von Karman Institute for Fluid Dynamics

Influence of Wind Turbine Wakes on the Flight of

6/05/2017

André Tristany Farinha

Abstract

The effect of the presence of the wake created by an Enercon E-82 E2 – 2300 kW wind turbine
on the flight of an endurance UAV used for civil support is evaluated. The flow in the wake of
the wind turbine is modelled by estimating the velocity deficit and the turbulence in the wake.
These quantities are calculated using analytical models that have been validated by
comparison with field measurements and CDF simulations where an actuator disc is used to
model the rotor. The standard deviation of the velocity oscillations scaled by the integral
length scale in the wake is used as a perturbation on the aircraft flight. This perturbation is
introduced in a simplified fixed wing aircraft flight dynamics model in order to assess the
occurrence of stall, lift loss and loss of climb rate. The effect on the rolling of the aircraft is
evaluated as a static rolling moment and compared to the reference non-dimensional values
in the literature.

1
1 Introduction

The main objective of this work is to create a simple tool that can to some extent quantify the
effect of the presence of wind turbines near an airfield on the flight of small aircraft. Besides
that, given the large number of accidents that occur every year on the inspection of wind
turbines, some more relevance is given to UAV’s which is a tool that could minimize the
number of inspections with human presence. Although rotorcrafts are a more obvious choice
for this task, high endurance fixed wing aircraft have been used for decades in surveillance
tasks, both for military and civil applications and are thus worth considering.

The effect of wind turbine wakes on aircraft has been studied in the past by different authors
and using a variety of different methods. Glabeke, et all [1] use the actuator disk theory to
model the wind turbine and use a RANS method to solve the flow in the wake. The results
obtained from the CFD analysis are then used to calculate the evolution of the standard
deviation of the velocity u’ assuming isotropic turbulence and the integral turbulent length
scale L downstream along the axis of the rotor. The aircraft inside the wake is then assumed to
be located in the centre of a vortex of diameter L and tangent velocity u’. Being estimated the
perturbation acting on the aircraft wing; the change in angle of attack, the roll movement and
resulting climb rate difference is estimated. The most hazardous effect was found to be the
roll movement.

On a different approach, Wang et all [2] use an adapted vortex wake model used to study
rotorcraft wakes for a wind turbine wake, developed by Kokurek [3], which shows good
agreement with field measurements of wind turbine wakes. The authors then performed
piloted flight simulations where the pilot’s control activities where recorded. It was found that
the generated disturbances were minimal, requiring small corrections from the pilots and that
the most hazardous effect was on the yaw degree of freedom.

Tomaszewski et all [4] take yet a different approach by using Large Eddy Simulations to
model the wake of a wind turbine that is modelled using the actuator disk method. Different
trajectories of the aircraft are tested, where the wake is taken as frozen turbulence and the
aircraft wing is divided into 8 segments so that the lift distribution is calculated and the size of
the aircraft is taken into account. The aircraft is considered as static and not interacting with
the flow, being the aerodynamic forces acting on the aircraft measured. The hazardous effect
of the presence of the wake is once more qualified as low.
2
2 Wind Turbine Wake Modelling

The modelling of wind turbine wakes is an active research topic on CFD, the results obtained
with this tool are often used to create analytical models that can predict the properties of the
wake. These models often describe the velocity deficit in the wake which is created by the
power that is extracted by the turbine from the wind. There are numerous of these models
being actively used in industry to optimise wind turbine placement [5], being one of the most
known the Jensen model. This model takes as input atmospheric boundary layer information
and the information on the wind. As for the modelled turbine dimensions and power
extracted; and calculates a wind deficit as a step that grows according with the wake.

2.1 Wake Velocity Deficit Analytical Model

The model used for this project is similar to Jensen’s model in that it calculates the wind
deficit in the wake and takes as input the characteristics of the present atmospheric boundary
layer and dimensions and power extracted by the wind turbine. The wind deficit is however
defined as a Gaussian profile whose standard deviation increases as we move downstream,
creating the growth of the wake. The wind deficit profile is axisymmetric relatively to the
wind turbine axis and is given by equation (2.1).

