Development of concept for silent UAV propulsion

(1)

Development of concept for silent UAV propulsion

Filip Sjöö Ingemar Jönsson

Master of Science Thesis TRITA-ITM-EX 2018:368 KTH Industrial Engineering and Management

Machine Design SE-100 44 STOCKHOLM

(2)

(3)

Examensarbete TRITA-ITM-EX 2018:368 Utveckling av koncept för tyst

framdrivning av UAV

Filip Sjöö Ingemar Jönsson

Godkänt Examinator Handledare

2018-06-11 Ulf Sellgren Ulf Sellgren

Uppdragsgivare Kontaktperson Filip Sjöö AB & Filip Sjöö &

KBK AB Ingemar Jönsson

Sammanfattning

Eftersom användningen av små UAV:s (Unmanned Aerial Vehicles) fortsätter att öka, har bullret från deras framdrivningssystem blivit ett ökande problem. Denna rapport är resul- tatet av ett masterprojekt med målet att utveckla en framdrivningsmetod med låga buller- nivåer för små UAV:s.

Projektet startade med en informationssökning där målet var att hitta information om bullerkällor i nuvarande system samt information om de fundamentala sätten på vilket luft- flöde kan skapas.

När informationssökningen var färdig, genererades ett stort antal olika koncept. Kon- ceptet som författarna ansåg ha mest potential, var en propeller med en ny metod för passiv kontroll av gränsskiktet. Konceptet har ett luftintag nära rotationscentrum. Efter att luften har kommit in i detta luftintag, leds den genom interna kanaler och accelereras radiellt utåt på grund av centrifugalkraften. Luften sprutas sedan ut genom en slits nära framkanten propellerbladets lågtryckssida. Denna ström av luft färdas över propellerbladet och sugs in genom en slits nära vingens bakkant. Därefter sprutas luften ut genom ett utlopp nära propellerbladets spets.

Tanken är att den beskrivna metoden ska fördröja eller förhindra avlösning. Detta skulle potentiellt möjliggöra högre lyftkraft vid lägre rotationshastigheter, vilket därigenom potentiellt sänker bullernivåerna. Förenklade modeller av det valda konceptet har utvecklats och analyserats med hjälp av CFD (Computational Fluid Dynamics) och jämförts med simu- leringar av en referensmodell utan gränsskiktskontroll. Resultaten indikerar att flödet i konceptmodellen strömmar genom kanalerna och över propellerbladet som det var tänkt.

Lyftkraften och effektiviteten ökade med 4.3 % respektive 1.9 %, jämfört med referensmod- ellen, vid samma rotationshastighet. Den möjliga minskningen av rotationshastigheten på grund av ökningen i lyftkraft resulterar i en minskning av bullernivån med 0.9 dB. Det bör noteras att resultaten från simuleringarna bör ses med försiktighet och att ytterligare arbete måste göras innan några definitiva slutsatser kan dras beträffande potentiella pre- standaökningar av konceptet jämfört med en konventionell propeller.

(4)

(5)

Master of Science Thesis TRITA-ITM-EX 2018:368 Development of concept for silent UAV propulsion

Filip Sjöö Ingemar Jönsson

Approved Examiner Suprevisor

2018-06-11 Ulf Sellgren Ulf Sellgren

Comissoner Contact person

Filip Sjöö AB & Filip Sjöö &

KBK AB Ingemar Jönsson

Abstract

As the use of small UAVs (Unmanned Aerial Vehicles) keeps increasing, the noise emitted from their propulsion systems have become an increasing issue. This report is the result of a master thesis project with the aim of developing a propulsion method with low noise emissions for small UAVs.

The project started with a background study, where the aim was to find information about sources of noise in current systems and information about the fundamental ways in which air flow can be created.

When the background study was finished, a large number of different concepts were generated. The concept that the authors considered having the most potential, was a propeller with a new method for passive circulation control. The concept has an air intake close to the rotational center. After air has entered this inlet it is led through internal channels and is accelerated radially outwards due to centrifugal forces. The air is then ejected through a slot close to the leading edge on the low pressure side on the propeller blade. This stream of air travels over the propeller blade and is the sucked in through a slot close to the trailing edge. After this, the air is ejected through an outlet close to the propeller blades tip.

The idea is that the method described should delay or prevent boundary layer separation.

This would potentially allow for higher thrust at lower rotational speeds, thus potentially lowering the noise emissions. Simplified models of the chosen concept have been developed and analyzed using CFD (Computational Fluid Dynamics) and compared to simulations of a baseline model with no circulation control. The results indicate that the fluid flow in the concept model flows through the channels and over the propeller blade, as intended. The thrust and efficiency were increased by 4.3 % and 1.9 % respectively, compared to the baseline model, at the same rotational speed. The possible reduction of the rotational speed due to the increase in thrust, results in a reduction of the noise level by 0.9 dB. It should be noted that the results from the simulations should be viewed with caution and the that further work needs to be done before any clear conclusions can be drawn regarding the potential performance increase of the concept compared to a conventional propeller.

(6)

(7)

NOMENCLATURE

This part describes the designations and abbreviations used in the report.

Designations

Symbol Description Units

A Area m²

α Angle rad

C_D Drag coefficient

C_f Skin friction

C_L Lift coefficient

D Drag N

d Diameter m

E Energy J

F Force N

F_T Thrust N

g Acceleration due to gravity m · s⁻²

h Height m

J Advance ratio

k_F_T Thrust coefficient

k_T Torque coefficient

l Length m

L Lift N

m Mass kg

˙

m Mass flow rate kg/s

M Mach number

µ Dynamic viscosity kg · m⁻¹· s⁻¹

n Rotational speed rev/s

η Efficiency %

P Pressure P a

Q Volumetric flow rate m³/s

r Radius m

ρ Density kg/m³

Re Reynolds number

t Time s

T Torque N m

τ_w Wall shear stress P a

θ Angular displacement rad

U Velocity m/s

u^∗ Friction velocity m/s

v Speed m/s

V Volume m³

W Sound power W

ω Angular velocity rad/s

W Sound power W

y Wall distance m

y⁺ Dimensionsless wall distance

z Difference in elevation m

(8)

Abbreviations

ANC Active Noise Control

AOA Angle Of Attack

CFD Computational Fluid Dynamics

CFJ Cow-Flow Jet

EFA Electrostatic Fluid Accelerator

FDM Fused Deposition Modeling

HPC High Performance Computing

LEV Leading Edge Vortices

MPD MagnetoPlasmaDynamic

MRF Moving Reference Frame

NACA National Advisory Committee for Aeronautics

NS Navier-Stokes

PCC Passive Circulation Control RANS Reynolds-Averaged Naiver-Stokes

UAV Unmanned Aerial Vehicle

WBS Work Breakdown Structure

(9)

1 Introduction

This chapter is about the background, problem description, purpose, goals and delimitations.

