Professional Documents
Culture Documents
10 1016@j Renene 2019 12 060
10 1016@j Renene 2019 12 060
10 1016@j Renene 2019 12 060
The design of a small lab-scale wind turbine model with high performance
similarity to its utility-scale prototype
PII: S0960-1481(19)31932-9
DOI: https://doi.org/10.1016/j.renene.2019.12.060
Reference: RENE 12769
Please cite this article as: B. Li, D.L. Zhou, Y. Wang, Y. Shuai, Q.Z. Liu, W.H. Cai, The design of a
small lab-scale wind turbine model with high performance similarity to its utility-scale prototype,
Renewable Energy (2019), https://doi.org/10.1016/j.renene.2019.12.060
This is a PDF file of an article that has undergone enhancements after acceptance, such as the
addition of a cover page and metadata, and formatting for readability, but it is not yet the definitive
version of record. This version will undergo additional copyediting, typesetting and review before it
is published in its final form, but we are providing this version to give early visibility of the article.
Please note that, during the production process, errors may be discovered which could affect the
content, and all legal disclaimers that apply to the journal pertain.
1 Harbin Institute of Technology, School of Energy Science and Engineering, 92 West Dazhi
Rd, Harbin, China
2 Northeast Electric Power Univeristy, School of Energy and Power Engineering, 169
Abstract
This paper discussed the procedure of an optimization design method involving a small lab-
scale wind turbine rotor. To simulate the aerodynamic performance of a prototype wind turbine
wake flow in small wind tunnels), an appropriate similar design of the small lab-scale rotor is
needed. However, the similar design of blade performance from a prototype wind turbine for
small scaled model is rare. In this paper, using a 2.5 MW utility-scale wind turbine as the
prototype, we set up a new optimization process of similar design for a model wind turbine
with a 320 mm rotor based on the Lifting-Line Theory (LLT) with wake-induced corrections.
Results show a dramatic deviation in performance of the geometric scaled model compared to
the objective values, which demonstrates the significance of the Reynolds number effects. With
the optimized blade, the distribution of normal thrust force is very similar to the objective
values. Geometric characteristics variations along the blade span are similar to their prototype.
1
Journal Pre-proof
The tip speed ratio runs at a roughly matched rated condition. The optimized rotor designed
model gives a better performance than other models used in previous wind tunnel studies.
1 Introduction
The development of sustainable renewable energy has become the strategy of energy
of wake development mechanism that would help to improve wind turbine or wind farm
optimal operation strategies [3,4]. Model experimental studies in a lab-scale wind tunnel is also
necessary for the validation of numerical simulation studies, and aid the design of wind turbines.
The rotor of a wind turbine, consisting of the blade and hub rotor assembly, operates in an
atmospheric boundary layer (ABL) characterized by highly turbulent airflow and is subjected
experiments can be an important method for studying those extremely complex interactions.
Different sizes of model wind turbines have been used, such as large-scale model turbine [5,6,7],
meter-scale model turbine [8,9,10], and small size model turbine [11,12]. Model experimental
studies with precise measurement methods, such as PIV (Particle image velocimetry), in wind
tunnels can also quantify the performance of a turbine and the correlation with its wake
development accurately [13]. However, both the large scale and the meter scale models are
oversized for most of the wind tunnels due to blockage effect, and its large wake area is
unfavorable for PIV measurement. Therefore, there are many studies adopt small sized model
2
Journal Pre-proof
Due to the different blade types used in wind tunnel experiments at small size of lab-
scale, related researches have generally investigated four types of wind turbine models namely;
direct geometric scale models [14,15], commercial propeller models [16], parametric design
models [17,18], and simple rotor models [19-21]. First, using a geometric scale model based
on the 2MW ERS-100, Yuan W et al. [14] study the influence of the opposite rotation direction
of blades on the wake flow of a series wind turbine. Taking 1/350 as the geometric scaled
factor, the rotor radius is 127mm and the hub height is 225 mm. Tian W et al. [15] use the same
model as Yuan et al. to study the influence of wind conditions on the wake characteristics and
dynamic wind load of the wind turbine model. The difference is that the geometric scale factor
is 1/320 with a smaller 140 mm rotor radius. Results show correlation between the tip vortex
shedding and inflow, but has no comparison with the prototype wind turbine. Second, without
a wind turbine prototype, Hu H et al. [16] installed a propeller model rotor MA0530TE while
studying the aerodynamic characteristics near the wake of the horizontal-axis wind turbine
model in the atmospheric boundary layer wind tunnel. The diameter of the impeller was 127mm,
the twisted blades ranged from 20° at the root to 10°at the tip, which was reversely installed to
simulate the wind turbine. Third, Mctavish et al. [17, 18] designed a three-bladed rotor
specifically for operation conditions under Reynolds number of order 104. It is a parametric
design that matches the tip speed ratio and thrust coefficient to common utility-scale wind
turbines. Fourth, studying the vertical transport of momentum and kinetic energy in the wind
turbine boundary layer, Cal R et al. [19] designed a simple wind turbine model for wind tunnel
experiments; the model blade was cut and twisted from uniform metal sheets. The twist angle
varied from 15° at the root to 10° at the top with a blade thickness of 0.48 mm. Lebrón J et al.