U  CT  
  exp   1   z  z 2  y 2  
 1 1        (2.1)
h

U 
 
8 k x d 0  0.2
*
 
  
 2 k x d 0  0.2 
*
   d0   d0   
 

Where U  is the wind velocity measured at hub height, CT the thrust coefficient, β is

function of CT , d 0 the diameter of the rotor, z0 the hub height and k * is the growth rate of the
wake. The growth rate of the wake is a parameter that needs to be fitted to experimental or
CFD results as it is done by the authors of the model in [6].

2.2 Wake Turbulence Analytical Model

Though there are numerous analytical models that estimate the velocity deficit in the wake of
wind turbines, the ones that model the turbulence properties in the wake are not so numerous.
One such model is developed by Crespo and Hernández in [7] where the expressions that

3
compose the model are created either through extending atmospheric boundary layer theory to
the wake or through comparison with experimental results or numerical results obtained with
the in-house developed CFD code UPMWAKE.

The value of the integral length scale in the wake is obtained assuming that the turbulence
spectrum in the wake has the same form as the one in the unperturbed atmospheric flow and is
given by equation (2.2) where k and ε are respectively the local turbulence kinetic energy and
the turbulent dissipation rate.

2.5  k 5.47 
32

L (2.2)


The turbulence kinetic energy and the turbulent dissipation rate are given by analytical
expressions developed for the near wake region d0  x  3  d0 and the far wake region

5  d0  x  15  d0 . At first the expressions for both regions were used for their domains, but
the discontinuity created by the different expressions created a large discontinuity for the
length scale that caused the later to be larger in the near wake region than in the beginning of
the far wake. A smoothing of the transition was attempted through the usage of the level rule
but the results didn’t improve sufficiently and a decision on not using the near wake model
was taken. The expressions for k are given as a turbulence intensity increase in equation (2.3),
being k calculated taking into account the turbulence anisotropy in the atmospheric flow:

k  0.95 TI U   .
2

0.32

    x 
0.8325
TI max  0.73  0.5 1  1  CT TI 0.0325
   (2.3)
 d0 

U 3  TI max
2.77
  0.14 (2.4)
z

The turbulent dissipation is given by equation (2.4), where z is the turbulent length scale in
the unperturbed atmospheric flow, given by the distance to the ground. Notice that these
expressions are not the turbulence intensity or dissipation rate profiles in the wake but the
values corresponding to the top region of the wake where the highest turbulence intensity
normally occurs. Thus z should correspond to the height at which the TI max is taken.

4
The velocity standard deviation u' can then be calculated as in equation (2.5), when the
«anisotropy in the atmospheric flow is taken into account.

u '  1.026k 0.5 (2.5)

2.3 Models’ Setup

The parameters necessary to setup the models described above are either wind turbine
parameters or atmospheric flow parameters. As for the wind turbine, the necessary parameters
are the hub height z0 = 108 m , rotor diameter d0 = 108 m and thrust coefficient CT which
was taken from the manufacturer’s website [8] and is shown in Figure 2.1. For the purpose of
simplifying the implementation of the models a 4th order polynomial fit of the curve was used.
Although the fit is considered sufficiently close to the original curve, it should be stressed that
there is a shift in the velocity at which the maximum CT occurs and that the curve is not used
outside the bounds of the data available.

0.6

0.5

0.4
CT

0.3

0.2

0.1

0
0 5 10 15 20 25
hub height velocity [m/s]

Figure 2.1: Thrust coefficient data as function of the hub height velocity as extracted from [8]
and fourth order polynomial fit.

As for the atmospheric flow parameters, the unperturbed velocity profile in not exactly
necessary to setup the models (only necessary at hub height) but it will be useful for the
description of the environment around the aircraft flight. The velocity profile in the
unperturbed flow is taken from the VDI 3783 standard [9] given by equation (2.6) where U ref

5
is the velocity measured at a reference height and  is a parameter that accounts for the
roughness in the ground.


U z  z  d0 
  (2.6)
U ref  z d
 ref 0 

The roughness level is taken as in the boundary between rough and moderately rough, as
construction in the vicinity of airfields is normally not completely prohibited, though
regulated (   0.18 ).

The choice of the wake growth parameter k * in the wake deficit model was made taking into
consideration the results in [6], where the growth parameter is chosen so that the calculations
match LES results of different wind turbines in terrains with different roughness levels. The
chosen value was k*  0.04 that corresponds to a wind turbine 80 m rotor diameter, 70 m hub

height in a terrain classified as moderately rough.

