See discussions, stats, and author profiles for this publication at: https://www.researchgate.

net/publication/224164862

Effects of WECS settings and PMSG parameters in the performance of a small

wind energy generator

Conference Paper · July 2010

DOI: 10.1109/SPEEDAM.2010.5542223 · Source: IEEE Xplore

CITATIONS READS

6 260

4 authors, including:

C. Carrillo Eloy Diaz-Dorado

University of Vigo University of Vigo
45 PUBLICATIONS   845 CITATIONS    35 PUBLICATIONS   877 CITATIONS   

SEE PROFILE SEE PROFILE

Some of the authors of this publication are also working on these related projects:

Leveraging Machine Learning Methods for Stochastic Optimization View project

All content following this page was uploaded by C. Carrillo on 11 September 2014.

The user has requested enhancement of the downloaded file.

 EPE Wind Energy Seminar 2010

Performance of a small wind energy generator

under different configurations and settings
C. Carrillo*, M. Silva-Ucha*, E. Diaz-Dorado* and F. Perez-Sabín*
* Department of Electrical Engineering, University of Vigo, ETSEI - Campus Universitario
Vigo (Spain)
Tel: +34 986813912
Fax: +34 986812173
Email: [email protected]; [email protected]; [email protected]; [email protected]

Equation Chapter 1 Section 1 Acknowledgements

This work was supported in part by the Ministerio de Ciencia e Innovación (Spanish Government) under
Contract ENE 2009-13074 and in part by Ministerio de Educación y Ciencia (Spanish Government) under
Contract ENE 2007-67473.

Abstract
Market of small wind turbine generators (SWTG) has been continuously growing during recent years. In
this context, the offer for components for this technology has been also increased. The main components of
SWTG are the rotor, the generator and the wind energy conversion system (WECS), typically formed by a
rectifier plus an inverter. The settings of WECS affects to the SWTG behavior, especially those related with
maximum and minimum voltages and the way to establish the inverter power curve. In this paper a method
to calculate the inverter power curve of a STWG is presented, the impact of different types of curves is also
evaluated.

Keywords
Wind power generation, AC-AC power conversion, permanent magnet generators

Introduction
Due to continued market growth of small wind turbine generators (SWTG) in recent years, the study of
their behavior to improve their performance is an important issue. Nowadays, the design and manufacture of
these SWTG’s tends to the use of permanent magnet synchronous generators (PMSG). Thus in recent years,
studies about them have been done, mainly about control and design methods [1,2,3,10].
PMSG’s require wind energy conversion systems (WECS) that are used as interface between the generator
(variable voltage and frequency) and the grid (fixed voltage and frequency).
There are several WECS configurations: with only diode rectifier and a inverter [1,2], with a DC/DC
converter between rectifier and inverter [4,5,6], with a controlled rectifier [7,8], and with a controlled
rectifier and a DC/DC converter [9]. However, the typical one is that composed by a rectifier, an inverter and
dump load resistors is shown in Fig. 1. Dump loads are used to limit the AC voltage in the generator side or
the DC voltage; usually they are connected or dimmed when a preset voltage value is achieved. In this
configuration, most of the setting values are implemented in the inverter, and the power curve values are one
of the most important settings, because they are directly related to the energy production.
In this paper, a study based on SWTG’s with PMSG is presented. This paper focuses on the study of the
influence of PMSG parameters and WECS configuration in energy production.