1.1 Background

It is not uncommon that machines include some sort of mechanism used to create air flow, the problem is that this process often creates noise. This problem is apparent in devices such as vacuum cleaners (75 to 85 dB (Jittanakatti, 2015)), computer fans, ventilation systems in data centers (can exceed 85 dB (Miljkovic, 2016)), table fans, hair dryers (65 to 90 dB (Akhmetov et al., 2014)), exhaust hoods in kitchens and propellers on aerial vehicles.

This thesis will focus on solving the problem in the last example and especially when it comes to small unmanned aerial vehicles (UAVs). The use of small UAVs becomes more and more common in urban areas. This has also made the noise emissions of such systems become more of an issue.

Some major companies in the world are considering using UAVs in product delivery, which might further increase the importance of this issue. In a study by NASA, it was found that the nuisance level that people experienced from UAVs, was systematically higher than the level experienced from traditional road vehicles (Christian and Cabell, 2017). Even if people eventually would get used to the sound of delivery drones, the noise would contribute to the noise pollution in cities, which could affect peoples health (Passchier-Vermeer and Passchier, 2000).

Even if the sound level of a product is not a direct health issue, the sound level can be an important factor in product purchasing decisions (Brooks, 2001) (Knöferle, 2001). Therefore, the reduction of the noise emissions of a product might in some cases correlate greatly with the sales of a product.

The authors believe that there are significant improvements to be made regarding the noise emissions from the propulsion systems for small UAVs. When observing how this problem is solved in nature it seems like today’s technology is far from the theoretical limits.

1.2 Purpose

The purpose of this master thesis is to facilitate the development of air flow based propulsion systems for small UAVs with lower noise emissions than systems of today’s standards. In order to do this, it will be discussed what causes noise in current systems and how these systems can be improved. In addition to that, the development of novel methods of creating silent air flows, for UAV propulsion, will be investigated.

The fundamental research question of this thesis is:

How can fluid flow be created in a silent way?

More specific research questions related to the chosen concept is presented in the begin- ning of the technology development phase section.

(12)

1.3 Goals

The objective of the master thesis is to create an innovative model based concept of a flow generating unit for a small UAV. The concept should utilize methods of creating air flow that are optimized for low noise emissions. The concept should also have commercial potential.

Deliverables

• CFD models of a flow generating unit for a small UAV.

• A new or improved method of creating air flow with low noise emissions.

• Master Thesis Report.

1.4 Delimitations

The master thesis is limited to the investigation of flow generating methods with low noise emissions. In many applications, air needs to be directed in a certain direction. In such cases, it is not only the flow generation that creates the noise. Obstacles in the way of the generated flow might also produce noise. To try to solve such problems is out of scope. The models that is created should only consist of the essential flow generating unit. The housing and its associated physical effects will not be taken into account. If a motor is needed to turn fan blades or similar, the noise emitted from the motor will not be taken into account.

It is also out of scope to consider other fluids than air.

1.5 Method

This project was conducted in four different slightly overlapping phases: the initiation and planning phase, the information search phase, the concept generation phase and the technology development phase. A illustration of the method is visible in figure 1 and some of the tools used are listed in table 1.

In the information search phase, information was gathered and sorted into different cat- egories using an application called Evernote. The information was mostly obtained through internet searches, but also from books and lectures. The information search phase had a very wide focus. This was to collect the necessary knowledge needed to come up with a wide range of ideas in the concept generation phase. As an aid in the information search, various questions related to the subject, were evaluated. These questions are described under the

“Frame of Reference” chapter.

The concept generation phase was started with a brainstorming session. The brainstorming session was used in order to come up with a large number of ideas on how to create air flow with low noise emissions. After the brainstorming, simple models of selected concepts were developed in order to make fast approximations of their potential. These models had many different forms, ranging from simple analytical models and physical models to models created in simulation softwares. In order to determine which concept to put more work into a weighted Pugh matrix was used.

The technology development phase involved making models of the concept chosen in the concept generation phase. In order to do this, the CAD software Solidworks and the simulation platform Simscale was used.

(13)

Figure 1: Work Breakdown Structure (WBS) of the entire project.

Table 1: Tools used in the project.

Tool Purpose

COMSOL Multiphysics Multiphysics modeling

Simscale CFD

MATLAB Numerical and analytical modeling

Solidworks Design of solid objects as an input to COMSOL and Simscale 3D-printers Manufacturing of test specimens for physical tests

Teamwork Projects Project management (task list, gantt chart etc)

Evernote Storage of collected information (articles, books, ideas and websites) Weighted Pugh Matrix Evaluation of ideas generated during brainstorming

Google Drive Storage of project files (reports, pictures etc) Overleaf Final report writing

(14)

(15)

2 Frame of reference

In the following chapter, some of the information that was obtained during the information search phase is summarized and discussed.

In order to be able to come up with potential solutions to the problem described in the introduction, a background study was conducted. This was done by trying to answer various questions. The idea was that the answers to these question eventually might lead to possible solutions. In the following sub sections, the most important questions that will be examined is presented. In order to answer these questions, existing knowledge was taken from fluid mechanics, acoustics, chemistry, electromagnetism, thermodynamics and biology.

However, since the time of the project is limited, so will the amount of knowledge that can be gathered. In additions the these questions, more general topics such as general fluid dynamics, CFD and state of the art, will be covered. The idea is that the amount of information should be enough to give answers to the questions, that are good enough to be a basis for the concept generation.

2.1 Some important concepts in fluid dynamics and CFD

Fluid dynamics deals with the behavior of fluids in motion, under this heading some basics of fluid dynamics and CFD will be discussed.

2.1.1 The Navier-Stokes Equations

The Naiver Stokes (NS) equations describes the relationship between velocity, pressure, temperature and density of moving fluids (NASA, 2015). They are one of the most important set of equations for modeling fluid flow.

In some cases one can use simplifications of the NS equations. For very low Reynolds numbers (R 1), the inertial forces in the NS equations can be neglected, since they are very small compared to the viscous forces.

At high Reynolds numbers, the inertial forces become dominant and the flow becomes turbulent. To make models of such flow regimes using the full NS equations is often very computationally demanding. So instead of using the full NS one can use the Reynolds- Averaged Naiver-Stokes (RANS) formulation, which averages the pressure and velocity fields in time, which makes to model require less computing power. In some cases one can also assume that the flow is incompressible, which means that the density is assumed to be constant. The variations in density are often so small that it can be neglected for M<0.3 (Comsol, 2015b).