As for the turbulence properties upstream of the wind turbine, the turbulence intensity is also
taken from the same standard and for the same roughness as in Figure 2.2, where the value
chosen is 10%.

Figure 2.2: Turbulence intensity values as in [9] for slightly rough, moderately rough, rough
and very rough terrain respectively from the left.

Engineering standards for atmospheric flow don’t currently include estimations for the
turbulent dissipation rate in the flow, the value used for this work was taken from [10].

6
Assuming a typical onshore atmospheric boundary layer height greater than 1 km, neutral and
stable boundary layer, the ε profile is given by equation (2.7). Where  is the von Kármán
constant and u* is the friction velocity given by u*  0.428k 1 2 when the typical turbulence
anisotropy in ABL is taken into account. Taking the previously presented relation between k
and TI, the value of  will vary with the incoming wind speed.

u*3
  1.24 (2.7)
z

As the TI is taken in the wake turbulence model as the highest value in the wake, which
occurs near the top, the value of turbulent dissipation rate is also taken at the top of the wake,
according to the wake growth predicted by the velocity deficit model. The wake radius at a
certain distance from the wind turbine was taken as the region where U U  0.99 .

3 Aircraft Model

The effect of the turbulence in the wake on the aircraft was divided into two effects that
completely decoupled: a static moment that acts on the rolling degree of freedom and the 2D
dynamics that do not include the pitching degree of freedom.

3.1 Roll

The procedure taken to calculate the roll moment on the aircraft is the same as in [1], where
the velocity standard deviation u' scaled down from the integral length scale of the turbulent
flow to the aircraft wing span, as shown in Figure 3.1.

Figure 3.1: Scaling of the velocity perturbation according to the turbulent length scale and
wing span. [1]

7
The velocity perturbation felt on the aircraft is the evaluated as given by equation (3.1).

b
uw'  u ' (3.1)
L

This perturbation is then added to the incoming velocity at the aircraft on one wing so that the
lift is increased and subtracted to the incoming velocity on the other wing so that the lift is
decreased. The lift difference between both wings causes a rolling moment centred in the
aircraft fuselage. This effect is not introduced in the dynamics model because of the
complexity of a three dimensional trajectory in an aircraft flight.

3.2 2D Dynamics

The dynamics model used for this work is an extremely simplified approximation of the two
dimensional dynamics problem for a fixed wing aircraft. Perhaps the greatest simplification is
the lack of any control actions and the lack of an equation for the pitch rotation DoF. Pitch
was not taken into account partially because the position of the pressure and mass centre of
the aircraft were needed (that would require a full description of the lifting surfaces) and
because it would considerably increase the complexity of the take-off trajectory. After these
key simplifications the model left is described by equation (3.2) that basically consists in the
motion equation for a mass particle moving in a 2D planar domain.

 T  v  cos    D  v  cos     L  v  sin     m  ax 

T, D and L are respectively the thrust drag and lift that are functions of the velocity vector on
the airfoil reference system. Theta  is the thrust force inclination relatively to the horizontal
direction, which means that as there is no equation for rotation the pilot is considered to
perfectly correct any pitching movement. Gamma  is different from the angle of attack at the
airfoil in that it corresponds to the mass particle instantaneous velocity, while the angle of
attack  takes into account the orientation of the airfoil relatively to instantaneous velocity
summed with the perturbations and atmospheric conditions:

 v p  ech 
   0    cos 1   (3.3)
 vp 
 

8
In equation (3.3)  0 is the angle of attack in levelled flight, so the angle between the chord

line of the airfoil and the trust direction; ech is the unit vector of the chord line and v p is given

by:

v p  v  uw'  U atm (3.4)

As considering the cross-wind would require a 3D model for the flight dynamics, only the
parallel component of the velocity of the atmospheric flow in equation (3.4) is taken into
account. This means that the hazardous effect of the cross-wind will not be present but that
the effect of the velocity deficit in the wake will be felt. Gamma  is then simply given by

  tan 1  v y vx  .

The lift and drag forces are given by equations (3.5) which require the lift and drag
coefficients to be calculated.

1
L  CL   v 2p  S
2
(3.5)
1
D  CD   v 2p  S
2

The wing surface area can be calculated from equation where AR is the wing area ratio and b
the wing span.

