Beruflich Dokumente
Kultur Dokumente
Renewable Energy
journal homepage: www.elsevier.com/locate/renene
Universidade Federal do Par, Faculdade de Engenharia Mecnica, Av. Augusto Correa, s/n e Belm, PA 66075-900, Brazil
Universidade Federal do Par, Faculdade de Engenharia Eltrica, Av. Augusto Correa, s/n e Belm, PA 66075-900, Brazil
Universidade de Brasilia, Faculdade de Tecnologia, Departamento de Engenharia Mecnica, Av. L3 Norte, Asa Norte, Braslia, Distrito Federal, Cep. 70.910-900, Brazil
a r t i c l e i n f o
a b s t r a c t
Article history:
Received 28 August 2012
Accepted 8 December 2012
Available online 30 January 2013
This paper presents a model to optimize the distribution of chord and twist angle of horizontal axis wind
turbine blades, taking into account the inuence of the wake, by using a Rankine vortex. This model is
applied to both large and small wind turbines, aiming to improve the aerodynamics of the wind rotor,
and particularly useful for the case of wind turbines operating at low tip-speed ratios. The proposed
optimization is based on maximizing the power coefcient, coupled with the general relationship between the axial induction factor in the rotor plane and in the wake. The results show an increase in the
chord and a slightly decrease in the twist angle distributions as compared to other classical optimization
methods, resulting in an improved aerodynamic shape of the blade. An evaluation of the efciency of
wind rotors designed with the proposed model is developed and compared other optimization models in
the literature, showing an improvement in the power coefcient of the wind turbine.
Published by Elsevier Ltd.
Keywords:
Aerodynamic optimization
Wind turbine
Wind energy
BEM model
1. Introduction
In wind turbine design, the maximization of the power coefcient is of fundamental importance in order to optimize the
extraction of energy from the wind. This paper addresses the
problem of the aerodynamic optimization of a horizontal axis wind
turbine rotor, considering the search of optimum shape design of
the blade. The optimum design of wind turbine can be achieved
from three different approaches that describe the energy conversion in the turbine blades. The rst one uses the classical Blade
Element Model (BEM), which describes this energy conversion by
means of force and moment balances in radial sections of the
blades (Glauert theory [7,18]). The second approach is based in
vortex methods [1], and the last methodology uses the modern
uid ow simulation tools (CFD approach [14]).
Presently, an important effort has been devoted to the development of methodologies of optimization using advanced genetic
algorithm and evolutionary computation [1,4,6] coupled to CFD
tools [14]. Global optimum design can be achieved by the use of
those methodologies, where each blade analysis requires a CFD
D.A. Tavares Dias do Rio Vaz et al. / Renewable Energy 55 (2013) 296e304
wind turbine blades, taking into account the inuence of the wake,
by using a Rankine vortex model, as proposed by Wilson and Lissaman [18]. This model is applied to both large and small wind
turbines, aiming to improve the aerodynamics of the wind rotor.
This is particularly useful for the case of wind turbines operating at
low tip-speed ratios, as for example the ones with multibladed
rotors. The proposed optimization is based on maximizing the local
power coefcient, coupled to the general relationship between the
axial induction factor in the rotor plane and in the wake. The results
show a modication in the chord and twist angle distributions,
resulting in an improved aerodynamic shape of the blade. An
evaluation of the efciency of wind rotors designed with the proposed model is developed and compared with Glauert [7] and
Stewart [15] optimization models. The results also show an
improvement in the extraction of wind energy due to the change in
the aerodynamic shape of the wind blade.
b
Bc
Cn
1a
4pr Fsin2 f
(5)
b0
Bc
Ct
1 a0
4pr Fsinfcosf
(6)
where a0 and b0 are the tangential induction factors at the rotor and
factor at the rotor wake, respectively, and dened by
a0
2. Mathematical model
b0
V0 v uh1 aV0
V0 v1 u1 h1 bV0
V u
a 0
V0
b
V0 u1
V0
(2)
(3)
(7)
2U
u1
(8)
2U
(1)
Ct
Fn
CL cosf CD sinf
1
rW 2 cdr
2
Ft
CL sinf CD cosf
1
rW 2 cdr
2
b2 1 a2
ba
(4)
Fig. 1. Simplied illustration of the velocities in the rotor plane and in the wake.