2.1.2 The Boundary Layer

One important concept in fluid mechanics is the boundary layer. The fundamental idea behind this, is that he fluid domain can be divided into two regions. One region outside the boundary layer, where viscous forces can be neglected, and one region inside the boundary layer where the viscous forces must be taken into account (Schlichting and Gersten, 2017, p.29). This division of the flow domain makes it possible to simplify the NS equations.

(16)

Further away from the wall, the velocity will gradually approach the free stream velocity.

The border between the boundary layer and the free stream is usually defined as the point where the velocity in the boundary layer reaches a certain percentage of the free stream velocity, usually 99 %. As a flow travels along the surface of an object, it gradually becomes thicker and thicker and after a while, it transitions from laminar to turbulent, as can be seen in figure 2.

Figure 2: The transposition from laminar to turbulent boundary layer, as a fluid flows past a surface (Nuclear-Power, 2017).

One important phenomena regarding boundary layers, is boundary layer separation. Bound- ary layer separation occurs when the flow over a convex surface stops following the shape of the surface and sucks the boundary layer away from the wall. In figure 3, the flow past a circular cylinder is shown. As the flow is traveling from point D to point E, the flow is accelerated and it’s pressure is transformed into kinetic energy. When the flow is traveling from point E to point F, the kinetic energy of the flow is converted into pressure again. The problem is that the particles in the boundary layer loses so much of their kinetic energy in the boundary layer due to frictional forces, that they can’t travel into the region of higher pressure. After the kinetic energy of these particles are decreased further, they eventually come to a stop and begins to travel backwards. This point is called the separation point (point S, in figure 2).

Figure 3: Boundary layer separation on a cylinder (Schlichting and Gersten 2017, p.39) and corresponding pressure distribution.

(17)

2.2 How does flow produce sound?

An air flow can produce noise when there is fluctuations in the flow (Comsol, 2015a). These fluctuations can be produced by different mechanisms. There are three fundamental ways in which kinetic energy can be transformed into acoustic energy: monopole sources, dipole sources and quadrupole sources (Lighthill, 1952; Made and Kurtz, 1970).

Monopole sources of sound are present when there are mass fluctuations in a fixed region of space. Examples of monopoles sources are loudspeakers, the in or out flow opening of piston machines and cavitation (Åbom, 2013; Lighthill, 1952).

Dipole sources occurs when there are fluctuations in the pressure in a fixed region of space. Examples where this sound source is apparent would be rotating fan blades, during flow separation and during vortex shedding (Åbom, 2013, 2006; Lighthill, 1952).

Quadruple sources are present when there are free jets flows, and are normally important only when the mach number of a flow approaches or exceeds one (Åbom, 2006).

2.3 Noise in existing UAV propulsion methods

Most UAVs use propellers for propulsion. In this section the dominant causes propeller noise will be discussed briefly. The speed that the propeller blades travel with are of course one of the parameters that influences that noise emissions from propellers the most. For propellers, the sound power W scales roughly with the speed U as follows (Åbom, 2013).

W ∝ U⁵ (1)

Hence, a reduction in the speed would greatly influence the sound power. It should be noted that the power that the speed is raised to, can vary from 4 to 6, depending on the situation.

The noise from propellers are usually divided into two parts, one tonal part and one broadband part. The tonal part occurs at harmonics of the blade passing frequency (the number of blades passing a fixed point per unit of time). The broadband part can be further divided into leading edge noise, noise from the interaction between the turbulent boundary layer and the trailing edge, and noise caused by boundary layer separation (Sinibaldi and Marino, 2013).

2.4 Which are the fundamental methods in which flow can be created?

In the following section the fundamental methods of creating fluid flow that have been identified, is discussed.

2.4.1 Mechanical flow generation

The most obvious way of generating flow in fluids is perhaps the mechanical way. This is probably the most common method used in engineering, as well as in biology. Some examples of this is propellers, impellers, wings and lungs. The principle behind the mechanical way of generating fluid flow, is that fluids are affected by moving boundaries, usually solid boundaries.

(18)

2.4.2 Thermodynamic flow generation

Since hot air has lower density than cold air, heated air tends to rises in an environment of colder air. This is one of the fundamental principles behind the creation of wind on earth.

This is also known as natural convection (Çengel, 2001). This process of creating fluid flow is for example used for electronics cooling and for indoor climate systems (Fontes, 2016).

This thermodynamic principle can even be used as a space propulsion device. This is the case with for example electrothermal propulsion devices (Sforza, 2012, p.545) such as Radio-frequency and microwave excited jets (Sforza, 2012, p.550).

2.4.3 Electromagnetic flow generation

Since fluids can be affected by magnetic or electrical fields, it is possible to utilize this in order to produce fluid flow. One example of this is the so called electrostatic fluid accelerator (EFA) (Krichtafovitch et al., 2005). The basic principle behind this is that high voltage creates an electric field in the air between one positively charged pole and one negatively charged pole.

The positively pole creates positively charged ions that pushes air molecules with them to the negatively charged pole, which produces an air flow. The advantages of electromagnetic fluid accelerators compared to traditional flow generators is that they produce a nearly laminar flow and have no moving mechanical parts, which can make them very quiet (Krichtafovitch et al., 2005; Foley, 2013).

Instead of just using electrostatic forces to accelerate particles, one can utilized the forces that arise when crossing magnetic and electric fields. This is done the magnetoplasmadynamic (MPD) accelerator for example (Sforza, 2012, p.559].

2.4.4 Chemical flow generation

Chemical reactions between different substances can create fluid flow, if the reaction produces a fluid. If this chemical reaction releases lots of energy, it can be used to create very powerful flows. This principle is utilized in chemical rocket propulsion. Currently, the only way to access space, is through the use of chemical rocket propulsion (Luigi T. DeLuca and Manzara, 2017, p.6).

When this principle is used in conventional rocket engines the chemical reaction is created from a mixture of two substances, one fuel part and one oxidizer part, together referred to as propellant. When the two substances react inside a combustion chamber hot gases is produced

2.5 How is fluid flow created in the animal king- dom?

There are a lots of examples of flow generating mechanisms among, such as bird wings, lungs, fish fins and hearts. In some cases, natural selection have optimized these mechanisms for low noise. Since many silently flying vertebrates operate at similar Reynolds numbers as small UAVs, they should be highly relevant to study, when the goal is to optimize for low noise.

The barn owl is an example of an excellent silent flyer. It’s flight is so silent that available microphones are at their limits, making it difficult to study (Kleinheerenbrink et al., 2017).

The barn owl is a night hunting bird and it is believed to be at least two reasons for it’s silent

(19)

flight. Firstly, since it is hunting at night, when it is dark, it depends much on its hearing to catch prey. If it couldn’t fly with low noise emissions, it would just hear it’s wings and not the prey. Secondly, the owl depend on being undetected by its prey, a noisy owl would scare it’s pray away. For some other birds, noisy flight is not a problems, since they depend on out speeding it’s prey instead of sneaking up two it, like the owls.

