STOCKHOLM SWEDEN 2016,
“Experimental Assessment of the Effectiveness of Active Flaps to Reduce the Blade Root Bending Moment of Wind Turbine Blades”
ADRIANO SANCHEZ PEÑA
KTH ROYAL INSTITUTE OF TECHNOLOGY SCHOOL OF ENGINEERING SCIENCES
KUNGLIGA TEKNISKA HÖGSKOLAN DEPARMENT OF MECHANICS
SWEDEN
“Experimental Assessment of the Effectiveness of Active Flaps to Reduce the Blade Root Bending
Moment of Wind Turbine Blades”
MASTER THESIS PROJECT BY
ADRIANO SANCHEZ PE Ñ A
A dissertation submitted in partial fulfilment of the requirement for the degree of MSc of Engineering Mechanics in Fluid Mechanics.
July 2016
Abstract
This thesis intends to asses the application of local flow control to wind turbines blades, in order to demonstrate that the normal root benging moment on the blades can be reduced, with the aim of increasing wind turbines life span, reducing regular maintenance and further to make a contribution to current research on wind turbines.
It develops the subject by first assessing the flow characteristics of the Großer Windkannal (GroWiKa) in the Hermann-Föttinger Institute (HFI) at the Technical University of Berlin and continues with establishing the effectiveness of local flow control on the Reasearch Wind Turbine (BeRT) designed by the institute.
It is divided into six chapters; the first chapter gives an overview of the literature review required, to study the different methodologies used during the experiments. In following the different experimental set-ups used will be explained, which will lead to the presentation of the different characteristics of the flow into the settling chamber of GroWiKa, so that its effects on the wind turbine can be evaluated by means of pitot tubes and hot-wires. Chapter 4 reflects the different effects of blockage calculated during the experiments and its corrections for a blockage of 40%, while chapter 5 gives an overview on the system dynamics of the flaps, where three servos has been used and compared with each other, in order to give an assesment on its effectiveness for the rotative system. The final chapter analyses the normal bending at the blade root, by yawing the wind turbine and setting the flaps into a simple prescribe motion, in pursuance of creating a 1 period disturbance and show how the normal bending at the root of the blade can be reduced by the action of local flow control.
To this end it was found that the servo system was not effective enough to respond to sudden changes in the flow field. However, it will be shown that after accounting for the servos delay, it is possible to reduce the blade normal to rotor plane root bending moment up to 31%, when the correct phase shift is applied to the flap’s prescribed motion.
Acknowledgements
I would like to take a space to thank my mother Fabiola Peña, the strongest woman I know, which I admire the most and my example in life. She has given away part of her life, in order for me to build one; that has no price and I will never be able to pay you back.
I would like to thank my father Luis Sanchez, my grandma Raquel Requena and all my family for their support all my life.
I would like to thank as well to every person that has been through out the path of this thesis, especially: Jamila Stute for supporting me and give me strength to put my head up and keep on going forward; my thesis supervisor Sirko Bartholomey for being always there for me when I needed him and most importantly, helping me to achieve this last step of my studies; Joseph Saverin for his insight on the wall pressure signature method.
Last but not least I would like to thank Sweden and the Küngliga Tekniska Högskolan (KTH), for giving me the possibility to live among you, discover the reach culture of Sweden and most of all, for giving me the best education I could ever ask for.
Thank you all!
July 2016 Adriano Sánchez Peña
Table of Contents
Abstract ... ii
Acknowledgements ... iv
List of Figures ... viii
List of Tables ... x
Introduction ... 12
1 Chapter 1: Literature Review ... 14
1.1
Bernoulli’s Equation and Application to Pitot Tubes. ... 14
1.2
Research Project ‘Berlin Research Wind Turbine’ (BeRT): ... 14
1.2.1
BeRT Wind Turbine: ... 15
1.3
Hot-wire Anemometry ... 25
1.4
Boundary Layers ... 26
1.4.1
Displacement Thickness: ... 27
1.4.2
Momentum Thickness: ... 28
1.5
Actuator Disc Theory ... 28
1.5.1
Axial Momentum Theory: ... 28
1.6
Blockage Correction Theories ... 31
1.6.1
H. Glauert (1933) ... 31
1.6.2
A. Pope (1964) ... 32
1.6.3
Hackett-Wilsden “Pressure Signature Method” (1960) ... 32
1.6.4
M. J. Werle (2010) ... 34
1.7
Control Theory ... 37
1.7.1
Control Methods ... 40
1.8
Q-blade Software ... 43
2 Chapter 2: Experimental Set-up ... 44
2.1
GroWiKa (Großer Windkannal) ... 44
2.2
Traverse Systems ... 46
2.2.1
Small Traverse System ... 46
2.2.2
Large Horizontal Traverse System ... 46
2.3
Wall Boundary Layer Study ... 46
2.4
Hot-wire Inflow and Wake Study ... 48
2.5
Blockage Effect Study ... 50
2.5.1
Measurement of Inflow and Wake Velocity: ... 50
2.5.2
Wall Pressure Signature: ... 51
2.6
