Applied Energy 190 (2017) 339–353

Contents lists available at ScienceDirect

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

Optimal capacitance selection for a wind-driven self-excited reluctance

generator under varying wind speed and load conditions
T.R. Ayodele, A.S.O. Ogunjuyigbe ⇑, B.B. Adetokun
Power Energy Machine and Drive (PEMD) Research Group, Electrical and Electronic Engineering Department, Faculty of Technology, University of Ibadan, Nigeria

h i g h l i g h t s

We determined & selected suitable excitation capacitance value for WDSERG.

This ensured a constant output voltage under changing wind speed & connected load.

Mathematical model of WDSERG has been developed from the dynamic model of SERG.

We developed an algorithm that searches for the optimal excitation capacitance.

There is an optimal capacitance value that would produce a constant output voltage.

a r t i c l e i n f o a b s t r a c t

Article history: This paper presents a methodology to determine and select a suitable excitation capacitance value for a
Received 10 July 2016 wind-driven self-excited reluctance generator (WDSERG), which would produce a constant output volt-
Received in revised form 7 December 2016 age under changing wind speed and connected load. A steady state mathematical model of WDSERG is
Accepted 27 December 2016
developed from the dynamic model of self-excited reluctance generator (SERG) and phasor diagram.
Available online 6 January 2017
This model is used to develop an algorithm that searches for the optimum excitation capacitance which
produces a desired output voltage level for any given wind speed and load within the operating limits of
Keywords:
the WDSERG. Different scenarios of variable speed, load and power factor are considered and an optimal
Self-excited reluctance generator
Wind speed
capacitance value is determined and selected for each cases. Published experimental data was utilised to
Constant output voltage validate the developed model. The results show that there is a distinct capacitance value that would pro-
Excitation capacitance duce a constant output voltage under any given operating conditions of wind speed and terminal load.
Load The procedure presented can form a basis for the design of a variable excitation capacitor to maintain
a constant output voltage under varying wind speed and load, which will invariably offer an effective
and low-cost solution for output voltage control of WDSERG.
Ó 2016 Elsevier Ltd. All rights reserved.

1. Introduction
The rise in energy demand and the need to mitigate the effects
of climate change due to global warming has informed various
research interests to harness renewable energy sources in recent
years. Viable renewable energy sources such as hydro, wind, solar,
and geothermal resources are receiving concentrated attention in
various parts of the world particularly in developed countries [1].
Moreover, the inadequate and erratic supply of electricity in most
developing countries has raised concern with increasing efforts
geared toward the exploration of available renewable sources [2].
⇑ Corresponding author.
E-mail addresses: tr.ayodele@ui.edu.ng (T.R. Ayodele), aogunjuyigbe@yahoo.
com, a.ogunjuyigbe@ui.edu.ng (A.S.O. Ogunjuyigbe), jesutunde@gmail.com
(B.B. Adetokun).
Thus the perennial exploitation of fossil fuel for power generation
is gradually being phased out for a more environmentally friendly
renewable energy sources.
Wind Energy Conversion Systems (WECS) are becoming popular
as a viable renewable alternative [1,3]. This is because of wind
availability in several parts of the world [4,5]. Cheng and Zhu [6]
presented a review of the modern wind energy conversion tech-
nologies, with particular focus on the various wind power genera-
tors and the control schemes. The authors compared different
types of wind energy conversion systems and the most commonly
used maximum power point tracking (MPPT) control to achieve the
optimal generator speed which would produce maximum energy.
Permanent Magnet Synchronous Generator (PMSG) has
received considerable attention for wind energy conversion both
in direct-driven mode and with single or multi-stage gearboxes
[6–9]. Due to the presence of the permanent magnet in the rotor

http://dx.doi.org/10.1016/j.apenergy.2016.12.137
0306-2619/Ó 2016 Elsevier Ltd. All rights reserved.
340 T.R. Ayodele et al. / Applied Energy 190 (2017) 339–353
of PMSG, it offers a high energy conversion efficiency, reduced
rotor thermal stress and high torque density [10,11]. However,
the major limitations of PMSG includes demagnetisation of the
permanent magnet material at high temperature, high cost of per-
manent magnet, and cogging torque effects [12]. The authors of
reference [8] evaluated the electromagnetic losses in PMSG-
based WECS, putting into consideration the stochastic nature of
wind speed. Losses and overload capabilities of different genera-
tors with the same rated power but different rated voltage were
compared. It was shown that an optimum efficiency and a high
overload capability can be achieved by adjusting the rated voltage
level. In [7], the authors presented the analysis of speed and torque
control structures for variable speed, fixed pitch wind energy con-
version systems with a view to determining the most suitable
structure which would improve the generator’s reliability and
robustness. More information on grid-connected PMSG is pre-
sented in [13], where the authors discussed different power con-
verter topologies presently adopted in PMSG-based WECS,
various control strategies and the important grid integration
requirements.
Furthermore, the design and application of Doubly-Fed Induc-
tion Generator (DFIG) has also been found to be an attractive
choice for variable speed constant frequency WECS due to its sig-
nificant advantages [14,15]. DFIG has double output capability
through both the stator and the rotor. It requires reduced rating
of power converters and harmonic filters and does not require
external reactive power compensation due to the flexibility of its
stator excitation from the rotor circuit [16]. Also, its active and
reactive power can be independently controlled [17]. However,
the presence of slip rings and brushes on the DFIG rotor necessi-
tates the need for regular maintenance. This makes the application
of DFIG for remote and off-shore applications less reliable and
more costly [12,18]. Also, the difficulty associated with grid fault
ride-through control of DFIG in grid-connected applications poses
a major challenge to manufacturers of wind turbine since both
the stator and rotor of the DFIG are connected to the grid [19]. In
addition, the large size, enormous complexity and high cost associ-
ated with the design, assembly and control of DFIG makes it
unsuitable for small stand-alone applications where low-cost is
of primary consideration [12].
In stand-alone, off-grid WECS, the key considerations in the
choice of wind turbine generators include cost and simplicity of
control. In this regard, self-excited generators has been identified
as a low-cost alternative to PMSG and DFIG. Therefore, significant
research and development of self-excited generators has been on
the increase in recent years [1]. The most commonly investigated
self-excited generator is the Self-Excited Induction Generator
(SEIG) with squirrel-cage rotor [12]. SEIG has been applied in
stand-alone wind energy conversion because of its inherent advan-
tages over the conventional synchronous generators. Its simplicity,
as well as its rugged construction makes it relatively inexpensive
and is therefore popular in isolated wind energy conversion system
applications. Furthermore, it is self-protected against excessive
overload and short-circuit contingencies [20,21], it requires no
external direct current (DC) supply for excitation and voltage reg-
ulation and has better transient performances [22]. However, it
suffers from poor voltage and frequency regulation with changes
in load and prime-mover speed, with an attendant need for voltage
and frequency stabilizing circuits. This invariably increases the
installation cost [23–25].
In order to overcome the aforementioned challenges of SEIG,
Self-Excited Reluctance Generator (SERG) has been explored as
alternative choice for use in isolated wind energy conversion
system. SERG has additional advantages of constant output
frequency in spite of variation in load and excitation capacitances
as well as enhanced steady state performances over a wide range
of operation [23,26–30]. This makes SERG suitable for wind energy
conversion.
Although, an SERG will operate at constant output frequency
with load variations at its output terminals, however, in a similar
way to the SEIG, the generated voltage of the SERG falls with
increasing load and therefore requires an increase in excitation
capacitance within the stable operating limits of the machine to
maintain the output voltage within acceptable limits. Also, the
output voltage changes considerably with variation in
prime-mover speed. Therefore, a leading VAR is required to
continuously regulate the output voltage under changing speed
and/or load. This can be accomplished using variable excitation
capacitors.
Since it is desirable to obtain an efficient energy production
with constant output voltage from a generator for most practical
applications, there is a need to keep the output voltage constant
at a desired level under changing speed and/or load. This paper
therefore presents a method for the selection of an optimal excita-
tion capacitance for SERG, which can be implemented in a con-
troller in order to maintain an efficient energy production at the
desired output voltage within predetermined limits.
Different approaches have been adopted to regulate the output
of wind turbine-based systems. For instance, the authors of refer-
ence [31] proposed the use of a supercapacitor bank as a dynamic
regulation system in a wind-powered small-scale seawater reverse
osmosis desalination plant. The system was designed to continu-
ously adjust its energy consumption as the generated wind power
varies. For this particular application, the authors illustrate the
possibility of using supercapacitors to regulate the energy con-
sumption of the system in spite of the varying wind speed and
wind power.
In the case of SEIG, different methods for determining the exci-
tation capacitance requirement for a regulated output voltage of
SEIG have been presented in several works [32–36]. A general anal-
ysis to determine the steady state performance of an isolated SEIG
feeding a balanced R-L load was presented in [33]. The analysis
included the machine core losses and their variation with air gap
flux. Predictions of the minimum value of excitation under no load,
resistive and inductive load were made as well as the capacitance
value required to maintain a constant terminal voltage under vary-
ing load, power factor and speed. The authors also show a mathe-
matical basis for the design of a static exciter which utilises a Fixed
Capacitor Thyristor-Controlled Reactor (FCTCR) to provide the vari-
able excitation capacitance. The results depict that the capacitance
requirement increases significantly with decreasing speed, load
impedance and power factor for a lagging load. The authors of
Ref. [32] presented a controlled shunt capacitor SEIG, which is con-
tinuously adjusted in order to maintain a constant output voltage
over a relatively wide range of load and rotor speed. The method
employed Insulated Gate Bipolar Transistor (IGBT) switches con-
nected back to back across the fixed excitation capacitors to
achieve variable excitation. It was shown that the proposed
method can achieve a high level of voltage regulation for a wide
range of load.
However, there is paucity of literature on the use of SERG for
stand-alone WECS since most works on stand-alone WECS are
based on PMSG and SEIG [6,13]. Moreover, most of the literature
works on capacitance requirement of SERG are concerned with
the determination of minimum excitation capacitance required
for self-excitation and voltage build up [20,26,37]. Therefore, this
present work advances the existing literature by proposing a
methodology to determine and select the optimal capacitance
required to maintain a constant air-gap flux, and hence, a constant
output voltage for a Wind Driven Self Excited Reluctance Generator
(WDSERG) under conditions of varying wind speed and changing
load.
 T.R. Ayodele et al. / Applied Energy 190 (2017) 339–353 341