(9)
(10)
where W is the relative velocity, r is the air density, Cl and Cd are the
lift and drag coefcients, which are usually obtained from wind
tunnel tests or numerical methods [5], and F is the Prandtl tip-loss
factor, as described in Ref. [17], which is dened as the ratio between the bound circulation of all blades and the circulation of
Cp
297
298
D.A. Tavares Dias do Rio Vaz et al. / Renewable Energy 55 (2013) 296e304
RU
V0
(11)
where R is the rotor radius. This result suggests that a more detailed
investigation for the validity of the use of the Prandt correction in
the case of X < 2 and for turbine ow is necessary. However, this
analysis is beyond the scope of this paper and will be considered in
future works.
Wilson and Lissaman [18] showed that it is possible to establish
a general relationship between the axial induction factors a and
b from the application of the continuity, momentum and energy
equations for the induced velocities in the streamtube shown in
Fig. 1. Thus, considering the hypothesis that the wake behaves like
a free vortex, Wilson and Lissaman [18] demonstrated that the axial
induction factor in the rotor plane has a non-linear relationship
with the axial inducing factor in the wake for low tip-speed ratios,
X, especially for values X < 2, as shown in Fig. 3. This relationship is
dened by:
b
b2 1 a
1
2
4X 2 b a
(12)
Mesquita and Alves [12] showed that Eq. (12) can be rearranged
in the form of a complete cubic equation in b, where only one of the
roots shows a consistent behavior with the physical constraints of
the problem. It should be noted that Eq. (12) can also be solved
numerically, with good results as shown in Ref. [16]. In this case the
analytical solution to Eq. (12) is
1
1
1 p
b S T a1 i
3S T
2
3
2
(13)
where
r
q
3
S
Z Q 3 Z2
r
q
3
Z Q 3 Z2
3a2 a21
9
(16)
(17)
a1
4X 2
a1
(18)
a2
12aX 2
1a
(19)
a3
8a2 X 2
a1
(20)
Cp
(14)
b1 a2
2Na 1 Nb
ba
0.8
0.6
X = 0.25
X = 0.50
X = 0.75
X = 1.00
X = 2.00
X = 3.00
X=
0.4
0.2
0
0
0.1
0.2
0.3
0.4
0.5
a
Fig. 3. Relation b/a for some values of X.
(15)
(21)
D.A. Tavares Dias do Rio Vaz et al. / Renewable Energy 55 (2013) 296e304
vq
Gr
1=n
2n
p
2 rc r n
(22)
where vq is the tangential velocity component, G is the vortex circulation strength, rc is the vortex core radius, n is an arbitrary
integer, and r is the local radial coordinate for the vortex tangential
velocity prole. For n 1 the Scully vortex is reproduced, for n 2
one obtains the Bagai-Leishaman vortex model, and for n/N then
Vatistas model reduces to the Rankine vortex model.
Young [19] has performed an extensive review on existing
experimental works available in the literature and, using a kinetic
energy conservation approach, concludes that all forms of the
Vatistas vortex model, included the Rankine vortex prole, match
reasonably well with the experimental data for the tip vortex. This
result is comprehensive because in the model implementation the
vortex strength or the core size is often adjusted on a trial and error
bases with experimental data. However, as already commented, for
the hub vortex the available experimental data are scarce.
On the theoretical analysis, this problem was analyzed by some
authors after the pioneer work from Wilson and Lissaman [18] in
the middle of the 70s. Wood [24] uses Vatistas vortex with n 1
and concludes that this model does not have a signicant effect on
the basic analysis that leads to the Betz limit [2], provided that the
core radius of the hub vortex is sufciently small and the tip-speed
299
ratio is sufciently high. For low tip-speed ratios the results become
not physically possible with innite values for the power coefcient
when the tip-speed ratio decreases to zero. Recently, Srensen and
Kuik [25] proposed a model that remedies this problem by
including the contribution from the lateral pressure and friction
forces in the axial momentum theorem, with result that the power
coefcient never exceeds the Betz limit [2] and tends to zero at zero
tip-speed ratio. Since the friction force and pressure on the lateral
boundary on the control volume are not known a priori, the authors
proposed a model, in which this force is proportional to the wake
expansion area and to the tip pressure drop multiplied by a small
coefcient giving the net inuence of the integrated pressure acting
on the lateral boundary of control volume. The determination of
this coefcient is yet open in this methodology.
Finally, it is important to analyze the inuence of the wake
rotation in the radial pressure gradient. Srensen and Kuik [25]
showed that the radial pressure gradient in the far wake is given by:
vp1
1 2
vu
r
vq u1 1
vr1
vr1
r1
(23)
where p1 and r1 are the pressure and the local radial coordinate in
the far wake, respectively. This gradient has to deliver the centripetal acceleration rv2q =r1 . This is satised when vu1/vr1 0.