S  b2 AR (3.6)

Note that CL and CD correspond to the finite wing lift and drag coefficients, which can be
calculated from the finite wing coefficients available in databases if the pressure distribution
in the wing can be considered elliptical [11]. Equation (3.7) gives us the finite wing drag
coefficient as function of the finite wing lift coefficient, where r is a constant that accounts for
non-elliptical pressure distribution along the span of the wing and is taken between 0.85 and
0.95 [11].

CL 2
C D  cd  (3.7)
 eAR

9
For the calculation of the CL is necessary to calculate the corresponding effective angle of

attack  eff that is a function of CL itself, which means that an iterative process is needed

when solving equation (3.8).

CL  eff 
 eff    (3.8)
 eAR

The available thrust, given the power supplied to the thruster is a complex problem that
cannot be addressed with reasonable accuracy with specifications on the thruster. Thus for
this work, the power available is maintained constant with P  v  T as long as the thrust does
not exceed a certain maximum value.

The algorithm used to solve the motion problem consists in ensuring dynamic equilibrium
given the velocity, inclination and position of the aircraft, by calculating the required
acceleration to balance the forces applied. The velocity and position used to calculate the
equilibrium in the next time step are obtained from equations (3.9).

vi 1  ai  t  vi
(3.9)
xi 1  vi  t  xi

While the velocity is used in next loop to calculate the forces acting on the aircraft, the
position vector is used to assess if the aircraft is inside the wake which on the other hand
influences the acting forces.

3.3 UAV Description

The aircraft chosen as a test case for this work is the PAIC Império SP1, a Portuguese
prototype Uninhabited Aerial Vehicle (Figure 3.2) designed for civil support missions such as
forest fire surveillance and coast patrol. This aircraft is an attempt of a high endurance UAV
designed for long flights and extended range. Some of the data that was not available about
the aircraft will be approximated based in the supposition that the designers took decisions
according to these objectives.

10
 Figure 3.2: PAIC Império SP1 UAV in flight. [12]

The information and specifications found on the UAV are the following:

 Wing span: 5 m

 Area ratio: 13

 Maximum take-off weight: 70 kg

 Cruise velocity: 33 m/s

 Maximum velocity 77 m/s

 Take-off velocity: 16 m/s

 Engine power: 17 hp

Based on the available information, the parameters to be calculated in order to setup the
model are the following: infinite wing thrust and drag coefficients, level flight required power
and angle of attack and maximum thrust.

Endurance drones are characterised by very large aspect ratio wings and considerably low
cruise velocities, so that the fuel consumption is as low as possible and its range as high as
possible [13],[14]. One airfoil that meets these requirements and has been previously used in
endurance UAV’s is the Selig S7075 (9%), which is a low Reynolds airfoil used mainly in
gliders.

11
 Figure 3.3: Selig S7075 (9%) low Reynolds airfoil.

The lift and drag coefficients were calculated using the widely used and tested code XFOIL
which uses an inviscid formulation with a liner vorticity function for the description of the
airfoil geometry. The Ncrit parameter that can be chosen in the software is related to the eN
method of linear stability, used to predict transition. The value taken for the calculation was 5.
Lift and drag curves for both cruise speed and take-off speed were taken (Figure 3.4 and ); but
implementing the CL and CD variation with freestream velocity increased the complexity of
iterating for the finite wing correction and thus the two curves were averaged and a 6 th degree
polynomial fit was used for the iterative procedure.

1.6
1.4
1.2
1
0.8
0.6
CL

0.4
0.2
0
-10 -5 -0.2 0 5 10 15
-0.4
-0.6
angle of attack [deg]

1.00E+05 5.00E+04 average

12
 0.12

0.1

0.08
CD

0.06

0.04

0.02

0
-10 -5 0 5 10 15
angle of attack [deg]

Figure 3.5: Average drag coefficient obtained for the S7075 (9%) airfoil with Reynolds
number equal to 100,000 and 50,000 and Ncrit 5.

Outside the range of angles off attack calculated the lift coefficient is taken as zero to account
for stall and the drag coefficient is considered as increasing linearly.

In level flight equation (3.2) can be simplified so that there is no acceleration in both
directions. The equations of motion are then reduced to equation (3.10), being possible to
calculate first the required lift for the level flight and thus the required angle of attack on the
airfoil. Calculated the angle of attack the drag can be calculated and thus the required thrust.