Flap Motion - Dynamic Behavior Analysis ... 53
2.7
Prescribed Flap Motion for Load Alleviation ... 53
3 Chapter 3: Inflow, Wake and Wall Boundary Layer Study ... 55
3.1
Wall Boundary Layer Study ... 55
3.2
Hot-wire Inflow and Wake Study ... 61
4 Chapter 4: Blockage Effect Study. ... 66
5 Chapter 5: Flap Motion - Dynamic Behavior Analysis ... 72
6 Chapter 6: Prescribed Flap Motion For Load Alleviation ... 78
Conclusion ... 80
Future work - Recommendations ... 82
Bibliography ... 83
Appendix A: Hot-wire Calibration Curve ... 85
Appendix B: Inflow and Boundary Layer Study ... 86
Appendix C: Wall Static Pressure Measurements and Corrected Ct Curves by Blockage effect Methods ... 89
Appendix D: System Dynamics Study ... 94
Appendix E: GroWiKa Calibration Curve ... 102
List of Figures
FIGURE 1 PITOT TUBE ... 14
FIGURE 2 CAD OF BERT POSITIONED INTO WIND TUNNEL ... 15
FIGURE 3 BERT SCHEMATIC OF THE MAIN FRAME SYSTEMS ... 16
FIGURE 4 CROSS-SECTIONAL VIEW OF BERT. ... 17
FIGURE 5 BERT WIND TURBINE MOUNTING INSIDE WIND TUNNEL. ... 18
FIGURE 6 CONTROL CABINET ... 19
FIGURE 7 TRAILING EDGE MODIFICATION BY THE MANUFACTURER ... 20
FIGURE 8 BLADE IN MOLD ... 21
FIGURE 9 UNASSEMBLED BLADE (LEFT) AND BOTH SHELL BEFORE BONDING (RIGHT). ... 22
FIGURE 10 POSITION AND ATTACHMENT TO SERVOS INTO PAYLOAD BAY ... 22
FIGURE 11 LINKAGE MECHANISM (LEFT) AND SERVOS AFTER PRINTING PROCESS (RIGHT). ... 23
FIGURE 12 SCHEMATIC OF POSITIONS FOR THE THREE HOLE PROBES. ... 23
FIGURE 13 SERVO CROSS-SECTION. [8] ... 24
FIGURE 14 VOLZ SERVO MANUFACTURER INFORMATION [9] ... 24
FIGURE 15 AXIAL STREAM TUBE [13] ... 28
FIGURE 16 CHANGES OF PRESSURE AND VELOCITY ACROSS THE ACTUATOR DISK [14] ... 29
FIGURE 17 POSITION OF SOURCES AND SINKS (UP) AND EFFECTS OF BLOCKAGE ON THE WIND TUNNEL WALL (LOW) ... 33
FIGURE 18 IMAGE METHOD MODELING ... 33
FIGURE 19 WERLE METHOD CONTROL VOLUME ... 35
FIGURE 20 DIAGRAM OF SYSTEM DYNAMICS CONCEPTS [24] ... 37
FIGURE 21 FIRST ORDER STEP RESPONSE (TIME DOMAIN) ... 39
FIGURE 22 SECOND ORDER RESPONSES (Ζ= 0.5,Ζ= 1,Ζ= 1.25) ... 40
FIGURE 23 SMITH'S METHOD RELATIONSHIP BETWEEN ... 41
FIGURE 24 PROCESS STEP RESPONSE AND SECOND ORDER TIME DELAY APPROXIMATION [27] ... 42
FIGURE 25 STEP RESPONSE OF FIRST ORDER SYSTEMS AND ESTIMATION OF THE 63.2% PERCENT METHOD [21] ... 43
FIGURE 26 AXIS OF WIND TURBINE IN WIND TUNNEL GROWIKA ... 44
FIGURE 27 3D CAD OF THE GROWIKA (INITIAL SET-UP) ... 45
FIGURE 28 NEW CONFIGURATION OF GROWIKA WITH 2 NEW WOX ... 45
FIGURE 29 SMALL TRAVERSE IN WIND TUNNEL ... 46
FIGURE 30 BOUNDARY LAYER TRAVERSE LOCATED AT -0.483D INSIDE WIND TUNNEL ... 47
FIGURE 31 MEASUREMENT GRID USED FOR WALL BOUNDARY LAYER STUDY ... 48
FIGURE 32 POSITION OF PITOT TUBE FROM THE WALL ... 48
FIGURE 33 RIG FOR HOT-WIRE CALIBRATION ... 49
FIGURE 34 MEASUREMENT GRID USED FOR HOT-WIRE STUDY ... 50
FIGURE 35 SET-UP FOR INFLOW AND WAKE VELOCITY ... 50
FIGURE 36 CALIBRATION CURVE WALL PRESSURE TRANSDUCERS ... 51
FIGURE 37 NATIONAL INSTRUMENTS DATA ADQUISITION SYSTEM (LEFT) ... 52
FIGURE 38 WALL PRESSURE TABS ON RIGHT WALL OF WIND TUNNEL ... 52
FIGURE 39 SEPARATION OF SMALL TRAVERSE TO THE WALL ... 56
FIGURE 40 COUNTOUR OF INFLOW FIELD AT THE ... 57
FIGURE 41 DEVELOPMENT OF BOUNDARY LAYER IN THE Z AXIS ALONG X ... 59
FIGURE 42 COMPARISON OF BOUNDARY LAYER PROFILE FOR ... 60
FIGURE 43 MEASUREMENT OF INFLOW VELOCITY AT X/D= -0.44 ... 62
FIGURE 44 MEASUREMENTS OF TURBULENCE INTENSITY AT X/D = -0.44 ... 63
FIGURE 45 MEASUREMENT OF INFLOW VELOCITY AT X/D= 0.5 ... 64
FIGURE 46 MEASUREMENTS OF TURBULENCE INTENSITY AT X/D = 0.5 ... 64
FIGURE 47 MEASUREMENTS OF HORIZONTAL VELOCITY AT X/D = 0.5 ... 65
FIGURE 48 WALL PRESSURE MEASUREMENT FOR EMPTY WIND TUNNEL (UP) AND BERT INSIDE WIND TUNNEL (DOWN)68
FIGURE 49 CP VS. TSR (COMPARISON AMONG: MEASURED UNCORRECTED DATA, ... 70
FIGURE 50 CP VS. TSR (COMPARISON AMONG: MEASURED UNCORRECTED DATA, ... 70
FIGURE 51 CP VS. TSR (COMPARISON AMONG: MEASURED UNCORRECTED DATA, ... 71
FIGURE 52 SERVO 3 COMPARISON OF METHOD’S STEP RESPONSES ... 77
FIGURE 53 RATIO OF THE NORMAL ROOT BENDING MOMENT OF THE BLADE ... 79
FIGURE 54 COMPARISON BETWEEN BEST CASE ∆Φ= 90° (RIGHT) AND WORST CASE ∆Φ= 270° (LEFT) SCENARIO . 79
FIGURE 55 DIFFERENCES OF COEFFICIENT OF LIFT FOR AIRFOIL CLARK-Y ... 79
List of Tables
TABLE 1 GENERAL BERT TURBINE CHARACTERISTICS ... 15
TABLE 2 MAIN CHARACTERISTICS OF BLADES ... 20
TABLE 3 SPACING OF TRAVERSE GRID (BOUNDARY LAYER STUDY) ... 47
TABLE 4 RESULTS OF CORRECTIONS FOR GLAUERT’S, POPE’S AND WERLE’S METHOD ... 69
TABLE 5 RESULTS INTO ROTATIVE SYSTEM ... 73
TABLE 6 SYSTEM DYNAMICS PARAMETERS SERVO 3 (UNLOADED CASE) ... 74
TABLE 7 SYSTEM DYNAMICS PARAMETERS SERVO 3 (LOADED CASE) ... 75
Introduction
Nowadays, it is important to look for new ways of generating energy. Many renewable options have been developed through the recent years, in order to decrease the amount of CO2 produced by burning fossils. Wind energy is currently one of the biggest industries in renewable energies, generating up to 433 Gigawatts around the world from 314k turbines at the end of 2015 [1].