The main factors that causes for the silent flight of the barn owl is it’s large wing area and slow flying speed. In addition to this it has a number of special features in its wings, which is thought to contribute to it’s silent flight.

Firstly, they have comb-like features on it’s leading edge (called serrations). According to Lockard and Lilley (2004), these serrations acts as vortex generators and thus increases the energy in the boundary layer, which minimizes the risk for boundary layer separation.

This enables the owl to fly at higher angles of attack, than would otherwise be possible.

Experiments on airfoils have shown that similar serrations does not affect the point where the flow separation occurs at the wing, however, they seem to move the point of reattachment closer to the leading edge, making the separation bubble smaller. In addition to this, it was shown that the noise emissions from the wake vortex shedding were also significantly reduced (Winzen et al., 2014).

In addition to the serrations on the leading edge, the barn owl wing has a fringe-like structure on it’s trailing edge. These fringes have been shown to reduce the trailing edge noise (Kleinheerenbrink et al., 2017). According to Lilley (Lilley, 1998; Kleinheerenbrink et al., 2017), who developed a theoretical model for the noise production for both air crafts and birds, the majority of the far-field noise is produced by the wings trailing edge. He claims that the trailing edge fringes can reduce the noise by 18 dB when flying at 6 m/s.

These fringes are thought to act as pressure release mechanisms (Lockard and Lilley, 2004), making the transition from the loaded wing to the free stream behind the wing less abrupt, which reduces the noise.

In addition to the two features mentioned above, the owls have velvet like surfaces on the suction side of its wings (Lilley, 1998). Experiments have shown that having a velvet-like material on the top of an airfoil (Klän et al., 2012), can reduce the size of the separation bubble.

Another animal with a comparable silent flight, is the bat. The aerodynamic character- istics of bats are quite different compared to that of birds. Bats have no feathers and the leading edge of their wings are very sharp, compared to the leading edge of bird wings. They also have lots of tiny hairs on the wings, which acts as sensors, allowing them to obtain detailed information about the behavior of the flow field. In addition to this, the bats have a very high morphing ability on their wings, which makes them able to adapt the shape of the wing to various aerodynamic circumstances (Hedenström and Johansson, 2015). The bats also creates so called leading edge vortices (LEVs), which is able to increase the lift by as much as 40%. These vortices are most likely formed with the help of the sharp leading edges of the bats (Muijres et al., 2008).

LEVs is actually used by some air planes, in order to generate lift at high angles of attack. These vortices is generated on delta planes that have sharp leading edges. The generated vortices sucks in air from above and forces in downwards, which generates lift.

The disadvantage of LEVs is that they increase the drag (Drela, 2014).

(20)

2.6 State of the art

2.6.1 The co-flow jet

As before, many flow generating mechanisms use airfoils. If one could reduce the speed while maintaining a high lift for a airfoil, one could reduce the sound level significantly. One interesting concept for producing high lift at low speeds is the co-flow jet (CFJ), seen in figure 4. The principle behind it is the use of a jet stream on the low pressure side of the airfoil, which increases the energy of the boundary layer and prevents flow separation (Zha and Paxton, 2006).

Figure 4: A modified NACA2415 airfoil where the CFJ principle is incorporated and an unaltered airfoil as a baseline (Zha and Paxton, 2006).

Some studies suggests that it is possible to increase the lift in this way while maintaining (or even increasing) the lift to drag ratio (L/D ratio) (Zha and Paxton, 2006; Xu et al., 2015).

In one study, a 47% increase in the lift coefficient was obtained, while the L/D ratio was maintained (Abinav et al., 2016).

2.6.2 Plasma boundary layer control

One way to avoid the noise produced by flow separation, is to use plasma actuators to control the boundary layer. Researchers at KTH have applied this method, in order to reduce the fuel consumption of a truck in a project together with Scania. With the help of plasma actuators mounted on the A-pillars on a truck cabin they have developed a solution in small scale where the plasma actuators is changing the direction of the flow around the cabin, in such a way that the fuel consumption could be reduced by up to 5% (Vernet, 2017).

2.6.3 EFA in cooling applications

Several companies exists which works on applying EFA in electronics cooling. One example of this is Tessera which had a working concept of how to implement the technology in for example “Ultrabooks” (Foley, 2013), however, not much seems to have happened regarding that since then. Other companies that have patents in similar technology from same time

(21)

period (around 2013) is Apple and General Electric (Foley, 2013). However, no public information about any of the companies future plans for the technology seems to exist.

2.6.4 Piezoelectric actuators

Piezoelectric actuator are made out of certain materials that changes shape when subjected to an electrical voltage. Piezoelectric actuators have been successfully used to actuate the flapping motion of small wings (Wood, 2007). Similar methods have also been used in order to provide silent cooling for electronic components (MIDE, 2015; Bimitech, 2017).

2.6.5 Active noise control

A method that can be used to reduce noise is to use passive noise control, for example in the form of acoustically absorbent materials. One limitation of this method is that these materials only are effective at comparably high frequencies (Farhang-Boroujeny, 2013).

An other noise control method that can be used is Active Noise Control (ANC). This method are using the fact that an acoustic signal can be nullified using an acoustic signal of opposite direction. This is done by using electromechanical systems such as microphones, speakers and active noise control filters.

The future use of ANC in flow generating devices shows great promise. However, it might be a better approach to reduce to reduce the noise emissions instead of canceling them. Since implementing noise canceling in systems, makes the systems more complex. Combining low noise flow generation with ANC might also be a viable option, at least until flow generation becomes sufficiently silent (if it will ever be).

(22)

(23)

3 Concept generation phase

The concept generation phase started with a pair of brainstorming sessions. In the following chapter, the most promising of the concepts that were generated are presented, as well as simple models intended to aid in the evaluation of the concepts. At the end of the chapter one of the concepts is chosen for further development, with the help of a weighted Pugh matrix.

3.1 Concept 1 - Continuous fluid accelerator

The idea behind this concept was to create air flow using a rotating spiral structure. The air would be sucked in at the top and then pushed through the spiral and exit straight down at the bottom. The advantage of this concept compared to a conventional propeller was thought to be that it produced a continuous flow at the outlet, which possibly would make this concept produce less tonal noise. The concept is shown in figure 5.

Figure 5: The CAD model of the continuous fluid accelerator, on the image to the right the model is sectioned, in order to display the internal channel.