DC bus
PMSG GRID

Rectifier Inverter

Dump load

Fig. 1: General scheme of a wind turbine

1
 EPE Wind Energy Seminar 2010

System description and modeling

The power configuration of the analyzed system is shown in Fig. 1. The wind-turbine used in this work is a
direct-drive system. The generator is connected to the network through a rectifier and an inverter (WECS).
The torque produced by the wind-turbine is used as the input torque to the PMSG and the voltage produced
by the PMSG is rectified and passed through an inverter to the network. A maximum-power strategy is
necessary to extract the maximum-power out of the system. In this case, the maximum power strategy is
implemented by means of the inverter power curve.
The main parameters of the simulated SWTG are summarized in TABLE I.
TABLE I. SWTG main parameters
ROTOR
Type Up-wind / fixed pitch
Number of blades 3
Diameter 2.8 m
GENERATOR
Type PMSG
Pole pairs 12
Nominal voltage 230 V
Nominal speed 220 rpm
Power 6600 W
Steady state resistance 1.63 Ω
Steady state inductance 22.46 mH
No-load voltage at nominal speed 190.3 V
RECTIFIER
Type Full wave / diode
Power 7,2 kW
Efficiency 99,5%
Max. input voltage 400 VAC
Dump load voltage 530 VDC
INVERTER
Efficiency 97%
Power 6 kW
Input voltage 50 – 560 VDC
Wind power extraction
The mechanical power (Pm) captured by the rotor can be calculated by means of the following equation:
Pm = 12 ρAC p ( λ ) v3ηmec (1)
Where ρ is the air density (typ. 1,225 kg/m3), A is the area covered by blades’ rotation, Cp is the power
coefficient and v is the wind speed (m/s) and ηmec is the mechanical performance. The power coefficient
depends on the tip speed ratio λ calculated with the following expression:
λ = ΩR v (2)
Where Ω is the rotational speed (1/s) and R is the blade radius.
The relationship between Cp and λ is non-linear and for fixed pitch rotor can be modeled with following
equation [11]:
RC
1  RC f  − k2 λ f
Cp = 2  − k1  e (3)
 λ 
Where Cf is a blade design constant, and k1 and k2 are constant parameters. These parameters have been
calculated from real data rotor behavior, and the result is shown in Fig. 2.

2
 EPE Wind Energy Seminar 2010

0.5

0.45

0.4

0.35

power coefficient
0.3

0.25

0.2

0.15

0.1

0.05

0
0 2 4 6 8 10 12 14
tip speed ratio

Fig. 2: Power coefficient curve

For each wind speed, there exists an optimum tip speed ratio λopt with its correspondent maximum power
coefficient Cp,opt (typ. 0.4-0.5), so the mechanical power has its maximum value Pm,opt for that wind speed.
From (1) and (2), the following equations can be obtained:
Ωopt = λ opt v R (4)
C p,max
Pm,opt = 12 ρAC p,opt v3 = 21 ρAR 3
(5) Ω3opt
λ 3
opt

This last equation establishes the well-known cubic relation-ship between rotational speed and mechanical
power used to maximize wind power production (see Fig. 3).
mechanical power in W

6000

4000

2000

0
3 4 5 6 7 8 9 10 11
wind speed in m/s
mechanical power in W

6000

4000

2000

0
50 100 150 200
rotational speed in rpm

Fig. 3: Maximum mechanical power curves

Permanent magnet synchronous generator (PMSG)
Permanent magnet synchronous generators (PMSG’s) are typically used in small wind turbines because due
to several reason (high efficiency, gearless, simple control…).
A simple steady estate model has been chosen in order to evaluate the behaviour of this kind of machines
[13], with the following assumptions:
• Saturation is neglected;
• Internal voltage or back emf is sinusoidal (harmonic components are neglected);
• Eddy currents and hysteresis losses are negligible.
In this case, a single phase equivalent circuit, as shown in Fig. 4, can be used and the following equation
determines the relationship between voltages and current:
E = U + ( R s + jωLs ) I (6)
Where E is the internal voltage, U is the voltage in the estátor, I is the current delivered, Rs and Ls are the

3
 EPE Wind Energy Seminar 2010

steady estate resistance and inductance, respectively, and ω is the frequency of voltage and currents in the
PMSG stator. In steady state, frequency is related to rotational speed by mean the equation:
ω = pΩ (7)
Where p is the number of pole pairs of PMSG.
Finally, the magnetic flux is supposed to be constant, so a linear relationship between rotational speed and
no-load voltage is considered, so:
E = K eΩ (8)
Where Ke is the relationship between E and Ω.
Ls Rs I

E U

Fig. 4: PMSG equivalent circuit

Wind energy conversion system (WECS)
In the SWTG considered in this paper, WECS is composed by a diode full-wave rectifier, dump load
resistors and an inverter. By means of these elements, power extracted from wind can be delivered to the
network in constant voltage and frequency conditions.
The rectifier has been modelled by mean the relationship between the input RMS voltage U and the DC
voltage UDC:
U DC = 3 3 2 π U (9)
In the rectifier input, voltage and current are in phase (cos φ = 1), so the following equations can be written:
E = U + ( R s + jωLs ) I (10)
E 2 I 2 = Pm2 + ω2 L2s I 4 (11)
A current source equivalent has been used to model the inverter. Its main parameters are the DC voltage
range and the power curve. Both parameters affect to the wind energy production for a given rotor and
PMSG. Power delivered by the inverter P in W can be written as:
P = 3UIηele (12)
Where ηele is the efficiency of WECS.