Through out the recent years, wind turbine sizes have been growing up exponentially from a 15m diameter in 1980 to a size of 123m diameter in 2009 [2]. Today, the SeaTitan 10MW wind turbine manufactured by AMSC (American Superconductor) has a diameter of 190m [3], meaning an increment of 54.47% in just three years. All this has been possible due to research into the area of materials, aeroelasticity, aerodynamics and related fields.
The conditions at which wind turbines operate in the outer conditions are not optimal and this means that the rotor will encounter many local flow changes; generated by gusts, turbulence from upstream turbines, among others; these will affect the normal bending at the rotor blade root in different ways; hence, incrementing the need of regular maintenance and reducement on the wind turbine’s lifespan.
As this disturbances impact the blade locally, the changes of pitch angles will not be able to cover for a local change of flow and furthermore, the large size of the blade generate high inertial effects that are not fast enough to counteract the sudden local disturbances of the flow. Under this consideration, it is of high necessity to implement local flow control devices, which could help to reduce the normal bending moment at the root of the blades.
The aim of this project thesis is to study the feasibility of implementing flaps at the trailing edge of the rotor blades for local flow control, in order to reduce the normal bending moment at the root of the blade.
The institute of fluid dynamics Hermann-Föttinger (HFI) of the TU Berlin has recently developed a utility scale wind turbine, which will be used as set up for the realization of the experiments. The rotor has 3 meters diameter and three blades. It is positioned in the Großer Windkannal (GroWiKa), which has a maximum speed at the test section of ~12 m/s, a height of 4.2m by a width of 4.2m.
The different parts of the thesis will be divided into five chapters; the first and second chapter will be dedicated to literature review and the experimental set-up together with an overview of test articles used during the experiments, respectively. The next four chapters will comprise results and discussions of each experiment performed.
The third chapter will give an insight of the inflow and near wall characteristics by the use of pitot tubes and hot wires at the test section.
The fourth chapter will explain in detail the blockage effect produced by the wind turbine inside the wind tunnel. The results will then be compared between unconstrained wind turbine performance curve, simulated in Q-blade, the Glauert’s method, the Pope’s method, the Werle’s method and the wall pressure signature method.
In the fifth chapter, three different methods will be applied to get the principal dynamic parameters of the flap system (from the set point at the computer until it reaches the final steady state). The methods applied are as follows:
• Smith’s method.
• Sundaresan and Krishnaswamy’s method (two point method).
• Krishnaswamy and Randanian’s method (three point method).
• Slope-intercept 63.2% Method.
The sixth chapter will be dedicated to explain the behavior of the flaps into a simple prescribed motion formula, where it will be possible to show the feasibility of decreasing the root bending moment at the wind turbine axis.
Finally, conclusions of the different studies are drawn, in order to state the effectiveness of the methodologies applied.
1 Chapter 1: Literature Review
1.1 Bernoulli’s Equation and Application to Pitot Tubes.
Pitot tubes are the most widely used technique to sense flow velocities. It consist of two measurements of pressure taken from the flow, total or stagnation pressure in front of the tube as shown in figure 1 and static pressure taken on the surface of the circumference of the tube behind the stagnating point. The result of the measurement, gives the dynamic pressure from where the velocity of the flow can be derived, by means of the Bernoulli’s equation, (1) shows the relation between the three pressures.
!" = !" +1
2 ! !! (1)
Where,
!" = Total or stagnation pressure.
!" = Static pressure.
! = density of the fluid.
! = velocity of the flow.
F ig u re 1 P ito t T u b e
1.2 Research Project ‘Berlin Research Wind Turbine’ (BeRT):
BeRT is a research group integrated by five universities of Germany, TU Berlin, TU Darmstadt and IAG University Stuttgart, RWTH Aachen University and Oldenburg University. The main objective is to validate numerical and experimental investigations on wind turbine blades of small scale, before it can be transferred into full-scale wind turbines.
TUBerlin takes part into the experimental part of the project, aiming to investigate load reduction potentials by implementing active flap control on the turbine blade.
Figure 2 shows the wind turbine CAD design and the WOX section, where measurements are undertaken. All data presented on the next section has been taken
from the last BeRT manufacturing report performed by the company SMART BLADE GmbH [4], for specific details on aeroelastic behaviors and blades technical data the reader is recommended to contact the Hermann Föttinger Institut (HFI) or SMART BLADE GmbH.
F ig u re 2 C A D o f B eR T p o sitio n ed in to w in d tu n n el
1.2.1 BeRT Wind Turbine:
The BeRT has been designed by SMART BLADE GmbH and the Technical University of Berlin. It has integrated a National Instruments data acquisition system called cRIO, located inside the hub of the rotor, that collect data from sensors installed in different systems of the wind turbine. The main characteristics from the deisgn can be found in table 1.
T a b le 1 G en era l B eR T T u rb in e C h a ra cteristics G e n e ra l B e R T T u rb in e C h a ra c te ristic s
R otor d iam eter 3.000m
R otor h u b rad iu s 0.215m
B lad e len gth 1.285m
T ow er h eigh t 2.100m
R ated p ow er 2kW
R ated sp eed 180rpm
R P M ran ge 100 to 250rpm
D esign T SR 4.6
F ig u re 3 B eR T S ch em a tic o f th e M a in F ra m e S y stem s
Mechanical Design:
o Tower:
It is made of regular pipe steel material with an outer diameter of 273mm and a wall thickness of 5 mm, so it serves as housing for the gearbox and the generator, as seen in figure 3. It has been designed stiff enough to avoid aeroelastic coupling between the different components of the wind turbine. All data was calculated from the CAD model, such as; center of gravity (COG), stiffness, mass and geometry.
o Rotor Hub:
The inner rotor hub structure has a hexagonal shape, designed to accommodate all the signal acquisition units and control instrumentation, it houses the main controller unit dominated cRIO-9068 manufactured by National Instruments with capacity for 8 slots.