3
Journal Pre-proof
[20] used the same blade design method to study the interaction between the wind turbine array
and the turbulent boundary layer. Hamilton et al. [21] also used a similar design as the scaled
model of a 1 MW horizontal axial wind turbine for the study of the boundary layer of wind
turbine arrays. However, due to Reynolds number effect, the performance of a small lab-scale
model of wind turbine can be significantly different from its prototype [22,23]. For the wake
studies on a small lab-scale model of wind turbine, whether the geometric scaled rotor,
commercial impeller, parametric design rotor, or simple rotor represent the real wind turbine
There is a lack of practical methods for the design of a model wind turbine at the lab-scale
to match the wake flow of its prototype real turbine. It greatly restricts the development of wind
tunnel experiments for relative studies of wake flow. To solve this problem, a design method
of wind turbine, especially the blade, at small scale with high performance of similarity to its
prototype should be put forward. In this study, an innovative design method is proposed. In
section 2, the conversion relationship between prototype and model parameters is deduced with
similarity conditions; the lifting-line theory based model with wake-induced corrections is set
up for the calculation of blade performance. In section 3, we use a similar hybrid blade as the
prototype with the optimization variable at each section of the model blade reaching the
objective value. Finally, a blade model with high performance simulation of the aerodynamic
2 Method
In this study, an onshore 2.5-MW turbine was used as the prototype. The lab-scale turbine
has a geometric scaling factor of λ = 1: 300. In view of the big difference in unit performance
4
Journal Pre-proof
between the direct geometric scale model and the prototype, an optimized method of blade
design is established in this paper. The general procedure of the design of a lab-scale wind
turbine model mainly consists of three steps (as shown in Fig.1). The first step, is to calculate
the detailed performance of the prototype wind turbine, and derive the objective values of unit
thrust force at each section following the similarity criterion. The Second, is to take the direct
geometric scale value as the initial condition, and set up constraints for optimization. The final
step, is to calculate and validate the results of the optimization. In details, the radial distribution
of chord length and twist angle of the blade were obtained from the optimization calculations
using a lifting-line theory based model with wake-induced corrections. Results of the model
and prototype wind turbines were compared with the fully coupled simulation tool (Qblade)
[24]. Finally, we can get an optimized model with aerodynamic performance similar to the
prototype.
Table 1 illustrates the primary parameters of the three-bladed horizontal-axis 2.5-MW turbine
5
Journal Pre-proof
used as prototype in this study. The diameter of the rotor is 96 m, and the rated operating speed
is 11 m/s with design tip speed ratio of TSR = 8 ~ 12. The twist angle of the blade ranges from
9.5 ° at root to -2 ° near the tip and 2 ° at the tip. It is a hybrid designed blade with a circle
cross-section at the root and a high aerodynamic performance airfoil near the tip. The chord-
based Reynolds number at the tip region is approximately 6.3 ×106. The turbine essentially
consists of three pitchable blades to control the aerodynamic torque; the collective pitch control
A similar design of a lab-scale wind turbine model to its referenced utility-scale wind
turbine should employ the following similarity conditions [25, 26]: geometric similarity,
Geometric similarity: It is the baseline for a similar design such that the model and its
prototype are geometrically similar. For linear scale parameters, geometric similarity
conditions must be satisfied, such as diameter, hub height, hub diameter, etc. The geometric
𝜆𝐿 = 𝐷m 𝐷P = 𝜆 (1)
Kinematic similarity: If the corresponding points of the prototype flow field and the model
flow field have the same velocity, and the velocity vector is geometrically similar, the kinematic
similarity is satisfied. Obviously, it is impossible to match the Reynolds number because of the
huge difference in lengths between the model and the prototype. However, blade tip speed ratio
(TSR = 𝜔𝑅 𝑉𝑖𝑛), which represents the dimensionless number of the ratio of the blade tip linear
velocity to incoming wind speed Vin, can generally characterize the kinematic similarity for the
