You are on page 1of 10

Energy 148 (2018) 169e178

Contents lists available at ScienceDirect

Energy
journal homepage: www.elsevier.com/locate/energy

The effects of blade pitch angle on the performance of small-scale


wind turbine in urban environments
P.A. Costa Rocha a, *, J.W. Carneiro de Araujo a, R.J. Pontes Lima a, M.E. Vieira da Silva a,
D. Albiero a, C.F. de Andrade a, F.O.M. Carneiro b
a
Solar Energy and Natural Gas Laboratory, Campus do Pici, Bl. 718, 60455-760, Fortaleza, CE, Brazil
b
University for International Integration of the Afro-Brazilian Lusophony, Palmares Academic Unit, Rodovia CE 060, Km 51, Acarape, CE, Brazil

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

Article history: Due to the growing importance of wind power as a clean and renewable energy source, the use of small-
Available online 19 January 2018 scale wind turbines in urban environments has increased lately. The blade pitch control is an effective
method to improve the aerodynamic response of a wind turbine, usually applied to large-scale wind
Keywords: turbines. This study presents the effects of varied blade pitch on the aerodynamic performance of a
Small-scale wind turbine small-size wind turbine. The blades were sketched out according to the Blade Element Moment (BEM)
Pitch angle control
theory, applying the aerodynamic profile NREL-S809 and designed for a tip speed ratio of eight. To
Urban environment
analyze the influence of the blade pitch angle on the energy conversion, a comparative study was carried
Blade element momentum theory
ANOVA
out varying the pitch angle to five different values. Using the analysis of variance (ANOVA), it was
possible to demonstrate that blade pitch control could be an effective method also for small-sized wind
turbines. A performance chart from the results of blade pitch experiments shows that the power coef-
ficient varies significantly when the angle changes. As conclusion, it is highlighted that an enhanced
behavior could be attained by the use of a pitch angle controller resulting in a better recovery of the
energy available in the wind.
© 2018 Elsevier Ltd. All rights reserved.

1. Introduction contained in flowing air into electricity [4].


Another way of producing electricity from renewable sources is
During the last decade, the interest in renewable energy sources through the use of photovoltaic (PV) technology, in which the
(RES) has been increasing all over the world due to concerns about incident solar radiation is directly converted into electrical energy,
the limited fossil fuel supply, and their harmful impact on the according to the principle of photoelectric effect. Despite the great
environment [1]. The support of RES is considered, besides the gain performance and cost efficiency of photovoltaic (PV) panels tech-
in power efficiency, a key measure to achieve the climate goals [2], nology in the electricity production, the use of wind turbines has
such as the reduction of CO2 emissions. Consequently, the share of been grown in the last years. For instance, throughout 2015 the
RES in primary energy supply has been intensified as a means to installed wind power across Europe outgrows 13,800 MW, that
minimize the impact of conventional energy generation on the means 5,4% more than in the previous year [5]. German was the
environment. largest market in terms of annual installations, setting up
The electric power is an essential factor for the development 6013.4 MW of new capacity.
and the evolution of human society whether in the field of the As a result of this growth, the wind turbines technologies are
living conditions improvement or in the industrial activities [3]. For being increasingly developed. New control methods have been
this reason, its production and distribution must be accessible to all proposed to reduce structural damage caused by high wind speeds
the people in many regions. One way to generating electricity from and to enhance the wind turbine efficiency, intensifying their use,
renewable sources is to use wind turbines that convert the energy for example, in high turbulence environments. Also, urban energy
generation has been grown. According to Bahaj et al., in contrast to
the traditional centralized energy supply, micro-generation tech-
nologies bring power generation close to the user to sustain their
* Corresponding author.
homes or buildings [6]. Economic studies have also demonstrated
E-mail address: paulo.rocha@ufc.br (P.A.C. Rocha).

https://doi.org/10.1016/j.energy.2018.01.096
0360-5442/© 2018 Elsevier Ltd. All rights reserved.
170 P.A.C. Rocha et al. / Energy 148 (2018) 169e178