Evaluation

In order to test the concept, a simple 3D model was made and then 3D printed. This prototype was then attached to a drill machine, to see if any flow could be produced. This test showed that it produced almost no flow at all, it actually produced more flow when turning it slowly by hand (at least the first rotation), than at high RPMs on the drill machine. There could be several reason for this, but since there are not coming any air out of the outlet, it certainly not coming in any air through the opening. One reason why there is no flow through the channels could be that high pressure air is trapped in the opening, because of centrifugal forces. This air would require energy in order to go “uphill” and out through the bottom opening. The flow is further restricted by the fact that the outlet is much smaller than the inlet. In order to determine if it was theoretically possible to produce flow in this way, some calculations were made.

(24)

Figure 6: Visualization of the principles behind the mathematical model.

Consider an Archimedes spiral where r = θ, where 0 < θ < 3π/2. r is the distance from the center and θ is the angle of rotation. A particle is placed at r_max = 3π/2, at time t = 0.

This particle is able to slide without friction along the spiral (see figure 6). This particle represents an air particle that is flowing into the inlet of the propeller. One could also think of the particle as a small metal ball traveling in a spiral grove on a rotating disk. At t = 0, the particle is stationary and then the disk instantly begins to rotate at the rotational speed ω. The speed of the particle relative to the line that it travels on is v = ω · r_max. Which means that it’s initial kinetic energy is

E_p = mv²

2 (2)

v = ωr_max → E_p = mω²r²_max

2 (3)

This kinetic energy needs to be larger than the energy required to reach the center. The energy required could be calculated as follows.

The centrifugal force acting on the particle is given by

F_c= mω²r (4)

The component of the centrifugal force that has a direction parallel, but opposite, to the velocity vector of the particle, can be calculated as follows.

l

r_max = Fc

F_p (5)

F_p = F_cr_max

l = mω²rr_max

l (6)

Where l is the curve length of the spiral (12.48 in this case). The total energy required can now be calculated by integrating Fp over l. Since this force decreases linearly from rmax to r=0, one could simply calculate the area of the triangle that is formed (see figure 6).

E_c= F_p· l · cos(α)

2 = mω²r_max²

l ·l · cos(α)

2 = mω²r_max²

2 · cos(α) (7)

where α = arcsin(r_max/l)

If the particle is to the reach the center the following must be true: E_p > E_c. mω²r_max²

2 > mω²r_max²

2 · cos(α)π (8)

1 > cos(α) → 1 > 0.93 (9)

According to the calculation above, it should actually be possible for the particle to reach the center, however more than 90 percent of it’s kinetic energy is would be lost, in the fight

(25)

against the centrifugal force. In reality the flow would be further slowed down by friction, which would make this device very inefficient.

Summary of concept 1

Feasibility Both the physical test and the calculations shows that the feasibility of this concept is questionable.

Size N/A

Noise level N/A

Efficiency N/A

Non complexity N/A

Innovation height N/A

3.2 Concept 2 - Bellow propulsion

The idea behind this concept was that one could reduce the noise emissions by using a series of bellows instead of rotating blades, in order to produce flow. The compression of the bellows could be accomplished using electromagnets that are turned on and off. As with the previous concept, the idea with this was to create a more steady flow. A conceptual model is visible in figure 7.

Figure 7: CAD model of the bellow propulsion concept.

(26)

Evaluation

Assuming one would use this propulsion technology for a drone in the size of a DJI Mavic Pro, one of the most popular commercial quadcopters, one would need a total of at least around 8 Ns of thrust, just to lift the vehicle, since it weighs about 734 grams (DJI, 2017).

In order to see if this is feasible, the necessary outlet velocity from the bellow is calculated.

First the following assumptions were made:

Incompressible flow.

Only one compression is considered.

F_T = 2 N , thrust per bellow.

V = 0.006 m³, the volume per bellow.

t = 1 s, the time it takes to empty the bellow.

ρ = 1.2 kg/m³, air density.

The thrust is given by the mass flow rate times the velocity

F_T = ˙m · U (10)

where ˙m = V · ρ/t

U = F_T

˙

m = 278 m/s (11)

The necessary outlet diameter can be calculated as follows. Firstly, the time it takes to empty the bellow through a specific area is given by

t = V

A · U = V

(D/2)²· π · U (12)

To obtain the necessary diameter, this could be rewritten as follows,

d =

r 4 · V

π · U · t = 0.0052 m (13)

Assuming that Bernoulli´s principle is applicable to this case, the force required to push out the fluid could be calculated as follows.

U₁²

2 + gz₁+P₁ ρ = U₂²

2 + gz₂+ P₂

ρ (14)

Since the air inside the bellow is approximately stationary, U1 is assumed to be zero. gz1

and gz₂ cancels out, since they should be very similar outside and inside the bellow. P₂ is assumed to be zero as well, since it is the pressure difference that are of interest in this case.

This makes it possible to simplify the equation to P₁ = 1

2U₁²ρ → 1

2273²· 1.2 = 46370.4 P a (15)

(27)

Assuming that a force is applied on an area of 0.04m² on each below, one would need the following force to reach the pressure state above.

P · A = 46370.4 · 0.04 = 1855 N (16)

According to this simplified calculation. It should be possible to produce enough thrust.

However, the exhaust speed of the outlet might be an noise issue and the efficiency of the contraption is doubtful.

Hence, this this concept is not impossible, but it might be impractical.

Feasibility This concept is probably feasible. Similar propulsion methods exists in nature, at least in water.

Size In order to get a sufficient thrust, one needs to have a high frequency between compressions or a very large bellow. In order to have low noise emissions, larger bellows would probably be favorable.

Noise level Most likely lower than conventional propellers, if de- signed correctly. However, the high speed air that exits the outlet might cause noise.

Efficiency This is somewhat unclear, but to force an air flow through a bellow might have more losses than the use of a regular propeller.

Non complexity The design can be made very simple, not much more than a bag and some electromagnets.

Innovation height The authors have not seen anything like it before.

(28)

3.3 Concept 3 - Thermal accelerator

The idea behind this concept was to accelerate air by using the principles of natural convection. As seen in figure 8 air would flow into a small channel, where it would be heated, this would accelerate the air upwards and force it through the channel up the a certain point.

Once the air gets to this point in the channel, it would be chilled which would accelerate the air downwards. The idea is that one could do this in many stages in order to reach high velocities.

Figure 8: Schematic of the working principle behind the thermal accelerator.

Evaluation

A COMSOL model of one stage was made, just to see it it was possible to accelerate air this way. The simulation consisted of a channel with a height of 20 mm. A heat source and a cooling source was placed as can be seen on figure 9.

Figure 9: Surface plot of the velocity of the flow field around the thermal accelerator, with one heating and cooling stage.

According to this simulation, the speed at the exit is approximately 20 mm/s, which is twice the velocity of the surrounding flow. The question now was what would happen if one placed more sections of this mechanism i series. A new simulation was made, with an added section (as seen in figure 10).