wind turbine design. In order to meet the similarity condition, the TSR of the model blade and
the prototype should be consistent that is, TSRp = TSRm. The rotating motion of the blade is
periodic. Thus, the model blade and prototype must keep the value of Strouhal number equal:
𝑉𝑝 𝑉𝑚
Sr𝑏𝑙𝑎𝑑𝑒 = = (2)
𝑛𝑝𝐷𝑝 𝑛𝑚𝐷𝑚
where n is the revolutions per minute (RPM). With 𝜔 = 2𝜋𝑛 60 for the blade, it has TSR =
𝜋𝑛𝐷 60𝑉in, then Eq.(2) indicates Sr = 𝜋 60 ∙ TSR. Therefore, the similarity of TSR can match
the similarity of Sr. Thus, the RPM of model turbine can be achieved by the design of reference
𝜆𝑉 = 𝑉m 𝑉P (3)
Dynamic similarity: Once we study the wake area of a wind turbine, the thrust coefficient
can be the most important parameter derived by many wake models [27, 28]. According to the
Jensen wake model [29], the velocity distribution in the wake area mainly relates to the axial
thrust coefficient of the wind turbine and assumes that the wake flow increases linearly with
the increasing distance from the rotor. According to Jensen’s wake model, the wake velocity
7
Journal Pre-proof
r
2
v v0 1 (1 1 CT ) (4)
r x
Where, v0 is the local wind speed, v and r are respectively the wake velocity and radius at the
distance x downstream of the wind turbine, CT is the thrust coefficient, is the dimensionless
Therefore, the thrust coefficient is an important coefficient for the wake flow. The Frandsen’s
wake model [27] also demonstrates a similar conclusion. Therefore, in this study, the thrust
combination of both the lift coefficient CL and the drag coefficient CD because of the triangle
of velocities. In all, aerodynamic parameters of the blade can be scaled down according to the
In wind turbine studies, the Blade Element Momentum (BEM) theory is widely used for
blade design of wind turbine, because of its high computational efficiency. However, it also
has limitations and is only applicable to the situation that the flow field around the airfoil is
achieve a tolerable accuracy. In fact, the wake of a rotor running naturally presents relatively
lagged unsteady aerodynamic characteristics [30]. Therefore, the calculation error of BEM in
rotor coupling analysis is large [31]. Moreover, the lifting-line theory [32], also known as the
related to the lift of blade. According to the geometric shape of the wing (the distribution of
chord, airfoil and twist angle along the aspect of the chord) and the flow condition, the lift force
8
Journal Pre-proof
distribution along the aspect of the chord is obtained. Afjeh [33] corroborated the high accuracy
of the lifting line model in his study of the influence of wakes on the aerodynamic performance
of rotors. In addition, to consider the effect of wake-induced corrections, the resultant flow
condition is updated to generate a more accurate blade performance. Therefore, in this study,
the LLT is improved to calculate the aerodynamic performance of the rotor. The new version
of QBlade (v0.963) also includes a new aerodynamic module with a new advanced Lifting-line
theory module.
Based on the generalized Lifting-line theory, with incompressible and inviscid flow, the
high-aspect-ratio wing is modelled as a single bound vortex line located at the 1/4 chord
position and an associated shed vortex sheet. The lifting line combines the attached vortices on
the blade into a vortex line with a variable strength. The lift of each section acts on the line. For
the first part, the wake model assumes that the wake is negligible and remains in a plane parallel
to the free stream. The model wake using single vortex sheet starting at the quarter chord, is
shown in Fig. 2.
Assuming that the Kutta-Joukowsky condition is satisfied at local sections, then the lift L
is expressed as:
9
Journal Pre-proof
𝑠
𝐿 ≈ 𝜌𝑉∞ ∫𝛤𝑑𝑦 (5)
―𝑠
the circulation (m²/s). The downwash at y due to entire wake is given by:
―𝑑Γ
𝑠 𝑑𝑦| 𝑑𝑦1
𝑦1 (6)
𝑤(y) = ∫ ―𝑠 4𝜋(𝑦 ― 𝑦1)
where w is the downwash velocity, y1 is the location of the section dy1, s is the half span.
vortex filaments in the wake area. Under induced flow, both the aerodynamic performance and
vortex filament nodes changes. Assuming it moves as Lagrangian lattices, the corrected
∂𝒙 ∂𝒙
𝑉' = 𝑉∞ + 𝑉𝑒 + 𝑉𝑇 = n( + ) (5)
∂𝜓 ∂𝜁
where, ψ is the azimuthal rotor position, ζ is the wake age, Ve is the induced effective velocity,
For the second part, the section model assumes that flow over each section are determined
by downwash at ¼ chord, according to the thin airfoil theory. The sectional lift coefficient (2D
𝑙 𝜌𝑉'Γ
C𝑙 = =
1 '2 1 '2 (6)
𝜌𝑉 𝑐 𝜌𝑉 𝑐
2 2
where c is the chord length of the local section,l is the lift force per unit span.
