Professional Documents
Culture Documents
Maximum Power Point Tracking of A Variab
Maximum Power Point Tracking of A Variab
Abstract—Optimum torque (OT)-based maximum power point Climbing/ Perturb and Observe(PO) method [11]. TSR con-
tracking (MPPT) is widely used in high power turbines because of trol uses the wind speed sensor, rotational speed sensor and
its simplicity. Several recent improvements claim better response the optimal tip speed ratio value, determined offline, to set
speeds by augmenting the basic OT algorithm with a suitably
scaled inertial torque. However, the underlying dynamic model for the speed reference [8], [9]. OT control uses a lookup table,
all these improved methods is premised on a rigid shaft, which decided offline, mapping turbine speed to optimal torque, which
ignores all torsional behavior. This lacuna is addressed in this sets the online generator torque reference [2], [5], [10]. Both
paper, where a small-signal system description is developed con- these methods may fail to track exact MPP during param-
sidering a more accurate flexible shaft model. It is shown that the eter variations caused by changes in air density, aging and
improvements have three possible forms, the third being proposed
in this paper. Close-loop stability with each of these is analyti- blade surface contamination etc [12]. The PO algorithm dis-
cally investigated using this accurate small-signal model. It is also penses with all offline data and uses only rotational speed and
proved that one of the reported approaches causes system insta- generator active power measurements. Even though the PO
bility while attempting even moderate improvement over the OT method is generic, its slow response for high inertia turbines
method. The problems in realization of the other two forms are renders it suitable only for small scale systems. High power
highlighted and a realizable alternative proposed, which requires
no additional sensor. Design of the proposed approach is presented modern turbines generally use the OT approach because of its
in appropriate detail. It is analytically established that the pro- simplicity [11].
posed method ensures superior dynamic response. All analytical Of the reported improvements on the OT algorithm, Power
conclusions are validated by numerical simulations. signal feedback(PSF) method is essentially similar except that
Index Terms—Maximum power point tracking (MPPT), torque the feedback and reference are active power instead of torque
control, wind turbine. [6]. An improvement [13], proposing a sub-optimal gener-
ator torque reference, is shown to result in more captured
I. I NTRODUCTION power in real turbulent wind. However, energy gains for a
given sub-optimal value depends heavily on the level of turbu-
Mechanical power, Pt , generated by a wind turbine, as a With increasing a, this leads to decrease in the effective iner-
function of its aerodynamic efficiency Cp is expressed as tia offered by the wind turbine with consequent reduction in
the mechanical time-constant. For positive values of a, the
1 ωt R generator torque reference is reduced when the turbine is
Pt = ρAυ 3 Cp (λ, β); λ = (1) accelerating and vice-versa, thereby improving the dynamic
2 υ
response. However, realization of this improved control law (5)
where, υ is the wind velocity, ωt is the rotational speed of the requires real-time updates of the aerodynamic torque, Tt .
turbine, R the radius of the turbine blade and β the pitch angle In literature [3]–[10], the drive train of the turbine has been
of the turbine, all expressed in consistent units [1]–[2]. modeled predominantly by a single lumped mass (stiff-shaft
The basic power equation of the turbine (1) is modified to model), which is an aggregate inertia of all the moving parts.
obtain the aerodynamic torque in the following form. For example, the aerodynamic torque estimation premised on a
1 1 stiff-shaft model, formulates (5) effectively as
Tt = ρCp (λ)πR5 3 ωt2 . (2)
2 λ dωg
Tg∗ = Kωg2 − a(Jt + Jg ) (7)
dt
For any wind velocity below rated value, the turbine torque
that corresponds to the maximum power lies on the trajectory where Jt and Jg are the turbine and generator inertia respec-
specified by, tively. Though this model is simple, the stiff-shaft model fails
to capture the complete mechanical dynamics since it totally
1 1 ignores all torsional dynamics due to flexibility of the shaft.