(29)

Figure 10: Simulation of the concept with two stages.

According to this simulation, the difference in the speed of the flow is negligible compared to the speed of the flow when using only one section. It didn’t even make a difference to add a third section, as seen in figure 11.

Figure 11: Simulation of the concept with three stages.

Feasibility It seems to be possible to create a flow using this method, which isn’t surprising since all the wind on earth are created using the same basic principles. The question is if this can provide a high enough velocity to be used for propulsion on earth.

Size Since it is uncertain if this can even provide enough thrust, so is it’s size. However, it would probably not be smaller than a conventional propeller.

Noise level If this method could produce enough flow, it would most likely be more quiet than conventional propellers.

Efficiency It would probably be very inefficient to heat up and cool down the air multiple times in this way. At least with the technology of today.

Non complexity All the heaters and cooling elements would make this concept very complex.

Innovation height As on the previous concept, the authors haven’t seen anything like this before.

(30)

3.4 Concept 4 - Ionic wind multiplier

The idea behind this concept was derived from the so called blade less fan from dyson. This product has a small fan in it’s base which sucks in air from the surrounding. The airflow created from this hidden fan is then directed up to a ring, where the airflow is ejected through a small slot that spreads around the entire ring. When the air is ejected through the slot, it starts to follow to curved shape of the ring, due to the coanda effect (it’s similar to what can be seen in figure 12). The flow that originates from the fan in the base, drags the air from the surroundings with it. The effect is that the amount of air that travels through the ring is up to 15 times larger than the amount of air the gets sucked in through the base (Telegraph, 2009). The main problem with current EFA, is that the thrusts achieved so far is too low to make it a viable alternative for propulsion on earth. So the idea was that a concept that used principles similar to the Dyson fan, would increase this thrust.

Figure 12: Surface plot of the velocity of the flow field around the ionic wind multiplier.

Evaluation

The only reason this concept might be good when it comes to propulsion, is if it increases the thrust. This, is unfortunately impossible, since all the thrust is produced by the high speed jets. What happens after this flow has been ejected shouldn’t affect the thrust. Even though the amount of airflow might increase.

Feasibility The feasibility of this is questionable. It would certainly not produce more thrust then the EFAs alone

Size N/A

Noise level N/A

Efficiency N/A

Non complexity N/A

Innovation height N/A

(31)

3.5 Concept 5 - Piezo wing

The idea behind this concept was to use a piezoelectric material in order to actuate artificial wings and produce a flapping motion. As mentioned in the chapter about state of the art, there are companies working on similar methods of creating airflow. A method like this might be a viable propulsion method for UAVs in the future. One could for example have a flexible wing with multiple piezoelectric actuators, the motion of this wing could then be optimized using an evolutionary algorithm (or similar) that is optimizing for low noise.

Feasibility It has been shown that is possible to create air flow using piezo electric actuators. As mentioned earlier, some people have actually managed to make flying machines using this technology. So it i certainly possible.

Size The size would probably be similar to that of conventional propellers.

Noise level The noise level would probably be lower compared to a propeller. Especially if the design uses multiple actuators, which can be used to morph the shape of the wings to shapes that are optimized for the specific flow conditions.

Efficiency If the design is made as described above, the efficiency would probably be very good.

Innovation height Since similar concepts have been tried before, it probably not be classified as an innovation.

3.6 Concept 6 - Ion wind propeller

In this concept the idea is to mount an EFA unit on low pressure side of a propeller blade, in order to minimize the effects flow separation, illustrated in figure 13.

Figure 13: A small section of the ion wind propellers airfoil.

Evaluation

The question is if EFA can produce a flow that is fast enough to prevent or delay the flow

(32)

between the two poles (Rickard et al., 2005). It has been suggested that EFA can contribute to the velocity of the surrounding flow, in velocities up to 30 m/s (Wen et al., 2016). How- ever, this contribution is very small for velocities over 4 m/s. One potential problem with this concept, is the high voltage needed (from 3 kV to more than 36 kV (Vernet, 2017)), which could be problematic to transfer to the blades of a propeller.

If one was to use this technology on the propeller of a DJI Mavic Pro, the EFA would probably need to be able to contribute to the flow speed at higher flow speeds than 4 m/s.

The rotational speed of the propellers of this UAV, when it’s hovering, is around 7000 RPM and it’s blade diameter is around 210 mm. This means that the velocity of the blade tips is 7000/60 · 0.21π = 77 m/s.

Feasibility The principle behind this concept has at least been proven to work when it comes to regular airfoils. It is plausible that it would work for propellers as well.

However, there are some challenges to be solved when it comes to transferring the high voltage to the propeller blades. Also, this concept might be too complex to evaluate properly in this thesis.

Size Similar to a conventional propeller.

Noise level The EFA should make the noise emissions from this concept lower, since it would reduce or delay boundary layer separation and allow for higher thrust at lower rotational speeds.

Efficiency Probably similar to a conventional propeller. However, there is a chance that the efficiency can be increased since this concept makes is possible to manipulate the boundary layer in order to fit different flow conditions.

Non complexity Definitely more complex.

Innovation height This has not, as far as the authors know, been done before, so it could probably be said to have a high innovation height.

(33)

3.7 Concept 7 - PCC propeller

This concept consists of a propeller with passive circulation control (PCC). The propeller (as seen in figure 14) has an air inlet close to the rotational center, the air that enters this inlet is lead through a channel and ejected through an outlet on the low pressure side of the propeller blade. The fundamental principle is similar to the co-flow jet, which is described earlier. The idea is that the flow from the outlet should delay or prevent boundary layer separation. This would allow the wing to operate at higher angles of attack, making it possible to create more lift at lower rotational speeds, which would create lower noise emissions.

Figure 14: The basic principle of the PCC propeller.

Evaluation

In order to investigate the effects of adding a jet flow to the low pressure side of an airfoil, two COMSOL simulations were made. In figure 15 a comparison between a normal airfoil and an airfoil with an added flow is shown. The velocity of the free stream is 1 m/s and the jet flow has a velocity of 5 m/s.

Figure 15: Surface plots showing the velocities obtained from simulations of a conventional airfoil and an airfoil with an added jet flow.

Table 2: Results from simulation.

Re 100000 100000

C_L 12.236 14.876

C_D 1.1 1.34

L/D 11.1 11.05

The results from this clearly shows that this increased the coefficient of lift, while maintaining a similar L/D ratio, as seen in table 2. Also, it seems to a reduce size of the wake significantly.

It should be noted that the k- turbulence model was used, which uses wall functions and therefore it is not the best choice when calculating lift and drag forces (Lyu, 2015). However,

(34)

As mentioned before, the co-flow method has been shown to increase the lift coefficient by as much as 47 % while maintaining the L/D ratio. Assuming similar performance is possible with this concept, the reduction in noise could be estimated as follows.