T  D  0 
  (3.10)
 L  W  0

The maximum thrust available from the motor + thruster was taken based on similar setups of
fuel motors with similar power, flying aircraft sensibly with the same mass. The value that
was finally taken was 150 N.

13
4 Results

4.1 Wake Description

The results obtained with the wake velocity deficit model yield results where as expected, the
highest velocity deficit is show in Figure 4.1 through Figure 4.3 to correspond to the wind
seep at which the thrust coefficient is higher, namely 8 m/s. The turbulence intensity results in
Figure 4.4 show that although both TI plots should eventually recover to the free stream
value, the 8 m/s case maintains the turbulence in the wake for a much longer distance. This is
in accordance with the results obtained with the velocity deficit model.

On the other hand, the turbulent length scale results show that L in decreasing in the free
stream direction. This is contrary to what is expected as in the wake is produced turbulence
with lower length scales than the atmospheric flow that should dissipate and thus swish the
highest energy scale for a larger value. In fact, the problem revealed to be extremely
dependant on the inflow turbulent conditions, different expressions from different literature
yielded extremely different results both in the size of L and in in its evolution. Some input
parameters showed an increase in L along the downstream direction but its size was not very
physical. Another attempt is shown in Figure 4.6, where the turbulence intensity is taken as 1
% instead of 10 % (an extremely low and unrealistic value). It can be seen the increase in L
for the 8 m/s case while for the 20 m/s case L increases only in the near wake.

Figure 4.1: Velocity profile at 2d0 when the wind speed at hub height is 20 m/s on the left and
when it is 8 m/s on the right.

14
Figure 4.2 Velocity profile at 5d0 when the wind speed at hub height is 20 m/s on the left and
when it is 8 m/s on the right.

Figure 4.3: Velocity at hub height as function of distance downstream from the wind turbine
for a wind speed at hub height of 20 m/s on the left and 8 m/s on the right.

Figure 4.4: Maximum turbulence intensity in the wake as function of distance downstream
from the wind turbine for a wind speed at hub height of 20 m/s on the left and 8 m/s on the
right.

15
Figure 4.5: Turbulent length scale at hub height as function of distance downstream from the
wind turbine for a wind speed at hub height of 20 m/s on the left and 8 m/s on the right.

Figure 4.6: Turbulent length scale at hub height as function of distance downstream from the
wind turbine for a wind speed at hub height of 20 m/s on the left and 8 m/s on the right, where
the incoming TI was taken as 1 %.

As the velocity perturbations decrease in the wake along with the length scale, it would be
possible that the perturbation in the wing would still decrease, but that is not what is observed,
as shown in Figure 4.7.

16
 Figure 4.7: Velocity perturbation scaled according to the aircraft wing span on turbulent
length scale along the direction downstream of the wake.

4.2 Wake Encounter in Perpendicular Level Flight

The first wake encounter case studied corresponds to a level flight perpendicular to the
direction of development of the wake and at a height of 100 m, as shown in Figure 4.8. As a
first comment, it can be seen in the following results that there is a very low lift force acting
on the aircraft at the beginning of the flight, which means that there is some numerical error in
the calculation of the required thrust for level flight.

Figure 4.8: Three dimensional domains where the trajectory of the UAV is shown in red
relatively to the wind turbine wake in blue.

17
As shown in Figure 4.9 through Figure 4.11 the perturbation introduced in the flow doesn’t
affect considerably the flight of the aircraft, although the effect of lift loss is easily
recognised. Besides that, it is shown how the effect on the aircraft is increasing with the
distance from the turbine, which is unrealistic and caused by the turbulent length scale.
Besides that, it is clearly visible the growth of the area in which the aircraft is affected by the
wake as the trajectories are moved downstream.

Figure 4.9: Trajectory of the UAV when crossing the wind turbine wake and rolling moment
felt by the aircraft during the flight at a distance of the WT equivalent to 3d0.

Figure 4.10: Trajectory of the UAV when crossing the wind turbine wake and rolling moment
felt by the aircraft during the flight at a distance of the WT equivalent to 7d0.

18
Figure 4.11: Trajectory of the UAV when crossing the wind turbine wake and rolling moment
felt by the aircraft during the flight at a distance of the WT equivalent to 7d0.