the viability of the domestic uses of wind power, mainly in urban turbine in urban environments.
areas [2,7,8]. Following this trend, several works are devoted to
small wind turbines acting in urban environments [6,9e14]. 2. Experimental methods
According to Lubitz [10], ambient turbulence and wind direction
variance both have significant impacts on small wind turbines. 2.1. Blade shape and materials
Several attributes of specific urban surroundings, such as the height
of buildings and shape (the geometric feature), contribute to The blade assembly constitutes one of the main components of a
generating typical ambient turbulence, which distinguishes it from wind turbine. The wind turbine efficiency depends on the capa-
open areas of wind farms. The wind field in the urban areas is bility of these devices to extract kinetic energy from the wind,
mainly impacted by the dynamics of airflow in the surroundings of converting it to a rotary motion. For that, the blades should be
buildings. Lubitz [10] highlights that the effect of turbulence on specially designed to transform the force of the wind into the tor-
power output is more difficult to generalize because turbulent que needed to generate useful power. Thus, the blade design is
gusts affect different features such as wind alignment, airfoil per- focusing onto maximize the rotor power output and to minimize
formance and power limiting. Zanforlin and Letizia [9] demon- the losses, ensuring higher performance.
strated that in the realistic urban context the wind approaches the There are many things to consider in designing blades, but most
roof with a strong vertical component (skewed wind). Grieser et al. of them fall into one of two categories: aerodynamics and structure
[2] exemplify the resulting turbulence as a vortex that mainly ap- [15]. Considering these two principal groups, some design param-
pears at the roof edges, in which the air streams impinge on the eters must be defined before starting the project, for instance, the
building wall and cause a so-called “separation bubble” above the operating mode, aerodynamics airfoil design, tip speed ratio,
roof edge. number of blades, and manufacturing materials.
The power output of a wind turbine varies with wind speed, and In this work, a variable speed-operating mode was applied to a
every wind power converter has a characteristic power perfor- three-bladed rotor. The three-bladed rotor was chosen considering
mance curve [15]. The cut-in speed is the minimum wind speed the remarkable applicability and proven efficiency of this rotor
necessary above which the machine will deliver useful power. In type. Three blades have the particular advantage that the polar
the same sense, the cut-off speed, usually limited by the engi- moment of inertia with respect to yawing is constant, and is in-
neering design and safety constraints, is the maximum wind speed dependent of the azimuthal position of the rotor [6]. According to
at which the turbine is allowed to deliver power. When the wind Hau [21], the three-bladed rotor type is the one that presents the
velocity reaches and overpasses the cut-off speed, the output po- best relation between the power coefficient (Cp) and the tip speed
wer must be limited reducing the aerodynamic efficiency [15]. ratio (l). This rotor type achieves power coefficients about to 0.5 at
There are different ways to achieve the over-speed control. Some the tip speed ratio between 7 and 8 [21], that is close to the Betz
intermediate-sized turbines used fixed-blade pitch and stall con- limit (0.593) [22].
trol. Most manufacturers use pitch control, and the general trend is The blades made of balsa wood conform to the aerodynamic
the increased use of pitch control, especially in larger machines airfoil NREL-S809 created by the National Renewable Energy Lab-
[1,15,16]. oratory [23]. The NREL-S809 airfoil, shown in Fig. 1, was specifically
The blade pitch control is a useful method to reduce overload developed for horizontal-axis wind turbines to allow higher lift and
when wind speeds exceed the limit designed for the turbine. Pitch lower drag. Balsa wood has been chosen because it is very soft and
control is then used to regulate the rotor speed, with the load light material, which provides lighter blades and facilitates the
torque held constant [16]. Morten et al. investigated the use of construction of profiles. Twenty airfoils were used to build each
pitching to obtain the maximum power output of a wind turbine blade. Fig. 2 depicts taper ratio curve, as well as twist curve is
[17]. The authors proposed the use of an optimized pitching profile depicted in Fig. 3.
that results in reduction of force and root mean square (RMS) tor- The dimensions of designed blade preview the maximum power
que variation by over 96% percent compared to any constant pitch coefficient at a tip speed ratio equal to eight, for which the blade
profile. Other studies investigate varied ways to use the pitch length setup of 1.5 m resulting to 3.0 m diameter rotors. The hub
control systems as a means to maximize the aerodynamic perfor- diameter is 0.30 m that is 10% of the diameter rotor.
mance [18] and improve the turbine power [19] of large turbines.
Pitch control systems are widely used to large wind turbines, 2.2. Measurements
but not so often to small wind turbines [20], consequently few
works approaches this subject in the literature. However, the per- A prototype of a small wind turbine was specially fabricated to
formance of a small wind turbine can also be enhanced by the use carry out the performance experiments. The assembling consists of
of pitch angle control system, as shown in the study conducted to a three-bladed rotor that is coupled with a rotating shaft in which a
evaluate pitch controlled small turbines under tempestuous gust torque transducer was connected. In order to provide any safety
winds or strong winds caused by typhoons blow [20]. The authors interruption of experiments, the shaft was also equipped with a
demonstrated that the pitch angle control was effective in these mechanical brake. The prototype is composed of an upwind, hori-
turbines to change their rotational speed as a function of wind zontal axis wind turbine, which was mounted on a 5 m tall, four
velocities to avoid damages from strong winds. inches in diameter, carbon steel tubular tower. A 5 m tall additional
Similarly the study conducted by Nagai et al. [20], the present carbon-steel tower, two inches in diameter, was installed a few
study aims to investigate the influence of pitch angle variation on meters upwind to the wind turbine tower to serve as support to
small turbines under high turbulence, but herein it is evaluated in install the anemometer, as shown in Fig. 4, which is the same as
typical urban environments. For this, a study on the power coeffi- [24].
cient of a three-bladed wind turbine was carried out. The wind The wind speed and direction simultaneously were recorded by
turbine performance was analyzed for five different pitch angle using a sonic anemometer WindMaster model 1590-PK-20. This
adjustments applying statistical methods. In addition, this work device can perform measurements in the wind speed up to
introduces a direct experimental correlation between the changes 45 m s1, resolution of 0.001 m s1 and accuracy of <1.5% RMS. The
in pitch angle and the output power coefficient, which suggests wind direction ranges from 0 to 359.9 , with resolution of 0.1. The
that a blade pitch control would be applicable for a small wind selected sampling rate is either 20 Hz or 32 Hz, depending on the
P.A.C. Rocha et al. / Energy 148 (2018) 169e178 171

Fig. 1. Geometry of the NREL-S809 airfoil.

Fig. 2. Taper ratio curve.


Fig. 4. Prototype installations [24].

Fig. 5. Torque transducer and tachometer assembly.


Fig. 3. Twist curve.