2. Wind-driven self-excited reluctance generator wind speed. A gear system is required to convert the tip speed of
the turbine blade to the required SERG speed. In terms of speed,
The schematic diagram of a Wind-Driven Self-Excited Reluc- the gear ratio, g is the ratio of the SERG speed xr (in rad/s) to
tance Generator (WDSERG) with variable excitation capacitance the turbine rotor blade tip speed, xt . With xt eliminated in Eq.
bank is shown in Fig. 1. It consists of wind turbine which serves (4) and further rearrangement made, the SERG rotor speed can
as the prime mover of the reluctance generator, the gear box which be related to the wind speed as follows:
ensures that the low speed wind turbine matches the high speed


120 gk
reluctance generator, the reluctance machine and the excitation Nr ¼ Vw ð6Þ
2pP R
capacitance.
where Nr is the SERG rotor speed in revolutions per minute (RPM), P
2.1. Wind turbine model is the number of poles on the rotor of the SERG. Fig. 2 depicts the
graph illustrating the variation of SERG rotor speed with the wind
The mechanical output power of a wind turbine (in Watts) is speed for different WDSERG design configurations denoted by the
given by [8,38]: values of R, g and P.
1
Pm ¼ C p ðk; bÞqAV 3w ð1Þ 3. Steady state mathematical modelling of reluctance generator
2
at different loading conditions
where C p is the performance coefficient of the turbine, k is the tip
speed ratio (TSR), b is the blade pitch angle, q is density of air in In order to determine the optimal excitation capacitance
kg/m3, A is the turbine swept area in m2, and V w is the wind speed required by a WDSERG to maintain a desired constant output volt-
in m/s. The performance coefficient (C p ) of the turbine is a measure age, steady state model of the SERG is developed, with the follow-
of the amount of mechanical power that can be obtained from the ing simplifying assumptions:
wind’s kinetic energy, and it connotes the aerodynamic efficiency
of the turbine. C p is a function of the tip-speed ratio (k) and blade i. Magnetic saturation is concentrated in the d-axis only.
pitch angle (b) given as [38]: ii. Core loss is negligible.

 C iii. Space harmonics in the air-gap flux and time harmonics in
C2 5
C p ðk; bÞ ¼ C 1  C 3 b  C 4 e k1 þ C 6 k ð2Þ the electromotive force and current waveforms are
k1
negligible.
where
The equations that describe the dynamic behaviour of a reluc-
ðb3 þ 1Þðk þ 0:08bÞ
k1 ¼ ð3Þ tance machine can be obtained from those of the wound field syn-
b3  0:028b  0:035k þ 1 chronous machine with the field winding terms omitted [39–41].
The coefficients C1 to C6 are C 1 ¼ 0:5176; C 2 ¼ 116; C 3 ¼ 0:4; The steady state equations of the SERG are obtained from the
C 4 ¼ 5; C 5 ¼ 21; C 6 ¼ 0:0068. The tip speed ratio is the ratio of dynamic model, with all the time rate of change of flux linkages
the turbine blade linear speed to the wind speed (or flux linkages per seconds) set equal to zero. Thus,
xt R V ds ¼ Rs Ids þ xr Kqs ¼ Rs Ids  xr Lq Iqs ð7Þ
k¼ ð4Þ
Vw
where xt is the rotor blade tip speed in rad/s, and R is the radius of
V qs ¼ Rs Iqs  xr Kds ¼ Rs Iqs þ xr Ld Ids ð8Þ
the turbine blade. The mechanical torque developed by the wind where V qs (V ds ) is the q-axis (d-axis) stator voltage, Iqs (Ids ) is the
turbine is given as: q-axis (d-axis) stator current, Rs is the stator resistance, Lq (Ld ) is
the q-axis (d-axis) stator inductance, Kqs (Kds ) is the q-axis
Pm 0:5C p ðk; bÞqAV 3w
Tm ¼ ¼ ð5Þ (d-axis) stator flux linkage, xr is the rotor speed. In terms of
xt xt
reactances and flux linkages per second, Eqs. (7) and (8) are
For a regulated turbine, the turbine’s output speed is controlled expressed in terms as:
to give a constant regulated tip speed such that the SERG operates
xr x
at a constant rotor speed. An unregulated turbine will however V ds ¼ Rs Ids þ W ¼ Rs Ids  r X q Iqs ð9Þ
produce a varying SERG rotor speed according to the variation in xb qs xb

Wind
3-phase
Load
Variable
Gear Box Capacitance
Wind
Bank
Reluctance
Machine

Wind Turbine
Fig. 1. A simple schematic diagram of a WDSERG with variable excitation capacitance bank.
342 T.R. Ayodele et al. / Applied Energy 190 (2017) 339–353

3500 2500
g = 14 R = 6m
3000 g = 16
2000 R = 7m
g = 18

SERG Speed (RPM)

SERG Speed (RPM)

2500 R = 8m
g = 20 R = 9m
2000 R = 6m 1500
g = 14
P=4
1500 P=4
1000
1000
500
500

0 0
0 5 10 15 0 5 10 15
Wind Speed (m/s) Wind Speed (m/s)

Fig. 2. Graph of SERG rotor speed against wind speed.

xr x
V qs ¼ Rs Iqs  W ¼ Rs Iqs þ r X d Ids ð10Þ
xb ds xb
Flux linkages per second and reactance are defined respectively
as