Consequently, u1 is constant in the fully developed wake. This result
allows the determination of a relationship between axial induction
factors in the rotor plane and in the wake, respectively. Therefore,
according to Srensen and Kuik [25], applying a combination of the
momentum, energy and mass balance equation gives:
b2 1 a
2Xb a
(24)
u
q2 2Xq
1a
V0
2b
(25)
Combining Eq. (24) with Eq. (25), one obtains Eq. (12). For the
power coefcient, Srensen and Kuik [25] showed that:
Cp 2Xq1 a
(26)
Substituting Eq. (24) in Eq. (26), one obtains Eq. (4). These results, show that the study developed by Srensen and Kuik [25],
about momentum theory at low tip-speed ratio, agree with the
300
D.A. Tavares Dias do Rio Vaz et al. / Renewable Energy 55 (2013) 296e304
db
db
d
2 Na3 2 2bN
bN 1 N a2 2b N 1
da
da
d
db
db
2
2
N 1 a b 1
N 1 N
b 1 7N
da
da
2b3 N 1 0
(27)
Dividing Eq. (27) by the term 2Ndb/da gives
a3 d1 a2 d2 a d3 0
copt
4prbFsin2 f
BCn 1 aopt
db
bN 1 N
da
db
N
da
2b
d2
db
db
N 1 b2 1 7N
N 1
da
da
db
2N
da
(38)
a0opt aopt
tanf
x
(39)
b0opt b
2bN
(37)
(28)
where
d1
Z*
(29)
(40)
aopt
The optimum twist angle is obtained from aopt the velocity diagram shown in Fig. 2, by Eq. (41). The velocities diagram was
obtained according to the classical theory of turbomachinery.
bopt fopt a
db
b2 1
N 1 N 2b3 N 1
da
d3
db
2N
da
a0opt
(41)
(30)
1
fopt tan
(31)
"
#
1 aopt
1 a0opt x
(42)
8a2 X 2 b 16aX 2 b b1 b 8X 2
db
da
4a2 X 2 b 8aX 2 b 3a1 a 2b1 a 4X 2
(32)
Note that Eq. (27) is a complete cubic equation in a, where only
one root presents a physically consistent solution, in a similar way
as demonstrated by Mesquita e Alvez [12]. The solution of Eq. (27) is
given by:
1
1
1 p
aopt S* T* d1 i
3S* T*
2
3
2
(33)
where
r
q
3
S*
Z* Q*3 Z*2
T*
(34)
r
q
3
Z*
Q*3
Z*2
(35)
Q*
3d 2 d 1
9
(36)
D.A. Tavares Dias do Rio Vaz et al. / Renewable Energy 55 (2013) 296e304
301
0.2
0.8
0.15
CD
0.6
0.1
0.4
0.05
0.2
10
0
0
15
10
15
Fig. 7. (a) Lift coefcient and (b) drag coefcient of the NACA 0012 [27].
1.6
1.4
1.2
Induction Factor
0.5
Cp
0.3
0.2
0.1
0
0
b
b
a
a
b = 1.3633
1
b = 0.7375
0.8
a = 0.6196
0.6
a = 0.3352
0.4
N=0
N=1
N=2
10
20
Number of iterations
30
(43)
0.7
0.4
1 vrvq 1 vvr !
!
!
e z zz e z 2wz e z
r vr
r vq
0.2
40
10
15
Number of iterations
20
Fig. 8. Convergence of the method: (a) Power coefcient. (b) Induction factors in the rotor plane and in the wake for the optimized procedure for X 1:57 and N 0.
302
D.A. Tavares Dias do Rio Vaz et al. / Renewable Energy 55 (2013) 296e304
Table 1
Estimates for N from measurements of Hu et al. [9].
X
zz max
wmax
3.0
3.5
4.0
4.5
94.5
110.2
126.0
141.7
50
50
50
50
25
25
25
25
3.8
4.4
5.0
5.7
Table 2
Estimates for N from measurements of Whale et al. [21].
X
zn max
zz max
wmax
3.0
4.0
5.0
6.0
8.0
8.6
11.4
14.3
17.1
22.9
0.8
0.8
0.8
0.8
0.8
6.7
6.7
6.7
6.7
6.7
3.3
3.3
3.3
3.3
3.3
2.6
3.4
4.3
5.1
6.9
4z d
zn z
V0
Cp(%)
P(W)
(44)
0.3
Glauert
Stewart
N 0
N 1
N 2
29.17
278
29.22
278
35.00
334
34.78
331
34.28
327
40
Stewart (1976)
Glauert (1926)
N=0
N=1
N=2
35
0.25
30
c/R
o
( )
0.2
0.15
0.1
0.05
0.3
Stewart (1976)
Glauert (1926)
N=0
N=1
N=2
0.4
0.5
25
20
15
0.6
0.7
0.8
0.9
10
0.3
0.4
0.5
r/R
Fig. 9. (a) Chord distributions. (b) Twist angle distributions.