where α0 is zero-lift angle of attack of that section (depends on the airfoil geometry),α is the
In this study, in order to make the aerodynamic performance of the blade more accurate,
10
Journal Pre-proof
According to the geometric, motion and dynamic similarity conditions given above, the
conversion relationship that each physical quantity should satisfy for the model turbine can be
deduced as below in Table 2. Thus, the model turbine has a rotor diameter of 320 mm, hub
height of 270 mm, and TSR of 7. With an 11 m/s incoming wind speed in the lab experiment,
the rated RPM can go up to 4650. The lab-scale experiment can implement a low incoming
wind speed to achieve a low RPM that is suitable with the use of small drive-motor, but it may
generate a low power coefficient. Moreover, the objective values of unit thrust force of all sections
on the blade can be derived based on the similarity scale factor as shown in Fig. 6. The chord based
Reynolds number at the tip region of the model turbine is scaled to approximately 2 ×104 which
The comparison of aerodynamic parameters among different types of blade design models
is illustrated in Fig. 3. Here, the theoretical scaled value at lab-scale is converted directly from
11
Journal Pre-proof
the prototype based on the scale factor; the geometric scaled value is the aerodynamic
performance of the lab-scale model blade which is geometrically scaled from the prototype;
DU96W180 and NACA4412 are two blades with corresponding uniform airfoil along the entire
span. It should be mentioned that the comparison in Fig. 3 is based on the same model size.
Moreover, we used Qblade to calculate the aerodynamic performance of the turbine at different
scales. As it can be seen from Fig. 3a, the Ct of the geometric scaled model is dramatically
smaller than the theoretically scaled one, which means that experimental results of the
geometric scaled model should have considerable deviations. This demonstrates the significant
Reynolds number effects. Therefore, the necessity of optimization of the direct geometric scale
model is obvious, especially for a wake area study. Fig. 3a also illustrates that Ct of blades with
uniform airfoil of DU96W180 and NACA4412 at their entire span are smaller than the
theoretical scaled one, but NACA4412 obviously has a better performance under this low-Re
condition. A similar conclusion can be drawn from Fig. 3b, where lift coefficients in other blade
models are much smaller than the theoretical scaled value. In addition, the NACA4412 model
gives a high Cl for entire blade with fine pitch at lab-scale, and the geometric scaled model
coming last shows significant deviation. From Fig. 3c, Cd of the geometric scaled model has
similar deviations as the DU96W180 and NACA4412, but illustrates a large gap compare to
the theoretical scaled model. In view of the huge difference in aerodynamic performance of the
geometric and the theoretical scaled model, undertaking an optimization design of the model
In the optimization procedure, we would like to embed the improved LLT-based model in
MATLAB as the main process tool of the performance calculation. However, a validation is
12
Journal Pre-proof
required and consists of two main steps in this test. The first step is to run a set up with the
prototype turbine under its rated condition. This gives a good agreement in terms of the rated
power output of around 2.5 MW. The second step compares the results of the MATLAB code
to Qblade. As shown in Fig. 4, the figure demonstrates that the results of the LLT based
MATLAB code and those of Qblade at utility-scale (e.g. theoretical scaled data) are quite
consistent. Moreover, there is a significant deviation of Ct near root area of the span at the lab-
scale (e.g. geometric scaled data). As the LLT based MATLAB code simulates a reasonably
similar performance between the geometric scaled blade and the blade with uniform airfoil of
DU96W180 in Fig. 3, and together with the deviation of lab-scale model in Fig.4, we would
not like to trust the Qblade under these low-Re conditions. Therefore, our optimization code
should embed the calculation of the LLT based model performance. At the same time, both
calculation methods show that there is a considerable variation in the thrust coefficient of the
a) Theoretical scaled
Geometric scaled
0.25 DU96W180
NACA4412
0.20
0.15
Ct
0.10
0.05
0.00
13
Journal Pre-proof
b) Theoretical scaled
c) 2.0
2.0 Geometric scaled Theoretical scaled
1.8
DU96W180 Geometric scaled
NACA4412 DU96W180
1.6 1.5
NACA4412
1.4
1.2
1.0
1.0
Cd
Cl
0.8
0.6 0.5
0.4
0.2
0.0
0.0
-0.2
0.0 0.2 0.4 0.6 0.8 1.0 0.0 0.2 0.4 0.6 0.8 1.0
r/R r/R
Fig. 3 Aerodynamic performance of different turbine models. Values of the theoretical scaled model
are directly converted from the values of the prototype turbine by scale factor; Values of the geometric
scaled model are given by the aerodynamic performance of the lab-scale model blade which is
geometrically scaled from the prototype; Values of the other two models are calculated from blades
consist of uniform airfoils of DU960W180 and NACA4412 respectively in the entire span.
Theoretical scaled-Qblade-BEM
0.25 Geometric scaled-Qblade-BEM
Theoretical scaled-Qblade-LLT
Geometric scaled-Qblade-LLT
Theoretical scaled-WInDS-LLT
0.20 Geometric scaled-WInDS-LLT
0.15
Ct
0.10
0.05
0.00
turbine to the prototype, a new optimized design method is developed in the current paper. In
addition, it should be noted that the utility-scale wind turbine design generally uses hybrid blade.
