Professional Documents
Culture Documents
Research Article
ISSN 1751-8687
Advanced fuzzy power extraction control of Received on 25th June 2015
Revised on 25th October 2015
wind energy conversion system for power Accepted on 1st December 2015
doi: 10.1049/iet-gtd.2015.0769
quality improvement in a grid tied hybrid www.ietdl.org
generation system
Chayan Bhattacharjee ✉, Binoy Krishna Roy
Electrical Engineering Department, National Institute of Technology, Silchar, India
✉ E-mail: chayan_kol@yahoo.co.in
Abstract: This paper introduces an intelligent extraction of optimum power and its dispatch by using fuzzy logic from a
grid tied hybrid generation system comprising of a permanent magnet synchronous generator based wind turbine and
a low concentration photovoltaic generator. For photovoltaic generator, maximum power point tracking control is
implemented using fuzzy logic under varying solar irradiance. Power extracted from wind turbine is designed as a
fuzzy function of the dc link voltage error, its rate of change and error in the direct axis current of the inverter. This
reduces high frequency oscillations in the wind extracted power. Such an extraction is considered as a novelty of this
paper. A failure mode and effect analysis is done for power converters and possible mitigation schemes are suggested
for different faults. A 1:1 delta wye-grounded transformer is used at the inverter output to eliminate the triplet
harmonics. Further, dynamic performances of both fuzzy proportional-derivative and integral (PD+I) controller and
classical proportional integral (PI) controller, to control the inverter currents, are compared. The proposed method
results in an enhanced power dispatch and improvement in distortions and oscillations in the converter currents. This
also reduces the probability of failures in the converter switches and other passive components.
1 Introduction algorithm based on the knowledge of direct (d) axis current error
of the inverter (Δid), deviation in dc link voltage (ΔVdc) with
Increased penetrations of renewable energy sources into the power respect to its reference and its rate of change (δ (ΔVdc)). This
grid arouse several challenges in integration not only amongst technique performs well to stabilise dc link voltage and is devoid
themselves but also between such sources and the grid. Though of high frequency components. The crucial challenge is fast
the energy obtained from such sources is environment friendly but damping of oscillations in dc link voltage during rapid changes in
the power and voltage obtained from such sources varies randomly wind speed under wind turbulence and change in solar irradiance
with the variation of weather. Furthermore, non-linear power while dispatching the extracted power, at the PCC. Furthermore,
converters, used for conditioning the power outputs from such sag ride through of the inverter voltage would not be possible if
sources, distort the waveform and hence degrade the quality of the inverter reactive power compensation for load variation is
dispatched power thereby affecting sensitive loads connected to inappropriate [12]. The earlier papers cited do not focus keenly
the grid. Exhaustion of fossil fuels, their hazardous influence on into these aspects altogether. The BESS control aims to regulate
environment and an increasing power demand result in an the dc link voltage (Vdc) as for a grid connected system fast power
increased penetration of renewable energy sources into the utility transfer at the PCC is possible if Vdc remains constant under all
grid [1]. Since availability of renewable energies is variable and circumstances [13, 14]. Voltage oriented control (VOC) of a
unpredictable so energy storage device is essential at the dc link. two-level fully controlled inverter is done as the load system
This device maintains the power balance at the dc link which considered requires no neutral grounding and switching losses may
results in a smooth power dispatch. Li-ion batteries though costly be reduced unlike higher-level inverters [15]. The VOC is
have higher power density and can withstand higher charging/ designed to achieve an improved quality power to be fed at the
discharging cycles than that of lead-acid batteries while NiMH grid, apart from load reactive power compensation for power
batteries are costlier and have carcinogenic effect on the factor improvement and voltage stabilisation at the load bus [16].
atmosphere in spite of having higher power density than Li-ion Low pass filtered references for both active and reactive current
batteries [2, 3]. Low concentration photovoltaic (LCPV) generator components are used to reduce power distortions at the PCC. This
used in this work has a degree of concentration of 2.2X which do method along with the passive L-C-L filter at the inverter output
not require any extra tracking mechanism and cooling together improves both voltage and current THD. A failure mode
arrangements for their cells [4]. The use of such cells can increase and effect analysis (FMEA) of the power conditioning interface
solar power extraction efficiency to 20–25% [5] than flat plate between the generators and the grid is also discussed. Possible
collectors, with the use of simple mirror reflection with good mitigation methods for system protection for faults in different
optical efficiency. In the proposed system maximum power converters are shown. This analysis is of importance in such
extraction is essential from both LCPV generator and wind energy hybrid system as the continuous heating of the power switches;
conversion system (WECS) for increasing system efficiency [6, 7]. improper heat sink and ageing may result in either open or short
Fuzzy logic control has proved better than classical control for circuit faults in converters switches, failures in line inductors and
maximum power point tracking (MPPT) particularly under varying capacitors. This hampers power conversion process, degrades
inputs for renewable energy sources [8–10]. Tracking of maximum power quality and incurs economic losses.
