IFB

>> Layout design for a Venturi to encase a wind turbine, integrated in a high rise <<

Diploma Thesis of cand. aer. Christina Beller

accomplished at Ris National Laboratory Wind Energy Department DK-4000 Roskilde Copenhagen, November 2007

this work is dedicated to my lovely family

IFB

Institut fr Flugzeugbau

Universitt Stuttgart

Stiftungslehrstuhl Windenergie (SWE) Prof. Dr. Dipl.-Ing. Martin Khn

Legal declaration

Herein I affirm that I draw up the submitted diploma thesis on my own and without improper supplies. The used sources are completely given in the references.

Rechtliche Erklrung
Hiermit versichere ich, dass ich die vorliegende Diplomarbeit selbstndig und ohne unzulssige Hilfsmittel verfasst habe. Die verwendeten Quellen sind vollstndig im Literaturverzeichnis angegeben.

... Ort, Datum [Christina Beller]

IFB

Institut fr Flugzeugbau

Universitt Stuttgart
12.03.2007

Stiftungslehrstuhl Windenergie (SWE) Prof. Dr. Dipl.-Ing. Martin Khn SWE.JJT.05: External diploma thesis in flow simulation

Layout design for a Venturi to encase a wind turbine, integrated in a high rise
Wind energy plants should enter urban zones to support the on- site power supply with regenerative energy. Integration of wind turbines in structures demands developments of new concepts and design methods. This research project is concerning with the optimization of a Venturi for a multi-storey building, surrounded by very low tilled area, which is planned with three encased turbines, integrated in the upper fifteenth part of the structure. The building could at this take on the tower` s function. On the contrary to conventionally construction methods, the turbines should be sheathed. Effects in this aspect are reflected in flow characteristics as well as tip losses. A wind tunnel model of the architecture, considering the accordant block outs, has already been manufactured. Measurement results based on wind tunnel experiments with different flow characteristics are on hand. Additional an adequate CFD model has been generated. The company Norwin A/S, assisting the project and providing components, plans to attach already existing rotor blades. The blades are not optimized for this application, yet. A blade `s retro- design is to be contemplated.
[Source: http://www.ifb.uni-stuttgart.de/~doerner/Hausrotor2.jpg; not according to the project]

Approach: - Enquiry of related, comparable topics - Planning and conception of the thesis (first week in May, with Norwin A/S, Risoe and SWE) - 2-D (rotationally symmetrical) Design of optimized casing and CFD- calculations, in respect to the attached blades, based on the existing model `s measurements and values, to maximize energy generation. - Retro- design of the attached blades with respect to Venturi- rotor- combination and therefore special tip loss situation Location: Beginning: Duration: Editing place is Roskilde, DK May 2007 ca. 6 months Betreuung extern Ris Christian Bak Ris National Laboratory Frederiksborgvej 399 DK-4000 Roskilde Denmark e-mail: christian.bak@risoe.dk

Betreuung Uni Stuttgart Juan Jos Trujillo, MSc Stiftungslehrstuhl Windenergie am Institut fr Flugzeugbau Allmandring 5B 70659 Stuttgart Tel. (0711) 685 6 8289 e-mail: juan-jose.trujillo@ifb.uni-stuttgart.de

IFB

Institut fr Flugzeugbau

Universitt Stuttgart

Stiftungslehrstuhl Windenergie (SWE) Prof. Dr. Dipl.-Ing. Martin Khn

Institut fr Flugzeugbau Pfaffenwaldring 31 70550 Stuttgart

cand. aer. Christina Beller Hospacher Steig 1 72401 Haigerloch- Gruol Matr.-Nr.: 2022015

Postadresse: D-70550 Stuttgart Hausadresse: Pfaffenwaldring 31 D-70569 Stuttgart Telefon: Telefax: 49(0)711/685-2402 49(0)711/685-2449

Diplomarbeit
Layout design for a Venturi to encase a wind turbine integrated in a high rise

Aufgabenstellung: Parameterstudie anhand CFD Simulationen zum Entwurf eines geeigneten Venturirohrs als Ummantelung fr eine Windenergieanlage die in Gebude integriert sein sollte. Umfang:

Literaturstudie verfgbarer Experimente von Venturirohren. Einarbeitung in die CFD Simulation (das CFD Simulationstool FIDAP 8). Erstellung eines zwei-dimensionalen und symmetrischen Models eines Venturirohrs in FIDAP 8 mit 5 bis 10 ausgewhlten Geometrien aus der Literatur und gleichmige Einstrmung ohne Windenergieanlage. Berechnung verschiedener Konfigurationen des Venturirohrs und Auswahl der besten geometrischen Parameter im Bezug auf Erhhung der Windgeschwindigkeit in dem Hals des Rohrs. Definition von Optimierungsparameter (Energiesteigerung, Hchstbeschleunigung, usw). Simulation einer Windenergieanlage durch Einbringung eines Aktuator- Disk- Modells in das Oben genannter Venturirohr. Effekten die auf den Blttern auftreten wegen verminderter TipVerluste sollen bercksichtigt werden. Juan Jos Trujillo, MSc (IFB)

Betreuer: Christian Bak (Ris) Arbeit ausgegeben am: Arbeit abgegeben am: Note festgelegt am: Stuttgart, den 1. Mai 2007

........................................... ........................................... ............................................................ Prof. Dr. Dipl.- Ing. Martin Khn

............................

