Delft University of Technology

Rotor / Wake Aerodynamics

Assignment 1: BEM

Patrick Duffy (4756444)

Bose Sumantraa (4774302)
Simone Tamaro (4714229)

March, 2018
Contents

Introduction 2

Code flow chart 3

Task 1 4
Turbine Rotor Specifications . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4
Main Assumptions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4
Results and Plots . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5
Axial and Tangential Induction Factors vs. r/R . . . . . . . . . . . . . . . . . . . . . . . . . . 5
Inflow Angle and Angle of Attack vs. r/R . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5
Force Distributions and Coefficients Along Rotor Blade . . . . . . . . . . . . . . . . . . . . . 6
Thrust and torque coefficients vs. r/R . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 7
Prandtl tip loss correction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 8
Influence of Number of Annuli and effects of clustered grids . . . . . . . . . . . . . . . . . . . . . . 9
Convergence History . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 10
Tip and root sections . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 10
Medium sections . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 11

Task 2 13
Theory of the optimum rotor . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 13
Optimization process . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 13
Chord distribution . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 13
Twist distribution . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 14
Pitch angle . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 15
Post process . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 15
Considerations on force distribution and circulation . . . . . . . . . . . . . . . . . . . . . . . . . . . 17
Comparison with the actuator disk theory . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 18

Conclusions 20

1
Introduction
The Blade Element Momentum Theory provides a method for analyzing the performance of a wind turbine
rotor. It combines analysis momentum interaction with an actuator disk with analysis of individual blade
elements. This is important because even if there exists a turbine rotor with uniform oncoming wind velocity
U0 , the wind velocity experienced at different elements along radial locations of the blade Urel due to the
changes in radius. Assuming 2-Dimensional flow at each element, the aerodynamic forces per unit length
can be calculated at each based on the local angle of attack and inflow angle. The per unit length forces are
then integrated along the blade to compute the total aerodynamic forces acting along the blade. The blade
torque and power are then computed which gives insight into the aerodynamic performance of a particular
rotor design. The key is that the computation of the local velocity triangle for each blade element relies on
an iterative solution process for the induction factor a and axial induction factor a0 . A computer program
facilitates this process and a flowchart of the code is presented in the next section.

2
Code flow chart

3
Task 1
In the first task the analysis of a specific rotor blade is conducted using a Blade Element Momentum code
outlined in the Code Flowchart based on the theory described in Burton’s ’Wind Energy Handbook’.

Turbine Rotor Specifications

The exact details of the investigated blade geometry and operational conditions are provided below in Table
. A single airfoil is used across the span of the rotor blade with consistent polars.

Property Value
Radius 50 [m]
No. Blades 3
Blade starts at 0.2 R
Twist Distribution 14 · (1 − r/R)
Blade Pitch Angle -2 ◦
Chord Distribution (3 · (1 − r/R) + 1) [m]
Airfoil DU 95-W-180
Yaw angle 0 [degrees]
U0 10 [m/s]
λ 6, 8, 10

Main Assumptions
The blade element momentum theory makes several physical assumptions must be recognized in a perfor-
mance analysis. The following summarizes the key assumptions (impacts explained below):

• Steady, incompressible flow, no mass flow through streamtube boundary

• Individual blade elements are independent of each other
• Uniform upstream wind velocity normal to the rotor plane

• Actuator disk is a rotor with an infinite number of blades

• Original Blade Element Momentum Theory ignores losses at the root and tip

No mass flow across the streamtube boundary and the fact that the flow is steady and incompressible means
that the streamtube can be broken up into annuli (sections). The blade elements that occur at each annuli
are independent of each other. This means that there is no spanwise flow along the blade (the flow over each
thin airfoil element is 2-Dimensional). This highlights the key assumption that the force of a blade element
is responsible for the change of momentum of the wind flowing through that respective element’s annulus.
This is technically true only when the axial induction factor does not vary with radial position along the
blade (uniform circulation), but Burton’s Wind Energy Handbook cites experiments demonstrating that this
assumption is acceptable. 3-Dimensional effects are therefore ignored.