A hybrid blade can combine advantages of both strong structure at root and high performance
at near tip sections. In fact, it is also important for lab-scale model design, because the use of
thin airfoil with high performance at root area would result in a very weak structure of blade at
14
Journal Pre-proof
small size. It may not run at high revolving speed and is difficult to manufacture. For this reason,
The aerodynamic design of blades is directly based on the geometric scaled model, that
is, the distribution of chord length and the twist angle along the span are geometrically similar
to its prototype. In addition, in the optimized design process, there are mainly three problems
worthy of explanation. Firstly, the design of blades should follow the above-mentioned
similarity conditions. Secondly, the blade optimal design should be carried out based on the
geometric similarity, e.g. blade shape and the airfoil transformation. Thirdly, due to the fact
that the first several sections of the 2.5-MW turbine blade do not contribute significantly to the
performance, and also because the chord length changes dramatically, we clipped them off in
order to simplify the optimization process. Therefore, the redesigned wind turbine maintains
those near root sections by geometrically scaling profiles. In addition, considering the general
method of blade design of a traditional wind turbine [34], we select chord length and twist angle
For the optimal calculation of the hybrid blade design, the characteristics of different
airfoils used in the hybrid blade are the most remarkable input parameters, that is, the design
of transition airfoils, and aerodynamic parameters of all airfoils. In addition, the prototype
turbine consists of 27 sections with 16 different airfoils. It changes from a circle at root to the
DU96W180 at tip gradually. The circle section improves the strength at root area with large
thickness airfoils; the DU96W180 airfoil improves the performance at tip area with its specific
aerodynamic characteristics. Accordingly, the model blade design considers a similar design
of airfoils as its prototype, for which geometry of model blade generally changes from a circle
15
Journal Pre-proof
to the NACA4412 airfoil. In particular, it should be noted that the NACA4412 is much thinner
than those transition DUINT-series airfoils that are used in the prototype blade. Therefore, we
design a new transition airfoil of TEST3415 (Camber 3.13%, at 39.2%, thickness 15.46%) to
transfer smoothly from DUINT190 to NACA4412. The comparison of airfoils in prototype and
model blade are shown in Fig. 5. The model turbine has a much simple structure with only 8
kinds of airfoils for its small size. To improve the performance of sections at middle area of the
blade (≤50% span), more NACA4412 sections are used. Then, aerodynamic characteristics of
all airfoils are evaluated by Qblade respectively as parameters of the optimization model.
Fig. 5 Comparison of airfoils in prototype and model blade; left is the prototype and right is the model.
The initial guess of the two variables for the optimization process is to make variables the
same as in the geometric scaled blade. However, in order to ensure the continuity of the chord
length and the twist angle of two adjacent sections of different airfoils, the least square method
is used to fit the data along the span wise distribution. Prior to this, Kim B [35] also performed
similar blade parameter fitting work. Thus, for those two geometric parameters of the geometric
scaled model blade in the current study, the span wise distribution of the chord length and the
twist angle can be roughly fitted into two third-order polynomials as shown in Fig. 6. In addition,
the last point of the chord length and the last two points of the twist angle are ignored in the
16
Journal Pre-proof
optimization process because of its small size and sharp change at the tip sections. Those
a)
b)
Fig. 6 Fitting of distributions for a) chord length and b) twist angle along the span wise direction
The optimization method is set up to search for an optimized result of the blade shape. On
the basis of the specific airfoil at each section, non-unique results of the combinations of chord
lengths and twist angles can generate a similar thrust force. However, ranges of optimized chord
lengths and twist angles are subjected to additional restrictions of similarity of the blade shape
to the prototype. Therefore, the goal of the optimization method is to both satisfy the similarity
of shape to the geometric scaled model and match the objective values. Here, the objective
value is the unit thrust force (t0) of the theoretical scaled values. As a basic rule of constrained
optimization problems, the mathematical optimization with m constrain equations uses the
min f ( X );
s.t. (10)
g ( x) 0, i 1, 2,..., m
i
Here we generally consider that the optimized blade should follow geometrically within a
constraint region similar to its prototype (for example positive chord length and twist angle).
17
Journal Pre-proof
Therefore, the specified optimization model for the blade thrust can be expressed as follows:
Nsec
T X t i 1, 2, , N
2
min f ( X ) i 0i sec
i 1
s.t. (11)
LB X UB
AX b
where Nsec is the number of sections of the blade in the span wise direction; t0i is the objective
value of unit thrust force at each section from the similarity law; LB and UB are the lower and
upper limits for X respectively. In addition, after several pretests to ensure similarities in
geometry and having a better optimal calculation speed, the lower and upper bounds of the
variable were set to the range of plus or minus 1.9 times of X0. Moreover, AX ≤ b are the
constraint equations that will be expressed in details later. Ti is the unit thrust force at each
section, depending on the chord length (ci) and twist angle (θi) as shown below:
Ti ( X ) Ti (ci , i )
ci x4 x3 ( r ) x2 ( r ) 2 x1 ( r )3
R R R (12)
i x8 x7 ( r ) x6 ( r ) 2 x5 ( r )3
R R R
X R8
In all, we established a constrained optimization problem, and the pattern search method
was good for the calculations. The setting of constraints for optimization equation comprises
of two main guidelines. First, it is essential to have similar trends between distributions of
geometric parameters of the optimized blade and the geometric scaled blade. Second, to be
reasonable, the chord length should be small at the tip, and the twist angle cannot be far negative;
the twist angle in the first section and the chord length at the tip should not be too large.