Tto = ρCpo πR5 3 ωt2 = Kωt2 . (3)
2 λo Validity of that model is therefore questionable, particularly so
when the focus is on large systems characterized by low-speed
Eq. (3) is the basis for optimal torque MPPT and is used to turbines with large torques and large inertias.
set the reference value, Tg∗ , for the electromagnetic (genera- A first consequence of the flexible-shaft model is that the
tor) torque. The control law of the conventional optimal torque rotational speeds of the shaft, measured at the generator (ωg )
MPPT is thus given by [3] and turbine (ωt ) ends, are not identically equal. Hence the net
accelerating torque is correctly defined as
Tg∗ = Kωg2 (4)
dωt dωg
+ Jg
Ta = Jt (8)
In steady state, under constant wind speed, the OT scheme dt dt
ensures that the generator torque becomes equal to the tur- which in effect is a linear combination of the acceleration at the
bine torque and the system reaches MPP. If a positive (neg- turbine and generator ends. In a large turbine, since Jg ≪ Jt ,
ative) wind perturbation is now applied, the generator speed the approximation of the accelerating torque used in (7) is not
increases (decreases) along the above trajectory and reaches accurate. This leads to the first alternative control law proposed
the new MPP. Hence the system has a tendency to reach in this paper.
This article has been accepted for inclusion in a future issue of this journal. Content is final as presented, with the exception of pagination.
YENDURI AND SENSARMA: MAXIMUM POWER POINT TRACKING OF VARIABLE SPEED WIND TURBINES 3
TABLE I
C ONTROL L AWS AND C ORRESPONDING S MALL S IGNAL T RANSFER F UNCTIONS OF THE S YSTEM
Jg ∆
a< = â(2) . (18)
(Jg + Jt )
YENDURI AND SENSARMA: MAXIMUM POWER POINT TRACKING OF VARIABLE SPEED WIND TURBINES 5
Fig. 4. The Pole-Zero plots of small signal transfer function of the system using a) OT control law b) method 1 c) method 2 d) Proposed method I as a varies
from 0.2 to 0.8.
TABLE III
R EAL PART OF THE D OMINANT P OLE ( S )
dωt dωg ∆
Tt − T g = J t + Jg = Ta (19) A. Relative Influence on Ta
dt dt
In a normalized representation with respect to the rated speed
By using the second and third rows of the state space model (ωb ) and rated torque (Tb ), the corresponding variables
(11), the above equation can be written as
ω̄g = ωg /ωb and T̄g = Tg /Tb (24)
d J g d2 ωg 1 dTg dωg
Ta = J t ωg + + + Jg . (20) are expected to acquire values with similar orders of magnitude.
dt Ksh dt2 Ksh dt dt
Using these normalized variables, (23) is expressed as
Considering zero initial conditions in (20), frequency domain
Ta ωb
description of the shaft dynamics is expressed as T̄a = = G1 (s) ω̄g (s) + G2 (s)T̄g (s). (25)
Tb Tb
Jt Jg 2 Jt s 2 G′1
Ta = s (Jt + Jg ) + s ωg (s) + Tg (s). (21)
Ksh Ksh
G1 G2
Relative contributions of each of the input variables (T̄g , ω̄g )
on T̄a are evaluated from the magnitude (Bode) plots of G′1
Obviously G1 has an anti-resonance at Ωg , with an extremely and G2 , shown in fig. 5. Since |G2 | ≪ |G′1 | in general, Tg
low minimum arising out of the negligible mechanical damp- has marginal effect on the estimated value of Ta . Hence G2
ing, where is excluded from the estimation law, which is now sought to be
solely determined by G′1 .
Jt Jg Since G′1 is not proper, a realizable estimation law is pro-
Ωg = (22)
(Jt + Jg )Ksh posed as
is the natural frequency of the turbine-generator shaft. ˆ G′1 (s) ∆ G1 (s) ωb
From a foregoing discussion on the relative time-scales of the T̄a (s) = ω̄g (s) = ω̄g (s) (26)
H(s) H(s) Tb
shaft speed and generator torque dynamics, (21) is expressed as
Gp (s)
Ta = G1 (s)ωg (s) + G2 (s)Tg (s). (23)
where,
Direct inclusion of (23) in (5) does not offer a realizable def- H(s) = 1 + b1 s + b2 s2 + b3 s3 . (27)
inition of Tg∗ , since the coupling transfer functions in (23) are
improper. This aspect is separately investigated to arrive at a Roots of the general third degree polynomial in (27) can
feasible estimation law. either be all real or one real and a complex conjugate pair.
This article has been accepted for inclusion in a future issue of this journal. Content is final as presented, with the exception of pagination.
Fig. 6. (a) stability trajectories at different ζ values and equilibrium speeds (b) Maximum possible normalized bandwidth at different ζ values. (c) The real part
of the dominant pole(s) at different equilibrium speeds for all control methods.