It is also assumed that the flow has a uniform velocity normal to the rotor plane. This means that the
upstream wind velocity is the same at any point in the rotorplane and there is no yaw angle. This combined
with the previous assumptions means that the mass flow rate is the same at each streamwise position in
the streamtube. This indicates the cross-sectional area of the streamtube expands after the rotorplane to
account for the velocity reduction. The uniform upstream velocity assumption allows for a simpler treatment
of each blade element because it does not matter where in the rotational cycle a blade element is, the velocity
experienced will depend only on the velocity triangle. If there were to be wind shear or velocity variation

4
over the rotor plane, the variation in oncoming velocities, and therefore forces and moments, would need to
be accounted for.

Original Blade Element Momentum Theory assumes a solid actuator disk which is equivalent to assuming a
rotor with an infinite number of blades. As this is not the case this must be accounted for with the Glauert
correction. An additional correction relates to assumptions of an ideal rotor in the original theory - Prandtl’s
tip and root correction are therefore also applied.

Results and Plots

When the Blade Element Momentum Theory with Glauert and Prandtl corrections is utilized to analyze the
performance of the rotor blade defined above, the following results are obtained. Plots of the key parameters
are displayed in the following subsections.

Axial and Tangential Induction Factors vs. r/R

It is observed that with increasing tip speed ratio (λ), the drop-in wind velocity at the rotor plane is
significant. However, it almost remains constant for major span of the blade at a fixed λ. The significant
drop in velocity at the rotor plane for higher λs explains that the rotor-disc starts behaving as a direct
obstruction which blocks the wind from flowing around the blades and hence no ideal flow conditions are
created around the aerofoil. This decreases the effective lift generated which is responsible for lower power
production. Plots of a and a0 vs. Rr are displayed in Figure ??.

0.8
0.3
=6
=6
0.7 =8
=8
=10 0.25 =10
0.6
0.2
0.5
aprime [-]
a [-]

0.15
0.4

0.1
0.3

0.2 0.05

0.1 0
0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1
r/R [-] r/R [-]
(a) a vs. r
R
(b) a0 vs. r
R

Figure 1

Inflow Angle and Angle of Attack vs. r/R

Inflow angles are found to be lower for higher tip speed ratios. This is because, inflow angle is inversely
proportional to the rotational speed. Increasing rotational speed of the rotor directly reduces the inflow
angle. But, for a constant value of rotational speed, it is highest at the inboard section of the blade and
gradually decreases until the tip of the blade. For inboard part of the blade, the magnitude of velocity due to
rotation of the blade is lower compared to the tip of the blades and hence inflow angles are much larger near
the hub. For the outboard part of the blade, the magnitude of rotational velocity of the blade element is

5
much larger due to increased distance from the rotational center. Hence, a very low inflow angle is observed
at the tip section of the blade.

20 30
=6 =6
18 =8 =8
=10 25 =10
16

14 20

[deg]
12
[-]

15
10

8 10

6
5
4

2 0
0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1
r/R [-] r/R [-]
r r
(a) α vs. R
(b) φ vs. R

Figure 2

For optimum operation for the rotor, each blade element must operate at Maximum lift to drag ratio. The
characteristics of an aerofoil (Cl, Cd) are very sensitive to changes in angle of attack. Hence, angle of attack
for different radial elements of the blade has to remain same to effectively have the same L/D ratio. Angle
of attack is influenced by both the pitch and twist distribution at each radial position. Values of pitch and
twist are tweaked in such a way to achieve a constant value of angle of attack at different r/R.

Force Distributions and Coefficients Along Rotor Blade

The normalized axial and tangiental forces along the blade are presented in Figures 3a and 3b. The normal-
ized axial force linearly increases with r/R, while the normalized tangiential force remains approximately
constant. The normalized axial force becomes high towards the tip region due to large inflow angles. The
importance of this will be highlighted in the optimization section of the report.

6
 1.8 0.16
1.6
0.14

Normalized Azimuthal Force

1.4
Normalized Axial Force

0.12
1.2

1 0.1

0.8 0.08 =6
=8
0.6 =10
0.06
0.4
=6
=8 0.04
0.2
=10
0 0.02
0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1
r/R [-] r/R [-]
(a) Normalized axial force Fax vs. a (b) Normalized azimuthal force Faz vs. a

Also presented are plots of the force coefficients at each radial location of the blade. These are used in the
calculation of the actual forces acting at each section of the blade.