The rotor hub as can be seen in figure 4, it is the component that transfers the rotational torque from the blades into the main shaft, situated at the rotational axis of the rotor. The blades are attached to it, through three clamping pieces, which are secured by 11 Nm torquimeter. It is possible to change the pitch angle of the blades manually by means of a nonius scale situated o the hub side, next to the holes for blade attachement.
This component has a total inner space of ~18750cm3, a wall thickness of 8mm, stiff enough to withstand the loads of the rotor and a radius to blade distance of 215mm. It is covered by the spinner, made of GFRP (Glass-Fiber Reinforced Plastic) infused in vacuum with epoxy composite and painted in black mate paint, to avoid reflection of light when taking PIV measurements.
F ig u re 4 C ro ss -se c tio n a l V ie w o f B e R T .
o Drivetrain:
The drivetrain as seen in figure 4 consist of the main shaft, the bearings, a synchronous belt drive, a gearbox and an elastic belt. It drives the torque generated from and to the generator. The different components that form this unit are explained as follows:
1. Main shaft: the main shaft has been designed hollow of diameter 50mm bore to allow passing cables and pressure readings from the hub to the slipring on the nacelle. It transfers the mechanical power to the transmission and the encoder disk, which allows for rotational speed measurements and the loads to the structure through the bearings.
2. Bearings: two bearings one floating and one fixed attach and transfer the loads from the shaft to the structure. They are spherical bearings of type SKF 20214 TN9 (main bearing) and SKF 22215 E (secondary bearing).
3. Transmission: the transmission lowers the torque and increases the speed, in order to keep the size of the generator small in a ratio of 1:8. It consist of two different systems: a synchronous belt drive, which transfers the power to a parallel shaft axis bellow the main shaft, through the elastic belt with a 1:2 ratio, and a bevel gearbox that transfers the power from this secondary shaft 90 degrees to the generator in a ratio of 1:4 as seen in figure 6.
4. Brake: The brake is a pretension spring, pneumatically released disk friction brakes, manufactured by NEXEN, type Air Champ SSE-1000, that is activated in case of emergencies, it is controlled by two redundant electrical control 3/2 way fast-switching valves, the valves are set to be in close position and in case of electrical cutout or by exceeding the maximal rotational speed depressurizing the system and thus braking the wind turbine. The fixed side of the brake is connected to the fix bearing, and the rotating part is cover by the brake housing,
when activated the brake produces a frictional force of 340 Nm, it is then designed to set the reaction braking time to 1.5 seconds.
5. Slip Rings: slip rings are a rotary coupling that allows the transfer of electrical current or pneumatic pressure from a stationary unit to a rotary unit [5] The wind turbine contains two types of slip rings: an electrical slip ring, which has 12 signal channels (5A) and 6 power channels (10A) manufactured by A-Drive, type SRH 50120 - 6P/12S and a pneumatic slip ring with just one channel manufactured by Deublin.
o Nacelle Mainframe:
The nacelle mainframe serves as structure to hold the drivetrain system in position and to house extra measuring units that can’t be attached into the hub, its cover by the nacelle painted mate black and it is made of GFRP. The rear shape of the nacelle is aerodynamically optimized to resemble a “Kamm” shape.
o Yaw Plate System:
The yaw system is used to create unsteady inflow conditions affecting the rotor blades, as it is necessary to keep maximum distance from the blade tip to the wall equal to each rotation side of the rotor plane; the yaw is carried out via a bearing plate of 1.1m diameter on the wind tunnel floor manufactured by Thyssen Krupp / Rothe Erde of type KD 600 with external gear teeth of type No. 061.20.1094.500.01.1503, figure 5 shows the BeRT mounted onto the plate inside the wind tunnel.
F ig u re 5 B eR T W in d T u rb in e M o u n tin g In sid e W in d T u n n el.
o Generator:
The generator is a three-phase asynchronous motor with a rated power of 3kW, four poles and it is controlled via a converter at the control cabinet. It is manufactured by Bonfiglioli, its specification number is the following “BN 100LB 4 230/400-50 IP54 CLF B5 FD 40 NB 230 E3 EN3”, more detail information can be found on the manufacture webpage [6].
Together with the generator, a speed encoder senses the speed of the motor to form a precise speed close control loop, the motor has included a lifted safety brake which is used as a fallback emergency brake in case of main braking system failure.
o Nacelle electronics:
The nacelle electronics consist of an electronic board where all sensing signals are dent to the control cabinet. On the nacelle frame, there is a safety speed sensing switch, manufactured by IFM of type DI6001, it senses the speed by reading the ear of a red encoding disc attached to the main shaft, the maximum rotational speed for the shaft is set to be 260 RPM, the sensor is hard programmed to switch off the electrics, in order to activate the emergency brake, thus stopping the rotor safely.
o Control cabinet:
The control cabinet shown in figure 8 is specifically used to house the basic electronics to operate the wind turbine and extra space for measurement instruments. It has general manual override control elements, electronics for frequency converter and controller, grid protection units 3 phases of 400VAC, safety system switches and 24 VDC power supply. It contains the compressed air unit that controls the pneumatic braking system, manufactured by FESTO of type MSB6-1/2:C4:J3:I3-WP.
The frequency converter for the generator is used to control the turbine manually from the control cabinet or a computer software whenever necessary, it is manufactured by KIMO of type Transomnik 7U2-18-480.
The cabinet has been designed by AWINCO Ingenieur GmbH and a more detail report with the description of the cabinet components and electronic units can be found at [7].
F ig u re 6 C o n tro l C a b in et
Blade design:
The blade has a length of 1285mm and a root attachment pin of 28mm that attach the blade to the clamps of the rotor hub. The blades are the main part under investigation during this project, for that reason they were designed hollow for weight reduction, to be able to insert different sensing systems, carry different active actuators such as servos and implement flow control devices such as flaps.