operating power point is done using source side variations which The remaining part of this paper is organised as, Section 2 deals
results in high frequency components in the dc link voltage [11]. with the system component modelling and their dynamics, Section
In this work optimum power is extracted from WECS using fuzzy 3 discuss the FMEA of the power conditioning interface of the
3
no tracking and extra cooling arrangement thus enabling power
extraction at a higher efficiency [18]. Basic one diode equivalent Te = P wr − Ld − Lq id iq (8)
circuit is used [16] in this work, where solar concentration of 2
2.2X is assumed to vary due to cloud intervention. The solar
irradiance is in per unit (pu), (Gpu) over its base value of 1 KW/ where vd, vq, id and iq are the respective voltage and current
m2. The ambient temperature, Ta [°C] and the cell temperature Tc components in d-q-axes while Ed and Eq are the d-q-axes
[°C] are dependent on Gpu as per (1) and (2), as both temperature back-emf components and Te is the electro-magnetic torque
and irradiance regulates the output extractable power from the experienced by the generator. Ld, Lq and rs are the stator leakage
LCPV collector. An increase in cell temperature decreases the reactances and winding resistance while jr and P is the constant
amount of possible power extraction unlike an increase in solar rotor flux and the number of pole-pair of the PMSG, respectively,
irradiance. whose rated power is 90 KW. The instantaneous disturbances in
the rotor torque and shaft speed which follows from wind speed
Ta = 22 + 0.05(Gpu − 1.2) (1) variations and shaft dynamics as in (9) are directly reflected in the
output voltage and power of the generator. The amount of ripples
in the dc side voltage after rectification of the PMSG ac voltage
G(Tcnom − Tanom ) decides the value of the dc bus line inductor which is responsible
Tc = Ta + (2)
Gnom for suppressing the ripples [20] as per (10).
where Tanom , G and Gnom are the nominal ambient temperature (20° dv
Tm = J + Bv + T e (9)
C), solar irradiance in KW/m2 and nominal irradiance of 0.8 KW/m2, dt
where,
extraction will be close to MPPT. For case III, when load demand is
low focus is given in lessening ΔVdc and power extraction is adjusted Vdc = f (wr ) (14)
accordingly.
The fuzzy OPPT control is modelled as shown in Fig. 4a where
FLC is the fuzzy logic controller and pulses are generated using Table 4 Rule matrix for fuzzy-MPPT control of SEGS
sinusoidal pulse width modulation.
CE E NB NS Z PS PB
Fig. 4 Different control schemes implemented for the WECS, battery and the inverter
a Fuzzy OPPT control
b Blade pitch angle control
c BESS bidirectional power control
d VOC of the inverter
Fig. 6 Input and output characteristics of wind turbine and LCPV generator
a Wind speed
b Rotor speed
c Solar irradiance for LCPV
d Solar power generated
4.3 Pitch angle control for wind turbine where d, IL and (Rs, Ls) are the duty cycle of the chopper, solar
generator load current and parameters of the boost converter,
As per IEC 61400-2, blade pitch control for limiting extraction of respectively. A basic hill-climb approach is used for tracking
power from wind turbine is essential when the generated power maximum power operating point where the control inputs
exceeds the rated power of the turbine to avoid its overloading considered are dPpv/dVpv and its rate of change [29] and change in
[28]. The required pitch angle, β is generated from the power duty-cycle as the output. Mamdani type fuzzy control with five
difference of turbine rated power and actual power. Fig. 4b shows triangular membership functions for the two inputs error (E) and
the blade pitch angle control where the power error, ΔP, is sent to change in error (CE) and the output which is perturbation in d are
a PI regulator that generates the reference pitch angle. When ΔP < considered with its rule matrix given in Table 4. Input
0, the limiter restricts the pitch angle reference to zero but when membership functions are considered tilted towards centre as
ΔP > 0, the pitch actuator actuates the necessary pitch angle maximum occurrences of errors are around the ‘0’ crisp value.
change of the blades to limit the extracted power. The dynamics of
the pitch angle actuator is non-linear and depends upon saturation
limits on pitch angle and pitch rate. 4.5 Bidirectional power flow control for BESS
1 dVpv
Vboost = {V − IL (Rs + Ls )} (15) ipvL = ipv − Cpv (17)
1 − d pv dt
where iRL, iLoad and ibatt are current delivered from both renewable 3
sources, inverter input current and battery current, respectively, at the Q∗ = (vin iin − vin in
q id ) (20)
2 d q
common dc link. Cpv is the output capacitor of the LCPV generator
while Cdc is the dc link capacitor. The control action comprise of an
inner current control loop and outer voltage control loop as shown in where P*, the inverter active power reference is the sum of extracted
ref
Fig. 4c [31]. Ibatt is the battery reference current generated by the power from the renewable sources while Q* is the load reactive
voltage regulator for a ΔVdc. Vcom, battery voltage Vbatt and Vdc power demand. The component of net active current reference
controls the duty ratio, dbatt of the converter [32] as per (18) which derived from (19) by neglecting vq is id1* while the other
in turn controls the battery charge. Battery charge control is component Id2* is obtained from instantaneous ΔVdc as shown in
constrained by the two current limits of +Isat and –Isat for Fig. 4d. Here id2* is added with id1* so as to increase the current
respective discharge and charge operations, while limits on Vcom reference when ΔVdc > 0 and decrease the same when ΔVdc < 0.