ABSTRACT

In this work a general CFD supported investigation of the potential of Venturi shaped turbine casings, compared to a constant in diameter tubing, is carried out. Inspired by and based on the skyscraper design Castle House, including turbines, integrated into the buildings structure, a parameter study about encased turbines in urban areas is developed. Former works were handling mostly so called shrouded turbines or diffuser augmented wind turbines (DAWT). The shrouds allow the flow to surround the casing. Shrouds, shaped like profiles, are creating lift and so forcing the approaching air to accelerate. Since a turbine is integrated in a building the blocking of the turbine and the shroud are additionally coupled to the blockage of the building itself. The simulation approach is validated by comparing pressures derived by modelling a DIN EN ISO 5167-3:2003 [1] Venturi device in the CFD environment of FIDAP with pressure values ascertainable with experimental modified theory equations given in the standard [2]. The deviation is 4.07%. Starting with this standardized Venturi geometry, combined with characteristic wind tunnel design rules [3], three tubing geometry are designed. The geometry up-scale and change of wind velocity leads to ReD 3.6 * 106, referring to the inlet diameter, and therefore exceed the scope of the Venturi ISO standard. In 1998 Tianjin University approved within experiments the applicability of the standard also for slightly bigger diameters, 1.4 m, and Reynolds numbers within a range up to ReD = 4.0 *106 [4]. The top floor of the building is simplified as a coin model with a hole in the middle, wherein the turbine is placed later on. This is to simplify the model to be axis- symmetric. A constant diameter for the hole means no constriction and represents the contraction (Co) of one. As soon as a Venturi curvature is inserted in the coin hole, the flow is contracted and the highest velocity will occur in the smallest diameter. Contractions of 1.00, 2.25, 4.00, and 9.00 are simulated with the help of the CFD tool FIDAP. The coin thickness, outer diameter and the holes diameter are constant parameters of the study. Analyses of the flow through the single contractions are carried out and by comparison the most effective geometry by energetic meaning is chosen to work on further with. In the next step, a turbine, represented by an actuator disc, is implemented by body forces. A focus is taken on positioning the turbine with varying thrust coefficients CT = 0.300; 0.363; 0.439; 0.531; 0.590; 0.623; 0.640 and 0.656 at three different characteristic Venturi depths, with respect to efficiencies, depending on increased blocking by the turbine in the device. The resulting flow behaviour is qualitative described and a discussion follows.

II

In dieser Arbeit ist eine generelle CFD untersttzte Untersuchung des Potenzials einer Venturi geformten Windenergieanlagenummantelung im Vergleich zu einer im Durchmesser konstanten Ummantelung durchgefhrt. Inspiriert durch und basierend auf dem Hochhausentwurf Castle House, welcher in die Huserstruktur integrierte Windenergieanlagen vorsieht, ist eine Parameterstudie ber ummantelte Turbinen in stdtischen Gebieten entstanden. Vorherige Arbeiten beschftigten sich mit den so genannten Shrouded Turbines oder Diffuser Augmented Wind Turbines. Die bisherigen shrouds erlauben es dem Medium, die Ummantelung zu umstrmen. Shrouds, die geformt sind wie ein Flgel- bzw. Blattprofil, erzeugen Auftrieb, wodurch die anstrmende Luft durch die Struktur beschleunigt wird. Sobald eine Windenergieanlage in ein Gebude integriert ist, ist der Widerstand durch die Mantelformung und die Turbine zustzlich an die Blockadewirkung des Gebudes gekoppelt. Der Simulationsansatz ist durch Vergleich von Drcken, ermittelt durch die Modellierung einer DIN EN ISO 5167-3:2003 [1] Venturi Messapparatur in der CFD Umgebung von FIDAP, mit Druckwerten, errechenbar mit experimentell modifizierten Gleichungen, welche in der Norm [2] gegeben sind, validiert. Die Abweichung betrgt 4.07 %. Ausgehend von der normierten Venturiform und kombiniert mit charakteristischen Windkanalentwurfsgesetzen sind drei Rohrgeometrien entworfen. Die geometrische Hochskalierung und die nderung der Windgeschwindigkeit, um von dem Validierungsmodel zum aktuellen Parametermodel berzugehen, fhrt zu einer auf den Einlassdurchmesser bezogenen Reynoldszahl von ReD 3.6 * 106 und berschreitet somit die Anwendbarkeit der Venturi ISO Norm. Im Jahre 1998 bewies die Universitt von Tianjin mittels Experimenten die Anwendbarkeit des Standards fr etwas grssere Durchmesser, 1.4 m, und fr Reynoldszahlen in einem Rahmen von bis zu ReD = 4.0 *106 [4]. Die oberste Etage des Gebudes ist zu einem Mnzen Model mit einem Loch in der Mitte, worin die Windenergieanlage spter eingesetzt wird, vereinfacht. Das wird gemacht, um das Model als achsensymmetrisches Problem definieren zu knnen. Ein konstanter Durchmesser fr das Loch bedeutet keine Einschnrung and reprsentiert das Kontraktionsverhltnis (Co) eins. Sobald eine Venturikrmmung in das Mnzenloch eingefgt wird, ist die Strmung verengt und die hchste Geschwindigkeit wird in dem kleinsten Durchmesser auftreten. Kontraktionsverhltnisse von 1.00, 2.25, 4.00 und 9.00 sind mit der Hilfe des CFD Werkzeugs FIDAP simuliert. Die Mnzendicke, der ussere Radius and der Lochdurchmesser sind konstante Parameter der Studie. Analysen der Strmung durch die einzelnen Kontraktionen sind durchgefhrt und durch Vergleich ist die effektivste Geometrie, bezglich energetischer Bedeutung, ausgesucht, um damit weiter zu arbeiten. In dem nchsten Schritt ist eine Windenergieanlage, die durch eine Aktuator Disk dargestellt wird, mittels Krperkrften implementiert. Der Fokus wird dabei auf die Positionierung der Turbine mit variierendem Schubkoeffizienten CT = 0.300; 0.363; 0.439; 0.531; 0.590; 0.623; 0.640 und 0.656 gelegt. Drei verschiedene charakteristische Venturitiefen, mit Bercksichtigung auf die Effektivitt, welche von dem zunehmenden Widerstand der Turbine in der Apparatur abhngt, sind untersucht. Das resultierende Strmungsverhalten ist qualitativ beschrieben und eine Diskussion folgt.