The components included on the blade, which are relevant to this study will be explained thoroughly in the next sections, for more specific details the author recommend the SMART BLADE GmbH report [4].
o Aerodynamic design:
1. Airfoil: the blades have one simple airfoil profile of type CLARK Y, which has been modified slightlly for manufacturing and molding reasons. The trailing edge thickness has been adapted and hence reduced to a 0.83%c and then smoothly transitioned to the original airfoil coordinates as shown in figure 7.
T a b le 2 M a in C h a ra cteristics o f B la d es M ain C haracteristics of B lades
M ax. T hickness 11.89%c
M ax. T hickness position 30.5%c
T E th ickn ess 0.83%c
F ig u re 7 T ra ilin g E d g e M o d ifica tio n b y th e M a n u fa ctu rer
2. Chord twist and thickness distribution: the chord thickness and twist follow continuous functions. Between 65% and 95% of the blade span the chord and twist follow a linear function for an easy integration of the flaps in the smart blade. Below 10% of the span the chord and twist are designed such that the molding process is viable to meet manufacturing processes, as shown in figure 8.
F ig u re 8 B la d e in M o ld
o Structural Design:
The blade is designed to have the maximum internal space and for that same reason it lacks of typical spar, thus the structure is a skin loaded shell made of GFRP, molded in vacuum by infusion process with a final bonding process, to join together both blade shells.
The materials are non-crimp E-glass fiber for molding manufacture in vacuum with multiaxial and epoxy resin form the blade shells. The fibers are placed unidirectional with equal distributed parallel fibers, the lay-up has been done triaxial at 0, +45 and -45 degrees of 830g/m2 and a fabric of twill architecture of 163g/m2. The epoxy resin and curing agent are of RIM135. The two shells are bonded by an overlapping piece at the leading edge (LE) and at the trailing edge (TE) by a bonding paste of type MGS BP 20.
The ribs joints are reinforced on the SS by overlamination and later application of the same boding paste, as can be seen in figure 9.
BWKA-1.5 TUB blade:
The BWKA-1.5 TUB blade is specifically designed to test active flap control on BeRT wind turbine, so that moments and loadings can be alleviated at the root of the blade generated by gusts or instabilities of the inflow flow. It has integrated different sensing systems, such as: pressure hole probes, at the surface of the SS and PS chordwise and extra pressure holes at different spanwise positions, three hole probes better known as “Corrad probes”, strain gauges at the root, three flaps, leading edge payload bay that accommodates the three servos and pressure transducers, and a three axis accelerometer at the tip of the blade.
The blade followed the same manufacturing process explained previously. However, due to the cutouts necessary for the integration of the LE payload bay and the TE flaps, it has been required to include an extra rigid polymeric block on the area of the cutouts and in order to increase the stiffness of the blade a carbon fiber belt has been included in between the lamination process. Figure 9 shows the blade on the mold before bonding both shells together, it can be seen clearly the carbon fiber belt, the polymeric block for reinforcement on the cutout area and the extra pressure holes on the surface of the blade.
F ig u re 9 U n a ssem b led b la d e (left) a n d b o th sh ell b efo re b o n d in g (rig h t) .
o Flaps:
The BWKA-1.5 TUB blade has attached three flaps at the TE from positions 60% to 90% (900mm to 1350mm) of rotor radius; each flap has the same spanwise length of 30%
of rotor radius, with a chord length of 150mm. The flaps are actuated by three servos situated at the LE payload bay connected through a linkage mechanism as shown in figure 10. The pivoting mechanism axis is attached at four different positions on the blade and hold in position by body pins, as can be seen in figure 11 (left). The flaps are manufacture by 3D printing process with polyamide material, cut out from the original CAD blade design and optimized for weight reduction hence reducing inertia during flap actuation, figure 11 (right) shows the flap after printing.
F ig u re 1 0 P o sitio n a n d A tta c h m en t to S e rv o s in to P a y lo a d B a y
F ig u re 1 1 L in k a g e M ech a n ism (le ft) a n d S e rv o s A fte r P rin tin g P ro c e ss (rig h t).
o Leading Edge Payload Bay:
The LE payload bay is the area designed to house the actuating servos for the flaps, the pressure transducers, which sense and transfer the electrical signal of the changes in pressure at the three hole probes and help to support the three hole probes as shown in figure 12. The payload bay has the same length as the flap cutout and chord length of 24% of radius. It is fabricated from 3D printing process with polyamide material as the flaps.
F ig u re 1 2 S ch em a tic o f P o sitio n s fo r th e T h ree H o le P ro b e s.
o Servos:
Servo motors are widely used in industry and robotic applications, as they represent a method for control movements of mechanical devices. Servos normally work by a pulse of definite length, which is modulated at the chip inside the servos to set the output position of the shaft. The signal line does not supply any current to the servo, it rather just supply a small amount of current to state the output, the servos are instead fed by the power line of maximum 5 volts. The ground line serves as ground for the signal line and the power line. Finally, some servos have built in position feedback lines as in this case, which forms a closed loop indicating to the computer at which position the servo is at any time. Figure 13 shows the inner components of a regular servo [8].
F ig u re 1 3 S e rv o C ro ss-se c tio n . [8 ]
The servo used for the study of flap dynamics, it is the DA14-05-12.P.B.ST, manufactured by Volz Servos GmbH & Co. Some specifications supplied by the manufacturer can be found in figure 14 [9]. The servos transfer the movement to the flaps through a lever mechanism that can be seen in figure 10.
F ig u re 1 4 V o lz S e rv o M a n u fa c tu re r In fo rm a tio n [9 ]
o Three hole probes:
This kind of probes are designed to allow for a precise way to measure angles of attack by sensing the differential of pressure between the holes. During this study three hole probes measured just pressure difference between the holes in front of the probes are used, so it is possible to calculate the angle of attack at which the blade is rotating.
o Pressure transducers:
Pressure transducers are devices that convert pressure into analog electrical signal.
This is done by means of strain force produced by the action of pressure on a diaphragm;
strain gauges are bonded onto the diaphragm and wired into a wheatstone bridge configuration. The strain gauge then changes its resistance according to the deflection of the diaphragm and sends a voltage signal back to the computer.
1.3 Hot-wire Anemometry
Hot wire anemometer is a widely used technique to measure mean and fluctuating velocities of flow, this is due to its high capability to deliver an analogue output at high measuring frequencies [10].