are imposed assuming maximum Vbatt and desired Vdc. The net reactive current reference, id* is obtained from (20). The
inverter voltage quality is fully controlled by the series arm
Vcom + Vbatt impedance of the L-C-L filter while the current quality may be
dbatt = (18) improved according to IEEE 519 using an additional active shunt
Vdc
compensator. The PI regulators for both id and iq are first tuned
classically for optimum performances. These values are then
4.6 VOC of the inverter utilised in fuzzy proportional-derivative and integral (PD+I)
controllers and performances compared with optimal classical
The inverter control is aimed at – (i) maximising the power dispatch, controllers. The rule array for the fuzzy PD+I control is given in
(ii) stabilising Vdc, (iii) maintaining synchronism with the grid Table 5. The system’s sampling frequency is set at 1.25 KHz to
operating conditions and (iv) improving THD of the power improve the precision of fuzzy control output.
delivered at the PCC [33, 34]. Decoupled vector control approach The linguistics for the abbreviation of membership functions used
of inverter direct-axis (d-axis) and quadrature-axis (q-axis) currents in the rule matrix given in Table 5 are the same as used before. ‘PS’
is implemented on a synchronously reference rotating frame in
phase with the grid voltage by using phase locked loop control,
where the d-axis voltage is considered aligned with the grid Table 6 Inverter voltage and current quality in terms of THD
voltage. Dynamics of the inverter power dispatched are given in THD in inverter Modes of operation
(19) and (20) where vin in in in
d , id and vq , iq are the respective d-axis
and q-axis voltage and current components of the inverter. MPPT OPPT OPPT-fuzzy
that OPPT-fuzzy is the best amongst the three modes for voltage droop at the inverter output. Fig. 10 shows the
improvement of power quality at the PCC as shown in Fig. 8. compensation done by mode 3 tracks the load requirement much
Figs. 8a and b shows that although the voltage THD is the best in closely with lesser oscillations than that mode 1. Figs. 10a and b
MPPT method but a voltage THD of 0.87 is quite acceptable as show the difference in the two modes in reactive power
obtained in OPPT-fuzzy as per standard IEEE-519. However compensation at the PCC. While Fig. 10a shows a much improved
Figs. 8c and d show a good improvement in the THD of the stability for the compensation by mode 3 resulting in average grid
inverter current for OPPT-fuzzy over that obtained in MPPT. A reactive power requirement to zero, Fig. 10b gives a picture of
summary of the results of THD in inverter voltage and current is poor stability by mode 1 with much higher oscillations. A
shown in Table 6 for all three modes of operations. comparison of load reactive power compensation as shown in
Therefore, OPPT-fuzzy proves to be the best method out of the Fig. 10c gives precise difference in the stability performances of
three in terms of power maximisation, reducing oscillations in the two modes.
system responses and reducing distortions in the inverter
dispatched power. Third harmonic and higher order triplets (3rd
order harmonics) are negated at source by the use of 1:1 delta-star 6 Conclusion
transformer (Fig. 1) which completely blocks the flow of triplets at
the PCC. This work emphasises to extract maximum power from the 2.2X
Operation by mode 2 although produces almost similar concentration LCPV solar generator and optimum power from a
oscillations in system responses but yields lesser power at the PMSG based direct driven wind power generator to provide
inverter output so onwards responses with this method is not necessary power to the local load and the grid. The power
shown. The active power management at the inverter output and extractions are done intelligently using fuzzy logic control but
dc power extracted from the wind generator for mode 1 and mode wind power being the dominant contributor amongst the two,
3, respectively, are shown in Figs. 9a and b. Fig. 9a shows active inject high frequency oscillations in the dc link power. Therefore
power dispatched by mode 3 is higher particularly after 2.5 s as extraction of power from this generator by the proposed method
compared with Fig. 9b for mode 1. Thus grid support is also focuses to stabilise the dc link voltage and reduce active power
higher in mode 3 reducing its active power burden of the local mismatch at the inverter output. This method utilises intelligently
load. Power extraction from WECS again is also higher by mode 3 the knowledge of transient power imbalance at the dc link and the
as shown in Fig. 9c particularly after 2.5 s. The dc link voltage variation of load current to extract and dispatch maximum possible
shown in Fig. 9d indicates mode 3 has lesser undershoot and power with an improved quality at the inverter output. The system
oscillations of the dc link voltage about its reference value functions in grid supporting mode by dispatching the active power
compared with mode 1. generated and reactive power compensation of the load wholly.
The grid inverter is supposed to compensate for the instantaneous Frequency and magnitude of the inverter output voltage is
load reactive power to improve load power factor and to control maintained at its desired nominal value without any voltage sags/