W ¼ xb K ð11Þ

X ¼ xb L ð12Þ

From Eqs. (9) and (10), the steady state equivalent circuit for the
SERG is developed with the excitation capacitor and RL load con-
nected. C, RL and LL represent the excitation capacitance, load resis-
tance and load inductance respectively. This is shown in Fig. 3 and
the corresponding phasor diagram is depicted in Fig. 4.
The phasor diagram is developed for SERG connected to a lag-
ging load. The load current, IL lags the voltage by an angle / called
the power angle while the current drawn by the capacitor desig-
nated as Ic leads the output voltage by 90 . The load angle d is spec-
ified as the angle between the q-axis and the phase voltage V.
Quantities associated with the load and excitation capacitance
are included to obtain a complete phasor diagram. Z d and Z q are
the equivalent d-axis and q-axis impedance respectively. The pha-
Fig. 4. Complete phasor diagram of RL-connected SERG.
sor diagram has been developed on the basis that the voltages and
currents vary sinusoidally in time and the rotating magnetomotive
force (MMF) wave produced by the stator currents vary sinu- The following relations are derived from the phasor diagram in
soidally in space [42]. Fig. 4:
V ds ¼ V sin d ¼ X c Ic sin d ð13Þ
Rs
V qs ¼ V cos d ¼ X c Ic cos d ð14Þ
I qs
RL I qL
I qC Ids ¼ IL sinðd þ /Þ þ Ic cos d ð15Þ
Vqs =VqC =VqL C
ωr
X I
ω b d ds Iqs ¼ IL cosðd þ /Þ þ Ic sin d ð16Þ
LL

Since V ¼ X c Ic ¼ Z L IL ð17Þ
(a) Rs Therefore, the ratio of the capacitive reactance to the load impe-
dance is given as:
IdL Ids
RL I dC X c IL
Vds = VdC = VdL ωr s¼ ¼ ð18Þ
C X I Z L Ic
ω b q qs
LL For SERG operating at rated (base) synchronous speed, such that
xr
xb ¼ 1 in Eqs. (9) and (10), the expression for the load angle d and
the d-axis reactance X d are obtained from Eqs. (9) and (10) (with
(b) xr
xb¼ 1), and (13)–(16). After some mathematical manipulations
Fig. 3. Steady state equivalent circuits of SERG (a) q-axis (b) d-axis. and rearrangements, d and X d are obtained respectively as:
 T.R. Ayodele et al. / Applied Energy 190 (2017) 339–353 343


Rs þ sðX q cos /  Rs sin /Þ Eqs. (19)–(31) describe an SERG operating at rated synchronous
d ¼ arctan ð19Þ
X c  X q þ sðRs cos / þ X q sin /Þ speed. To account for the effect of variable speed operation, we
set RL ¼ RaL , Rs ¼ Ras , and X c ¼ Xa2c ; where a is the per unit speed. The
Xd ¼
X c þ Rs ½tan d þ sðcos /  sin / tan dÞ
ð20Þ value of a applies to both the wind speed v w and the SERG speed
1  sðsin / þ cos / tan dÞ xr , that is, a ¼ vv w ¼ xxr , where v wb and xb are the base wind and
wb b

An extended derivation of Eqs. (19) and (20) is shown in the SERG speeds respectively. Thus, (30) and (31) becomes
appendix. A representative expression for X d is obtained by substi- !
aðR2L þ a2 X 2L ÞRs þ aX c ðX q RL  Rs X L Þ
tuting the value of d in (19) into (20). Thus, we have: d ¼ arctan ð32Þ
ðR2L þ a2 X 2L ÞðX c  a2 X q Þ þ X c ðRs RL þ a2 X q X L Þ
R2s 2 s þ ½2Rs X c cos / þ ðX c X q  2R2s Þ sin / ½R2s
s þ þ ðX c  X q ÞX c 
Xd ¼ h i
X q s2  ½ðX c  2X q Þ sin /s þ ½X c  X q  R2s þ 2RL Rs þ a2 X q X L þ Z 0 X 2c  ½a2 ð2R2s X L þ X q Z 0 ÞX c þ a2 R2s Z 0
ð21Þ Xd ¼
½a2 ðX L þ X q ÞX 2c þ ½a2 ð2a2 X L X q þ Z 0 ÞX c  a4 Z 0 X q
d and X d expressions for load conditions other than the RL-case ð33Þ
specified in Fig. 4 are obtained from (19) and (21) respectively as
follows: where Z 0 ¼ R2L þ a2 X 2L
i. For no load (open circuit) condition, load current IL ¼ 0;
therefore s ¼ 0. Hence, 4. Determination of optimal excitation capacitance


Rs
d ¼ arctan ð22Þ The steady state equations evolved in the previous section are
Xc  Xq
utilised to compute the steady state performance characteristics
of the SERG for different operating conditions. This procedure is
R2s
Xd ¼ Xc þ ð23Þ extended to develop the algorithm which determines the optimal
Xc  Xq
excitation capacitance that would produce and maintain the termi-
ii. For a purely resistive load, the load voltage and the load cur- nal output voltage at the desired level and within predetermined
rent are in-phase, thus / ¼ 0, cos / ¼ 1; and sin / ¼ 0. limits at different operating wind speeds and connected load.
Hence, For specified values of load impedance, wind speed (and hence

 SERG rotor speed), and power factor, the corresponding values of
Rs þ sX q
d ¼ arctan ð24Þ load angle d and d-axis reactance X d are obtained from Eqs. (32)
X c  X q þ sRs and (33) respectively. The corresponding values of d-axis current
h i Ids for each value of X d are obtained from the magnetization curve
R2s s2 þ ½2Rs X c s þ R2s þ ðX c  X q ÞX c of the experimental machine, represented mathematically by a 4th
Xd ¼ ð25Þ order polynomial approximation as:
X q s2 þ ½X c  X q 
iii. For a purely inductive load, the load current lags the load Ids ¼ 1:7964  108 X 4d þ 7:2876  106 X 3d  7:8623
voltage by / ¼ 90 , thus cos / ¼ 0, and sin / ¼ 1. Hence,  104 X 2d  4:1151  102 X d þ 9:3067 ð34Þ


Rs ð1  sÞ Substituting (18) into (15), the capacitor current, Ic can be
d ¼ arctan ð26Þ obtained as:
X c  X q ð1  sÞ
Ids
Ic ¼ ð35Þ
R2 s2 þ ðX c X q  2R2s Þs þ ½R2s þ ðX c  X q ÞX c  cos d  s sinðd þ /Þ
Xd ¼ s ð27Þ
X q s2  ðX c  2X q Þs þ ½X c  X q 
Putting into consideration the variation in wind speed, values of
iv. For a capacitive load, the load current leads the load voltage capacitor current, Ic , terminal phase voltage, V load current, IL and
by /, thus, / is set equal to /. Hence, stator current, Is are obtained as follows:
pffiffiffiffiffi

 aIds Z 0
Rs þ sðX q cos / þ Rs sin /Þ Ic ¼ pffiffiffiffiffi ð36Þ
d ¼ arctan ð28Þ a Z 0 cos d  X c sinðd þ /Þ
X c  X q þ sðRs cos /  X q sin /Þ
X c Ic
R2s s2 þ ½2Rs X c cos / þ ð2R2s  X c X q Þ sin /s þ ½R2s þ ðX c  X q ÞX c  V¼ ð37Þ
Xd ¼ a
X q s2  ½ð2X q  X c Þ sin /s þ ½X c  X q 
ð29Þ V
IL ¼ pffiffiffiffiffi ð38Þ
Z0
Given that s ¼ XZLc , cos / ¼ RZLL , and sin / ¼ XZLL , and since qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
Z 2L
¼ þR2L X 2L ,
a new expression is obtained for d and X d in terms Is ¼ I2ds þ I2qs ð39Þ
of the load resistance and reactance as:
! 5. Algorithm development
ðR2L þ X 2L ÞRs þ X c ðX q RL  Rs X L Þ
d ¼ arctan ð30Þ
ðR2L þ X 2L ÞðX c  X q Þ þ X c ðRs RL þ X q X L Þ The algorithm developed in this paper utilises the mathematical
equations developed in the last section. The computational proce-
2
½ðRs þ RL Þ þ X q X L þ X 2L X 2c  ½2R2s X L þ X q ðR2L þ X 2L ÞX c þ R2s ðR2L þ X 2L Þ dure is carried out for different values of excitation capacitance
Xd ¼
½X L þ X q X 2c þ ð2X L X q þ R2L þ X 2L ÞX c  X q ðR2L þ X 2L Þ from a specified starting value, 5 lF to a suitable final value such
as 160 lF. The capacitance values which satisfies the terminal
ð31Þ
phase voltage criterion of 220  1 V for each wind speed and load
344 T.R. Ayodele et al. / Applied Energy 190 (2017) 339–353