This highly measurement frequency capabilities are possible, due to the heat transfer from the thin wire exposed directly to the flow. As the fluid is being accelerated, by heat transfer law the temperature of the wire will be reduced, hence the resistant to current of the wire will be reduce as well, so more current will be able to pass through it and this relation its what makes possible to estimate the velocity of the flow, by sensing the quantity of current necessary to keep the wire at the same temperature or keep the same current going through the wire, depending of the probe being used.
There are two types of control circuits for hot wire measurements:
• Constant current: During the realisation of this thesis, this type of control circuit was used, it maintains the same current at the hot wire as the resistance of the wire is lowered by the transfer of heat to the flow, for this reason; however this kind of circuits have a drawback, if fast velocity changes occur rapidly, the changes of temperature at the wire will lag behind due to its own temperature inertia.
• Constant temperature: this type of control circuit maintains the temperature of the wire constant by amplifying the current of the bridge as the resistance of the wire lowers by the action of the flow, the advantages of this type is that nonlinearities by heat inertia is almost removed from the output.
The hot-wire anemometry system comprises different devices to transform the heated wire output to a readable signal for post-processing. The system includes the heated wire, which will be exposed to the flow, a sensor and electronic systems. The wire material used for measurements was of tungsten covered with a thin layer of aluminum to avoid corrosion of the tungsten,
The system is then governed by the first law of thermodynamics, where the internal energy of the system, equals the work done by the system minus the heat lost to the surroundings, as exemplifies formula (2).
!"
!" = ! − ! (2)
Where,
! = !ℎ!"#$% !"!#$% !" !ℎ! !"#$
! = !! !
!! = ℎ!"# !"#"!$%& !"#$%&#% !" !ℎ! !"#$
! = !"#$% !"#"$%&"' !" ℎ!"#$%&
! = !! !!
!!= !! !!
! = ℎ!"# !"#$%&'""'( !" !"##$"%&'%( !"#$
Where the !"/!" is the energy change in time of the system, ! the work done by the system and ! the heat transferred to the surroundings.
The second law governing the system is the heat transfer law, which is the sum of the three types of heat transfer: conduction, convection and radiation. It is implicit that the heated wire will release heat to the surroundings by radiation to the air, by conduction to the supports and by convection to the flow field.
! = !"#$%!&'"# !" !"#$% + !"#$%!&'"# !" !"##$%&! + !"#$"%$&' !" !"##$"%&'%(!
There exist two types of sensor probes:
• Thin wire probes: consist of a thin cylindrical wire of around 3 - 5!m diameter and small length in order to reduce the area of measurement to a more punctual basis and reduce heat looses to the surroundings.
• Thin film probes: consist of a thin cylindrical film laid on a quartz or ceramic substrate to increase mechanical strength, the diameter is much bigger than the hot wires version and it is mostly used in applications where the frequency responses are lower.
• 2D and 3D probes: are a combination of several hot wire probes mostly mounted in X for 2D or perpendicular to the flow directions, in order to sense more than one direction of the flow field.
1.4 Boundary Layers
In 1904 Prandtl published the theory of boundary layer, which is still today’s main reference when talking about this subject. He defined the boundary layer into two main flow areas, an outer inviscid flow, where viscous effects are negligible and an inner very thing layer where viscous effect need to be taken into account.
At this inner region exist normal gradients of velocity to the wall; which are reduced by shear stress to zero, what is normally know as the no-slip condition, where the
velocity at the wall must be zero. Prandtl (1904) stated that at this thin boundary layer three major conclusions could be identified:
1. The flow outside this thin bounded area is unaffected by viscous effect and can be fairly described by ideal flow theory.
2. The changes of pressure along the layer are negligibly small, so it could be said that the pressure its maintain as !! from the edge of the boundary layer to the wall.
3. The flow in the boundary layer could be treated by simpler boundary layer equations.
Under the third condition, Prandtl then stated that it is not necessary to solve for the whole Navier-Stoke’s equations, rather a simplify version of it could be used [11].
With this simplification was then possible to determine that boundary layers grow with the distance due to viscous diffusion. This growth of boundary layer give an opening to the boundary layer thickness; which is considered to be the distance from the wall where the velocity gradient starts to develop, so its universally taken as the 0.99!!. The boundary layer thickness hence develops as explained by the distance ! and the conditions of the flow defined by the Reynolds number (!" = !"/!), where ! is the reference distance, ! the flow speed and ! the kinematic viscosity of the flow (!/!).
The Reynolds number is a dimensionless quantity that stated the conditions of the flow. It is consider as the ratio of dominance between inertial forces over viscous forces.
Boundary layers are divided into three main classes:
1. Laminar layer: this type of layers presents organized flow structure divided by layers of flow.
2. Transitional layer: on this type of layer occurs the transition between laminar and turbulent layer.
3. Turbulent layer: turbulent layers are mainly chaotic, where fluid particles move randomly and momentum is interchanged between flow layers, this enhances mixture and boundary layer growth. These boundary layers are mainly expected on flow with Reynolds numbers around 1 x 106.
1.4.1 Displacement Thickness:
Displacement thickness is by definition the distance by which the plate or wall could be moved so that the volumetric wouldn’t be affected. It is calculated by the following formula:
!∗= 1
! ! − ! !"
!
!
The limit ! is established so that is large enough not to affect the result of the displacement thickness, meaning that it does not matter whether the layer thickness it taken to be 90% or 99%; this consideration implies that under visual investigation of boundary layers the final result of the displacement thickness is not highly affected by measuring errors [12].
1.4.2 Momentum Thickness:
Momentum thickness is defined to be the difference of momentum represented by a viscous flow, where boundary layer effects are taken into account, against the momentum of an ideal inviscid flow theory with the same volumetric flow [12]
! = 1
!! ! ! − ! !"
!
!
1.5 Actuator Disc Theory
Actuator disc theory uses the basics of momentum conservation theory into a control volume constrain by an upper and lower streamline, where no flow should overpassed this boundaries, in order to keep conservation of mass along the control volume.
This theory is divided into two different methods: axial momentum theory which will be used later on for blockage effect calculations and rotating annular momentum theory.