Therefore, it is factorized and written in another general choice of (a1 , a2 , ζ) decide the estimator bandwidth, ωb ,
form as defined as
s s s2 ωb = sup{(ω/Ωg ) : 0.9 ≤ |1/H(jω)|≤ 1.1}. (33)
H(s) = 1 + 1+ζ + 2 2 . (28)
a2 Ωg a1 Ωg a1 Ωg
Bandwidth of the quadratic term in the √ denominator of
The first term in (28) ensures a band-limited differentia- (32) increases monotonically with ζ(ζ ≤ 2). Also, for
tor component in Gp , while the second introduces a complex | log(a2 /a1 )| ≫ 0, either the linear factor minimally affects
or real pole-pair, based on the damping coefficient, ζ. An bandwidth or ωb is drastically reduced. Thus an exhaustive
undamped resonator (ζ = 0) is not considered to avoid inter- numerical search for the maximum bandwidth is carried√ out
nal instability. Substituting (26) and (28) in (5), the proposed within the localized parameter space 0.4 ≤ ζ ≤ 2, 0.5 ≤
control law for the reference generator torque is obtained as (a2 /a1 ) ≤ 2, following the steps detailed below. The parameter
space is discretized using
Tg∗ = Kωg2 − aGp (s)ωg (s). (29)
ζ[i] − ζ[i − 1] = 0.01; (a2 /a1 )[n] − (a2 /a1 )[n − 1] = 0.01
Transfer function between Tt and ωt is derived from the small-
signal model of (29) and state space model (11), linearized to allow numerical computation.
about the equilibrium point (12). Thus, Step 1: Over the parameter space, record the normalized
5 6 bandwidth (ωb′ ), defined in (33), of
δωt (s)
m
k 1 1
= cm s dk s (30) = (34)
δTt (s) H(jω ′ ) jω ′
m=0 k=0 1+ a2 /a1 1 − ω ′ 2 + jζω ′
where the coefficients cm , m ∈ [0, 5] and dk , k ∈ [0, 6]
where, ω ′ = ω/(a1 Ωg ). The results are arranged in an ordered
are detailed in Appendix A. Design of the parameter set
array {ω ′ [i, n], ζ[i], (a2 /a1 )[n]}.
(a, ζ, a1 , a2 ) is based on stability as well as dynamic perfor-
Step 2: For a given equilibrium speed, overall system stability
mance criteria and is detailed subsequently.
boundary in the (a2 /a1 − a1 ) plane is determined by apply-
ing Routh-Hurwitz criterion on the denominator polynomial of
B. Design of Estimator Parameters (30). This results in a second array {a1 [i, n], ζ[i], (a2 /a1 )[n]},
During a wind-gust, the torsional shock load on the shaft corresponding to a particular equilibrium speed. A family of
increases with a. Hence its choice is subject to the condition such boundaries are obtained for different equilibrium speeds,
ranging from cut-in to rated. A few representative results are
a≤ ξ T̂sh T̂t − 1 (31) plotted in fig. 6a, where feasible solutions are in the region to
the left of these curves. As the equilibrium speed is seen to have
where, T̂t is the maximum allowable aerodynamic torque on minimal effect on the stability boundary, it is sufficient to con-
the given turbine, T̂sh is the torsional yield limit and ξ a factor sider the specific small-signal model at cut-in speed for optimal
of safety. These are design inputs and a detailed discussion is estimator design.
beyond the scope of this paper. The value of a (=0.6) is chosen Step 3: For every ζ[io ], io ∈ {i}, the local maximum band-
in this paper. width, evaluated at cut-in speed, is obtained as
Investigating the transfer function
ω̂b [io ] = max{ωb′ [io , n] × a1 [io , n]} (35)
n
Gp (s) 1 1
= = , (32) and the result is plotted in fig. 6(b). The global maximum band-
G1 (s)
′ H(s) 1+ s
1 + ζ a1sΩg + s2
a2 Ω g a21 Ω2g width is seen to occur at ζ[î] = 0.59, which corresponds to
This article has been accepted for inclusion in a future issue of this journal. Content is final as presented, with the exception of pagination.
YENDURI AND SENSARMA: MAXIMUM POWER POINT TRACKING OF VARIABLE SPEED WIND TURBINES 7
Fig. 8. (a) Wind speed (m/s) vs time (s); (b) Cp vs time (s); (c) generator speed (rad/s) vs time (s); (d) generator torque (N-m) vs time (s); (e) generator rms
current (A) vs time (s); (f) shaft torque (N-m) vs time (s); (g) Cp histogram; (h) λ histogram; (i) month-wise incremental energy (as % of baseline).