L = ρv²ACL

2 (17)

v = s

2L ρACL

(18) Assuming that C_Lis independent of speed, the velocity needed to achieve the same lift with the co-flow jet, as with the baseline, would be the following.

v_pcc = s

2L

ρAC_L1.47 (19)

From this follows that

v

v_pcc = 1

1.47 = 0.68 (20)

This means that the speed can be decreased by 32 %, would translate to the following reduction in sound power:

W¯_pcc

W¯₀ = 0.68⁵

1 = 0.145 (21)

The sound level reduction in decibel would be

10 · log(0.145) = 8.39 dB (22)

However, this is a very simplified calculation that assumes that the air intake doesn’t increase the drag significantly and that this method can be used to produce as much flow as the co- flow method. In order to evaluate the contribution on the torque from the air intake the following calculations were made. The simplified geometry of the propeller blade used in the calculations can be seen in figure 16.

First, the torque from the blade without the intake can be calculated, by integrating the drag force over the radius of the propeller blade.

T_prop = Z 0.12

0

1

2ρv²C_DA · r = Z 0.12

0

1

2ρv²C_D · h₁· r dr (23) Since v = r · ω, we have

T_prop = Z 0.12

0

1

2ρω²r²C_D· r · h₁ dr = 1

2ρω²C_D Z 0.12

0

r³dr = 2 · 10⁻⁷1

2ρω²C_D (24) A similar calculation can be made for the air intake.

T_intake = Z 0.02

0.01

1

2ρω²r²C_D· r · h₂ dr = 1

2ρω²C_D Z 0.02

0.01

r³dr = 7.5 · 10⁻¹⁰1

2ρω²C_D (25) The relationship between the torque generated by the intake and the torque generated by the rest of the propeller would be

(35)

Figure 16: Simplified geometry of the PCC propeller. Where h₁ = 0.01 m and h₂ = 0.02 m.

T_intake

T_prop = 0.00375 (26)

According the these calculations, with these assumed dimensions, the intake duct would have a very small effect on the total necessary torque. It should be noted that the integration is simplified. The first integration should ideally be divided into two integrals, one from 0.01 to 0.02 and one from 0.02 to 0.12. But nevertheless, the result shows that the contribution from the intake is very small. Another concern if is the intake would take in enough air, since the airspeed close to the center is low, compared to further away from the center. The average outlet velocity can be estimated as follows.

ω = 314 rad/s r₁ = 0.015 m r₂ = 0.075 m v1 = ω · r1 = 6.28

A₁ = 0.02 · 0.02 = 0.004 m² A₂ = 0.05 · 0.001 = 0.00005 m²

Q = v1· A1 = 4.71 · 0.004 = 0.002512 m³/s (27) From this v₂ can be calculated

v₂ = Q

A₂ = 50.24 m/s (28)

The results shows that the average outlet velocity is at least double the speed of the propeller at that specific point, if friction, pressure and centrifugal forces are neglected. Even if this calculation if very simplified and probably far from the truth, it shows at least that it is not impossible to get enough flow in an idealized case.

(36)

Feasibility According to the simple simulations and calculations done of this concept, it does seem to have potential when it comes to noise reduction.

Size The size would probably be similar to that of a conventional propeller.

Noise level Should be lower, since it theoretically allows for lower rotational speeds.

Efficiency The idea is that the increased torque from the intake would be compensated by higher thrust, thus at least keeping the efficiency constant compared to conventional propellers.

Non complexity Since no parts are added, one might argue that the complexity is changed. At least if one defines complexity as number of parts. However, the shape is definitely more complicated and probably more difficult to manufacture.

Innovation height To add an flow to the suction side of an airfoil is nothing new. However, to do it using air intakes on a propeller has not been investigated, at least to the best of the authors knowledge.

3.8 Concept 8 - Hub-less propeller

This concept consists of a propeller without a shaft in the center. Instead the blades are connected to a rim the rotates, similar to what can be seen of figure 17. This concept as actually not entirely new. Similar concepts have been proven to work well for underwater applications, regarding the noise emissions (Yan et al., 2017). The idea is that this concept would create a less disturbed flow, since it doesn’t need any structure to be in the way of the flow to hold the blades, like on a conventional propeller. In addition to this, no central shaft is needed. These things combined could potentially make the propeller less noisy.

The rotation of the rim where the blades are mounted could be created in various ways. The simplest way might be to make the entire device into an electric motor, where the rim is the rotor and the frame that it’s rotating in is the stator.

(37)

Figure 17: Hub-less propeller on a subsea thruster (Copenhagen Subsea A A/S, 2014).

Evaluation

Feasibility Since this concept has been proven to work under water with good results, it might be worth investigating its potential for propulsion in air as well.

Size The size of this is probably similar to that of a regular propeller.

Noise level Probably slightly lower.

Efficiency Roughly the same probably.

Non complexity Similar.

Innovation height Since it has been done before, but for a different fluid, the innovation height is low.

3.9 Concept 9 - Saw tooth propeller

As mentioned previously, one of the most important sources of noise when it comes to airfoils, are the noise the originates from the trailing edge. This is due to the abrupt end of the trailing edge. This problem has been solved in nature by using fringe like structures, which makes the transition from airfoil to free stream more continuous. The idea behind this concept is to use these principles on a propeller, which actually has been done before.

However, one could always try to do it in a better way.

Evaluation

In order to evaluate the potential of this concept, two simple 3D models were made and tested (seen in figure 18). They were both based on the Tiger motor 15x5 carbon fiber

(38)

hadn’t. They both had close to the same surface area. The original propeller had a surface area of 3092 mm² and the propeller with saw teeth had a surface area of 3188 mm².

Figure 18: 3D printed propeller with and without saw teeth on the trailing edge.

The propeller was attached to the shaft of a table fan from Clas Ohlson. The thrust was measured using a kitchen scale placed perpendicular to the axis for rotation. The noise level was measured using a cellphone, and the measurement was done 300 mm from the rotational center of the propeller directly downstream of the flow.

Table 3: Comparison between propeller with and without saw teeth.

Type Thrust [N] Sound level [dBA]

Smooth 0.36 66.4

Saw tooh 0.265 64

According to the results from the test (shown in table 3), the fringed propeller was 2.4 dB quieter than the baseline propeller, however the thrust was significantly lower. There is a lot of possible explanations for the large difference in thrust. The propellers are 3D printed using inexpensive FDM printers, which sometimes produce inconsistently. This might be be part of the explanation, since the difference between the surface roughness for the propellers were significant.

Feasibility Since the basic principles have been proved to work, this concept in certainly feasible.

Size Similar.

Noise level Lower than a conventional propeller.