1.5.1 Axial Momentum Theory:
The axial momentum is an idealize theory which allows to calculate different physics aspects along the streamtube. Figure 15 shows a representation of the streamtube and the different positions use for the explanation of the theory, where 1 is taken as the inflow, 2 the position right before the rotor, 3 the position right behind the rotor and 4 somewhere in the wake [13].
F ig u re 1 5 A x ia l S trea m T u b e [1 3 ]
This theory assumes the flow to be frictionless between 1-2 and 3-4, p1=p4 as the pressure in the far wake must equal the inflow pressure and V2=V3 as the velocity of the flow does not change at positions 2 and 3.
The changes on height of the streamlines along the annular control volume is negligible, hence it is possible to apply Bernoulli’s equation to any streamline. It is possible to find then:
!!
! +1
2 !!! = !!
! +1
2 !!! !!
! +1
2 !!!= !!
! +1
2 !!! (3/4) As p1=p4, it is possible to equal equation (3/4) and after some algebra reach to (5);
!!− !! = 1
2 ! (!!!− !!!) (5)
F ig u re 1 6 C h a n g es o f P ressu re a n d V elo city A cro ss th e A ctu a to r D isk [1 4 ]
Figure 16 exemplifies the changes of pressure and velocity as the flow passes through the rotor disc. It is possible to see at the disc a drastic change of pressure as a result of the extraction of kinectic energy from the wind. This change on pressure can be related to the thrust that the rotor applies on the wind, as force is equal to pressure times the area of application.
! = !!− !! !! (6)
! = 1
2 ! !!!− !!! !! (7)
Where !! is the cross-sectional area of the disc. After some math is then possible to clonclude, that the velocity at the disc is the average between the inflow velocity and the wake velocity.
!! = !!=!!+ !!
2 (8)
And gives a step into the inclusion of the axial induction factor definition; which is the fractional decrease in wind velocity between the free stream and the rotor plane [14].
The introduction of this variable is necessary to give a prior understanding of the disc behavior. Where;
! = !!− !!
!! (9)
So that;
!! = !!= !! 1 − ! (10)
!! = !! 1 − 2! (11)
This result demonstrate that the higher the induction factor higher the blockage effect.
The introduction of dimensionless variables makes necessary the definition of thrust and power into dimensionless quantities. It s then at this point that the coefficients of thrust and power enter the theory defined as;
!!= 2 !
! !!! !! (12)
which defines the ratio between the thrust applied to the wind and the dynamic pressure of the free stream applied on the rotor cross-sectional area.
Power is defined to be the force of thrust times the velocity of the flow at the rotor
!!.
! = ! !! (13)
and the coefficient of power as;
!!= 2 !
! !!! !! (14)
which defines the extracted power versus the work done by dynamic pressure at freestream velocity !!.
These two variables are hence defined by the induction factor ! by;
!! = 4! (1 − !) (15)
where ! = 0 means that there is no kinetic energy extraction at the rotor and !! = !!; in this same manner Cp is defined as;
!!= 4! (1 − !)! (16)
These two conclusions give as result the two roots;
! = 1/3 ! = 1
by differentiating the problem when;
!!!
!" = 4 1 − ! 1 − 3! = 0
where the maximum can only be ! = 1/3, that correspond to what is know as the Betz limit with a maximum power coefficient of;
!! !"#=16
27≈ 0.593
1.6 Blockage Correction Theories
1.6.1 H. Glauert (1933)
H. Glauert [15] was one of the first aerodynamics to work on blockage effects for airscrews in wind tunnels. His publication from 1933, “Wind Tunnel Interference on Wings, Bodies and Airscrews”, explains the application of correction methods for thrust generated by a propeller in wind tunnels.
As it is well known, the flow must be accelerated aft the propeller and this flow acceleration produce a decrease in static pressure behind it; by Bernoulli’s theorem, it is then possible to know that lower static pressures means lower velocities outside the slipstream, this means that the velocity outside the slipstream would be slower of that upstream the propeller. This effect is produce by the restriction of the walls, so it then reacts back to the propeller producing a higher thrust.
He explains in his paper that the correction accounts for this restriction by means of momentum theory, making a remark that the pressure within the slipstream must be equal to that outside the slipstream. Under this consideration he then states the correction formula (17), where the corrected velocity !′ is the velocity in free stream, corresponding to the wind tunnel velocity !, at which the propeller produces the same thrust and torque as there were no wall restrictions.
!′
! = 1 − ! !
2 1 + 2! (17)
Where;
!′ = Corrected inflow speed
! = Measured inflow speed
! = is the ratio between the cross-sectional of the propeller divided by the cross- sectional area of the wind tunnel
! =!"
!"
where, !" is the propeller swept area and !" is the cross-sectional area of the wind tunnel.
! = !
! !" !!
where, ! is the thrust of the propeller and ! is the density of the air.
Glauert states that the values of !! must be lower of those measured upstream the propeller. Cases where thrust is negative are not shown in the paper; however, several research teams, among them J. Ryi, W. Rhee, U. C. Hwang, J. Choi [16] have been able to apply Glauert corrections successfully to wind turbines.
1.6.2 A. Pope (1964)
A. Pope [17] in his book “Wind Tunnel Testing” makes reference to many methodologies when introducing a model in a wind tunnel, among them, he dedicate a part to blockage effects produce by models and its effect in the surrounding flow.
The author states that blockage is form by two parts and solid blockage, which is the blockage of the model and wake blockage, which is the blockage produce by the expansion of the wake, he sums them up as in equation (18)
! = !!"+ !!" (18)
Where, ! is the total blockage, !!" is the solid blockage and !!" is the wake blockage.
However, Pope (1964) states: “that equation (18) corresponds closely to assuming that the model frontal area is one-fourth effective, or equation (19) yields a correction quite close to the theoretical development, and may be used for rough mental calculations or when the model shape defies a theoretical approach.”
! =1 4
!"#$% !"#$%&' !"#!
!"#$ !"#$%&' !"#! (19) So from this approach is possible to derive exact equations for easier use, where the subscript ! are uncorrected data measured in the wind tunnel;
• Velocity:
! = !!(1 + !) (20)
• Dynamic pressure:
! = !!(1 + 2!) (21)
• Reynolds number:
! = !!(1 + !) (22)
1.6.3 Hackett-Wilsden “Pressure Signature Method” (1960)
The Hackett-wilsden [18] method is a blockage correction method that uses static pressure signature at the walls of the wind tunnel. It consists on the modeling of the body by a source-sink-source method, which accounts for the solid blockage and the blockage produce by the wake of the body.