Power Curve Optimization

Inverter Power Curve
The inverter power curve relates an input variable (typ. rotor speed, DC voltage or frequency) with the
desired inverter power output. Usually, frequency of the voltage in the PMSG side fr or DC bus voltage UDC
is used as input for the mentioned curve. This curve, in commercial inverters, can be defined in several ways,
typically by means a polyline [0 or by third degree polynomial [1,2,12]. In any case, inverter power curve
must be designed to maximize the SWTG power production.
In certain inverters, inverter power curve is divided in two parts, the lower power part the power curve is
linear and in the upper part the power curve tends to be cubic (see Fig. 5). The upper high power part can be
expressed with a set of polyline vertices (frequency-power or DC voltage-power):
{( X0 , P0 ) ; ( X0 , P0 ) ;...; ( X n , Pn )} (13)
Where X is the input variable (fr or UDC) and n the number of point. Another way to define the power curve
is by means the coefficients (a0, a1, a2 and a3) of a third degree polynomial expression:
Ppol = a 0 + a1X + a 2 X 2 + a 3 X 3 (14)

4
 EPE Wind Energy Seminar 2010

Power in W
max
min
mid

DC voltage or
frequency

Fig. 5: Inverter power curve definition

So, the main inverter settings are: inverter power curve defined with (13) or (14), a minimum DC voltage to
start generation (UDC,min), a middle DC voltage or power as a limit between the low and high power parts of
the power curve parts and a maximum DC voltage or power when the power is fixed at the nominal value.
For the shake of simplicity lower part of the curve has been not taken into account in this paper.
Power Curve Optimization
In large WTG’s, an optimum power curve, as shown in (5), is usually used to define desired power
production. In this case, the inverter power curve can be defined as:
Popt = K opt ω3 (15)
Where:
C p,max
K opt = 12 ρAR 3
ηmec ηele (16)
p3λ 3opt
And Popt is the optimum power delivered by the WTG. A similar equation can be found to relate power and
DC voltage.
In SWTG, the cubic equation shown in (15) could not be the most appropriate due to the high impedance
values of PMSG. The maximum mechanical power for a given wind speed could not imply the maximum
electrical power. In this paper, the maximum electrical power Pmax delivered by the SWTG is obtained by
using (3), (7), (8), (10), (12) and (15) with an algorithm of maximization [0 in Matlab (Mathworks Inc.,
Natick, US). The result is shown in Fig. 6, where the curve called “opt.” is that obtained from (15), and the
curve called “max.” is obtained with the optimization process mentioned above. Furthermore, the maximum
wind speed vlim at which the power delivered by the inverter can have the values given in (15) is calculated
with the equation:
λ opt 1 1
vlim = K e (17)
RAρC p ηmec 3Ls p
For the data shown in TABLE I, this value is 10.23 m/s which implies an electrical power of approx. 4855W
that is lower than the nominal one (see Fig. 6).

6000

5000
power in kW

4000

3000

2000
max.
1000 opt.
nom.
0
4 6 8 10
wind speed (m/s)

Fig. 6: Maximum power and optimum power curves

5
 EPE Wind Energy Seminar 2010

Power Curve Calculation: polynomial approximation

In the previous paragraphs, the maximum achievable power Pmax has been obtained. This power can be used
as reference for the inverter, so the maximum energy from wind can be obtained. To be implemented in the
inverter, this curve can be approximated by a third degree polynomial equation, as that shown in (14). In
order to be implemented in the inverter, the input values must be the frequency fr or the DC voltage UDC. By
means a least-squared based algorithm, the approximated polynomial curves can be obtained, as shown Fig.
7. The different between the curves is quasi negligible, so the power achieved by the SWTG with this
inverter power curve must be very close to the maximum one. For this reason, the polynomial approximation
is taken as reference in this paper.

6000
power in kW

4000

2000
polynomial approx.
max.
0
100 150 200 250 300 350 400
DC voltage (V)

6000
power in kW

4000

2000
polynomial approx.
max.
0
10 15 20 25 30 35 40 45 50 55 60
frequency (Hz)

Fig. 7: Maximum power curve and its polynomial approximation.