YENDURI AND SENSARMA: MAXIMUM POWER POINT TRACKING OF VARIABLE SPEED WIND TURBINES 9
c1 = Ksh A + 2Kωg0 ; c0 = Ksh ; [11] S. M. Raza Kazmi, H. Goto, H.-J. Guo, and O. Ichinokura, “Review and
critical analysis of the research papers published till date on maximum
d6 = CJt Jg ; d5 = 2Kωg0 CJt + BJt Jg − aJt (Jt + Jg )/Ω2g ; power point tracking in wind energy conversion system,” in Proc. Energy
Convers. Congr. Expo. (ECCE), Sep. 12–16, 2010 pp. 4075–4082.
d4 = C(Jt + Jg ) + 2Kωg0 BJt + AJt Jg ; [12] Z. M. Dalala, Z. U. Zahid, W. Yu, Y. Cho, and J.-S. Lai, “Design and
d3 = Ksh (2Kωg0 C + (Jt + Jg )B) + 2Kωg0 AJt + Jt Jg analysis of an MPPT technique for small-scale wind energy conversion
systems,” IEEE Trans. Energy Convers., vol. 28, no. 3, pp. 756–767, Sep.
− (Jt + Jg )(aJt + aKsh /Ω2g ); 2013.
[13] K. E. Johnson, L. J. Fingersh, M. J. Balas, and L. Y. Pao, “Methods for
d2 = 2Kωg0 BKsh + (Jt + Jg )AKsh + 2Kωg0 Jt ; increasing region 2 power capture on a variable-speed wind turbine,” J.
Sol. Energy Eng., vol. 126, no. 4, pp. 1092–1100, 2004.
d1 = 2Kωg0 AKsh + Ksh (1 − a)(Jt + Jg ); d0 = 2Kωg0 Ksh ; [14] L. J. Fingersh and P. W. Carlin, “Results from the NREL variable-speed
test bed,” in Nat. Renew. Energy Lab., Golden, CO, USA, Rep. NREL/CP-
where A = ζ/a1 Ωg + 1/a2 Ωg , B = 1/a21 Ω2g + ζ/a1 a2 Ω2g and 500-23811, 1998.
C = 1/a21 a2 Ω3g . [15] W. E. Leithead and B. Connor, “Control of variable speed wind turbines:
Design task,” Int. J. Control, vol. 73, no. 13, pp. 1189–1212, 2000.
[16] K.-H. Kim, T. L. Van, D.-C. Lee, S.-H. Song, and E.-H. Kim, “Maximum
output power tracking control in variable-speed wind turbine systems
A PPENDIX B considering rotor inertial power,” IEEE Trans. Ind. Electron., vol. 60,
PMSG PARAMETERS & W EIGHTING F UNCTIONS no. 8, pp. 3207–3217, Aug. 2013.
[17] B. Boukhezzar and H. Siguerdidjane, “Nonlinear control of a variable-
speed wind turbine using a two-mass model,” IEEE Energy Convers.,
vol. 26, no. 1, pp. 149–162, Mar. 2011.
[18] I. Girsang, J. Dhupia, E. Muljadi, M. Singh, and L. Pao, “Gearbox and
drivetrain models to study dynamic effects of modern wind turbines,” in
Proc. IEEE Energy Convers. Congr. Expo., 2013, pp. 874–881.
[19] L. Soderlund, J.-T. Eriksson, J. Salonen, H. Vihriala, and R. Perala, “A
permanent-magnet generator for wind power applications,” IEEE Trans.
Magn., vol. 32, no. 4, pp. 2389–2392, Jul. 1996.
[20] J. Y. Chen, C. V. Nayar, and X. Longya, “Design and finite-element anal-
ysis of an outer-rotor permanent-magnet generator for directly coupled
wind turbines,” IEEE Trans. Magn., vol. 36, no. 5, pp. 3802–3809, Sep.
2000.
[21] I. Munteanu, A. I. Bratcu, and E. Ceanga, “Wind turbulence used as
searching signal for MPPT in variable-speed wind energy conversion
systems,” Renew. Energy, vol. 34, no. 1, pp. 322–327, Jan. 2009, doi:
10.1016/j.renene.2008.03.001.
[22] C. Nichita, D. Luca, B. Dakyo, and E. Ceanga, “Large band simulation
of the wind speed for real time wind turbine simulators,” IEEE Trans.
Energy Convers., vol. 17, no. 4, pp. 523–529, Dec. 2002.
[23] R. Krishnan, Permanent Magnet Synchronous and Brushless DC Motor
Drives. Boca Raton, FL, USA: CRC Press, 2009.
