltage
amplitude to about 311 V, which almost reached the phase
voltage value in Table 2 without a significant increase in
phase resistance and machine weight (only 0.173%).
Finally, the relation between the induced voltage and the
magnet thickness is proportional in a specific range of
magnet thickness (up to 5.5 mm), beyond which, the phase
voltage amplitude will decrease. Moreover, an increase in (c)
Fig. 5. Sensitivity analysis Results
the magnet thickness of the linear design – by only 0.50 mm Induced voltage amplitude versus (a) inner to outer ratio
– raises the phase voltage amplitude to 312 V, which exactly (b) Number of turns per phase
achieves the desired phase voltage in Table 2. The drawback (c) Magnet thickness
in the previous option is the inevitable small increase in the
overall machine weight (by 1%) and the cost of magnets. Therefore, a suggestion of a new hybrid machine
with lower stacking factor and without saturation was
VI. CONCULSION presented. Lastly, a sensitivity analysis of several design
In this paper, preliminary equations of Savonius water variables was carried out to discover which design variable
turbines and AFPM machines were presented to design a is more influential in compensating the reduction in the
2019 8th International Conference on Modeling Simulation and Applied Optimization (ICMSAO)

phase voltage concluding that increasing the number of


turns per coil by only 4.4% was the most effective and
economical solution in compensating the reduction in the
phase voltage.
VII. REFRENCES
[1] S. Iio, Y. Katayama, F. Uchiyama, E. Sato and T. Ikeda1
“Influence of Setting Condition on Characteristics of Savonius
Hydraulic Turbine with a Shield Plate,” Journal of Thermal Science,
vol. 20, no. 3, pp. 224-228, 2011.
[2] N. Taran and M. Ardebili, “A novel approach for efficiency and
power density optimization of an axial flux permanent magnet
generator through genetic algorithm and finite element analysis,”
IEEE 23rd international symposium on industrial electronics(ISIE),
pp. 709–714, 2014.
[3] A. Mahmoudi, H. Wooi Ping and N. Abdul Rahim, “A Comparison
between the TORUS and AFIRAxial-Flux Permanent-Magnet
MachineUsing Finite Element Analysis,” IEEE Electric Machines&
Drives(IEMDC), pp. 742–247, 2011.
[4] D.Hanselman, “Brushless permanent magnet motor design,”Magna
Physics Publishing, second Ed, 2006.
[5] J. H. Kim, W Choi, and B. Sarlioglu, “Closed-form Solution for
Winding Typesof Axial Flux Permanent Magnet Machines,”IEEE
Energy Conversion Congress and Exposition (ECCE), vol. 42, no. 5,
pp. 5877–5884, 2014.
[6] S. Arslan, E. Kurt, O. Akizu, J. M. Lopez-Guede, “Design
optimization study of a torus type axial flux machine,” Journal of
Energy Systems, vol. 2, Issue 2, pp. 43-56, 2018.
[7] S. E. Skaar and R. Nilssen, “Distribution, Coil-Span and Winding
Factors for PM Machines with Concentrated Windings,” In
Proceedings of the 17th Conference on Electrical Machines (ICEM),
China, Crete Island, Greece, 2006.
[8] R. Huzlik, C. Ondrusek, “Design of axial corelesspermanent
magnetgenerator for small hydro power plant,”IEEE XXII
International Conference on Electrical Machines (ICEM) , pp. 1539-
1544, 2016.
[9] W. A. M. Ghoneim, A. Hebala, and H. Ashour, “A Comparative
Study of Winding Configuration Effect on the Performance of Low
Speed PMSG Using FEM,” The Eighteenth International Middle
EastPower Systems Conference (MEPCON), pp. 348–352, Cairo,
2016.
[10] Mabrouki, Z. Driss, M. S. Abid“ Experimental Investigation of the
Height Effect of Water Savonius Rotors,” International Journal of
Mechanics and Applications, pp. 8-12, 2014
[11] A. Mahmoudi, N. A. Rahim and W. P. Hew,“Axial-flux permanent-
magnet machine modeling, design, simulation and analysis,”
Scientific Research and Essays, Vol. 6(12), pp. 2525-2549, 2011
[12] J. F. Gieras, R. J. Wang, and M. J. Kamper, “Axial Flux Permanent
Magnet Brushless Machines," Kluwer Academic Publisher, 2008.
[13] S. Kahourzade, A. Gandomkar, A. Mahmoudi, N Abd Rahim, W. P.
Hew, and M. N. Uddin, “Design optimization and analysis of AFPM
synchronous machine incorporating power density, thermal analysis,
and back-EMF THD,” Progress in Electromagnetic Research,PP.
327–367, 2013.

You might also like