III

IV

INDEX OF CONTENTS

ABSTRACT INDEX OF CONTENTS NOMENCLATURE FIGURES TABLES

I V VII XI XIII

INTRODUCTION 1.1 1.2 1.3 HISTORY UNTIL NOW CASTLE HOUSE TURBINE POSITIONS

1-1 1-3 1-5 1-9 2-1 2-2 2-4

2-4 2-7

VENTURI 2.1 2.2 EFFECT VENTURI ISO STANDARD

2.2.1 2.2.2 ISO 5167-1:2003 ISO 5167-3:2003

2.3

APPLIED GEOMETRY
2.3.1 2.3.2 DESIGN LAWS PARAMETER MODELS

2-12
2-13 2-17

FIDAP 3.1 INTRODUCTION AND VALIDATION

3.1.1 3.1.2 3.1.3 FI-GEN FI-PREP FI-POST

3-1 3-3
3-4 3-12 3-15

MODEL WITHOUT TURBINE 4.1 SETTINGS

4.1.1 4.1.2 4.1.3 DOMAIN BOUNDARY CONDITIONS INPUT

4-1 4-2
4-2 4-2 4-3

4.2

RESULTS
4.2.1 4.2.2 4.2.3 4.2.4 RESULTS CO1.00 RESULTS CO2.25 RESULTS CO4.00 RESULTS CO9.00

4-3
4-4 4-7 4-9 4-11

4.3 5

CONCLUSION

4-12 5-1 5-2

5-3 5-3 5-4 5-4

MODEL WITH TURBINE 5.1 ACTUATOR DISC SIMULATION IN FREE FIELD/ VALIDATION
5.1.1 5.1.2 5.1.3 5.1.4 DOMAIN BOUNDARY CONDITIONS INPUT RESULTS AD FREE

5.2 5.3

VENTURI ENCASED ACTUATOR DISC SIMULATION SETTINGS

5.3.1 5.3.2 5.3.3 DOMAIN BOUNDARY CONDITIONS INPUT

5-5 5-6
5-7 5-8 5-8

5.4

RESULTS
5.4.1 5.4.2 RESULTS CO2.25AD RESULTS CO1.00AD

5-9
5-9 5-18

5.5 6

CONCLUSION

5-20 6-1 6-2 6-4 6-5

SUMMARY 6.1 6.2 6.3 STATE OF THE ART FUTURE PROSPECTS ACKNOWLEDGEMENTS

REFERENCES APPENDIX

A a

VI

NOMENCLATURE

ABBREVIATIONS
2-D 3-D AD CFD DAWT EN Fig. FI-BC FI-GEN FI-MESH FI-POST FI-PREP FI-SOLV GUI SP Tab. : two dimensional : three dimensional : actuator disc : computational fluid dynamic : diffuser augmented wind turbine : European norm : figure : Initial and Boundary Condition Processor Module/ boundary conditions : FIDAP mesh generation module/ generation : Mesh Generation Module/ meshing : Postprocessor Module/ post processing : Problem Data Specification/ preparation : Computational Simulation/ solver : graphical user interface : stagnation point : table

LATIN LETTERS
Symbol a A Aeff Ai Ainlet Aoutlet Arotor Athroat b c C CT CT_f CP CP_f CP* Co Co1.00 Co2.25 Co4.00 Co9.00 Co1.00pos1 Co1.00pos2 Co1.00pos3 Meaning induction factor cross sectional area effective cross section area cross section are at position i =1; 2 inlet cross section area outlet cross section area rotor cross section area throat cross section area constant outer radius of the coin model pressure loss due to the Venturi device flow rate coefficient thrust coefficient, referring to the rotor area free turbine ideal thrust coefficient, referring to the rotor area power coefficient, referring to the rotor area free turbine ideal power coefficient, referring to the rotor area power coefficient, referring to the inlet area contraction model, contraction 1.00 model, contraction 2.25 model, contraction 4.00 model, contraction 9.00 model, contraction 1.00, turbine at position one model, contraction 1.00, turbine at position two model, contraction 1.00, turbine at position three Unit [-] [m2] [m2] [m2] [m2] [m2] [m2] [m2] [m] [Pa] [-] [-] [-] [-] [-] [-] [-] [-] [-] [-] [-] [-] [-] [-]