R EFERENCES [24] M. Chinchilla, S. Arnaltes, and J. C. Burgos, “Control of permanent-
magnet generators applied to variable-speed wind-energy systems con-
[1] D. S. Zinger and E. Muljadi, “Annualized wind energy improvement nected to the grid,” IEEE Trans. Energy Convers., vol. 21, no. 1,
using variable speeds,” IEEE Trans. Ind. Appl., vol. 33, no. 6, pp. 1444– pp. 130–135, Mar. 2006.
1447, Nov./Dec. 1997. [25] W. Leonhard, Control of Electrical Drives. New York, NY, USA:
[2] E. Muljadi and C. P. Butterfield, “Pitch-controlled variable-speed wind Springer, 1997.
turbine generation,” IEEE Trans. Ind. Appl., vol. 37, no. 1, pp. 240–246, [26] S. Butterfield, W. Musial, and G. Scott, “Definition of a 5-MW reference
Jan./Feb. 2001. wind turbine for offshore system development,” Nat. Renew. Energy Lab.,
[3] R. Datta and V. T. Ranganathan, “A method of tracking the peak power Golden, CO, USA, Rep. NREL/TP-500-38060, 2009.
points for a variable speed wind energy conversion system,” IEEE Trans. [27] A. Steinberg and H. Holttinen, “Analysing failure statistics of wind
Energy Convers., vol. 18, no. 1, pp. 163–168, Mar. 2003. turbines in Finland,” in Proc. Eur. Wind Energy Conf., Apr. 2010,
[4] K. Tan and S. Islam, “Optimum control strategies in energy conversion pp. 20–23.
of PMSG wind turbine system without mechanical sensors,” IEEE Trans. [28] I. Van der Hoven, “Power spectrum of horizontal wind speed in the fre-
Energy Convers., vol. 19, no. 2, pp. 392–399, Jun. 2004. quency range from 0.0007 to 900 cycles per hour,” J. Meteorol., vol. 14,
[5] S. Morimoto, H. Nakayama, M. Sanada, and Y. Takeda, “Sensorless no. 2, pp. 160–164, 1957.
output maximization control for variable-speed wind generation system [29] Western Regional Climate Center. (2016, Jan 07). Average wind speeds
using IPMSG,” IEEE Trans. Ind. Appl., vol. 41, no. 1, pp. 60–67, Jan./Feb. data [Online]. Available: http://www.wrcc.dri.edu/htmlfiles/westwind.
2005. final.html .
[6] E. Koutroulis and K. Kalaitzakis, “Design of a maximum power track-
ing system for wind-energy-conversion applications,” IEEE Trans. Ind.
Electron., vol. 53, no. 2, pp. 486–494, Apr. 2006.
[7] S. M. R. Kazmi, H. Goto, H.-J. Guo, and O. Ichinokura, “A novel
algorithm for fast and efficient speed-sensorless maximum power point
tracking in wind energy conversion systems,” IEEE Trans. Ind. Electron.,
vol. 58, no. 1, pp. 29–36, Jan. 2011.
[8] T. Thiringer and J. Linders, “Control by variable rotor speed of a fixed- Kalyan Yenduri received the B.Tech. degree in
pitch wind turbine operating in a wide speed range,” IEEE Trans. Energy electrical engineering from the Jawaharlal Nehru
Convers., vol. 8, no. 3, pp. 520–526, Sep. 1993. Technological University, Hyderabad, India, and the
[9] R. M. Hilloowala and A. M. Sharaf, “A rule-based fuzzy logic controller M.Tech. degree in power electronics and drives from
for a PWM inverter in a stand alone wind energy conversion scheme,” National Institute of Technology, Warangal, India, in
IEEE Trans. Ind. Appl., vol. 32, no. 1, pp. 57–65, Jan./Feb. 1996. 2007 and 2009, respectively. He is currently pursu-
[10] M. E. Haque, M. Negnevitsky, and K. M. Muttaqi, “A novel control ing the Ph.D. degree in electrical engineering, Indian
strategy for a variable-speed wind turbine with a permanent-magnet syn- Institute of Technology Kanpur, Kanpur, India. His
chronous generator,” IEEE Trans. Ind. Appl., vol. 46, no. 1, pp. 331–339, research interests include renewable energy integra-
Jan./Feb. 2010. tion and power converters.
This article has been accepted for inclusion in a future issue of this journal. Content is final as presented, with the exception of pagination.