0.4 1.2
=6
0.35 =8 1.1
=10
0.3
1
0.25
0.9
C n [-]
C t [-]

0.2
0.8
0.15
0.7
0.1
=6
0.05 0.6 =8
=10
0 0.5
0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1
r/R [-] r/R
r r
(a) Ct vs. R
(b) Cn vs. R

Figure 4

Thrust and torque coefficients vs. r/R

The thrust coefficient is a measure of blade loading. CT is not directly related to aerodynamic efficiency, but
indirectly shows how much the energy extraction affects the fluid flow and hence is of practical importance
for structural considerations. The wind turbine imparts a force on the wind which slows the wind down to
extract kinetic energy, the reaction force to this is the thrust force the wind imparts on the turbine and is
significant for aerodynamic loads. The ideal value of thrust coefficient that maximises Cp is 8/9, according
to the actuator disk theory. For higher λs, thrust coefficient exceeds the ideal value of 8/9 and hence
suffers lower Cp . The higher value of thrust coefficients can be explained on the same lines as increasing
axial induction factor with increasing λs induces flow blockage. At a certain point in rotor design marginal
increases in the extracted power are associated with significant increases with the thrust forces, leading to

7
a cost tradeoff evaluation regarding structural requirements for increases in performance. The local thrust
and torque coefficients along the blade radius are presented below.

10-4
1.5 3.5
=6
=8 3
=10
Thrust coefficient C T [-]

2.5

C Q [-]
1
1.5

=6
0.5 =8
=10
0.5 0
0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1
r/R [-] r/R [-]
r r
(a) Thrust coefficient CT vs. R
(b) Torque coefficient CQ vs. R

Figure 5

Prandtl tip loss correction

The tip loss can be defined as the high local inflow that is produced at the tip of the blade due to the trailing
vortices at the tip. This correction factor accounts for the changes in the induced velocities at the tip due to
finite number of blades. The tip of the blades shed vortices, and this changes the local axial induction factor
which must be accounted for. The effect is clearly pronounced near the hub and tip part of the blade, as this
is where the vortices are shed. High values of axial induction factor near the hub and tip section explains
the reduced magnitude in correction factor.

0.9

0.8
=6
0.7 =8
=10
F [-]

0.6

0.5

0.4

0.3

0.2
0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1
r/R [-]
r
Figure 6: Prandtl’s Tip Correction F vs. R

8
 0.9 0.9

0.8 0.8
Thrust coefficient C T [-]

Thrust coefficient C T [-]

0.7 0.7

0.6 0.6

0.5 0.5

0.4 =6 0.4 =6
=8 =8
=10 =10
0.3 0.3
0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1
r/R [-] r/R [-]
(a) Thrust coefficient CT vs. r/R (b) Thrust coefficient CT vs. r/R no tip/root correction

Figure 7

Influence of Number of Annuli and effects of clustered grids

Figure 8: Sketch of one annulus δr on the blade inspected.

Spacial discretization was inspected by varying the number of annuli considered in the BEM process. Two
different grids were considered. First a linearly spaced one, and secondly a clustered grid for higher precision
in the tip region. This approach allows to focus computational power availability on the most efficient
outboard sections. In this way, inner regions can be computed with a coarser grid with saves in computational
time with a low loss in precision.
In this report, a sine function was computed between 0 and π/2 to obtain the discretization displayed in
Figure 9.

9
 1

-1
0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1
r/R
Figure 9: Clustered grid inspected with 40 point.

The results are visible in Figure 10.

0.5

0.45

0.4

0.35
C p [-]

0.3
linear
0.25 clustered grid

0.2

0.15

0.1
100 101 102 103
Number of annuli

Figure 10: Influence of the number of annuli for λ = 9, pitch angle of -2◦ .

It can be observed that in order to obtain a correct solution for this blade, at least 32 elements are required
for both techniques. It is evident that although the convergence seems to be slightly faster, the use of a
clustered grid does not provide any relevant benefit. This is mainly because for this blade one single airfoil
was considered, and therefore no distinctions can be made between the medium and the tip sections of the
blade.