VII

Symbol Co2.25pos1 Co2.25pos2 Co2.25pos3 Co1.00AD Co2.25AD d dm dAi D Dm Do dm/dt dE/dt E fturbulent fx F g G J Jmax Jmin k l1 l2 lc L Lanal Ldiffuser LISA LVenturi length m M1 M2 Ma n o O Pi pi_stat pJ pJm pJmax pJmin pQ pQm pQmax pQmin pos1 pos2 pos3 P q Q Qmax

Meaning model, contraction 2.25, turbine at position one model, contraction 2.25, turbine at position two model, contraction 2.25, turbine at position three model, contraction 1.00, turbine implemented model, contraction 2.25, turbine implemented throat diameter model throat diameter annular cross section area i = 1; 2 inlet diameter model inlet diameter original inlet diameter mass flux energy flux throat length one, inlet to measurement point Q turbulent friction factor force density throat length two, measurement point Q to diffuser gravitation constant empirical constant reference pressure measurement point, given in the standard bore J ending bore J beginning turbulent kinetic energy Karman constant construction line one construction line teo help construction line length analogous length diffuser length length of the ISA-1932 nozzle all over length of the Venturi device length of the Venturi and thickness of the coin mass R1 centre point R2 centre point Mach number dimensionless parameter for the power law velocity profile construction parameter osculation point Power in a cross section at position I = 1; 2 static pressure at position i = 1; 2 static pressure at reference point J mean value of pJmin and pJmax static pressure at bore J ending static pressure at bore J beginning static pressure at reference point Q mean value of pQmin and pQmax static pressure at bore Q ending static pressure at bore Q beginning position one for the turbine position two for the turbine position three for the turbine power heat flux reference pressure measurement point, given in the standard bore Q ending

Unit [-] [-] [-] [-] [-] [m] [m] [m2] [m] [m] [m] [kg/s] [W] [m] [-] [m/s2] [m] [m/s2] [-] [-] [m] [m] [-] [-] [-] [-] [-] [m] [m] [m] [m] [m] [m] [kg] [-] [-] [-] [-] [m] [-] [W] [Pa] [Pa] [Pa] [Pa] [Pa] [Pa] [Pa] [Pa] [Pa] [-] [-] [-] [W] [K/s] [-] [m]

VIII

Symbol Qmin r R R* R1 R2 Rrotor Red ReD s T u u+ u* u uw vi vincompressible vinlet vm vm_mean vout vo_mean vthroat vthroat_tubed vthroat_CFD vw vAD VAD xo xsep_Co4.00 xsep_Co9.00 xst_Co1.00 xst_Co2.25 xst_Co4.00 xst_Co9.00 xR1 xR2 y+ y1 y2 yc yo xsep_Co4.00 xsep_Co9.00 yst_Co1.00 yst_Co2.25 yst_Co4.00 yst_Co9.00 yR1 yR2

Meaning bore Q beginning radial coordinate radius construction circle radius first Venturi inlet radius second Venturi inlet radius, blending tangentially first radius and throat rotor radius Reynolds number, referring to the diameter, here throat diameter Reynolds number, referring to the diameter, here inlet diameter distance between y1 and y2 thrust velocity dimensionless velocity friction velocity shear rate wall velocity mean velocity at position i = 1; 2... highest maximal local velocity to consider the flow as incompressible mean velocity inlet mean velocity model mean velocity mean velocity exiting the device original mean velocity mean velocity in the throat mean velocity in throat, in case device is supported by tube mean velocity in throat, in case device is not supported by tube, CFD result mean wind velocity mean velocity in actuator disc cross section actuator disc volume osculation point centre x- coordinate contraction four separation point x- coordinate contraction nine separation point x- coordinate contraction one stagnation point x- coordinate contraction 2.25 stagnation point x- coordinate contraction four stagnation point x- coordinate contraction nine stagnation point x- coordinate R1 centre x- coordinate R2 centre x- coordinate element Reynolds number first construction line equation second construction line equation help construction line equation osculation point centre y- coordinate contraction four separation point y- coordinate contraction nine separation point y- coordinate contraction one stagnation point y- coordinate contraction 2.25 stagnation point y- coordinate contraction four stagnation point y- coordinate contraction nine stagnation point y- coordinate R1 centre y- coordinate R2 centre y- coordinate

Unit [m] [m] [m] [m] [m] [m] [m] [-] [-] [m] [N] [m/s] [-] [m/s] [(Nm/kg)] [m/s] [m/s] [m/s] [m/s] [m/s] [m/s] [m/s] [m/s] [m/s] [m/s] [m/s] [m/s] [m/s] [m3] [m] [m] [m] [m] [m] [m] [m] [m] [m] [-] [-] [-] [-] [m] [m] [m] [m] [m] [m] [m] [m] [m]

IX

GREEK LETTERS
Symbol

Meaning diameter ratio diffuser opening angle normal distance from the wall
boundary layer thickness

Unit [-] [] [m]