Headquarters (Germany), measuring nominal torque up to 200 Nm,


configured output. nominal rotational speed up to 9000 rpm, accuracy class 0.5. The
The delivered power from the wind turbine was measured by Digital tachometer model TS-TADIG measures the axis rotation
tracking the torque and rotational velocity transmitted to the using a micro-controlled unidirectional pulse counter operating in
rotating shaft. In order to obtain these experimental measure- the 4 MHz range. The model has an inductive proximity sensor
ments, a torque transducer and a digital tachometer were coupled running in the range of 0e9999 rpm with 0.8% error.
to the rotating shaft positioned near to the three-bladed rotor, as The aerodynamic performance of the small wind turbine pro-
shown in Fig. 5. totype could be evaluated by comparing the power available in the
The Torque transducer is an HBM-T22 fabricated by HBM wind and the useful power delivered from the wind turbine. Using
172 P.A.C. Rocha et al. / Energy 148 (2018) 169e178

the three variables obtained from the wind convertor system, it was
possible to calculate these values and establish a correlation be-
tween the variation of the blade pitch angle and wind turbine
performance.
All devices provide analog outputs, which have been converted
into digital signals before the data transferring to a computer for
long-term storage and graphical analysis. A schematic view of the
experimental apparatus is shown in Fig. 6. Note that since the wind
simultaneously reaches both towers, data recorded from the
tachometer and torque transducer are synchronized to the
anemometer signal.

2.3. Instrument calibration Fig. 7. Torque standard machine.

Since analog outputs need to be converted into digital signals, all shaft speed measurement obtained by the two methods, the
devices must be calibrated before starting the measuring test. In- tachometer accuracy is determined. Tachometer calibration curve is
strument calibration checks the accuracy of instruments and pro- plotted applying several rotational shaft speeds as a function of
vides the output conversion curve to a digital signal. The analogic signal outputs from the tachometer. The maximum
proprietary manufacturer calibrated the anemometer beforehand tachometer accuracy was 0.8%, which is in accordance to the
and the following conversion equation was indicated: manufacturer catalogue.
h i
U ¼ 8$V  20 m s1 (1) 3. Experimental data

where U is wind speed, V is output voltage [V]. As previously mentioned, the main objective of this work is to
The torque transducer calibration evaluates the torque mea- analyze the impact of the pitch angle variation on the aerodynamic
surement using a lever deadweight system, i.e., a torque generation power output of a small wind turbine, under urban environmental
method using gravity. For this, the torque transducer coupled to the conditions. For this purpose, initially the nominal designed angle
shaft was assembled in a test bench. Triggering the mechanical set up the initial blade position at 25.63 , called “zero angle,” since
brake is prevented from any axis motion when applying dead- it was directly obtained by the BEM equations. During the experi-
weight on the lever. Fig. 7 shows the torque standard machine, in mental procedure this angle was varied four times. Twice the pitch
which the lever nominal full-length is 1 m. The lever was coupled to angle was turned in the clockwise direction, and twice moved in
the shaft at the central position to result null torque at initial counterclockwise rotation from the zero angle by discrete in-
calibration. Deadweight loading was positioned at 0.486 m of the crements of 3 . As a result, the blades have been positioned in five
lever center then the resultant torque was measured. pitch angles, i.e., 31.63 , 28.63 , 25.63 , 22.63 and 19.63 . In order
Torque transducer accuracy is determined by comparing the to simplify the sampling nomenclature and easy understand, each
measured torque with calculated value using the gravitational force angle was named by the normalized angular displacement from the
and lever length (t ¼ mgl). Since the device delivers proportional initial blade position (called of 0 ). Thus, the five experimental
analogic signals (voltage), it is possible to plot a torque calibration angles were called as 6 , 3 , 0 , þ3 , þ6 , corresponding the
chart varying the applied deadweight. Fig. 8 depicts the calibration pitch angles of 31.63 , 28.63 , 25.63 , 22.63 and 19.63 . Experi-
curve showing the torque as a linear function of the output voltage. mental data obtained for each one of these five positions were
Applying similar methodology, a tachometer calibration curve recorded. Each dataset contains the instantaneous wind speed (U),
(Fig. 9) was plotted using stroboscopic flashlight to check the axial the rotational speed (U) and the torque (t) delivered from the wind
rotation of a shaft coupled to an electric motor. Reference point is turbine. Complete experimental procedure has recorded 3516 data.
marked installing a little steel bar at the end of the shaft. The Table 1 shows the average values of these parameters.
tachometer was installed near to the shaft showing both rpm and The performance of a wind turbine can be characterized by the
analogic signal output (mA). Starting from the highest pulse of the variation of three main indicators e power, torque and thruste as a
flashlight, the flash rate is adjusted down until a single image of the function of wind speed [8]. The available power from wind is pro-
shaft is seen. The flashlight frequency at the first unique stop mo- portional to the speed of wind cubed. It can be calculated for un-
tion observed corresponds to the true shaft speed. Comparing the disturbed airstream by the expression P ¼ ½rAU3, where (r) is the
air density, (A) is the cross-sectional area and (U) is wind speed. The
power output of a wind turbine (Pw) is a portion of this power that
is extracted by the converter machine. The ratio between the me-
chanical power extracted by the converter and that of the undis-
turbed air stream is called the “power coefficient” (CP). Therefore,
power coefficient may be calculated by the following equation:

Pw Pw
Cp ¼ ¼1 ½Dimensionless (2)
P r
2 AU
3

The power transmitted from the wind turbine (Pw) to the shaft
is:

Pw ¼ t$U ½W (3)


Fig. 6. Schematic view of the experimental apparatus.
P.A.C. Rocha et al. / Energy 148 (2018) 169e178 173

Fig. 8. Torque transducer calibration.