Convergence History
Convergence history was inspected for the thrust coefficient CT . The obtained results were divided in three
categories because of the different trends observed. Specifically, root, tip and medium sections of the blade
showed typical behaviours for CT convergence.

Tip and root sections

In these areas the computation of the thrust coefficient is highly affected by the Prandtl corrections.
Figure 11a shows convergence for the root sections. It can be seen that the very first sections have quicker

10
convergence although the steps in each iterations are quite significant. The thrust coefficient is seen to
increase along the blade span, as long as the number of iterations required to achieve convergence.
Figure 11b inspects the tip area of the blade. For this region, the thrust coefficient is seen to decrease when
the tip of the blade is approached. A smoother convergence is observed if compared with the first sections
of the root area.

0.8 0.78

0.75 0.77
Computed thrust coefficient

Computed thrust coefficient

0.7
0.76
0.65
0.75
0.6 R

0.74
0.55

0.5 0.73
R

0.45 0.72
0 2 4 6 8 10 12 0 5 10 15 20
Iteration number Iteration number

(a) Root sections. (b) Tip sections.

Figure 11: History convergence for the thrust coefficient CT with a tolerance of 10−5 .

Medium sections
The obtained convergence history for the medium sections is plotted in Figure 12.

0.78
R
0.76
Computed thrust coefficient

0.74

0.72

0.7

0.68

0.66

0.64

0.62
0 5 10 15 20
Iteration number

Figure 12: History convergence for the thrust coefficient CT in the medium sections of the blade with a tolerance of
10−5 .

These sections of the blade show a typical converging trend. The number of iterations required is pretty
much constant for all sections inspected. However, it can be noticed that the increase in CT achieved in
every iterations is higher close to the root, while littler variations per iteration are obtained when outboard

11
sections are considered.

12
Task 2
Theory of the optimum rotor
For an optimum operational state of a wind turbine rotor, each radial section must ideally operate at the
following conditions:
• maximum L/D ratio for each annulus
• constant value of a over a significant part of the blade
• small tangential induction factor a0 over a significant part of the blade.

An equation for the optimal chord distribution can be found in literature [2]:

8πaR
c(λ, r/R) = (1)
Bλ2 (r/R)C l

Where B is the number of blades, r/R indicates the non dimensionalized span of the blade, λ is the design
value of tip speed ratio and c is the chord length.
From the equation above it is clear that in order to maintain total rotor lift at the fixed optimum value, the
local chord must vary inversely with the square of the tip speed ratio.
An optimal twist distribution is also found to be:
 1−a 
θ(r/R) = atan − αmax,Cl (2)
λr/R(1 + a0 )

The lift produced by an aerofoil section represents the bound circulation which is virtual over the span of the
blade but becomes a real vortex at the tip of the blade. These tip vortices are modelled by the Prandtl tip
loss correction factor in the BEM theory. It is not desirable to have rapid changes in the span-wise variation
of lift force (the gradient of lift is directly proportional to the trailing vortices). If this is not satisfied, then
more trailing vortices are shed, and this causes more induced drag. Hence, a sudden change in chord is
undesirable both structurally and aerodynamically. This also confirms why Cl should not abruptly change
since the relative inflow velocity varies gradually over the span of the blade.

Optimization process
As part of the optimization process, chord, twist and pitch angles were considered separately. The linear
distributions for chord and twist were chosen as benchmarks to compare the results, as long as the suggested
pitch angle of -2◦ . Apart for the optimal distributions mentioned in Equations 1 and 2, other functions were
considered.

Chord distribution
Four different profiles were obtained by using different equations. A chord of 3.4 m at 20% of the blade
span length and a final chord of 1 m at the tip were used as reference points. The obtained geometries are
represented in Figure 13.

13
 3.5
linear
x-1
3 logarithm
parabola 1
Chord distribution [m]
parabola 2
2.5

1.5

0.5
0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1
r/R [-]

Figure 13: Plot of the different chord distributions obtained.

The functions plotted above are described in Table 1.