[m]

p pISO pstat pFIDAP_averaged pFIDAP_discrete x i w i i

viscous dissipation differential static pressure between J and Q differential static pressure between J and Q, derived from the standard differential static pressure differential static pressure between J and Q derived by FIDAP, averaged differential static pressure between J and Q derived by FIDAP, discrete actuator disc thickness expansion number adiabatic coefficient static pressure ratio density at position i = 1; 2 shear stress wall stress dynamical viscosity at position i = 1; 2 kinematical viscosity at position i = 1; 2 relative pressure loss

[-] [Pa] [Pa] [Pa] [Pa] [Pa] [m] [-] [-] [-] [kg/m3] [N/m2] [N/m2] [Ns/m2] [m2/s] [-]

FIGURES
Fig. 1: 2: 3: 4: 5: 6: 7: 8: 9: 10: 11: 12: 13: 14: 15: 16: 17: 18: 19: 20: 21: 22: 23: 24: 25: 26: 27: 28: 29: 30: 31: 32: 33: 34: 35: 36: 37: 38: 39: 40: 41: 42: 43: 44: 45: 46: 47: 48: 49: 50: 51: 52: 53: 54: Title Page

520H Aeroturbine, to install on roof tops, perpendicular to the main wind direction [19] ... 1-3 WTC, conceptual design[20] .. 1-4 WTC, conceptual design[20] .. 1-4 WTC, erection [21] . 1-4 Castle House, conception [22] .... 1-5 Castle House simulation [23] ..... 1-6 Zoom on the field of interest .... 1-7 Geometrical benchmarks, 15th part ... 1-7 Coin, Co1.00 ......... 1-7 Coin contraction inlay; contraction 2.25 ...... 1-8 Theoretical streamline tube of a bare turbine[24] .. 1-8 Principle of the Venturi effect[25] ..... 2-3 Classical Venturi, geometry profile[1] .. 2-5 Schematic Venturi ISO 5167-3 device integration ... 2-6 Venturi nozzle, geometry profile[2] .. 2-7 Laminar and turbulent flow in a constant tube [28] ... 2-9 Static pressure along the tube including the Venturi device[1] 2-12 Drawing of a model wind tunnel, Eiffel type, University of Stuttgart [30] ... 2-13 Curvature of the inlet geometry, given in ISO standard ... 2-14 Thickening boundary layer, to the left, and influence of a slight opening angle, to the right .. 2-14 Straight throat diffuser interface, left, and throat section as gradual diffuser initiation, right . 2-14 Slight expansion near the nozzle exit ... 2-15 Throat subdivisions with constant cross sections . 2-16 Diffusers primary parameters ....... 2-17 Tangential transitions at the inlet ..2-18 Inlet curvature construction .. 2-18 Co1.00 model .. 2-20 Co2.25 inlay ..2-20 Co4.00 inlay . 2-20 a) Co9.00 inlay, left; b) Co1.00 model, middle; c) Co9.00 model, right . 2-20 Graphical user interface, GUI, FIG- GEN module opened ...... 3-1 Venturi device in supporting tube, geometry plot .... 3-4 Computational domain, elongated tube 3-5 FI-POST plot: velocity profiles in a tube at 0m; 2m; 7m; 11,5m; vmean= 30 m/s; D = 0.5 m ... 3-5 Boundary elements ... 3-6 Turbulent flow regions . 3-8 Function graphs of the velocity profiles for the viscous sub layer, transitional region, core flow .. 3-9 FI-POST plot: y+ values along the ENTITY wall, device starts at ARC LENGTH 12.00 3-10 Paved mesh, detail from the inlet region .. 3-10 Significant geometrical deflections .. 3-11 Characteristic meshing regions ..... 3-11 FI-POST plot: Convergence history for the uncoupled quantities, velc=10-5, resc=10-5 ... 3-13 ISO measurement positions vs. FIDAP measurement positions .. 3-15 Superimposed pressure plots of plane 1 and plane 2 3-15 Relative pressure distribution, above: isobars in the device, below: pressure solid patterned . 3-16 Plot of the velocity distribution in the standard Venturi device ... 3-17 Cylindrical velocity profiles at plane1 and plane2 ....3-17 a) Static pressure plot along the wall b) Static pressure plot along the axis .....3-18 Pressure along the device wall with geometrical important points emphasized ...... 3-19 Discrete measurement points and averaging measurement holes .3-19 Diagram of the computational domain, including dimensions, without turbine ...... 4-2 Boundary groups identification, without turbine .. 4-2 Velocity values at discrete points along the radius .. 4-3 Streamline plot; Co1.00 .... 4-4

XI

Fig. 55: 56: 57: 58: 59: 60: 61: 62: 63: 64: 65: 66: 67: 68: 69: 70: 71: 72: 73: 74: 75: 76: 77: 78: 79: 80: 81: 82: 83: 84: 85: 86: 87: 88: 89: 90: 91: 92: 93: 94: 95: 96: 97: 98: 99: 100: 101: 102: 103: 104: 105: 106: 107: 108: 109: 110: 111: 112:

Title

Page

Velocity profile development inside Co1.00 at 0 m, 0.5 m, 1.0 m, 1.5 m, 2.0 m and 2.5 m... 4-4 Relative static pressure plot along the face wall; Co1.00 . 4-5 Shear rate plot along the inner device wall; Co1.00 ..... 4-5 Separating velocity profile [33] .... 4-6 Cyl. velocity distribution plot & velocity profile inside device at 0.0, 2.7, 6.3, 16.36 m; Co1.00 4-6 Relative pressure distribution; Co1.00 . 4-6 Streamline plot; Co2.25 .... 4-7 Cyl. velocity distribution & velocity profile inside device at 0.0, 1.6, 3.4, 5.8 m, 16.3 m; Co2.25 4-7 Relative static pressure plot along the face wall; Co2.25 ..... 4-8 Relative pressure distribution; Co2.25 ..... 4-8 Streamline plot; Co4.00 .... 4-9 Shear rate plot along the diffuser wall; Co4.00 .... 4-9 Stagnation point determines effective inlet ... 4-9 Relative static pressure plot along the inlet wall; Co4.00 4-10 Cyl. velocity distribution & velocity profile inside device at 0.0, 2.3, 3.7, 5.5 m, 16.3 m; Co4.00 4-10 Relative pressure distribution; Co4.00 . 4-10 Streamline plot; Co9.00 4-11 Cyl. velocity distribution & velocity profile inside device at 0.0, 3.1, 4.0, 5.2 m, 16.4 m; Co9.00 4-11 Relative pressure distribution; Co9.00 . 4-11 Aspects influencing the efficiency of contraction and therefore accelerating, but blocking flow ... 4-12 Comparison of the four geometries; blockage, acceleration, energy amplification . 4-12 Sketch: AD implementation ..... 5-2 Diagram of the computational domain, including dimensions, bare turbine ... 5-3 Boundary groups identification, bare turbine ...... 5-3 Velocity and static pressure plot along the symmetrical axis, AD in free field, Betz optimum 5-4 Ainlet is 2.25 times bigger than Arotor for a contraction of 2.25 5-5 Ainlet corresponds to Arotor for a contraction of one .. 5-5 Diagram of the computational domain, including dimensions, device with turbine .... 5-7 Grid plot of the Co1.00 domain; 51742 nodes.. 5-7 Grid plot of the Co 2.25 domain, 46918 nodes .... 5-7 Boundary groups identification, device with turbine .. 5-8 vAD plotted over the corresponding CT for pos1, pos2 and pos3; Co2.25AD 5-8 Static pressure, large picture, and cyl. velocity plot, small picture; Co2.25pos1, CT = 0.44 ... 5-10 Static pressure, large picture, and cyl. velocity plot, small picture; Co2.25pos2, CT = 0.44 ... 5-10 Static pressure, large picture, and cyl. velocity plot, small picture; Co2.25pos3, CT = 0.44 ... 5-10 Static pressure plotted along the axis; structure inlet at 0m; CT = 0.53&0.59: Co2.25pos2&3 ....... 5-11 Static pressure distribution, 3-D plot: pressure shield in front of the device, bare structure 5-11 Pressure shield in front of the device, CT = 0.53: upper pic.: Co2.25pos1; lower pic.: Co2.25pos3 5-12 Cylindrical velocity plot; Co2.25pos2, CT = 0.59 ................... 5-13 Cylindrical velocity plot; Co2.25pos3, CT = 0.59 ....... 5-13 Superposition of cyl. velocity and streamline plot; Co2.25pos3, CT = 0.59 .... 5-13 Cylindrical velocity along axis plot: CT = 0.53, pos2, pos3; CT = 0.59, pos3 .. 5-14 Static pressure distribution, 3-D plot: pressure shield behind the device, bare structure . 5-14 Static pressure field behind device, CT = 0.53: upper pic.: Co2.25pos2; lower pic.: Co2.25pos3 .. 5-15 CP over CT for pos1, pos2, pos3; Co2.25AD ... 5-16 Streamline, AD velocity profile, pressure&cylindrical velocity plot; Co2.25pos1, CT = 0.44 5-16 Streamline, AD velocity profile, pressure&cylindrical velocity plot; Co2.25pos1, CT = 0.53 5-16 Streamline, AD velocity profile, pressure&cylindrical velocity plot; Co2.25pos1, CT = 0.59 5-17 Streamline, AD velocity profile, pressure&cylindrical velocity plot; Co2.25pos1, CT = 0.62 .5-17 Streamline, AD velocity profile, pressure&cylindrical velocity plot; Co2.25pos1, CT = 0.64 5-17 Streamline, AD velocity profile, pressure&cylindrical velocity plot; Co2.25pos1, CT = 0.66 5-17 vAD plotted over the corresponding CT for pos1, pos2 and pos3; Co1.00AD ... 5-18 Cylindrical velocity profile plot at pos1; CT = 0.30, CT = 0.44, CT = 0.53; Co1.00pos1 . 5-18 Streamline plot: CT = 0.30, CT = 0.36, CT = 0.44, CT = 0.53; Co1.00pos1 ... 5-19 CP over CT for pos1, pos2, pos3; Co1.00AD ... 5-19 Pressure distribution , pos1, pos2, pos3, Co1.00AD .....5-20 Venturi tube integrated ..... 6-2 Technical model drawings: first Co2.25; second Co4.00; third Co9.00 .. h