Fig. 9. Tachometer calibration.

Table 1 to the quotient between blades' tip velocity and the wind speed (Eq.
Experimental data for each pitch angle. (4)), it enables the evaluation of these parameters (Cp and l) from
Pitch angle U [m s1] U [rad s1] t [Nm] Turbulence intensity [%] each dataset, i.e. from U, t and U, then to plot the rotor power
6 4.03 2.20 4.00 22.54
characteristic curve for each tested pitch angle.
3 3.76 7.99 3.26 22.39
U$r
0 3.32 9.61 3.27 35.49
l¼ ½Dimensionless (4)
þ3 3.25 8.29 3.77 36.41 U
þ6 4.35 7.07 4.15 38.26
where (r) is blade rotor radius [m].

where ðtÞ is the torque [Nm] and ðUÞ the rotational speed [rad s1].
In general, rotor power characteristics are represented by 3.1. Statistical analysis
dimensionless parameters, such as rotor power coefficient (Cp)
versus the tip speed ratio (l) curves [25]. Since tip speed ratio refers The rotor power characteristics, i.e., rotor power coefficient (Cp)
174 P.A.C. Rocha et al. / Energy 148 (2018) 169e178

and tip speed ratio (l), were calculated and statistically analyzed. isolate the specific differences it must be performed complemen-
First, the data obtained for each pitch angle were grouped into tip- tary tests, e.g. Fisher's least significant difference (LSD) method.
speed intervals of unitary amplitude, from l ¼ 0 to l ¼ 9 performing Other methods and comparative discussion can be found in
nine class intervals. It is worth noting that the endpoint convention Ref. [27]. LSD compares the means of all treatment pairs with
in this case includes the left endpoint of all class intervals and ex- rejected null hypothesis H0 ¼ mi ¼ mj (for all isj) using a t-statistic,
cludes the right, except for the rightmost class interval, which in- for which LSD is defined as:
cludes both of its endpoints. vffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
u !ffi
Descriptive statistics summarize the dataset treatment, which is u 1 1
ranked into each tip-speed intervals by the pitch angle. This data LSD ¼ ta 2;Na tMSE $
=  (8)
ni nj
subgrouping not only provides the comparison between the effects
of each pitch angle on the expected rotor power coefficient, but also
Assuming that it is an unbalanced design, N is total sampling
determines the wind turbine behavior as a function of tip speed
size, a is the number of treatment levels, a/2 the confidence in-
operating on a specific pitch angle. Table 2 summarizes all
tervals, MSE is mean squares of each treatment, n is sampling size of
descriptive statistics, including means, the valid assignments, and
each i and j pair. The pair of means, mi and mj, can be declared
measures of variability: variance, the maximum and minimum
significantly different if:
values; in addition, some description about the shape of the dis-
 
tribution such as skewness and kurtosis is presented.  
mi  mj  > LDS (9)

3.2. The analysis of variance e ANOVA

The analysis of variance consists in evaluating whether there is


4. Results and discussions
statistically significant difference between mean values of several
treatments [26]. Assuming that the pitch angle is the main factor
Data acquisition was recorded throughout the experimental
that influences the rotor power coefficient, the ANOVA helps to
period in which the wind turbine was kept on a typical urban
understand the magnitude of its impact, as well as determining
environment. Processing data filter was carried out as a means to
which angles are more beneficial to enhance the power coefficient.
eliminate some noising data resulting in 3454 valid samples from
It is considerate that Cp has a normal distribution resulting to a
the 3516 initial data. The imposed criterion aims to adapt each data
single factor of interest, pitch angle displacement, which has five
treatment to a normal distribution that is evaluated by the skew-
levels (6, 3, 0, þ3, þ6). Thus, the one-way analysis of variance
ness and kurtosis. According to the literature [26,27] and other
can be applied to test the hypothesis in which the mean of each
statistical works [28e31], in general, skewness and kurtosis values
level has equality to each other.
greater than 2 and fewer than 2, represent great deviation from
The hypotheses are:
the normal distribution, then it must disregard the hypothesis of
H0 ¼ m1 ¼ m2 ¼ m3 / ¼ mn (5) normality of the data. The ANOVA was carried out to all the treat-
ment into the class intervals with sampling size of more than 3.
Table 3 presents the ANOVA results (for which a ¼ 0.05) including
Ha ¼ At least one pair mi smj (6)
the conclusion about the H0 test of hypotheses. Note that for almost
In the case of null hypothesis H0 is rejected, the test assumes the all class intervals of tip speed ratio, the p-value is considerably
hypothesis alternative is true, which indicates that a difference small except to the class interval [6e7) in which the p-value is
exists between the treatment means. It signifies that pitch angle greater than a (p-value ¼ 0.0527). This is a strong evidence to
impacts on the power coefficient. conclude that the hypothesis of equality between means of treat-
The ANOVA divides the total variability in the sample data into ment is not true for the majority of treatments. Then the results
two component parts: a sum of squares of differences between lead to conclude that pitch angle really influences the wind power
treatment means and the grand mean, which is denoted as behavior in the conditions of the experiment.
SStreatments, and a sum of squares of differences of observation As previously mentioned, since ANOVA does not identify which
within a treatment that is denoted SSE. Thereafter, ANOVA method intervals are different from each other, it is necessary to carry out a
analyzes the ratio between the means of squares (MS) of two par- comparative test of paired treatments. Following the adopted
titions; MSTreatment with (ae1) degrees of freedom (MSTreat- methodology, the Fisher's LSD test was applied. Table 4 shows the
ment ¼ SSTreatment/(ae1)) and MSE with n (a-1) degrees of freedom results of the test range into each class interval of tip speed ratio.
(MSE ¼ SSE/n (a-1)). The null hypothesis H0 is true if the ratio be- The Fisher's LSD criteria establish that a pair of treatment is
tween MSTreatment and MSE has an F-distribution with (ae1) and n statistically different since the difference of means is greater than
(a-1) degrees of freedom. the LSD value (Eq. (9)). Based on this statement, some pairs of
The designed experiment has a single factor with “a” levels treatments could be interpreted as having the same average power
and“n” observations. Note in Table 2, that the experimental design coefficient. At the initial interval of tip speed ratio only pairs
has varying levels (a) and sample sizes (n), and that they are including the displacements of þ6 and 6 are different, but direct
different in each treatment. In this case, it is called unbalanced comparison between these angles has no significant difference,
design for which slight modifications must be made in the sums of once the turbine is still starting up and the Cp values are low. Very
squares' equation, considering that total sampling size (N) is: small l indicates that the rotational velocity is lesser than wind
speed (i.e. l is proportional to U/U). It suggests that initial motion of
X
a
the wind turbine is influenced by these pitch angles, perhaps they
N¼ ni (7) only help to break down inertial forces. Analyzing the diagram of
i¼1
forces acting on the blade (Fig. 10), the þ6 is aerodynamically
The ANOVA indicates whether there are significant differences preferable than 6 because negative pitch angle displacements
between treatments. However, it does not identify how many or may cause very low (or even negative) torque, since the effective
which average values are different from each other. In order to angle a increases resulting in stall over the most part of the blade.
P.A.C. Rocha et al. / Energy 148 (2018) 169e178 175