Power Curve Calculation: Polyline approximation
In many commercial inverters for WECS, the power curve must be defined as a set of vertices of a polyline,
being the difference between different manufacturers the number of vertices allowed. The way to obtain
these points and the size of the set determine de WECS behavior. In this paper, an optimization method to
obtain these points is presented. The process to obtain the optimum set of (Xi, Pi) vertices has the following
restrictions:
1) All the vertices must form an increasing curve.
X 0 < X1 < ... < X n
(18)
P0 < P1 < ... < Pn
2) Power associated to the last vertice must be the nominal one Pnom.
Pn = Pnom (19)
3) The maximum distance di between each segment formed by the vertices {(Xi, Pi); (Xi+1, Pi+1)} and
the Pmax curve must have a value lower than a given value -dmin (see Fig. 8).
Finally, the polyline formed by the vertices (Xi, Pi) is approximated to the maximum power curve by a
least-squares based algorithm with the mentioned above restrictions.
Pmax Pmax
Xi+1,Pi+1
Power in W
Power in W

di>0 Xi+1,Pi+1

di<0

Xi,Pi Xi,Pi

DC voltage or frequency DC voltage or frequency

Fig. 8: Distance between Pmax curve and the polyline defined by (Xi,Pi) points.

6
 EPE Wind Energy Seminar 2010

The resulting vertices, i.e., the polyline approximation of Pmax curve, is shown in Fig. 9 for a set size of two
vertices and using DC UDC voltage as input. Similar results can be obtained using frequency fr as input.

6000

power in kW
4000

2000
polyline approx.
max.
0
100 150 200 250 300 350 400
DC voltage (V)

Fig. 9: Maximum power curve and its approximation by means a polyline defined by two vertices.

Results
Once the inverter power curves have been obtained (polynomial and polyline approximation), they must
tested comparing their wind energy production in different wind regimes. For this purpose, a probability
distribution of Rayleigh has been used for wind speed [16].
Results are shown in TABLE II, where the resulting mean power production has been obtained for different
mean wind speeds (3-9 m/s), using the polynomial approximation and the polyline approximation (with 2, 3
and 4 vertices). The difference between de mean power obtained with the polyline approximation of the
inverter power curve and the mean power obtained with the polynomial approximation one is also shown in
the table. The difference between them is only significant when the wind speed is low (capacity factor < 0.2)
and the number of vertices (N. points) used in the polyline approximation is two. In Fig. 10 the mean power
ratio (mean power obtained with the polyline approx. / mean power obtained with the polynomial approx.) is
presented when the number of vertices for the polyline approximation is only two.
TABLE II. Mean power results
Mean Polynomial Polyline approx.
Wind approx. N. points: 2 N. points: 3 N. points: 4
Speed Mean Power Mean Mean Mean
Diff. Diff. Diff.
(m/s) (W) Power (W) Power (W) Power (W)

3 278 ,48 25 2,78 -9,23% 277,75 -0,26% 278,48 0, 00%

4 680 ,63 65 3,53 -3,98% 679,41 -0,18% 680,63 0, 00%
5 1.235 ,16 1.21 1,40 -1,92% 1 .233,72 -0,12% 1.235,16 0, 00%
6 1.845 ,49 1.82 5,75 -1,07% 1 .844,07 -0,08% 1.845,49 0, 00%
7 2.415 ,04 2.39 8,83 -0,67% 2 .413,74 -0,05% 2.415,04 0, 00%
8 2.879 ,37 2.86 6,00 -0,46% 2 .878,21 -0,04% 2.879,37 0, 00%
9 3.208 ,93 3.19 7,81 -0,35% 3 .207,92 -0,03% 3.208,93 0, 00%

mean power polyline approx. / mean power polynomial approx.

1

0.95

0.9
3 4 5 6 7 8
mean wind speed (m/s)

Fig. 10: Mean power ratio

Apart from energy production, the working limits of power curve are an important issue because they affect
to the WECS dynamic behavior (see TABLE III). The relationship between maximum and minimum values
of DC voltages or frequencies of the power curve is a measure of available capability of the inverter to
follow the power curve. In this case, the number of points of the polyline approximation is three; its behavior
is very close to that with the polynomial approximation.