⇑ (continued)

Algorithm
L, and initialise load impedance value counter, k ¼ 1
Step 2: Generate total number of wind speed values, M,
and initialise wind speed value counter, i ¼ 1
Step 3: Generate total number of excitation capacitance
values, N, and initialise excitation capacitance value
counter, j ¼ 1
Step 4: Initialise counter for number of optimal excitation
capacitances for load impedance, Z Lk at wind speed V wi :
qk;i ¼ 0
Step 5: Compute load angle, dk;i;j and d-axis reactance
X dk;i;j capacitor current, Ick;i;j ; terminal phase voltage, V k;i;j ,
load current, ILk;i;j and stator current, Isk;i;j for the
corresponding load impedance, Z Lk , wind speed, V wi and
excitation capacitance, C j
Step 6: With excitation capacitance, C j , check if the
output voltage meets the voltage criterion of V rated  e
(where e is the desired tolerance limit for the output
voltage, which is taken to be 1V in this paper)
If V k;i;j ¼ V rated  e, then
qk;i ¼ qk;i þ 1
Coqk;i ¼ C j (where Co is initialised as an empty array
used as a temporary storage location)
Else
Go to Step 7
End
Step 7: Utilise the next excitation capacitance value:
C j ¼ C jþ1
If C j 6 C N , then
Go to Step 5
Else
Go to Step 8
End
Step 8: If there are one or more excitation capacitance
values that meet the voltage criterion for load impedance,
Z Lk and wind speed V wi , select the minimum value as the
optimal excitation capacitance, C optk;i
If qk;i P 1, then
C optk;i ¼ minðCoÞ
Fig. 5. A representative flowchart to determine optimal excitation capacitance. Else
For the specified voltage criterion, no optimal
excitation capacitance exists for Z Lk at V wi
impedance constitute the set of required optimal capacitance val- End
ues. The algorithm is built in such a way that if there are more than Step 9: Find the optimal excitation capacitance for load
one capacitance values that give the specified voltage criterion for impedance Z Lk at the next wind speed value
any wind speed value within the specified range, the minimum V wi ¼ V wiþ1
value is selected as the optimal capacitance (Copt). If V wi 6 V wM , then
The above procedure is repeated for other load impedance val- Go to Step 3
ues to obtain the characteristic relationships between the optimal Else
capacitance, wind speed, and the load. Fig. 5 depicts a representa- Go to Step 10
tive flow chart for the algorithm that determines Copt for each End
speed value at specified load impedances. The wind speed is varied Step 10: Find the optimal excitation capacitance for the
in steps of 0.1 m/s. next load impedance Z Lk
Z Lk ¼ Z Lkþ1
Algorithm If Z Lk 6 Z LL , then
Input: Machine parameters, array of load impedances, power Go to Step 2
factor, desired rated voltage, Vrated, array of wind speed Else
values, and array of excitation capacitance values. Stop
Output: Optimal excitation capacitance, Copt End
Start
Step 1: Generate total number of load impedance values,
 T.R. Ayodele et al. / Applied Energy 190 (2017) 339–353 345

Fig. 5 depicts a representative flow chart for the algorithm that 7

determines Copt for each speed value at a specified load impedance. Id versus Xd
6 4th degree
The wind speed is varied in steps of 0.1 m/s.

d-axis Current, Id (A)

Polynomial
This algorithm can be built into a controller for a variable three- 5 Approximation
phase excitation capacitor. The controller takes in wind speed and
load impedance as the inputs. Then the controller takes decision 4
according to the developed algorithm, which varies the variable
3
excitation capacitor to the required capacitance value. The excita-
tion capacitance variation can be accomplished by utilising a Fixed 2
Capacitor Thyristor-Controlled Reactor (FCTCR) static exciter as
1
proposed in Malik and Haque [33]. Alternatively, the variable exci-
tation can be achieved by employing Insulated Gate Bipolar Tran- 0
sistor (IGBT) switches connected back to back across fixed 50 70 90 110 130 150 170 190 200

excitation capacitors [32]. d-axis Reactance, Xd(Ohms)

Fig. 6. Magnetization curve of SERG: d-axis current against d-axis reactance.

6. Results and discussion

40
This section presents the results obtained from the mathemat- Simulated
ical models developed in the previous section under constant and 35
Experimental
varying conditions of wind speed, load impedances and power fac-

Minimum Excitation
30

Capacitance (uF)
tor. Important findings are discussed.
25
The analysis is carried out for an experimental machine, 4-pole,
1.5 kW, 50 Hz, 415 V reluctance machine with the parameters 20
summarised in Table 1. The magnetising characteristics of the 15
machine is given by the Id-Xd curve shown in Fig. 6.
10
A 4th order curve fitting polynomial has been used to approxi-
mate the relationship. This is used to account for the saturation 5
effect in the analysis. 0
Fig. 7 shows the experimental and simulation results of the 1000 1500 2000 2500 3000 3500
minimum excitation capacitance as it relates with the SERG speed SERG Speed (RPM)
(influenced by the wind speed) for a minimum output voltage gen-
Fig. 7. Variation of minimum excitation capacitance with SERG speed at no load
eration. The experimental results are as presented in [23]. These
(experiment and simulation).
results are utilised to verify the simulation results of this paper.
The figure shows a good agreement between the experimental
and the simulation results, thereby, providing a basic experimental illustrated in Fig. 8d start to drop as the loading increases beyond
validation of the mathematical model and the analysis presented 0.6 A.
in this paper to determine the optimal excitation capacitance for However, a constant output voltage is desired for most practical
a desired output voltage of the WDSERG. applications under changing load. Improved energy production
capability and efficiency are also desirable features needed in the
6.1. Comparison of steady state performances under fixed and variable harnessing of wind energy resource. Fig. 9 illustrates a way to
excitation capacitances achieve these by employing a variable excitation capacitance. The
variable capacitance bank adjusts itself according to the variation
6.1.1. Load characteristics in load to provide the optimal excitation required to maintain
Figs. 8 and 9 illustrate the load characteristics under fixed and the output voltage at the desired level, which for this case is
variable excitation capacitances respectively. This is for a unity 220 V (Fig. 9a). The variable excitation capacitance is shown in
power factor and a nominal speed of 1500 RPM which for this par- Fig. 9b. It is observed that for any given load within the maximum
ticular WDSERG design configuration is produced by a wind speed load limit, the output voltage remains constant with a variation not
of 10 m/s. This constant nominal speed may also be achieved under exceeding ±1 V, according to the voltage criterion in the developed
varying wind speed by a regulated turbine. It is observed from algorithm. In addition, under this variable excitation, the output
Fig. 8a that the terminal phase voltage falls considerably with power production and efficiency is enhanced as illustrated in
increase in load under a constant excitation. A steady fall of about Fig. 9c and d respectively. While under constant excitation, the
20% is observed between a load drawing a current of 0.3 A and energy production in terms of output current is limited to about
another drawing 0.6 A under the same constant excitation of 0.6 A (Fig. 8c), however, with variable excitation, output power is
30 lF (Fig. 8b). Under this condition of excitation, the generated delivered beyond 1.0 A with sustained efficiency of more than
voltage tends to collapse as the loading increases. Also, both the 70% as shown in Figs. 9c and d.
output power characteristics shown in Fig. 8c and the efficiency
6.1.2. Speed characteristics
The speed characteristics of the WDSERG feeding a unity power
Table 1 factor load of 300 X under fixed and variable excitation capaci-
SERG machine parameters [23]. tances are shown in Figs. 10 and 11 respectively. Fig. 10a shows
Machine rs Unsaturated Unsaturated Unsaturated that with a constant excitation shown in Fig. 10b, there is a signif-
rating Xq Xd saliency icant variation in the generated phase voltage with changes in the
ratio SERG speed. The variation of the SERG speed is produced by
Kilowatt Volts rpm Ohms Ohms Ohms
changes in the wind speed for an unregulated turbine. Under this
1.5 415 1500 10.12 49.11 181.8 3.70
condition of fixed excitation and at a speed of 1500 RPM, the gen-
346 T.R. Ayodele et al. / Applied Energy 190 (2017) 339–353

300 31

Terminal Voltage (V)

Capacitance(uF)
250 30.5

Excitation
200 30

150 29.5

100 29
0 0.2 0.4 0.6 0.8 0 0.2 0.4 0.6 0.8
Load Current (A) Load Current (A)
(a) (b)
400 80
Output Power (W)

Efficiency (%)
300 60

200 40

100 20

0 0
0 0.2 0.4 0.6 0.8 0 0.2 0.4 0.6 0.8
Load Current (A) Load Current (A)
(c) (d)

Fig. 8. Load characteristics under fixed excitation capacitance for a nominal SERG speed of 1500 RPM.