Label Function c(r)

linear c = 3(1 − r/R) + 1
30 2
∼ r−1 c= +
r 5
logarithm c = −3.48log10 (r) + 6.88
parabola 1 c = −0.001r2 − 0.015r + 3.63
parabola 2 c = −0.002r2 + 0.06r + 3

Table 1: Functions inspected for chord distribution

Twist distribution
The same procedure was followed for the twist distribution θ(r). In this case, an optimal distribution was
also inspected [1]. The other profiles were obtained trying to inspect different levels of overall and local
twisting. 20◦ was chosen as maximum twisting due to manufacturing and structural constraints. The results
are visible in Figure 14.

14
 20
linear
optimal( )
parabolic
15 logarithmic

Twist angle [°]

10

0
0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1
r/R [-]

Figure 14: Plot of the different twist distributions obtained.

Once again, the functions plotted are summarized in Table 2.

Label Function θ(r)

linear θ = 14(1 − r/R)
2/3
optimal θ= −8
λ(r/R)
parabolic θ = 0.005r2 − 0.55r + 15
logarithmic θ = −28.59log10 (r) + 48.57

Table 2: Functions inspected for twist distribution.

Pitch angle
The optimal pitch angle was inspected for each combination of twist and chord distribution by looping over
an appropriate interval. The best pitch angle for all configurations turned to be the one that provided the
optimal angle of attack α in terms of the lift to drag ratio. This is because α is almost constant in the
medium sections (40% - 90%) and therefore setting it to the optimal value is crucial to increase the power
coefficient. Specifically for the DU95W180 airfoil, the optimal angle of attack is 8.4◦ .
The optimization design should therefore be focused on obtaining an angle of attack close to the optimal.
For a real blade, this design driver should be satisfied especially in the outboard sections where airfoils are
more efficient. It is however important to underline that this approach does not take into account structural
issues and constraints.

Post process
To maximize the Cp for a fixed thrust coefficient of 75%, the different distributions described above are
combined together. Starting with a TSR of 9, different chord distributions were combined with linear twist
distributions. The pitch angle was set to approximately obtain a CT of 0.75. The same procedure was
followed by varying the twist distribution with a linear chord distribution. The results are visible in Table
3.

15
 Chord distribution Twist distribution [◦ ] Pitch angle [◦ ] Cp CT
r−1 linear -5.6 0.4651 0.7535
logarithm linear -4.2 0.4714 0.7548
parabola 1 linar -1.6 0.4601 0.7503
parabola 2 linear -1 0.4574 0.7537
linear linear -2.7 0.468 0.7546
linear optimal 2.7 0.4689 0.7546
linear parabolic -0.8 0.469 0.7499
linear logrithmic -3.5 0.4582 0.7519

Table 3: Optimization process for λ = 9

The best chord and twist distributions were at this point combined to inspect their interaction. The pitch
angle was again set to limit CT to 0.75. The results are displayed in Table 4 for two different combination.

Chord distribution [m] Twist distribution [◦ ] Pitch angle Cp CT

logarithmic parabolic -2.3 0.4745 0.7502
logarithmic optimal 1.3 0.4741 0.7487

Table 4: Optimal combinations of chord and twist distributions

The most efficient combination was chosen to be further analyzed. Hence, the logarithmic chord distribution
and the parabolic twist distributions were chosen. Figure 15 shows the trend of the main angles along the
blade span.

20
inflow

opt
15
twist

pitch
10

-5
0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1
r/R [-]

Figure 15: Plots obtained with a logarithmic chord distribution and a parabolic twist distribution. The angles are
expressed in degrees.

It can be seen that the angle of attack α is pretty much constant in the medium sections of the blade. Its
offset with the optimal angle of attack which maximize L/D is due to the imposition of CT = 0.75. It can be
however said that the angle of attack obtained is still in a the most efficient range for the inspected airfoil.

16
Considerations on force distribution and circulation
In order to inspect the quality of the optimal configuration obtained, the axial force obtained was plotted.
The result is visible in Figure 16

6000

5000

4000
F axial [N]/[m]

3000

2000 F axial
linear fit

1000

0
0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1
r/R [-]

Figure 16: Plot of the axial force for the optimal configuration.