Table 2
Descriptive statistics from the experimental dataset.

Tip speed ratio Intervals Angle displ. Average (l) Power coefficient (Cp)

Valid n Mean Variance Min. Max. Skewness Kurtosis

[0.00e1.00) 6 0.72 519 0.030 0.0002 0.003 0.082 1.07 1.84


3 0.77 17 0.041 0.0008 0.006 0.112 1.37 1.58
0 0.65 28 0.043 0.0009 0.006 0.123 1.22 0.73
þ3 0.50 34 0.037 0.0009 0.001 0.107 1.06 0.16
þ6 0.79 6 0.021 0.0001 0.001 0.034 0.92 0.48

[1.00e2.00) 6 1.06 69 0.037 0.0002 0.015 0.083 0.96 1.34


3 1.50 54 0.079 0.0011 0.027 0.181 0.90 0.86
0 1.57 114 0.117 0.0046 0.033 0.318 0.98 0.27
þ3 1.61 222 0.093 0.0013 0.028 0.243 1.08 1.40
þ6 1.70 182 0.072 0.0012 0.001 0.204 1.29 1.52

[2.00e3.00) 6 2.70 3 0.176 0.0105 0.102 0.293 1.54 *


3 2.75 204 0.071 0.0007 0.028 0.157 1.23 0.95
0 2.53 119 0.216 0.0090 0.046 0.535 0.56 0.20
þ3 2.58 60 0.232 0.0032 0.052 0.327 1.17 1.67
þ6 2.39 336 0.098 0.0024 0.033 0.293 1.38 1.85

[3.00e4.00) 6 3.60 2 0.127 0.0028 0.089 0.164 * *


3 3.43 428 0.141 0.0032 0.060 0.367 1.05 0.84
0 3.49 115 0.210 0.0269 0.032 0.572 0.69 0.66
þ3 3.31 23 0.319 0.0170 0.077 0.524 0.70 0.42
þ6 3.45 83 0.132 0.0046 0.051 0.363 1.30 1.63

[4.00e5.00) 6 4.36 2 0.237 0.0102 0.165 0.308 * *


3 4.35 128 0.294 0.0090 0.086 0.562 0.34 0.18
0 4.51 73 0.185 0.0294 0.053 0.586 1.29 0.05
þ3 4.56 47 0.372 0.0174 0.162 0.581 0.04 1.45
þ6 4.43 45 0.210 0.0083 0.087 0.503 1.21 1.55

[5.00e6.00) 6 5.49 1 0.339 * 0.339 0.339 * *


3 5.37 15 0.320 0.0111 0.201 0.534 1.12 0.34
0 5.49 56 0.246 0.0191 0.106 0.581 1.01 0.20
þ3 5.47 94 0.383 0.0170 0.026 0.587 0.15 1.01
þ6 0 * * * * * *

[6.00e7.00) 6 0 * * * * * *
3 6.19 4 0.345 0.0105 0.210 0.457 0.62 1.39
0 6.46 104 0.318 0.0100 0.166 0.579 0.85 0.30
þ3 6.55 51 0.362 0.0127 0.076 0.569 0.57 0.03
þ6 0 * * * * * *

[7.00e8.00) 6 0 * * * * * *
3 0 * * * * * *
0 7.5 80 0.451 0.0075 0.216 0.589 0.49 0.35
þ3 7.48 70 0.321 0.0043 0.156 0.484 0.24 0.14
þ6 0 * * * * * *

[8.00e9.00] 6 0 * * * * * *
3 0 * * * * * *
0 8.39 39 0.481 0.0023 0.343 0.598 0.67 1.64
þ3 8.41 27 0.359 0.0054 0.206 0.548 0.33 1.68
þ6 0 * * * * * *

*No data recorded.