250 45
Optimal Excitation
Phase Voltage (V)

Capacitance (uF)

240
40
230
35
220
30
210

200 25
0 0.2 0.4 0.6 0.8 1 0 0.2 0.4 0.6 0.8 1
Load Current (A) Load Current (A)
(a) (b)
800 80
Output Power (W)

Efficiency (%)

600 70

400 60

200 50

0 40
0 0.2 0.4 0.6 0.8 1 0 0.2 0.4 0.6 0.8 1
Load Current (A) Load Current (A)
(c) (d)

Fig. 9. Load characteristics under variable excitation capacitance for a nominal SERG speed of 1500 RPM.

erated phase voltage is about 188 V whereas the voltage rises to
about 359 V at 1800 RPM. This shows a marked change in gener-
ated voltage under varying speed. Fig. 10c and d shows the varia-
tion of the output power and efficiency with the SERG speed. It
can be observed from this Fig. 10d that the efficiency falls as the
SERG speed increases under the condition of constant excitation.
Since it is desirable to maintain a constant output voltage, the
WDSERG can be provided with a variable excitation shown in
Fig. 11b, such that the output voltage, current and hence power
remains constant in spite of the speed variation. This is illustrated
in Fig. 11a and d for the same WDSERG feeding the same constant
load of 300 X. This figure reveals that for any variation in the SERG
speed within the speed limits, an optimal capacitance can be
determined and selected which would produce the required
voltage level as depicted in Fig. 11a. Moreover, unlike in the case
of constant excitation in Fig. 10, under this condition of variable
excitation, the efficiency increases as the SERG speed increases.
This is shown in Fig. 11d.
6.2. Optimal capacitance requirement under constant loads, varying
speed
Optimal capacitance requirement for a constant output phase
voltage of 220 V under a constant unity power factor load of
300 X has been illustrated and discussed above. For an operating
wind speed of 10 m/s (which for a WDSERG configuration of gear
ratio = 14; turbine radius = 6 m; and number of SERG rotor
poles = 4, the SERG speed is 1500 RPM), the capacitance value that
would produce 220 V output is 32:5 lF.
 T.R. Ayodele et al. / Applied Energy 190 (2017) 339–353 347

600 31

Phase Voltage (V)

Capacitance(uF)
30.5

Excitation
400
30
200
29.5

0 29
1200 1400 1600 1800 2000 1200 1400 1600 1800 2000
SERG Speed (RPM) SERG Speed (RPM)
(a) (b)
2000 85
Output Power (W)

Efficiency (%)
1500 80

1000 75

500 70

0 65
1200 1400 1600 1800 2000 1200 1400 1600 1800 2000
SERG Speed (RPM) SERG Speed (RPM)
(c) (d)

Fig. 10. Speed characteristics under fixed excitation capacitance for a unity power factor load of 300 X.

250 60
Phase Voltage (V)

Optimal Excitation
Capacitance (uF)

240
40
230
220 20
210
200 0
1200 1400 1600 1800 2000 1200 1400 1600 1800 2000
SERG Speed (RPM) SERG Speed (RPM)
(a) (b)

600 90
Output Power (W)

Efficiency (%)

500
80
400
70
300

200 60
1200 1400 1600 1800 2000 1200 1400 1600 1800 2000
SERG Speed (RPM) SERG Speed (RPM)
(c) (d)

Fig. 11. Speed characteristics under variable excitation capacitance for a unity power factor load of 300 X.

However, Fig. 12 illustrates the capacitance requirement for dif-
ferent load impedance value. It can be observed that for load impe-
dance of 200 X and 220 X, the capacitance characteristics changes
at a particular wind speed value for each load cases. A new capac-
itance characteristic curve is obtained after 12.5 m/s for a load of
250 X and after 9.5 m/s for a load of 200 X. This wind speed value
represents the critical value for a given WDSERG configuration and
load condition. This anomalous behaviour is explained in what
follows.
It is a well-established fact that the magnetization (saturation)
characteristics of a self-excited generator plays a significant role in
the voltage build up of the machine. The saturation curve of the
experimental machine utilised for simulation is shown in Fig. 6.
The minimum unsaturated value of the d-axis reactance (the value
that corresponds to a zero d-axis current) is 181.8 X. Also, from the
graph, the minimum saturated d-axis reactance within the satura-
tion region is about 50 X.
If we take a load of 220 X as an example, the graphs of Fig. 13a
to h are obtained. It is observed from this figures that at a wind
speed of 12.5 m/s, the operating characteristics changes remark-
ably with the establishment of new operating points. Variation of
d-axis reactance with wind speed is illustrated in Fig. 13c. It is
observed that the d-axis reactance values are within the saturation
region (between 50 and 181.8 X) until a wind speed of 12.5 m/s.
Beyond this wind speed value, the d-axis reactance falls sharply
to about 14 X. The corresponding d-axis current is about 9.75 A
as depicted in Fig. 13h. Thus, a new saturation characteristics is
formed, which is an extrapolation of the previous one. This is
shown in Fig. 13b. As observed from Fig. 13d, the load current
remains almost constant with only a very slight increase.
In order to maintain a constant output voltage, the q-axis cur-
rent falls correspondingly to a negative value of about 3.0 A as
observed from Fig. 13g. The stator current, being the algebraic
sum of the d-axis and q-axis currents, also rises accordingly to a
348 T.R. Ayodele et al. / Applied Energy 190 (2017) 339–353

160 value of 10.2 A as shown in Fig. 13f. Most of this current is

absorbed by the variable excitation capacitor since there is a corre-
Optimal Excitation Capacitance (uF)

140 sponding decrease in the capacitive reactance of the capacitor after

the sharp increase in excitation capacitance shown in Fig. 13a. The
120
current drawn by the capacitor at this point is 10.15 A as observed
100 from Fig. 13e. The load current is the square root of the difference
in the square of the stator and capacitor currents, which is about
200 Ohms
80 1.0 A. The practical implication of this is that at this new operating
220 Ohms
conditions, the capacitance bank supplies a large amount of lead-
60 250 Ohms
ing VAR and the generator’s output power exceeds the nominal rat-
300 Ohms
ing of the machine.
40 350 Ohms
Furthermore, the sudden increase in the stator current, which
20 has now exceeded the current rating of the stator windings, leads
to a significant increase in the stator copper loss and the heating up
0 of the stator windings. Therefore, this condition is invariably
8 9 10 11 12 13 14 15
unsafe for the system and should be avoided. A way to avoid this
Wind Speed (m/s)
condition is to reduce the load at such critical speed. This load
Fig. 12. Capacitance requirement under varying speed for different load reduction is illustrated in Fig. 14a to h. The speed is monitored
impedance. such that at the critical speed of 12.5 m/s (1905 RPM), the load is
automatically reduced (which in this case is halved). Therefore, a

150 10
Optimal Excitation

d-axis Current (A)

Capacitance (uF)

100
5
50

0 0
8 10 12 14 16 0 50 100 150
Wind Speed (m/s) d-axis Reactance (Ohms)
(a) (b)

150 1.05
d-axis Reactance

Load Current (A)

100
(Ohms)