It is shown in Figure 4.12 that the aircraft system is stable as the effect of the perturbation
wears off in time leading to the accelerations on the aircraft to converge to zero and that even
inside the perturbed region the aircraft tends to stabilise.

Figure 4.12: Horizontal acceleration acting on the UAV when it is crossing the wind turbine
wake in a perpendicular trajectory.

4.3 Wake Encounter in angle level Flight

In an angle level flight the effect the velocity deficit in the wake should be felt besides the
perturbation in the flow, thus increasing even further the lift loss. This case corresponds to the
aircraft in level flight 100 m above the soil and with an inclination of 45º relatively to the
wake direction.

19
 Figure 4.13: Three dimensional domains where the trajectory of the UAV is shown in red
relatively to the wind turbine wake in blue.

Some observations can be made on the flight of the aircraft, after the lift drop in the wake the
aircraft has the tendency to climb even faster than at the beginning of the trajectory due to the
increased velocity in the airfoil. It can be seen that after the aircraft leaves the wake the
accelerations become negative which means this increased climb rate will slowly drop until it
reaches the initial one.

Figure 4.14: Trajectory of the UAV when crossing the wind turbine wake and roll moment
felt by the aircraft during an angle flight across a wind turbine wake.

The effect of having the velocity deficit in the model is seen mainly in the velocities and
acceleration lots in Figure 4.15, where the complexity of the movement has increased.

20
 Figure 4.15: Horizontal and vertical velocities and accelerations, respectively on the left and
right felt during an horizontal level flight with 45º of inclination relatively to the wind turbine
wake.

4.4 Wake Encounter in Perpendicular Take-off Flight

The first take-off trajectory to be studied is the trajectory perpendicular to the wake at 7
diameters from the rotor, as shown in Figure 4.16. As first comment, it can be seen from
Figure 4.17 the take-off run is greater than 100, an extremely large value for such a light
aircraft. Besides that, the take-off speed is 17.4 m/s and not 16 m/s. This might be an
indication that a higher lift airfoil was used and not the S7075 (9%).

There is also no visible effect on the trajectory of the aircraft although in the vertical velocity
plot (Figure 4.18) it is clearly visible the slight effect of the wake between 12 and 17 seconds
of flight. The initial velocity oscillations are present because the weight force is implemented
as a step function when the aircraft leaves the ground.

21
 Figure 4.16: Perpendicular take-off trajectory at 7d0 from the wind turbine.

Figure 4.17: Trajectory of the UAV when crossing the wind turbine wake and roll moment
felt by the aircraft during a perpendicular take-off flight across a wind turbine wake.

22
 Figure 4.18: Vertical velocity of the aircraft when crossing the wake in take-off trajectory.

4.5 Wake Encounter in Angle Take-off Flight

In figure Figure 4.19 is shown the angle take-off trajectory taken by the aircraft. As shown in
Figure 4.20 and Figure 4.21, the results are quite similar to the perpendicular flight. In fact,
due to the increased thrust relatively to the level flight, the effect of the wake in the flight of
the aircraft is greatly reduced.

Figure 4.19: Take-off trajectory with an angle of 45º relatively to the turbine wake.

23
 Figure 4.20: Trajectory of the UAV when crossing the wind turbine wake and roll moment
felt by the aircraft during an angle take-off flight across a wind turbine wake.

Figure 4.21: Vertical velocity of the aircraft when crossing the wake in take-off trajectory.

4.6 Wake Encounter in Landing

As previously seen, when the thrust is higher, the effect of the wake on the aircraft is reduced.
With that in mind, as the landing approach is normally made will low thrust, it was decided to
take the extreme case and calculate the effect of the wake on an unpowered landing. In order
to have a more realistic landing approach, there is an induced pilot correction on the pitch of
the aircraft: the pitch angle changes linearly from 0 to 8 degrees as the aircraft descends. It is
shown in Figure 4.22 that the aircraft starts at the left side of the domain, with an altitude of
100 m and takes 1 km to land (Figure 4.23). The final approach angle is close to 6º which is
quite low for an unpowered landing; this good performance is most likely related to the airfoil

24
that was chosen for the aircraft model. It is shown in Figure 4.24 that the loss of lift in the
wake causes the vertical velocity modulus to increase considerably, which causes a slight
change in the trajectory of the aircraft, Figure 4.23.