the twist angles should be within the range of (-9, 12) for the reliable Cl and Cd data, and the
18
Journal Pre-proof
chord length should be small but positive. Therefore, these constraint conditions can be
b 9 12 0
T
Solving the optimization problem, we deduced the optimized distribution of the chord
length and twist angle, as shown in Fig. 7. Comparatively, we also plotted the geometric scaled
values with the corresponding fitting lines. The values for the optimized blade parameters are
also listed in Table 3. Moreover, Fig. 8 illustrates simulation results to compare the
performance difference between the optimized model turbine and the objective values
(theoretical scaled values). The thrust coefficient of the optimized blade is approximately the
same as that of the objective value. The power coefficient also is a relatively good match to the
prototype. It is important to note that the power coefficient for the lab-scale model is far less
than the theoretical scaled value for the Reynolds effects [22]. Therefore, the optimization
method of this study gave a good comparative results for the model turbine to the objective
values. In addition, as shown in Fig. 9, the optimized model turbine in this study has a higher
power coefficient than previous studies [14][36]. The observed inconsistency in Cp of the
optimized model and the prototype is as a result of the Reynolds number effect, and it requires
more implementations of high performance low-Re airfoils in the future. Moreover, it gives a
similar range of high performance of Cp as the prototype 2.5-MW turbine at utility-scale. The
of the model turbine as we mentioned before. It ensures that the optimized turbine operates at
the optimum TSR with the maximum power output under designed incoming flow condition.
28 35
Chord length (Optimized)
Chord length (Geometric scaled)
24 Chord length (fitting) 30
20
25
16
Chord length (mm)
-4
5
-8
0
-12 Twist angle (Optimized)
Twist angle (Geometric scaled)
-16 Twist angle (fitting) -5
-20
0.0 0.2 0.4 0.6 0.8 1.0
r/R
Fig. 7 Distribution of chord length and twist angle in the span wise direction
Table 3 Optimized model turbine blade parameters
Chord Chord
Twist angle Twist angle
Nodes r/R length Airfoils Nodes r/R length Airfoils
(deg) (deg)
(mm) (mm)
1 0.000 8.00 9.50 circle 15 0.627 11.53 5.44 NACA4412
2 0.036 8.00 9.50 circle 16 0.678 10.77 4.97 NACA4412
3 0.058 8.00 9.50 circle 17 0.714 10.40 4.73 NACA4412
4 0.093 8.00 9.50 circle 18 0.760 9.38 3.98 NACA4412
5 0.130 8.00 9.50 circle 19 0.808 8.84 3.59 NACA4412
6 0.171 9.63 9.50 DUINT760 20 0.833 8.46 3.32 NACA4412
7 0.219 23.14 11.12 DUINT280 21 0.857 7.80 2.88 NACA4412
8 0.271 21.26 9.84 DUINT230 22 0.884 7.15 2.48 NACA4412
9 0.329 19.27 8.79 DUINT190 23 0.910 6.34 2.01 NACA4412
10 0.384 17.53 8.00 TEST3415 24 0.935 5.41 1.52 NACA4412
11 0.447 15.28 7.09 NACA4412 25 0.959 4.44 1.04 NACA4412
12 0.497 14.62 6.82 NACA4412 26 0.980 3.45 0.00 NACA4412
13 0.533 13.43 6.34 NACA4412 27 0.995 0.29 1.82 NACA4412
14 0.579 12.50 5.93 NACA4412
20
Journal Pre-proof
16 140
14 120
12 100
(N/m)
(N/m)
(W/m)
(W/m)
10 80
force
force
power
Unit power
8 60
thrust
Unit thrust
Normal
6 40
Normal
4 20
2 0
Objective values Objective values
Model values Model values
0 -20
0 0.5 1 0 0.5 1
r/R r/R
Fig. 8 Performance comparison between optimized model blades and prototype (target value)
0.6
Offset case-Hamilton2015
Base case-Hamilton2015
0.5 Tian2014
Present model
Prototype 2.5 MW turbine
0.4
0.3
Cp
0.2
0.1
0.0
0 2 4 6 8 10 12 14 16 18
TSR
Fig. 9 Comparison of the optimized model turbine with the CP of existing model turbines
As the development of wake can give a good demonstration of the model performance,
wake development simulations of the model and prototype turbines was deployed. It gives a
further verification of the similar design which can be seen intuitively from Fig. 10. The
development of the two wakes is roughly similar at near wake area, which is important for near
wake studies in wind tunnel. In addition, it should be mentioned that inflow velocity of both
model and prototype simulations are the same (11 m/s). Assuming that the wake development
length is 2.5 times the rotor diameter, the developing time of the model turbine wake is around
0.073 s, and that of the prototype wake is around 21.82 s. This gives a significant difference in
terms of the downstream wake structure at the far wake area as shown in Fig. 10, because of
21
Journal Pre-proof
Overall, from the above analysis, it can be concluded that the optimization model
established in this paper improves the similarity between the model turbine at lab-scale and the
Fig. 10 Comparison of the wake development of the model turbine and prototype turbine
4 Conclusions
It is very important to have a similar design of the model with wind tunnel experiments