Table 3
This phenomenon could be observed in the experiment since there
ANOVA results. is no Cp values recorded from 6 displacement above l ¼ 2.
Also, 3 seems to mitigate the power coefficient above l ¼ 2.
Tip speed ratio Intervals F p-value H0 ¼ m1 ¼ m2 ¼ m3 … ¼ mk
Analyzing the rotor power characteristic curves, which is
[0e1) 7.93 0.0000 False depicted in Fig. 11 for all angle displacements, it can be observed
[1e2) 46.70 0.0000 False
[2e3) 270.68 0.0000 False
that the best results came from 0 to þ3 angles. These angle
[3e4) 44.72 0.0000 False sometimes produces the average Cp value, for instance at tip speed
[4e5) 21.66 0.0000 False intervals of [0e1), [2e3) and [6e7). They provide a very profitable
[5e6) 19.28 0.0000 False ratio of the power extraction from the available wind power. It is
[6e7) 3.00 0.0527 True
interesting to note that after [2e3) interval this ratio decreases for
[7e8) 104.44 0.0000 False
[8,9] 66.01 0.0000 False the zero angle, but it is growing up for the þ3 angle. This behavior
is once more alternated when l reaches 6 to 7 until l ¼ 9.
176 P.A.C. Rocha et al. / Energy 148 (2018) 169e178

Some physical insights should be detailed in addition to the


Significant difference

statistic calculations, mainly for the 0 and þ3 angles. Despite


some results reveal difference between the performances at low
rotations (0 <l < 2), this finding does not seem significant because
it is based on average value of Cp, which is still very limited. Any
little deviation leads to large variations. Therefore, the use of þ3
FALSE

FALSE
FALSE
TRUE
TRUE
TRUE

TRUE
TRUE

TRUE

TRUE
angle is indicated to provide highest torque. This fact could be
explained by the highly effective angle of attack (a), which can
more easily stall the blade at 0 . Regarding Fig. 10, a angle increases
0.0173

0.0714
0.0747
0.0153
0.0116
0.0097

0.0160
0.0126

0.0434

0.0300
LSD

when b diminishes, which can cause the detachment of the


boundary layer in a situation of low rotations. Accordingly, for the
starting regime, the displacement of þ3 which supplies more
Difference of means

advantageous torque inclusively, is predicted.


After overcome the braking inertial forces, the blades accelerate,
as the tip speed ratios ranged from 2 to 7, increasing the component
0.0155

0.0631
0.0741
0.1338
0.1183
0.0272

0.1610
0.1455

0.1372

0.1219
of the rotational speed vector (U.r). In spite of there are no statistic
observations, the value of Cp for þ3 angle visibly appears superior
for all the varied l values. The resultant speed vector diminishes
þ6 vs þ3

þ6 vs 3

þ3 vs 3

with the angle of the rotation plan, as well as the effective angle of
þ3 vs -3
þ3 vs 0

0 vs 3

þ3 vs 0

þ3 vs 0
þ6 vs0

0 vs -3

attack. In this condition the ratio L/D goes to the optimum value.
[2e3)

[5e6)

[8e9]
Pairs

The þ3 position reaches this condition before, beginning from


l ¼ 2 until nearly to the projected value (i.e. l ¼ 8). From this point,
the blades operating at 0 position achieve enhanced performance
Significant difference

because this is the arrangement designed by BEM for optimized


operations. The effective angle of attack equals the ratio L/D along
the blade length. On the other hand, the angle of attack for the
displacement of þ3 become small, reducing the lift forces in all
FALSE

FALSE
TRUE
TRUE

TRUE
TRUE
TRUE
TRUE
TRUE
TRUE
TRUE

TRUE

TRUE
TRUE
TRUE
TRUE

TRUE

sections and mitigating the generating component of torque (see


the projected vector F on the rotation plane, Fig. 10).
It is important to highlight that in the present study, Cp does not
0.0127

0.0487
0.0082
0.0098

0.0116
0.0094
0.0124
0.0113
0.0135
0.0125
0.0149

0.0536

0.0486
0.0481
0.0479
0.0424

0.0249

decrease gradually. This finding clearly differs from other studies


LSD

for large or small-sized wind turbines conduced in controlled


conditions, e.g. wind tunnels. It can occur due to the urban envi-
Difference of means

ronment, which is characterized by the highly unsteady flow. As the


tip speed ratio becomes high, the wind turbine spins very rapidly
causing a blocking effect in which the wind diverts around the
turbine [16]. Thus, there is no wind interaction with the blades able
0.0073

0.0254
0.0217
0.0452

0.0350
0.0235
0.0144
0.0567
0.0379
0.0802
0.0423

0.1620

0.0623
0.1874
0.0997
0.0877

0.1297

to generate useful torque.