7
 EPE Wind Energy Seminar 2010

TABLE III. Voltage and frequency limits

Polynomial Polyline approx.
approx. N. points: 2 N. points: 3 N. points: 4
Vdc Min (V) 103 205 120 10 6
Vdc Max (V) 356 361 358 36 1
Vdc Max/Min 3,5 1,8 3,0 3 ,4
fr Min (Hz) 11,5 20,5 13,0 1 1,5
fr Max (Hz) 45,2 45,4 45,4 4 5,4
fr Max/Min 3,9 2,2 3,5 3 ,9

Another question that must be considered is how the WECS behavior is affected by the variation of PMSG
parameters. In this case, variations of ±10% and ±20% have been supposed in the steady estate inductance Ls
and in the internal voltage E (derived from the no-load voltage). The inverter power curves (fr-P and VDC-P)
obtained by polyline approximation (n. points: 2) are compared to the maximum available power Pmax at
different Ls and E values.
In Fig. 11 and Fig. 12 (see left) the resulting maximum power curves and the polyline power curve at
different Ls and E values are shown. In these cases, DC voltage is used as input for the power curve. In Fig.
11 and Fig. 12 (see right) a similar analysis is shown but using frequency as input. Main conclusions are:
• Polyline inverter power curve must be designed to manage PMSG parameters variations, because
when Pmax curve is lower than the polyline one, unstable situations can appear during SWTG
operation.
• Variations on steady state inductance Ls are more significant when frequency is used as input in the
power curve.
• Variations on open voltage E are more significant when DC voltage is used as input in the power
curve. In this case, to avoid unstable conditions the polyline power curve must be gently displaced to
higher voltages.
max. max.
6000 ∆Ls+10% 6000 ∆Ls+10%
∆Ls-10% ∆Ls-10%
∆Ls+20% ∆Ls+20%
∆Ls-20% ∆Ls-20%
5000 polyline approx. 5000 polyline approx.

4000 4000
power in kW

power in kW

3000 3000

2000 2000

1000 1000

0 0
100 150 200 250 300 350 400 10 15 20 25 30 35 40 45 50
DC voltage (V) frequency (Hz)

Fig. 11: Maximum power with Ls variations and polynomial power curve with DC voltage (right) and
frequency (left) in the X-axis.

8
 EPE Wind Energy Seminar 2010

max. max.
6000 ∆Eg+10% 6000 ∆Eg+10%
∆Eg-10% ∆Eg-10%
∆Eg+20% ∆Eg+20%
∆Eg-20% ∆Eg-20%
5000 polyline approx. 5000 polyline approx.

4000 4000
power in kW

power in kW
3000 3000

2000 2000

1000 1000

0 0
100 150 200 250 300 350 400 10 15 20 25 30 35 40 45 50
DC voltage (V) frequency (Hz)

Fig. 12: Maximum power with E variations and polynomial power curve with DC voltage (right) and
frequency (left) in the X-axis

Conclusions
In this paper a SWTG with a WECS formed by a rectifier, dump load resistors and an inverter is
considered.
The paper is focused in the way that the power curve is implemented in the inverter in order to optimize the
wind energy production. The following situations have been analyzed:
• Power curve is implemented by means a third degree polynomial function;
• Power curve is implemented by means a set of point, i.e., a polyline approximation;
• Voltage in the DC bus is used as input for the power curve;
• The frequency in the PMSG side is used as input.
The polynomial approximation has been calculated using a least-squared based algorithm. For the polyline
approximation a optimization method is proposed that takes into account the distance between the maximum
power curve and the proposed polyline approximation.
Using a third degree polynomial function a good agreement between maximum power curve and the
approximated power curve is obtained. However, many commercial inverters do not use this method to
implement its power curve and the curve is implemented by means a polyline.
When approximated power curves to be implemented in the inverter have been obtained different analysis
have been done. Firstly, the impact of the number of vertices used to define the polyline is evaluated. When
this number is greater than two a good approximation power curve is obtained.
The margin of input values (frequency or DC voltage) is also evaluated as function of the number of
vertices. In this case the behavior of the polynomial approximation and the polyline one are very close when
the number of vertices is greater than three.
Finally, the impact of variation in the PMSG parameters is considered. This impact depends on which input
the inverter power curve has. When the input is the DC voltage, the inverter curve must be gently displaced
to avoid intercept the maximum power curve; which may derive in unstable behavior. So, using frequency as
input for the inverter power curve a more stable can be expected against different PMSG parameters.