It can be noticed that the trend is extremely linear in the medium sections of the blade. This is a good sign
because it means that uniform loading is present on the blade. Uniform loading is an important aerodynamic
feature of a wind turbine blade. This comes from the actuator disk theory when the force distribution is
uniform.
It is known that uniform loading takes to constant circulation along the blade. This is showed below.
Two airfoils in a blade experience a normal loading as showed in Figure 17. Γ is the circulation on the airfoil,
Ω is the rotation speed of the rotor.

Figure 17: Axial force for a root and a tip section airfoils NREL s809 and s817.

17
If the normal loading is uniform, we obtain that:

FN,B FN,A FB FA
= =⇒ =
2πrB δr 2πrA δr rB rA

− ρΓB Ω
r
B − ρΓA Ω
r
A
=
r
B
 r
A

ΓB = ΓA
This result means for an aerodynamically ideal rotor the circulation must be constant all along the span of
the blade. This guarantees that trailing vortices are shed only at the tip and at the root of the blade. As
expressed by the actuator disk theory, this feature ensures that no jumps of energy are present inside the
streamtube of the turbine. The plot of the circulation for the optimal design inspected is showed in Figure
18.

0.9

0.8

0.7

0.6

0.5
* [-]

0.4

0.3

0.2

0.1

0
0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1
r/R [-]

Figure 18: Normalized circulation plotted against the normalized radial position.

The same procedure was followed for λ = 8 as well. In this case it is relevant to mention that the con-
figurations that proved to be the best for λ = 9 do not prove to be the best. Generally the values for Cp
obtained are lower, but still an improvement can be obtained when a parabolic function is chosen for the
twist distribution and the chord distribution is kept linear.

Comparison with the actuator disk theory

By applying the actuator disk theory, the rotor is seen as a disk with which applies a constant force distri-
bution to interact with the flow. This theory provides an equation for Cp and CT as functions of the global
axial induction factor a which can be evaluated from the comparison between the upstream and downstream
wind speeds.
By combining the actuator disk model with the aerodynamics of local blade annuli, BEM can be effectively
used to inspect the aerodynamics of a wind turbine. For an optimal rotor with constant circulation and
hence constant axial induction factor a in the inner sections, it can be assumed that the main differences
between BEM and actuator disk model are represented by the Glauert and Prandtl corrections.
To inspect this, the average value of a was inspected in the medium section of the blade (20% - 80%) for the
optimal design described above. The result is an axial induction factor of 0.28. This approach is meaningful

18
because the trend of a is pretty much constant along the medium sections. One can use this value of a to
compute the power coefficient and thrust coefficient with the actuator disk theory. The results are:

CT = 4a(1 − a) = 0.81

Cp = 4a(1 − a)2 = 0.58

The results obtained are optimistic as it is expected. In fact, the Prandtl tip correction introduces relevant
losses at the blade tips. With BEM, the power coefficient is seen to be reduced of more than 10%. The
thrust coefficient is also reduced of around 6%.

19
Conclusions
The application of a BEM code can provide good information regarding a wind turbine performance. Even
if the implementation of the code is relatively simple, some valuable outputs are provided by this method.
The application of the Galuert and Prandtl corrections allow to estimate the power extracted from the wind
as long as structural parameters like thrust force coefficient with a good precision.
The limitations of the method are due to the heavy hypothesis imposed like, for instance, the assumption
of radial independence of the annuli. In case more precise output are needed, BEM cannot be the method
chosen.

The optimization process allowed to inspect different blade designs. The driver towards the design of the
optimal rotor was mainly aerodynamic. No economical considerations were made. The thrust coefficient was
limited due to structural needs, while the twist and chord distributions were limited by possible manufac-
turing constraints.
It could be observed that the different parameters are strictly interconnected and the optimal combination is
most times not obvious. However, a main optimization driver can be a constant circulation along the blade
span, as was explained in the relative section of the report.

20
References
[1] TU Delft Course AE4W09: Wind Turbine Design, System design and scaling, February 2018
[2] Innovation in wind turbine design Jamieson P., Hassan G., page 31, first edition, Wiley, 2001.

21