1
50

0 0.95
8 10 12 14 16 8 10 12 14 16
Wind Speed (m/s) Wind Speed (m/s)
(c) (d)

15 15
Capacitor Current (A)

Stator Current (A)

10 10

5 5

0 0
8 10 12 14 8 10 12 14
Wind Speed (m/s) Wind Speed (m/s)
(e) (f)

4 10
q-axis Current (A)

d-axis Current (A)

0 5

-2

-4 0
8 10 12 14 8 10 12 14
Wind Speed (m/s) Wind Speed (m/s)
(g) (h)

Fig. 13. Operating characteristics under varying wind speed at constant load of 220 X.
 T.R. Ayodele et al. / Applied Energy 190 (2017) 339–353 349

100 3

Optimal Excitation
Capacitance (uF)

d-axis Current (A)

2
50
1

0 0
8 10 12 14 16 110 120 130 140 150
Wind Speed (m/s) d-axis Reactance (Ohms)
(a) (b)

160

Load Current (A)

d-axis Reactance

1
140
(Ohms)

0.8
120
0.6

100 0.4
8 10 12 14 16 8 10 12 14 16
Wind Speed (m/s) Wind Speed (m/s)
(c) (d)

4 4

Stator Current (A)

Current (A)
Capacitor

3 3

2 2

1 1
8 10 12 14 8 10 12 14
Wind Speed (m/s) Wind Speed (m/s)
(e) (f)
4 2.5
d-axis Current (A)
q-axis Current (A)

3 2

2 1.5

1 1

0 0.5
8 10 12 14 8 10 12 14
Wind Speed (m/s) Wind Speed (m/s)
(g) (h)

Fig. 14. Operating characteristics with load reduced to 440 X at 12.5 m/s.

new operating characteristics is established for the new load of 60

Optimal Excitation Capacitance (uF)

440 X, which is now within the SERG normal stable operating lim- pf = 0.8(lag)
its as depicted in the graphs of Fig. 14a to h. 50 pf = 0.9(lag)
The effect of power factor on Capacitance-Speed characteristics
pf = unity
at a load impedance of 300 X is illustrated in Fig. 15. This shows 40 pf = 0.9(lead)
that an inductive load generally requires a higher capacitance pf = 0.8(lead)
value to maintain the output voltage at the desired value. Con- 30
versely, the capacitance requirement is significantly reduced for a
leading load of the same rating, since the load, being connected 20
in parallel to the capacitance bank, increases the effective excita-
tion capacitance of the system. 10
Fig. 16 shows the effect of the variation in power factor on the
capacitance-speed characteristics separately for lagging loads and 0
8 9 10 11 12 13
leading loads. In Fig. 16a, for an inductive load, it can be observed
that the more inductive the load (signified by reduced power fac- Wind Speed (m/s)
tor), the more the capacitance required to maintain a constant out- Fig. 15. Capacitance-speed characteristics at different power factor.
put voltage of 220 V. It is however seen that variation in power
factor of an inductive load has a minimal effect on the optimal
capacitance-wind speed characteristics. As seen from the figure, 56:4 lF and 53:4 lF respectively. Whereas, at a wind speed of
at a wind speed of 8.5 m/s, the capacitance requirements for a 11 m/s, the capacitance requirement for all the specified power
300 X inductive load with power factor of 0.6 and 0.9 are factor is approximately 30 lF. Thus, the power factor of the lagging
350 T.R. Ayodele et al. / Applied Energy 190 (2017) 339–353

70 60
pf = 0.6(lag) pf = 0.6(lead)
pf = 0.7(lag) pf = 0.7(lead)
60 50

Optimal Excitation Capacitance (uF)

Optimal Excitation Capacitance (uF)

pf = 0.8(lag) pf = 0.8(lead)
pf = 0.9(lag) pf = 0.9(lead)
50 40

40 30

30 20

20 10

10 0
8 9 10 11 12 13 8 9 10 11 12 13
Wind Speed (m/s) Wind Speed (m/s)
(a) (b)
Fig. 16. Capacitance-speed characteristics at different power factor for (a) lagging load and (b) leading load.

load has a minimal effect on the capacitance requirement for a con-

80
stant output voltage of 220 V.
Optimal Excitation Capacitance (uF)

The case of a leading load is also considered. This is shown in 8m/s

70
Fig. 16b. It is seen that for a capacitive load, the power factor of 9m/s
the load has a more significant effect on the capacitance require- 10m/s
ment. For instance, at a wind speed of 10 m/s, the capacitance 60 11m/s
requirements for capacitive load of 300 X with power factor of
0.6 and 0.9 are 20:3 lF and 26:8 lF respectively. In particular, as 50
the operating speed increases, it follows that the more capacitive
the load as signified by lower power factor (leading), the lower 40
the excitation capacitance requirement to maintain the output
voltage at 220 V.
30
On the overall, comparing the inductive load with the capacitive
load in Fig. 16a and b, it can be observed that a lower optimal exci-
tation capacitance is required for a capacitive load than for a lag- 20
0 0.2 0.4 0.6 0.8 1 1.2
ging load of the same rating.
Load Current (A)

Fig. 17. Optimal capacitance requirement versus load current for different wind
6.3. Optimal capacitance requirement under steady Speed, changing speed.
load

50
Optimal Excitation Capacitance (uF)

Fig. 17 shows the variation of the optimal excitation

capacitance with the load at unity power factor under different pf = 0.8 (lag)
steady wind speed conditions. It can be observed from the figure 45 pf = 0.9 (lag)
pf = unity
that the capacitance requirement increases as the loading
40 pf = 0.9 (lead)
increases under any operating speed. Moreover, higher capaci-
pf = 0.8 (lead)
tance values are required for lower operating speeds. For a
steady wind speed of 9 m/s, the range of capacitance require- 35
ment from no load to full load is about 22:2 lF (34:4 lF to
56:6 lF) while for a steady wind speed of 10 m/s (which for 30
the specified WDSERG configuration produces an SERG speed of
1500 RPM), the range of capacitance requirement from no load 25
to full load is about 15 lF (26 lF to 41 lF). Therefore, this infor-
mation can be helpful in the selection of a suitable capacitance 20
0.2 0.4 0.6 0.8 1 1.2
value and the rating of the switching devices required to achieve
Load Current (A)
variable capacitance.
Similar analysis is carried out putting into consideration varia- Fig. 18. Optimal capacitance requirement versus load current for different power
tion in the power factor of the load. The effects of the power factor factor.
on the capacitance-load characteristics is illustrated in Fig. 18. It
can be observed from the figure that capacitive loads have lower A 0.9 power factor capacitive load requires approximately 25 lF at
optimal capacitance requirement, with an additional advantage no load and 32 lF while a capacitive load with a power factor of 0.8
of reduced range of excitation capacitance from no load to full load. requires about 25 lF  1 lF for all loading conditions. However, as
 T.R. Ayodele et al. / Applied Energy 190 (2017) 339–353 351

seen in the figure, lagging loads require higher optimal capaci- ing increases, the capacitance required for energy production at
tances and a higher capacitance range for energy production with constant output phase voltage of 220 V increases. This shows that
constant output voltage. in spite of variations in wind speed and load conditions, a constant
output voltage can be obtained by varying the excitation capaci-
6.4. Optimal capacitance requirement under varying speed, changing tance accordingly.
load
6.5. Evolution of the output frequency under constant and varying load
Fig. 19 illustrates a three-dimensional plot of optimal excitation conditions
capacitance for changing wind speed and varying load. It can be
observed that for every load, the required capacitance value Fig. 20 illustrates the effect of load variation on the frequency of
reduces with increasing wind speed. It is also seen that as the load- SERG. It can be observed that in spite of the variation in connected
Optimal Excitation Capacitance (uF)

60

50

40

30

20

10
500
450 15
14
400 13
12
350 11
10
Load Impedance (Ohms) 300
9 Wind Speed (m/s)
8

Fig. 19. Optimal capacitance requirement under varying speed and changing load.