for wind turbine studies. However, a similar design of blade performance for a small scale wind
turbine is rare, especially for scaled factor less than 1: 200. This paper describes the similar
design of the blade to enhance the similarity in aerodynamic performance and wake
characteristics of lab-scale wind turbine to its prototype. It improves the comparability between
the wind tunnel experiment of wind turbines at small lab-scale and the field measurements of
its prototype wind turbine at utility-scale, especially, the PIV measurement of entire wake flow
Due to the Reynolds number effect in the wind tunnel experiments for the studies of wake
area of wind turbines, a big performance gap exists between small lab-scale models and utility-
scale prototype. There is also lack of a practical method for the similar design of a small lab-
22
Journal Pre-proof
scale wind turbine model to match the wake flow of its prototype turbine. In this study, a new
method of blade design is proposed, using a real 2.5-MW turbine as the prototype. The model
rotor is 320 mm under a geometric scale ratio of 1:300. The comparison of aerodynamic
parameters of different types of design with objective values illustrates a dramatic deviation of
the geometric scaled model and a preferable performance of NACA4412 at low-Re condition.
From the validation, the application of the lifting-line theory with the improvement of wake-
the rotor. For the blade optimization design, a similar hybrid blade just like the prototype was
implemented, together with transition airfoil of TEST3415 design. The variation in geometric
characteristics along the span of the blade maintains similarities to its prototype. The tip speed
ratio runs at a roughly matched rated condition. In all, the model wind turbine with optimized
hybrid blade depicts a high performance in similarity to the prototype turbine, and gives a better
ACKNOWLEDGEMENTS
This work has been supported financially by the National Natural Science Foundation of
China (Grant No.51708151) and the China Postdoctoral Science Foundation (Grant
No.2019T120274), which is gratefully acknowledged by the authors. We are also grateful for
the academic help provided by Dr. Jiarong Hong, Mr. Bingzheng Dou, Dr. Yun Liu, and Dr.
References
[1] Saidur R, Islam M, Rahim N, Solangi K. A review on global wind energy policy, Renewable and
[2] Hua Y, Oliphant M, Hu EJ. Development of Renewable Energy in Australia and China: a
https://doi.org/10.1016/j.renene.2015.07.060
[3] Rockel S, Peinke J, Hölling M, et al. Wake to wake interaction of floating wind turbine models
in free pitch motion: An eddy viscosity and mixing length approach, Renewable Energy, 85 (2016)
666-676. https://doi.org/10.1016/j.renene.2015.07.012.
[4] González-longatt Fall P, Terzija V. Wake Effect in Wind Farm Performance: Steady-state and
https://doi.org/10.1016/j.renene.2011.08.053
[5] Hand M, Simms D, Fingersh L, et al. Unsteady aerodynamics experiment phase Vi: wind tunnel
test configurations and available data campaigns. National Renewable Energy Laboratory, Golden,
[6] Snel H, Schepers J, Montgomerie B. The Mexico Project (model experiments in controlled
conditions): the database and first results of data processing and interpretation. Journal of Physics:
https://iopscience.iop.org/article/10.1088/1742-6596/75/1/012014
[7] Cho T, Kim C. Wind Tunnel Test Results for a 2/4.5 Scale Mexico Rotor. Renewable Energy,
[8] Bottasso C, Campagnolo F, Petrović V. Wind tunnel testing of scaled wind turbine models:
beyond aerodynamics, Journal of Wind Engineering and Industrial Aerodynamics, 127 (2014) 11-
28. https://doi.org/10.1016/j.jweia.2014.01.009.
scale-model wind turbine on a floating offshore wind turbine platform//ASME 2014 33rd
[10] Ryi J, Rhee W, Hwang UC, et al. Blockage effect correction for a scaled wind turbine rotor by
https://doi.org/10.1016/j.renene.2014.11.057
[11] Lignarolo L, Ragni D, Krishnaswami C, et al. Experimental analysis of the wake of a horizontal-
https://doi.org/10.1016/j.renene.2014.01.020.
[12] Villegas A, Diez F. On the quasi-instantaneous aerodynamic load and pressure field
[13] Chamorro L, Porté-Agel F. Effects of thermal stability and incoming boundary-layer flow
[14] Yuan W, Tian W, Ozbay A, et al. An experimental study on the effects of relative rotation
direction on the wake interferences among tandem wind turbines. Sci. China-physics. Mechanic &
[15] Tian W, Ozbay A, Hu H. Effects of incoming surface wind conditions on the wake
characteristics and dynamic wind loads acting on a wind turbine model, Physics of Fluids 26 (2014)
125-108. https://doi.org/10.1063/1.4904375
[16] Hu H, Yang Z, Sarkar P. Dynamic wind loads and wake characteristics of a wind turbine model
https://doi.org/10.1007/s00348-011-1253-5.