The authors stated, that due to the wall restriction, an acceleration is produced around the body, thus producing a decrease in static pressure on the axial axis of the wall; this decrease can then be compared to the static pressures along the wall of the empty wind tunnel, in order to get an approximation of the changes of static pressure around the body and finally correct it to values of free stream wind speed.
This method employs the modeling of the body as two pairs of source and sink, where each is named solid blockage pair and wake blockage pair. Figure 17 shows a sketch that reassembles the location of the sources and sinks along the axial axis of the wind tunnel.
F ig u re 1 7 P o sitio n o f S o u rc es a n d S in k s (u p ) a n d E ffects o f B lo c k a g e o n th e W in d T u n n el W a ll (lo w )
By using potential flow theory, the model is capable of calculating the influence of each source/sink that will end up matching the experimental data measured of the model in the wind tunnel. Equation (23) shows the summation of influence of each source/sink.
! = !!!!!!
4!(!!!+ !!)(!!!+ !!!+ !!)!/!+ !! 2!"
!
!!!
(23)
Where,
Q = source/sink strength.
!!= wake source strength.
B = wind tunnel width.
H = wind tunnel height.
Then a method of images is applied, in order to model the walls constrain, where each source/sink is modeled longitudinally and vertically as shown in figure 18. As the method depends on seven different variables, the author recommends an engineering method for blockage estimation, where the problem reduces to calculate for source strength and span, axial position of strengths and source and sink separation.
F ig u re 1 8 Im a g e M eth o d M o d elin g
In cases where the separation bubble is large compared to the viscous wake blockage, the peak of the curve will coincide with the center point of the solid blockage source and sink pair. By applying such a simplification, it is then possible to take the half width and the peak height and use the “Carpet of determination” as called by the author, in order to determine the values of !!/! source and sink spacing and then use the result to get the u/(!!/!!). By means of this result is then possible to equate, in order to get the changes of velocity along the wind tunnel axial axis, as stated in equation (24). Then the positions of source and sink relative to the model will be calculated by equation (25) and (26).
!!
!!!!= (! !!)!"#
[! (!! !)]!"# (24)
!
!
!
= !!
! −!!
! −1 2
!!
! (25)
!
!
!
= !!
! −!!
! +1 2
!!
! (26)
Where,
!! = distance between source and sink strengths.
!! = suction peak relative to the tunnel datum.
!! = model position relative to the tunnel datum.
For viscous blockage determination the authors Hackett-Wilsden state that by modeling the viscous source at the model position, it seems reasonable to neglect the span and use a linearize pressure coefficient change along the wind tunnel. This is because the viscous blockage correction, represents a small part of the total correction, so by applying equation (27) allows correct for dynamic pressure at the model.
∆!
! =1
2∆!! (27)
1.6.4 M. J. Werle (2010)
M. J. Werle [19] has followed the theory of Glauert [15] and in 2010 developed a correction methodology for wind turbines in wind tunnels, by means of momentum theory and actuator-disc theory.
In contrast to a propeller the flow behind a wind turbine decelerate the flow as kinetic energy is extracted from it, by momentum theory is known that the more energy is extracted from the incoming wind, the lower will be the speed of the flow behind the wind turbine. In this case, the slipstream expands instead of contracting as happens for propellers in wind tunnels. The expansion of the slipstream contracts the flow outside the slipstream reducing the static pressure, hence accelerating the flow in this area due to the presence of the walls.
Like Werle (2010) introduces the control volume as shown in figure 19, where momentum, mass and energy are applied over it, assuming that:
1. Inviscid incompressible flow is through a conduit of constant cross-sectional area A!.
2. Flow with specified uniform pressure and velocity enters at upstream infinity.
3. Flow exits at downstream infinity at uniform pressure p! and a slipstream occurs between the mainstream fluid exiting at velocity V! and the fluid exiting at velocity V!, which passed through the actuator-disc model of the rotor. The three variables V!, V! and p! are threated as unknowns.
4. The cuts around the propeller shrink to its surface. (Pag. 1318)
The governing equations for the methodology are as follow:
• Continuity:
!!!!= !!!! (28)
!!!! = !!!!+ (!!− !!)!! (29)
• Bernoulli:
!!+1
2!!!!= !!+1
2!!!! = !!+1
2!!!! (30)
!!+1
2!!!! = !!+1
2!!!! = (31)
• Momentum:
!!!!+ !!− !! !!− !!!!= !!!!!!+ ! !!− !! !!!− !!!!!! (32)
F ig u re 1 9 W erle M eth o d C o n tro l V o lu m e
Combining equations(28-31), the author reach to the conclusion that;
!!− 1 2!!+ !!− 1 = !(!!!− !!!) (33) where, the nondimensional variables are stated as follows;
!! =!!
!!; !!=!!
!!; ! =!!
!!;
and the speed at the rotor becomes;
!!=!!
!!= !! !!+ !!
2!!+ !!− 1 (34)
So that the coefficient of power and thrust can be calculated to be;
!!=!! !!+ !! !(!!− !!)
2!!+ !!− 1 (35)
!! = !!!− !!! (36)
The author then applies this method to a set of different blockage ratios from 5% to 25% and shows how the power curves shifts for the different ratios, due to the effect of blockage, stating that is then not clear how to correct the power curves from the wind tunnel measurements to an unconstrained case.
Then in order to establish an exact correction formula to the methodology, Werle demonstrates the impact of ! on the maximum power extractable from the freestream, reaching to the inference that;
(!!/!!!)!!!≈ (!!/!!!)!!! (37)
where !!! is the maximum is the maximum power extractable, which gives as a result that the correction formula for the coefficient of power becomes;
(!!)!!!≈ 1 − ! !(!!)!!! (38)
And so the remaining parameters turn out as;
(!!)!!!≈ 1 − ! !
1 + ! (!!)!!! (39)
(!!)!!!≈ (1 − !)(!!)!!! (40)
Concluding that equations (37-40) are exact at the maximum power condition and that the results are approximations for values of !!/!!! within 0.1 up to !!/!!!= 1.25, where the method is highly sensitive to the generalized area ratio due to its dependency to a power-law.