Figure 4.22: Landing trajectory of the UAV in an angle trajectory when crossing the wake.

Figure 4.23: Aircraft landing trajectories when perturbed and not perturbed by the wake.

25
Figure 4.24: Vertical velocity plots for perturbed and unperturbed landing trajectories.

26
5 Conclusion and Future Work

A wake velocity deficit and a wake turbulence method were used to describe the wake of a
wind turbine near an airfield. It was shown that the wake turbulence model is not performing
as well as it was hoped, possibly due to the inflow parameters used.

The model should thus be thoroughly tested for test cases where the incoming atmospheric
boundary layer is fully described in order to understand how the turbulent length scale
behaviour changes with the incoming flow.

The rotation of the wake in not taken into account in this work but it is expected to have a
large impact on the flight of the aircraft, both on increasing the rolling moment and in the
pitching movement that can be added to the aircraft dynamics model.

The influence of the wake on the aircraft is calculated with the evolution of turbulence
intensity and turbulent length scale downstream of the wake. With these quantities, a scaled
perturbation for the size of the wings of the aircraft is considered as acting in the direction of
decreasing the lift. The results show that the effect in not dangerous for the aircraft on roll
moment as Croll  0.1 for all the calculations performed and that although the effect on the

trajectory of the aircraft is definitely visible, it is not very significant.

In order to implement the rotation degree of freedom in the equations of motion a more
detailed description of the aircraft would be necessary: aerodynamic coefficients of all the
lifting surfaces (including moment coefficient), drag of the aircraft fuselage.

Increasing the complexity of the aircraft will be a hopeless task if no control system is
implemented for the aircraft, so some compromise should be reached on how far one is
willing to go on complexity.

27
References

1. Glabeke, G., et al., The Influence of Wind Turbine Induced Turbulence on Ultralight
Aircraft, a CFD Analysis, in Europe's premier wind energy event. 2012, von Karman
Institute for Fluid Dynamics: Copenhagen, Denmark.
2. Wang, Y., M. White, and G.N. Barakos, Wind-Turbine Wake Encounter by Light
Aircraft. Journal of Aircraft, 2016. 54(1): p. 367-370.
3. Kocurek, D.S. and S.E.R. Institute, Lifting Surface Performance Analysis for
Horizontal Axis Wind Turbines. 1987: Solar Energy Research Institute.
4. Tomaszweski, J.M., et al., Do wind turbines pose roll hazards to light airgraft? 2017,
University of Colorado.
5. Göçmen, T., et al., Wind turbine wake models developed at the technical university of
Denmark: A review. Renewable and Sustainable Energy Reviews, 2016. 60: p. 752-
769.
6. Bastankhah, M. and F. Porté-Agel, A new analytical model for wind-turbine wakes.
Renewable Energy, 2014. 70: p. 116-123.
7. Crespo, A. and J. Herna´ndez, Turbulence characteristics in wind-turbine wakes.
Journal of Wind Engineering and Industrial Aerodynamics, 1996. 61(1): p. 71-85.
8. ENERCON. 216 20/4/2017]; Available from: http://www.enercon.de/en/products/ep-
2/e-82/.
9. Ingenieure, V.D., Environmental meteorology Physical modelling of flow and
dispersion processes in the atmospheric boundary layer Application of wind tunnels.
2000.
10. Han, J., et al., An Estimation of Turbulent Kinetic Energy and Energy Dissipation Rate
Based on Atmospheric Boundary Layer Similarity Theory, NASA, Editor. 2000, North
Carolina State University: Langley Research Center.
11. Anderson, J.D., Introduction to Flight. 2005, New York: McGraw-Hill.
12. AVIA. PAIC Imperio SP1. Especificações. Foto. 2016 26/04/217]; Available from:
http://pt.avia.pro/blog/paic-imperio-sp1-tehnicheskie-harakteristiki-foto.
13. Gano, S. and J. Renaud, Optimized Unmanned Aerial Vehicle with Wing Morphing for
Extended Range and Endurance, in 9th AIAA/ISSMO Symposium on Multidisciplinary
Analysis and Optimization. 2002, American Institute of Aeronautics and Astronautics.
14. Hua, J., et al., Optimization of Long-Endurance Airfoils, in 21st AIAA Applied
Aerodynamics Conference. 2003, American Institute of Aeronautics and Astronautics.

28