400
Impedance
(Ohms)
Load

300

200
0 0.2 0.4 0.6 0.8 1 1.2 1.4 1.6 1.8 2
Time (sec)
(a) Load Variation

100
Frequency
(Hz)

50

0
0 0.2 0.4 0.6 0.8 1 1.2 1.4 1.6 1.8 2
Time (sec)
(b) Output Frequency

200
Voltage (V)
Phase

-200

0 0.2 0.4 0.6 0.8 1 1.2 1.4 1.6 1.8 2

Time (sec)
(c) SERG Voltage

Fig. 20. Illustration of constant frequency under load variation and constant voltage under optimally-selected excitation capacitance.
352 T.R. Ayodele et al. / Applied Energy 190 (2017) 339–353

load as shown in Fig. 20a, the SERG operates at constant output fre-
The WDSERG operates at constant frequency under both con-
quency as observed from Fig. 20b. Furthermore, Fig. 20c depicts the stant and varying load conditions. In addition, WDSERG can
SERG voltage under optimally-selected excitation capacitances operate at a desired constant level with optimally-selected exci-
which maintains the voltage within the desired limits. From the tation capacitances, under both variation in wind speed and
figure, a constant load of 200 X is supplied from 0.0 s to 0.5 s. It load conditions.
is observed that the frequency rises rapidly within a very short
WDSERG under optimally-selected capacitances can find real
duration of about 0.04 s and attains a constant value of 50 Hz. applications in providing low-cost and good quality power sup-
The system continues under this condition of constant load and ply to the rural communities where the high cost of transmis-
constant frequency until 0.5 s when the load is suddenly changed sion from a centralised grid is a challenge.
to a value 350 X. This sudden increase in the load impedance
produces a transient increase in the generated voltage, which then
stabilises back to a constant value of about 220 V. In spite of this Appendix A
change in load, it can be observed that the frequency remains con-
stant. The frequency remains at a constant value of 50 Hz for fur- Derivation of the Expressions for d and X d in Eqs. (19) and (20)
ther changes in load conditions. using Mathematica Software are as follows:
This result indicates a significant advantage of SERG, which xr
Eqs. (9) and (10) with x ¼ 1 are as follows:
b
therefore makes it suitable for low-cost, isolated applications,
where simplicity of voltage and frequency control is required. V ds ¼ Rs Ids  X q Iqs ðA1Þ
Thus, WDSERG under optimally-selected capacitances can find real
applications in rural electrification for providing low-cost and good V qs ¼ Rs Iqs þ X d Ids ðA2Þ
quality power supply to the rural communities, where the high Eqs. (13)–(16):
cost of transmission from a centralised grid has been a major con-
cern for extending electric power to the rural communities. This V ds ¼ V sin d ¼ X c Ic sin d ðA3Þ
SERG-based WECS also offers a low-cost and simple solution for
powering agricultural activities such as farm irrigation. The energy V qs ¼ V cos d ¼ X c Ic cos d ðA4Þ
derived from this system can be efficiently stored using optimised
storage systems and then utilised to power farm irrigation Ids ¼ IL sinðd þ /Þ þ Ic cos d ðA5Þ
systems.
Iqs ¼ IL cosðd þ /Þ þ Ic sin d ðA6Þ
7. Conclusion Equating (A1) with (A3) and (A2) with (A4), we have:
X c Ic sin d ¼ Rs Ids þ X q Iqs ðA7Þ
A method to determine the optimal excitation capacitance
required to maintain the output voltage of WDSERG at a desired
X c Ic cos d ¼ Rs Iqs þ X d Ids ðA8Þ
level has been presented and discussed. The steady state model
was used to develop an algorithm that searches for the optimum Substituting the expression for Ids and Iqs in (A5) and (A6) into
excitation capacitance which produces a desired output voltage (A7) and (A8),
under conditions of varying wind speed and changing load within
X c Ic sin d ¼ Rs ½IL sinðd þ /Þ þ Ic cos d þ X q ½IL cosðd þ /Þ þ Ic sin d
the operating limits of the WDSERG. A basic experimental valida-
tion has been provided using published experimental data. From ðA9Þ
the results, the following conclusion can be made:
X c Ic cos d ¼ Rs ½IL cosðd þ /Þ þ Ic sin d þ X d ½IL sinðd þ /Þ þ Ic cos d

There is an optimal excitation capacitance value that would ðA10Þ
produce a constant output voltage under given operating condi-
tions of wind speed, terminal load and power factor. After dividing (A9) and (A10) through by Ic and putting s ¼ IILc ,

For a given load, the optimal capacitance requirement reduces the resulting expressions are solved using Wolfram Mathematica
as the wind speed increases. However, at certain critical speed, Software as follows:
the capacitance requirement increases considerably. There is
thus a corresponding increase in stator current which brings Solve½fX c  Sin½d þ Rs  ðt  Sin½d þ p  Cos½dÞ
about an increase in the stator copper loss, heating up of stator X q  ðt  Cos½d þ p þ Sin½dÞ ¼¼ 0; X c  Cos½d
windings and hence a significant reduction in the machine’s
efficiency. Therefore there is a practical limit beyond which þ Rs  ðt  Cos½d þ p þ Sin½dÞt
such increase in excitation capacitance can be utilised to þ X d  ðt  Sin½d þ p  Cos½dÞ ¼¼ 0g; fX d g ðA11Þ
achieve control of output voltage.

A lower optimal excitation capacitance is required for a leading where t, d and p are used to denote tau (s), delta (d) and phi (/)
load than for a lagging load of the same rating. Moreover, the respectively.
power factor of the load has a more significant effect on the Thus Xd is obtained as:
capacitance requirement for a capacitive load than for an induc-  
tive load. tCos½d þ pRs  Sin½dRs  Cos½dX c
Xd ¼ ðA12Þ

The optimal excitation capacitance requirement increases as Cos½d þ tSin½d þ p
the loading increases under any given speed. Also, the range This expression for Xd is further simplified by expanding cos[d
of excitation capacitance required to achieve energy production + p] and sin[d + p]. After dividing through by cos[d], we obtain Eq.
with a constant output voltage reduces with increasing speed. (20) in the revised manuscript as:
This information on the range of excitation capacitance require-
ment can be helpful in selecting a suitable capacitance bank and X c þ Rs ½tan d þ sðcos /  sin / tan dÞ
the switching devices required to achieve variable excitation. Xd ¼ ðA13Þ
1  sðsin / þ cos / tan dÞ
 T.R. Ayodele et al. / Applied Energy 190 (2017) 339–353
The equations in the Solve function above (Eq. (A11)) are fur-
ther expressed in terms of Tan[d] by expanding Cos[d + p] and
Sin[d + p] and dividing through by Cos[d]. Thus we have:
Solve½fX c  Tan½d þ Rs  ðt  ðTan½d  Cos½p þ Sin½pÞ  1Þ
 X q  ðt  ðCos½p  Tan½d  Sin½pÞ þ Tan½dÞ ¼¼ 0;
 X c þ Rs  ðt  ðCos½p  Tan½d  Sin½pÞ þ Tan½dÞ
þX d  ðt  ðTan½d  Cos½p þ Sin½pÞ  1Þ ¼¼ 0g; fd; X d g ðA14Þ
This gives:
((
R2s  t 2 R2s þ 2tSin½pR2s  2tCos½pRs X c  X 2c þ X c X q  tSin½pX c X q
Xd ¼
X c þ tSin½pX c þ X q þ t2 X q  2tSin½pX q
Thus, from the d expression in (A15), we have:


Rs þ sðX q cos /  Rs sin /Þ
d ¼ arctan ðA16Þ
X c  X q þ sðRs cos / þ X q sin /Þ
The Xd expression in (A12) and (A15) are mathematically equal.
Thus, the Xd expression is as given in (A13).
References
[1] Ayodele TR, Ogunjuyigbe AS. Wind energy resource, wind energy conversion
system modelling and integration: a survey. Int J Sustain Energy
2015;34:657–71.
[2] Ogunjuyigbe ASO, Ayodele TR, Akinola OA. Optimal allocation and sizing of PV/
wind/split-diesel/battery hybrid energy system for minimizing life cycle cost,
carbon emission and dump energy of remote residential building. Appl Energy
2016;171:153–71.
[3] Schweizer J, Antonini A, Govoni L, Gottardi G, Archetti R, Supino E, et al.
Investigating the potential and feasibility of an offshore wind farm in the
Northern Adriatic Sea. Appl Energy 2016;177:449–63.
[4] Ayodele TR, Jimoh AA, Munda JL, Agee JT. Wind distribution and capacity factor
estimation for wind turbines in the coastal region of South Africa. Energy
Convers Manage 2012;64:614–25.
[5] Afanasyeva S, Saari J, Kalofen M, Partanen J, Pyrhönen O. Technical, economic
and uncertainty modelling of a wind farm project. Energy Convers Manage
2016;107:22–33.
[6] Cheng M, Zhu Y. The state of the art of wind energy conversion systems and
technologies: a review. Energy Convers Manage 2014;88:332–47.
[7] Carranza O, Figueres E, Garcera G, Gonzalez-Medina R. Analysis of the control
structure of wind energy generation systems based on a permanent magnet
synchronous generator. Appl Energy 2013;103:522–38.
[8] Erikkson S, Bernhoff H. Loss evaluation and design optimisation for direct
driven permanent magnet synchronous generators for wind power. Appl
Energy 2011;88:265–71.
[9] Meo S, Zohoori A, Vahedi A. Optimal design of permanent magnet flux
switching generator for wind applications via artificial neural network and
multi-objective particle swarm optimization hybrid approach. Energy Convers
Manage 2016;110:230–9.
[10] de Freitas TRS, Menegáz PJM, Simonetti DSL. Rectifier topologies for
permanent magnet synchronous generator on wind energy conversion
systems: a review. Renew Sustain Energy Rev 2016;54:1334–44.
[11] Yenduri K, Sensarma P. Maximum power point tracking of variable speed wind
turbines with flexible shaft. IEEE Trans Sustain Energy 2016;7:956–65.
[12] Alnasir Z, Kazerani M. An analytical literature review of stand-alone wind
energy conversion systems from generator view point. Renew Sustain Energy
Rev 2013;28:597–615.
[13] Tripathi SM, Tiwari AN, Singh D. Grid-integrated permanent magnet
synchronous generator based wind energy conversion systems: a technology
review. Renew Sustain Energy Rev 2015;51:1288–305.
[14] Hossain MM, Ali MH. Future research directions of the wind turbine generator
system. Renew Sustain Energy Rev 2015;49:481–9.
[15] Mohammadpour H, Zadeh SG, Tohidi S. Symmetrical and asymmetrical low-
voltage ride through of doubly-fed induction generator wind turbines using
gate controlled series capacitor. IET Renew Power Gener 2015;9:840–6.
[16] Naidu NKS, Singh B. Experimental implementation of doubly fed induction
generator-based standalone wind energy conversion system. IEEE Trans Ind
Appl 2016;52:3332–9.
353
[17] Meenakshi R, Muthu R. Doubly fed induction generator for wind energy
conversion system - a survey. In: 2013 international conference on energy
efficient technologies for sustainability (ICEETS); 2013. p. 617–28.
[18] Schulz EM, Betz RE. Use of doubly fed reluctance machines in wind power
generation. In: Power electronics and motion control conference, 2006. EPE-
PEMC 2006. 12th International; 2006. p. 1901–6.
[19] Song Z, Xia C, Shi T. Assessing transient response of DFIG based wind turbines
during voltage dips regarding main flux saturation and rotor deep-bar effect.
Appl Energy 2010;87:3283–93.
[20] Alolah AI. Capacitance requirements for three-phase self-excited reluctance
generator. IEE Proc-Gener Trans Distrib 1991;138:193–8.
[21] Attoui I, Omeiri A. Modeling, control and fault diagnosis of an isolated wind
 ))
Rs  tSin½pRs þ tCos½pX q
; d ¼ ArcTan ðA15Þ
tCos½pRs þ X c  X q þ tSin½pX q
energy conversion system with a self-excited induction generator subject to
electrical faults. Energy Convers Manage 2014;82:11–26.
[22] Kuo SC, Wang L. Analysis of isolated self-excited induction generator. IEEE Proc
Gener Trans Distrib 2002;149:90–7.
[23] Rahim YHA, Fletcher JE, Hassanain NEAM. Performance analysis of salient-pole
self-excited reluctance generators using a simplified model. IET Renew Power
Gener 2010;4:253–60.
[24] El-Khouly FM, Osheiba AM. Brushless induction and reluctance generators
with DC excitation: experimental verification. Electric Power Comp Syst
2004;32:839–53.
[25] Rahim YHA, Mohamadein AL, Al-khalaf AS. Comparison between the steady-
state performance of self-excited reluctance and induction generators. IEEE
Trans Energy Convers 1990;5:519–25.
[26] Abu-Elhaija WS. Capacitance requirement of single-phase self-excited
reluctance generator using eigenvalues criterion. In: Proc. IEEE PES
transmission and distribution conf. exposition (TD 2005-000019), Dallas,
Texas; 2006. p. 21–5.
[27] Obe ES, Anih LU. Influence of rotor cage on the performance of synchronous
reluctance generator. Electric Power Comp Syst 2010;38:960–73.
[28] Abu-Elhaija WS, Muetze A. Self-excitation and stability at speed transients of
self-excited single-phase reluctance generators. IEEE Trans Sustain Energy
2013;4:136–44.
[29] Singh B, Niwas R, Chandra A, Miloud R. Voltage control and load leveling of
synchronous reluctance generator based DG set. In: 2014 6th IEEE power India
international conference (PIICON); 2014. p. 1–6.
[30] Ogunjuyigbe ASO, Jimoh AA, Nicolae DV, Obe ES. Analysis of synchronous
reluctance machine with magnetically coupled three-phase windings and
reactive power compensation. IET Electr Power Appl 2010;4:291–303.
[31] Carta JA, González J, Cabrera P, Subiela VJ. Preliminary experimental analysis of
a small-scale prototype SWRO desalination plant, designed for continuous
adjustment of its energy consumption to the widely varying power generated
by a stand-alone wind turbine. Appl Energy 2015;137:222–39.
[32] Al-Saffar MA, Nho E, Lipo TA. Controlled shunt capacitor self-excited induction
generator. IEEE; 1998.
[33] Malik NH, Haque SE. Steady state analysis of an isolated self-excited induction
generator. IEEE Trans Energy Convers 1986;1:133–9.
[34] Premalatha K, Sudha S. Self-excitation and voltage control of an induction
generator in an independent wind energy conversion system. Int J Modern Eng
Res (IJMER) 2012;2:454–61.
[35] Seyoum D, Rahman F, Grantham C. Terminal voltage control of a wind turbine
driven isolated induction generator using stator orientated field control. In:
IEEE APEC Conf. Rec, Miami Beach; 2003. p. 846–52.
[36] Sivakami P, Karthigaivel R, Selvakumaran S. Voltage control of variable speed
induction generator using PWM converter. Int J Eng Adv Technol (IJEAT)
2013;2:30–5.
[37] Chan TF. Steady state analysis of self-excited reluctance generator. IEEE Trans
Energy Convers 1992;7:223–30.
[38] Ogunjuyigbe ASO, Ayodele TR, Adetokun BB, Jimoh AA. Wind driven self excited
reluctance generator for rural electrification. In: Aissaoui AG, editor. Wind
Turbine - Design, Control and Applications. InTechOpen; 2016. p. 231–52.
[39] Krause P, Wasynczuck O, Sudhoff S. Analysis of electric machinery and drive
system. John Wiley & Son Publication; 2002.
[40] Ogunjuyigbe ASO, Jimoh AA, Nicolae DV, Obe ES. Performance analysis of a
capacitance compensated dual stator winding synchronous reluctance
machine. Int Rev Electr Eng 2010;5:437–46.
[41] Novotny DW, Lipo TA. Vector control and dynamics of AC
drives. Oxford: Clarendon Press; 1996.
[42] Honsinger VB. Steady state performance of reluctance machines. IEEE Trans
Power Apparatus Syst 1971;PAS-90:305–17.