[17] Mctavish S, Feszty D, Nitzsche F. A Study of the Performance Benefits of Closely-spaced Lateral
https://doi.org/10.1016/j.renene.2013.03.032
https://onlinelibrary.wiley.com/doi/10.1002/we.1648
[19] Cal RB, Lebrón J, Castillo L, et al. Experimental Study of the Horizontally Averaged Flow
Structure in a model wind-turbine array boundary layer. Journal of Renewable and Sustainable
[20] Lebron J, Castillo L, Cal RB, et al. Interaction between a wind turbine array and a turbulent
boundary layer. 48th AIAA Aerospace Sciences Meeting Including the New Horizons Forum and
[21] Hamilton N, Tutkun M, Cal R. Low-order Representations of the canonical wind turbine array
boundary layer via double proper orthogonal decomposition, Physics of Fluids, 28 (2) (2016)
25
Journal Pre-proof
025103. https://doi.org/10.1063/1.4940659.
[22] McTavish S, Feszty D, Nitzsche F. Evaluating Reynolds number effects in small-scale wind
turbine experiments, Journal of Wind Engineering and Industrial Aerodynamics, 120 (2013) 81-90.
https://doi.org/10.1016/j.jweia.2013.07.006.
http://en.cnki.com.cn/Article_en/CJFDTOTAL-KQDX201104024.htm
of a Nonlinear lifting line-free vortex wake module within the wind turbine simulation code Qblade.
Journal of Engineering for Gas Turbines and Power, 138(7) (2016) 72601.
https://doi.org/10.1115/1.4031872
[25] Martin H, Development of a scale model wind turbine for testing of offshore floating wind
https://digitalcommons.library.umaine.edu/etd/1578
[26] Du W, Design and Analysis of a Model Wind Turbine Blade for Wave Basin Test of Floating
http://www.wanfangdata.com.cn/details/detail.do?_type=degree&id=D575663
[27] Frandsen S, Barthelmie R, Pryor S, et al. Analytical modelling of wind speed deficit in large
[28] Göçmen T, Van der Laan P, Réthoré P E, et al. Wind turbine wake models developed at the
technical university of Denmark: A review. Renewable and Sustainable Energy Reviews, 60 (2016)
752-769. https://doi.org/10.1016/j.rser.2016.01.113.
[29] Shakoor R, Hassan M Y, Raheem A, et al. Wake effect modeling: A review of wind farm layout
optimization using Jensen’s model, Renewable and Sustainable Energy Reviews, 58 (2016) 1048-
1059. https://doi.org/10.1016/j.rser.2015.12.229.
[30] Leishman J G. Challenges in modeling the unsteady aerodynamics of wind turbines, Wind
[31] Chen Q. Three-dimensional PIV wind tunnel experimental study on wake characteristics of
(2001).
[33] Afjeh A. Wake effects on the aerodynamic performance of horizontal-axis wind turbines:
[34] Lanzafame R, Messina M. Fluid dynamics wind turbine design: critical analysis, optimization
https://doi.org/10.1016/j.renene.2006.12.010.
[35] Kim B, Kim W, Lee S, et al. Development and verification of a performance based optimal
design software for wind turbine blades, Renewable Energy, 54 (2013) 166-172.
https://doi.org/10.1016/j.renene.2012.08.029.
[36] Hamilton N, Melius M, Cal R. Wind Turbine Boundary Layer Arrays for Cartesian and Staggered
Configurations‐part I, Flow Field and Power Measurements, Wind Energy, 18 (2) (2015) 277-295.
http://dx.doi.org/10.1002/we.1697
27
Journal Pre-proof
article.
intellectual content.
published
Journal Pre-proof
1 Harbin Institute of Technology, School of Energy Science and Engineering, 92 West Dazhi
2 Northeast Electric Power Univeristy, School of Energy and Power Engineering, 169
Corresponding author:
Address: School of Energy and Power Engineering, 169 Changchun Rd, Jilin City, China.
Address: School of Energy Science and Engineering, 92 West Dazhi Rd, Harbin, China
Acknowledgments
This work has been supported financially by the National Natural Science Foundation of
China (Grant No.51708151) and the China Postdoctoral Science Foundation (Grant
No.2019T120274), which is gratefully acknowledged by the authors. We are also grateful for
the academic help provided by Dr. Jiarong Hong, Mr. Bingzheng Dou, Dr. Yun Liu, and Dr. Teja
A new small lab-scale wind turbine similar to the prototype has been designed.
The design process provides a hybrid model blade similar to the prototype.
Performance of geometric scaled model is significantly less than the objective value.