The power coefficient variation observed throughout the ex-
periments as a function of the changing in the pitch angle clearly
þ6 vs þ3

þ6 vs 6

þ3 vs 3
þ3 vs 6

3 vs 6

þ6 vs þ3

þ6 vs 3

þ3 vs 3
þ6 vs -3

þ6 vs 0
þ3 vs 0

0 vs 3
0 vs 6

þ3 vs 0

0 vs 3

þ3 vs 0

confirms its utility. It indicates the possibility of the application of


þ6 vs0
[1e2)

[4e5)

[7e8)

blade pitch control for a small wind turbine in the severe conditions
Pairs

such as urban environments. Fig. 12 depicts hypothetical curves


Significant difference

FALSE
FALSE
FALSE

FALSE

FALSE

FALSE
FALSE
TRUE
TRUE
TRUE

TRUE

TRUE
TRUE

TRUE
TRUE

TRUE
TRUE
TRUE

TRUE
0.0131
0.0082
0.0095

0.0098

0.0211

0.1061
0.1041
0.0141
0.0144
0.0152

0.0057

0.0062
0.0079

0.0415
0.0254

0.0402
0.0377
0.0185

0.0349
LSD
Difference of means

0.0091
0.0058
0.0037

0.0022

0.0088

0.0174
0.0261
0.0165
0.0224
0.0202

0.0074

0.0132
0.0111

0.1869
0.0775

0.1094
0.1781
0.0687

0.0435
Fisher's LSD test.

þ6 vs þ3

þ6 vs 3

þ3 vs 6

3 vs 6

þ6 vs þ3

þ3 vs 3
þ6 vs -6

þ3 vs -3

þ6 vs -3

þ3 vs -3
þ3 vs 0

0 vs 6

þ3 vs 0

0 vs 3

þ3 vs 0
þ6 vs0

0 vs -3

þ6 vs0

0 vs -3
[0e1)

[3e4)

[6e7)
Table 4

Pairs

Fig. 10. Diagram of acting forces on the blade rotor plane.


P.A.C. Rocha et al. / Energy 148 (2018) 169e178 177

Fig. 11. Rotor power characteristic curves.

(traced red lines), which is suggested for the acting pitch angle energy capture throughout time-operation. Traced red lines is a
control, compared to polynomial regression curves from zero proposal operation curve and, depending of structural conditions of
and þ3 angles. the project, it is possible to allow increased turbine rotation
Following the pitch control curve (Fig. 12), it is possible to infer following the results obtained by angle profiles calculated from
the gain of energy capture. The choice to use the displacement BEM (i.e. 25.63 represented in this work by 0 ).
of þ3 to starting the wind turbine (l ¼ 0) until the tip speed ratio
reaches nearly the projected value (l ¼ 7) results in superior value 5. Conclusion
of Cp than the designed angle (0 ). Consequently, it results in a
larger energy capture. As the wind turbine accelerates there will be Statistical methods were applied to study the influence of blade
an angle’ changes to modulate the energy capture, but this transi- pitch angle on the performance of a three-bladed wind turbine in
tion does not concern the subject of this work. This matter can be urban environmental conditions. Experimental measurement ac-
investigated in a specific study. From l ¼ 7 the blades operate at quisitions from a wind turbine prototype, specially constructed for
25.63 (projected angle) for which Cp is higher providing greater this objective, provided a dataset that was statistically treated and

Fig. 12. Pitch control curve.