XII

TABLES
Tab. 1: 2: 3: 4: 5: 6: 7: 8: 9: 10: Title Equations to derive the whole geometry, depending on d Geometrical characters for the contractions Co1.00, Co2.25, Co4.00, Co9.00 Bare structure simulation results overview Applied thrust coefficients, referring to the rotor area, for the model Co2.25AD and Co1.00AD Actuator disc mean velocities depending on position and thrust of the actuator disc; Co2.25AD Actuator disc mean velocities depending on position and thrust of the actuator disc; Co1.00AD Power values for the Co2.25pos1 and Co1.00pos1 models Power coefficients and power values for bare turbines same size of encased ones, comparison FI-GEN READ in file to generate the geometry, computational domain and mesh; Co2.25 FI-PREP READ in file to set fluid properties, boundary conditions, implement AD; Co2.25 pos1 Page 2-9 2-19 4-13 5-6 5-9 5-18 5-20 5-21 a g

XIII

XIV

1 INTRODUCTION

Urban landscapes changed due to new architectures. Buildings are built much higher, therefore roofs are reaching in high wind speed layers, planned urban structure is applied, which can be used to direct mainstreams, and consciousness about environmental friendly systems is more and more present, which increases the tolerance and even stimulates desire of private renewable energy. Based and inspired by a skyscraper design, including turbines, integrated into the building structure, meant for the onside power supply, a parameter study about encased turbines in urban areas developed.

1-1

INTRODUCTION

1.1

HISTORY UNTIL NOW

In the seventies US American engineers from Northrop Grumman Corp. discovered within wind tunnel tests the energy harvesting increasing by a factor of approximately six by an augmented wind turbine compared to a common bare wind turbine [5]. Based on that experiments the company Vortec Energy Ltd. located in Auckland, New Zealand, took in 1994 the challenge to build a full scale augmented turbine to savour the mass flow increasing and within that the increasing power harvesting effect. In 1997 the 17 meter high demonstration system Vortec-7, 7.3 meter in rotor diameter with its 22 tons heavy hopper, was installed on a circular rail, to follow the wind. After the set up did not reach the predicted performance, a lot of money was poured into the project to improve the diffuser design until it finally run out of money and the whole chapter Vortec-7 and the future project Vortec-23 were closed. Anyway, the idea of breaking the Betz` power coefficient limit (= 16/27) [6] by using a diffuser or shroud like encasing never disengaged engineers since the 1920s and prevailed in the background until today. So for example a new design feature, a flanged diffuser, was experimentally and numerically examined in Japan [7]. Applying a flange generates a large separation area behind the diffuser and therein a very low-pressure region and appeared to draw more flow through the shroud, compared to a diffuser layout without flange. Their investigations resulted in power coefficient four times higher for a diffuser-shrouded wind turbine compared to the same turbine without the shrouding. Furthermore they came up with the conclusion, that normalising the performance to the velocity just behind the turbine, both the bare and diffuser shrouded wind turbines returned almost the same peak performance. Flow structures behind the blades were generally the same, while completely different trends were seen in the region downstream of the wind turbine. In particular rapid destruction of the vortex structure was observed for the shrouded version, while similar vortex structures in the far downstream region occurred. Almost simultaneously T. Matsushima et al. [8] also came up with a flanged diffuser encased turbine study. First they simulated with the thermo hydrodynamic analysis software I-DEAS the bare structure, without implementing a turbine to find out about the geometrical parameters influence on the maximum velocity in the entrance of the diffuser. Comparisons were made between structures with the same inlet diameter, diffuser opening angle and flange width. The longer the diffuser was the higher were the achieved velocities until a constant value was approached. Another parameter study was handling geometries keeping the length constant, but varying the diffuser angle. In the range from zero to four degrees for the diffuser opening angle a steep increase of velocity amplification was gained, flattening towards an opening angle of six degrees representing the maximum amplification of 1.7. Larger opening angles still created elevated velocities in the diffuser inlet, but in decreasing manner due to separations. Remarkable is that a diffuser opening angle of twelve degrees had the same amplifying effect as an opening angle of three degrees, but of course with the same length a much more space and structure demanding outlet diameter. The last parameter they examined is the flange width with the result that it does not really influence the wind speed amplification as soon as a flange is applied at all. One of the geometries was chosen to encase a five bladed turbine, 0.95 meter in diameter. The combined system was erected on a pylon, free to follow the wind. Next to it was a turbine of the same type but without diffuser encasing installed. It turned out that indeed a higher power performance can be achieved, but highly depending on the stability of the wind direction. When both turbines where free to follow the wind, the total