Efficiency Even though the thrust was significantly reduced in the tests described above. It has been shown the principle can be used without significantly reducing the efficiency.

Non complexity Similar, even though the fringes might make this a little bit harder to manufacture.

Innovation height This has been done before, so the innovation height is low.

(39)

3.10 Concept 10 - Magnus propeller

The idea behind this concept was that one could utilize the Magnus effect to create airflow.

This would be done using rotating cones, as can be seen in figure 19. This could potentially create more airflow at lower rotational speeds, then a conventional propeller, which might reduce the noise emissions.

Evaluation

A simple 3D model was created and 3D printed. This model was then attached to a screw- driver in order to see if any flow was created. The air flow was visualized with incense sticks.

Unfortunately it didn’t create any airflow in the desired direction, instead, in sucked air in from the back and front side, and ejected it out to the sides, due to centrifugal forces.

The reason it didn’t create any airflow, might be because the cylinders rotated too slow around their local axes of rotation, compared to the rotation around the main axis. Addi- tionally, it would be better if the cones were coned the other way, in order for the Magnus effect to be effective throughout the entire radius.

Figure 19: 3D printed model of the Magnus propeller.

(40)

Feasibility The basic principle behind this does certainly work. The question is how well it works when it is applied to propellers.

Size Probably a lot bulkier than a normal propeller.

Noise level There is a chance that the noise level might be lower, if the gearbox can be made sufficiently quiet.

Efficiency This design would probably create a lot of turbulence, in addition to this, the gearbox needed would most likely make the design less efficient than a regular propeller.

Non complexity Definitely more complex.

Innovation height The authors haven’t seen this concept used as a propeller, so the innovation height is considered high.

3.11 Weighted Pugh Matrix

In order to help with the selection of a concept, a weighted Pugh matrix was made (shown in table 6). Seven different criteria were selected and each of these criterion was assigned a weight between 1 and 5, based on its perceived importance, where 5 means that it is very important (shown in table 4). In table 5, some pictures of the different concepts are shown, as a complement to the Pugh matrix.

Table 4: Evaluation criteria with weight and motivation.

criterion Weight Motivation

Feasibility 3 It is of importance that the concept is possible to realize within the bounds of the thesis.

Size 1 One can probably live with a large size if most of the other criteria are met.

Noise level 5 This is the most important criterion, since the aim of the thesis is to develop a concept that is as quiet as possible.

Efficiency 4 Efficiency is very important when it comes to UAVs, as it it directly correlated to it’s flight time and range.

Non complexity 3 Complexity often leads to higher costs and lower re- liability.

Innovation height 4 Since the aim of the thesis is to develop something that has commercial potential, the innovation height is of high importance.

(41)

Table 5: All concepts up for evaluation.

Illustration

Concept Concept 2 Concept 3 Concept 5

Bellow propulsion Thermal accelerator Piezo wing

Illustration

Concept Concept 6 Concept 7 Concept 8

Ion wind propeller PCC propeller Hub-less propeller

Illustration

Concept Concept 9 Concept 10

Concept Saw tooth propeller Magnus propeller

Table 6: Weighted Pugh matrix - evaluation of generated concepts.

Concept

Criteria weight ref 2 3 5 6 7 8 9 9

Size 1 - 0 0 0 0 0 0 -

Noise level 5 R + + + + + + + +

Efficiency 4 E - - + 0 0 0 0 -

Feasibility 3 F 0 - 0 - 0 0 0 0

Non complexity 3 E 0 - - - 0 0 0 -

Innovation height 4 R + + 0 1 + 0 0 +

P + E 2 2 2 2 2 1 1 2

P 0 N 2 1 3 2 4 5 5 1

P − S 2 3 1 2 0 0 0 3

Net score 20 4 -1 6 3 9 5 5 1

The concept with the highest score in the weighted Pugh matrix was the PCC propeller.

Therefore, this concept was chosen for further development.

(42)

(43)

4 Technology development Phase

In this chapter, the development and furhter investigation of the chosen concept is described.

In order to aid in the development the PCC propeller, a number of questions were formulated:

• Is possible to create flow through channels in a propeller blade, entirely driven by the rotation of the propeller?

If the answer to the questions above is that it is possible, the questions below follows.

• Can these flows be directed so that the outflow is tangential to the rotation of the propeller?

• What are the major factors that contribute to the flow through the channels?

• Is it possible to arrange the rotation driven flow in a fashion similar to a co-flow jet, with combined injection and suction on the upper surface of the propeller?

• Can the jet flow delay flow separation?

• Can the principle behind the PCC propeller be used in order to reduce noise emissions?

4.1 General simulation and post processing setup

In order investigate the questions presented above, a number of CFD simulations were made.

Most of these simulations had the same general simulation setup and they were all done in Simscale. The design of the simulation domain was taken from a study by Rajendran and Kutty (2017), partly because the study was about a similar propeller to the one that some of the models described later are based on, but also because their results matched well with experimental data.

The simulation domain (seen in figure 20) consisted of cube with a length that was eight times the propeller diameter. In the center of this cube a cylindrical domain was placed, which was intended to be as base for a MRF (Moving Reference Frame) zone. This zone was is intended to produce the rotational effects in the model.

Figure 20: Setup of the simulation domain.

(44)

The different models that were simulated were meshed slightly differently, regarding mesh resolutions and mesh refinements. This is described in more detail when it is considered relevant.

The turbulence model the was chosen was the k-ω SST. The reasons for this is that the model is able to predict the fluid behavior all the way down to the wall and that the model is much more stable than the original k-ω turbulence model, since it uses the k- turbulence model for the free stream (Frei, 2013). It has also be shown to produce results the correspond well with experimental data for similar applications, such as for airfoils (COMSOL, 2018;

Douvi C. Eleni, 2012).

The post-processing of the simulations in the technology development phase were mostly done in the post-processing software Paraview. The velocity fields that are shown in the plots in this phase, are all modified so that the velocities are in relation to the propeller blade.

4.2 First approximation of flow field around the PCC propeller

In order to examine if it was possible to create any flow through the channels of the concept, a 3D model was created in Solidworks, seen in figure 21. The design of the model was based on a propeller called APC Slow flyer 8x3.5, due to the availability of experimental data. The idea was that this would ease a possible future comparison of performance. The dimensions of the propeller is also well suited for small UAVs (it has a diameter of 203.2 mm).

Figure 21: Early CAD model of the PCC propeller. The intended flow path is displayed in the bottom picture.

After the 3D model was created, it was inserted into Simscale, where the boundary conditions in table 7 were applied. The chosen parameter values are chosen to match the conditions in wind tunnel tests of the APC Slow Flyer 8x3.5 (Brandt et al., 2015).

Development of concept for silent UAV propulsion