178 P.A.C. Rocha et al. / Energy 148 (2018) 169e178

compared. Test of hypotheses using the analysis of variance e [5] Corbetta G, Mbistrova A, Ho A. Wind in power: 2015 European statistics. The
European Wind Association; 2016. p. 3e13.
ANOVA was applied to confirm the dependence of rotor power
[6] Bahaj AS, Myers L, James PAB. Urban energy generation: influence of micro-
coefficient on the pitch angle. wind turbine output on electricity consumption in buildings. Energy Build
Blade pitch angle set consists in the displacement of the first 2007;39:154e65.
section pitch angle (determined by BEM) by changing blade [7] Sogukpinar H, Bozkurt I. Analysis on meeting the electric energy demand with
small wind turbine in Turkey. Int J Eng Technol 2015;1(2):51e5.
orientation to up and down from this point. Thus, the experimental [8] Simic Z, Havelka JG, Vrhovcak MB. Small wind turbines e a unique segment of
dataset possesses five angular displacements of þ6, þ3, 0, 3 the wind power market. Renew Energ 2013;50:1027e36.
and 6 . It was observed that both the rotations of þ6 and 6 [9] Zanforlin S, Letizia S. Improving the performance of wind turbines in urban
environment by integrating the action of a diffuser with the aerodynamics of
mitigate the rotor power coefficient. That means, from the point of the rooftops. Energ. Proc. 2015;82:774e81.
view of the wind turbine efficiency, an impracticable (non-desir- [10] Lubitz WD. Impact of ambient turbulence on performance of a small wind
able) application angle. However, considering the application of a turbine. Renew Energ 2014;61:69e73.
[11] Yang Z. Optimization and simulation of hybrid renewable energy system for
pitch control, a displacement to þ6 is appreciable to start the wind urban building clusters [Dissertation]. Wageningen Univ.; 2016.
turbine. On the other hand, a displacement of þ3 shows more [12] Gawad AFA. New, simple blade-pitch control mechanism for small-size,
appropriate, which recorded higher Cp values for tip speed ratios horizontal-axis wind turbines. J Energ Power Eng 2013;7:2237e48.
[13] Bianchini A, Ferrara G, Ferrari L. Pitch optimization in small-size Darrieus
ranged from 3 to 7. As expected, the BEM designed angle (0 ) wind turbine. Energy Proc 2015;81:122e32.
provides better efficiency for the tip speed from seven onwards. In [14] Kekezoglu B, Tanrıo€ven M, Erduman A. A new wind turbine concept: design
addition, statistical analysis indicates that the angles of 0 and þ 3 and implementation. Acta Poly Hung 2015;12(1):199e211.
[15] Manwell JF, McGowan JG, Rogers AL. Wind energy explained e theory, design
result in the best wind turbine efficiency for all tested tip speed
and application. second ed. Sussex, UK: John Wiley & Sons Ltd; 2002.
ratios, suggesting that other angles are unfeasible to be applied. [16] Burton T, Sharpe D, Jenkins N, Bossanyi E. Wind energy e handbook. John
The most important conclusion of this study is addressed to the Wiley & Sons, Ltd; 2011.
use of a pitch angle control for a small wind turbine in urban [17] Morten BB, Kraning JM, Boyd S, Stoustrup J. Wind turbine pitch optimization.
In: 2011 IEEE international conference on control applications (CCA); 2011.
environment. Statistical tests from ANOVA confirm the physical [18] Lee J, Son E, Hwang B, Lee S. Blade pitch angle control for aerodynamic per-
insight of the dependence of the rotor power coefficient on varied formance optimization of a wind farm. Renew Energ 2013;54:124e30.
pitch-angle values. The tests have shown that the pitch angle in- [19] Kusiak A, Zheng H, Song Z. Power optimization of wind turbines with data
mining and evolutionary computation. Renew Energ 2010;35:695e702.
terferes on the performance of the wind turbine, since it was [20] Nagai BM, Ameku K, Roy JN. Performance of a 3 kW wind turbine generator
detected significant difference between the Cp averages. It suggests with variable pitch control system. Appl Energy 2009;86:1774e82.
that the controlling of this angle would be aerodynamically effi- [21] Hau E. Wind turbines e fundamentals, technologies, application, economics.
3th Ed. London, UK: Springer; 2013.
cient for the small wind turbine under urban conditions, and the [22] Betz A. Schraubenpropeller mit geringstem Energieverlust. Mit einem Zusatz
performance gains must be compared to financial costs, to guar- von L. Prandtl. Nachrichten von der Gesellschaft der Wissenschaften zu
antee positive economic results. Go€ttingen. Mathematisch-Physikalische Klasse 1919:193e217.
[23] Somers DM. Design and experimental results for the S809 airfoil. National
Renewable Energy Laboratory; 1997. p. 1e98.
Acknowledgments [24] Rocha PAC, Albiero D, Fernandes MPG, Fernandes JL. Desempenho aero-
din^amico de perfis NACA de quatro dígitos em aerogeradores de pequeno
porte para a agricultura familiar. 2015. p. 1227. https://doi.org/10.5433/1679-
This research was jointly supported by the Fundaç~ ao Cearense ^ncias Agrarias (Online) v. 36. .
0359.2015v36n3p1227. Semina. Cie
de Apoio ao Desenvolvimento Científico e Tecnolo gico (FUNCAP), [25] Muljadi E, Butterfield CP. Pitch-controlled variable-speed wind turbine gen-
~o de Aperfeiçoamento de Pessoal de Nível Superior ̆
the Coordenaça eration. National Renewable Energy Laboratory; 1999. p. 1e8.
(CAPES), and the Conselho Nacional de Desenvolvimento Científico [26] Montgomery DC, Runger GC. Applied statistics and probability for engineers.
3th Ed. New York, US: John Wiley & Sons, Inc.; 2003.
e Tecnolo gico (CNPq), Brazilian governmental agencies. The sup- [27] Montgomery DC. Design and analysis of experiments. fifth ed. New York: John
port received is gratefully acknowledged. Wiley & Sons; 2001.
[28] Albiero D, Maciel AJS, Milan M, Monteiro LA, Mion RL. Proposal of using
exponentially weighted moving average for studies about longitudinal dis-
References tribution of seeds. Rev Cie ^nc Agron 2012;43(1):86e95.
[29] Fonseca MR. Influe ^ncia do a^ngulo de pitch no desempenho de um aerogerador

[1] Yilmaz AS, Ozer Z. Pitch angle control in wind turbines above the rated wind de pequeno porte projetado com o perfil aerodin^ amico NREL S809. Master’s
speed by multi-layer perceptron and radial basis function neural networks. thesis. Mechanical Engineering Department, Federal University of Ceara ;
Expert Syst Appl 2009;36:9767e75. 2012.
[2] Grieser B, Sunak Y, Madiener R. Economics of small wind turbines in urban [30] Lopes AS. Projeto e testes de um aerogerador de pequeno porte e de um
settings: an empirical investigation for Germany. Renew Energy 2015;78: sistema de mediça ~o de eficie ^ncia mec^ anica. Master’s thesis. Mechanical En-
334e50. gineering Department, Federal University of Ceara ; 2011.
[3] Abdelkafi A, Krichen L. New strategy of pitch angle control for energy man- [31] Sousa IDP. Projeto e Ensaios de Pa s para Turbinas Eo licas de Pequeno Porte
agement of a wind farm. Energy 2011;36:1470e9. com Diferentes Correço ~ es Geome tricas na Ponta e na Raiz da Pa . Master’s
[4] Stootweg JG, de Haan SWH, Polinder H, Kling WL. General model for repre- thesis. Mechanical Engineering Department, Federal University of Ceara ;
senting variable speed wind turbines in power system dynamics simulations. 2012.
IEEE Trans Power Syst 2003;18(1):144e51.

You might also like