0.6
0.7
r/R
0.8
0.9
D.A. Tavares Dias do Rio Vaz et al. / Renewable Energy 55 (2013) 296e304
0.4
303
350
0.35
300
P (W)
Cp
0.3
0.25
0.2
Stewart (1976)
Glauert (1926)
N=0
N=1
N=2
0.15
0.1
1
1.2
1.4
1.6
250
Stewart (1976)
Glauert (1926)
N=0
N=1
N=2
200
150
5
1.8
5.5
6
V0 (m/s)
6.5
Fig. 10. (a) Power coefcient as a function of tip-speed ratio. (b) Output power as a function of speed.
0.8
0.7
0.7
0.6
Induction Factor
0.8
Cp
0.5
0.4
0.3
b
b
a
a
b = 0.667
0.5
b = 0.410
0.4
a = 0.333
0.3
a = 0.207
0.2
0.2
N=0
N=1
N=2
0.1
0
0
10
20
Number of iterations
30
0.1
0
40
10
15
Number of iterations
20
Fig. 11. (a) Power coefcient as a function of the number of iterations and (b) induction factors in the rotor plane and in the wake for the optimized procedure for X 4:18 and
N 0.
0.35
40
Stewart (1976)
Glauert (1926)
N=0
N=1
N=2
0.3
0.25
Stewart (1976)
Glauert (1926)
N=0
N=1
N=2
35
30
25
c/R
o
( )
0.2
0.15
20
15
0.1
10
0.05
0
0
5
0.2
0.4
0.6
0.8
0
0
0.2
r/R
0.4
0.6
0.8
r/R
Table 4
Power coefcient and output power for the designed wind turbines X 4:18.
Cp(%)
P(W)
Glauert
Stewart
N 0
N 1
N 2
37.23
631
36.10
612
39.57
671
39.57
671
39.57
671
304
D.A. Tavares Dias do Rio Vaz et al. / Renewable Energy 55 (2013) 296e304
0.4
650
0.35
600
0.3
550
P (W)
Cp
700
0.25
0.2
Stewart (1976)
Glauert (1926)
N=0
N=1
N=2
0.15
0.1
2.5
3.5
4.5
500
Stewart (1976)
Glauert (1926)
N=0
N=1
N=2
450
400
350
5
5.5
6
V0 (m/s)
6.5
Fig. 13. (a) Power coefcient as a function of tip-speed ratio. (b) Output power as a function of speed.
The output power obtained using the rotor designed with the
proposed model is around 12.73% higher than those obtained with
the Glauert [7] and Stewart [15] optimizations models, for a wind
speed of 6.5 m/s (see Fig. 13b). The curves for N 0, N 1, and N 2
coincide. This result was expected, because for high tip-speed ratios
the model present the same response.
4. Conclusions
The proposed mathematical model represents a simple alternative tool for the optimal design of wind rotors, especially for turbines with multibladed rotors, where the proposed optimization can
improve the efciency for both high and low tip-speed ratios, in this
case, an improvement of 18% was obtained for X < 2, when compared
with the classical models. The main advantage of the proposed
model is to take into account the effects of the wake rotation using
a Rankine vortex model. The method shows better efciency when
compared to the Glauerts [7] and Stewarts [15] optimizations, satisfying the condition described by Betz [2], where the maximum
energy to be extracted from the ow is 59.26% of its original energy.
For low tip-speed ratios the behavior of the axial induction factors in
the wake is completely non-linear and needs to be considered. This
fact considers the slow operation of a turbine, commonly used in
water pumping systems. The comparisons show that the developed
model improves the aerodynamics of the wind rotor. However, some
limitations need to be considered, such as the difculty in accurately
determining the maximum rotation of the wake vortex formed
behind the rotor, which indicates that more detailed experiments on
the wake structure for the hub region are necessary, as well as the
validity of the Prandt tip-loss correction, where more investigation is
also needed. For high tip-speed ratios, the presented model shows
that the inuence of the wake rotation is negligible.
Acknowledgments
The authors would like to thank CNPq (480135/2010e0), PROPESP/UFPA (EDITAL PARD 04/2011), GEDAE, INCT e EREEA and
ELETRONORTE for nancial support.
References
[1] Benini E, Toffolo A. Optimal design of horizontal-axis wind turbines using
blade-element theory and evolutionary computation. ASME Journal of Solar
Energy Engineering 2002;124:357e63.