References
[1] Patsios, C., Chaniotis, A., Rotas, M., Kladas, A.G.: A Comparison of Maximum-Power-Point Tracking Control Techniques for
Low-Power Variable-Speed Wind Generators, 2009 8th International Symposium on Advanced Electromechanical Motion Systems
and Electric Drives Joint Symposium, ELECTROMOTION 2009
[2] Tan, K., Islam, S.: Optimum Control Strategies in Energy Conversion of PMSG Wind Turbine System without Mechanical
Sensors, IEEE Transactions on Energy Conversion 19 (2), pp. 392-399, 2008
[3] Yaoqin, J., Zhaoan, W., Zhongqing, Y.: Experimental Study of Control Strategy for Wind Generation System, PESC Record -
IEEE Annual Power Electronics Specialists Conference, art. no. 4342164, pp. 1202-1207, 2007

9
 EPE Wind Energy Seminar 2010

[4] Haque, M.E., Negnevitsky, M., Muttaqi, K.M.: A Novel Control Strategy for a Variable Speed Wind Turbine with a Permanent
Magnet Synchronous Generator, Conference Record - IAS Annual Meeting (IEEE Industry Applications Society) , art. no. 4659162,
2008
[5] Haque, M.E., Muttaqi, K.M., Negnevtsky, M.: A Control of a Stand Alone Variable Speed Wind Turbine with a Permanent
Magnet Synchronous Generator, IEEE Power and Energy Society 2008 General Meeting: Conversion and Delivery of Electrical
Energy in the 21st Century, PES , 2008
[6] Hong, Y.-Y., Lu, S.-D., Chiou, C.-S.: MPPT for PM wind generator using gradient approximation, Energy Conversion and
Management 50 (1), pp. 82-89, 2009
[7] Bharanikumar, R., Yazhini, A.C., Kumar, A.N.: Novel Maximum Power Point Tracking Controller for Wind Turbine Driven
PMG, 2008 Joint International Conference on Power System Technology POWERCON and IEEE Power India Conference,
POWERCON 2008
[8] González, L.G., Figueres, E., Garcerá, G., Carranza, O., González-Espín, F.: Synchronization techniques comparison for
sensorless control applied to Wind Energy Conversion Systems, 2009 13th European Conference on Power Electronics and
Applications, EPE '09, 2009
[9] Chinchilla, M., Arnaltes, S., Burgos, J.C.:“Control of Permanent-Magnet Generators Applied to Variable-Speed Wind-Energy
Systems Connected to the Grid, IEEE Transactions on Energy Conversion 21 (1), pp. 130-135, 2006
[10] Wei, L., Abbey, C., Joós, G.: Control and Performance of Wind Turbine Generators based on Permanent Magnet Synchronous
Machines Feeding a Diode Rectifier, PESC Record - IEEE Annual Power Electronics Specialists Conference,
[11] Tang, Le, Zavadil, Robert: Shunt capacitor failures due to windfarm induction generator self-excitation phenomenon, IEEE
Transactions on Energy Conversion, vol. 8, n. 3, pp. 513-519, 1892.
[12] González, L.G., Figueres, E., Garcerá, G., Carranza, O.: Modeling and Control in Wind Energy Conversion Systems (WECS),
2009 13th European Conference on Power Electronics and Applications, EPE '09 , 2009
[13] Gieras, JF; Mitchell, WI: Permanent magnet motor technology. Design and Application, Marcel Dekker, New York, 1892.

[14] Pillay P.; Knshnan, R.: Modeling of Permanent Magnet Motor Drives, IEEE Transactions On Industrial Electronics, vol. 35,
no. 4, pp. 537-541, 1988
[15] Forsythe, G. E.; Malcolm, M.A. and Moler, C. B.: Computer Methods for Mathematical Computations, Prentice-Hall, 1976.

[16] Carta, J. A.; Ramírez, P. and Velázquez, S.: A review of wind speed probability distributions used in wind energy analysis. Case
studies in the Canary Islands, Renewable and Sustainable Energy Reviews, nº. 13, pp. 933-955, 2009.

10

View publication stats