1-3

INTRODUCTION energy production for the entire day of the shrouded turbine was 1.16 times the production of the bare turbine. As they fixed the turbines in yaw angle the ratio was 1.65. An analytical approach did Gerard J. W. van Bussel [9] in 2007. He found, based on simple momentum theory, that a pressure drop over the rotor disc equal to 8/9 times the local dynamic pressure always states the optimum and the amount of energy extracted per unit of volume with a DAWT (diffuser augmented wind turbine) is the same as for an ordinary bare wind turbine. As long as the power coefficient is referred to the diffuser outlet area, he says, the Betz limit is not surmountable and in practice the performance is even worse, due to viscous effects. A direct impact on the performance has the pressure at the diffuser outlet and the ratio of the velocity in the inlet over the undisturbed wind velocity. In his theoretical approach he did not consider the turbines location, separation and assumes loss free operation. Experiments, analytical definitions, simulations and patents related to encased turbines are plenty. One step further was gone by S. J. Watson et al. [10] with the modelling of the performance of a building-mounted ducted wind turbine. Computational fluid dynamic (CFD) simulation of a cross flow turbine, ducted by a Venturi like shape, mounted on a simplified building, showed that one dimensional theory can not predict the complex interactions of the building. Moreover is the flow around a building unique as its design, its position, its surroundings and its wind situation. On these circumstances focused Malcom A. Heath and John D. Walshe [11] to come out with the conclusion, that the wind is very strongly sheared at rooftop height, isolated small buildings can profit of the wind speed-up seen over the rooftop ridge and in urban surrounding this effect is destroyed by adjacent buildings. They see in general a chance for urban turbines on high buildings and isolated houses. Few ideas of urban turbines, encased [12], with vertical axis [13] and moreover unconventional concepts [14], [15] are realized. Some designs are of the category house hold turbines and are meant to supply the house, with no respect to the integration of the wind turbine into the flow situation around the building itself [16], [17]. Some others are already regarding the upwind on a house faade and are placed on logical places aligned to the main wind direction [18], [19]. Fig. 1 for example shows a drag type turbine meant to be mounted on high riser roof edges. All of them were so far fixed on an already existing building. The potential of considering a turbine already in the conceptual house design pays of in saving the structure for a turbine tower, in increased wind velocity and direction by smart blockage of the building and with this a higher power harvesting and especially the so long controversial encased turbine contra arguments can be cancelled out. No additional heavy structure is necessary, because it can be integrated in the architecture and the wind can be directed towards the turbine, fixed in yaw angle. It is definitely a process of compromises, on the side of the architectural freedom and the side of the technical efficiency.

Fig. 1: 520H Aeroturbine, to install on roof tops, perpendicular to the main wind direction [19]

1-4

INTRODUCTION

1.2

CASTLE HOUSE

---------------------------------THIS CHAPTER IS CONFIDENTIAL--------------------------------Since a few conceptual designs for turbines integrated in architectural structures appeared, one project is in the state of implementation (World Trade Center in Bahrain, see Fig. 2- 4).

Fig. 2 and Fig. 3: WTC, conceptual design [20]

Fig. 4: WTC, erection [21]

The idea of using the high wind velocities at high risers combined with shaping of buildings to even increase and direct the wind, are opening the path into urban areas for wind turbines and offers an on-site energy supply. A new project, Castle House, is planned for the year 2009. The high riser shall be erected in London, surrounded by low buildings (see Fig. 5).

Fig. 5: Castle House, conception [22]

Norwin A/S, Roskilde, Denmark, is supporting the project with the turbines. New aspects like increased safety in dense populated areas, small turbine diameters, short distances to people impacted by noise, have to be considered. Norwin A/S already dealt with the aspects about small turbines in urban areas as the responsible partner in the design and construction of the World Trade Centre in Bahrain. 1-5

INTRODUCTION

1.3

TURBINE POSITIONS

Next to the interest, which contracting curvature for the inlay delivers benefits or drawbacks, lies the interest how the turbine positioning influences the flow behaviour. Fig. 10 shows such an inlay, here for the contraction 2.25. It consists of the contracting, smooth shaped inlet, the constant diameter section called the throat and the diffuser, expanding from the throat diameter back to the inlet diameter. Due to the Venturi effect, explained detailed in chapter two, the highest velocity are expected to occur in the section with the smallest diameter and therefore in the throat. How the curvature of the single contractions are found and motivated follows in chapter 2.3. Gerald J. W. Van Bussel [9] did in his one dimensional analysis not consider the turbine location. Within this simulation here viscous effects are considered and the velocity profile inside the tube is consequently depending on radial and axial coordinates. Three distinctive positions along the axis are chosen to find the most attractive variant in terms of the interaction of the blockage effect of the contracted duct itself and the blockage of the turbine. Position one, pos1, is chosen at the interface of the inlet and throat. Because losses due to viscous effects are not negligible, this position is richest in energy and the relatively long throat following gives the flow the opportunity to develop to a more uniform profile and therefore reduces the risk of separation in the diffuser. Separation in the diffuser causes significant pressure losses. The second position, pos2, is defined in the throat, where a developed velocity profile exists. Because the critical Reynolds number is far exceeded, the velocity profile is a blunt turbulent one with mainly cylindrical velocity components. With an almost constant core velocity, the flow approach towards the turbine is comparable to the situation for a bare turbine in a free field and in consideration of the suggested conventional turbine desirable. To attend the theoretical streamline tube (see Fig. 11) of a bare turbine, position three, pos3, was found.

Fig. 10: Coin contraction inlay; contraction 2.25

By decelerating the flow, the stream tube enlarges in diameter. Of course, the situation for an encased turbine is different by all means, but the velocity is decelerated by the turbine as well and the diffuser allows or supports, respectively, the expansion and so contributes a smooth flow through the device. Another idea, but not accomplished within this work, is the positioning in half the depth of the diffuser, to include the whole stream tube shown in Fig. 11 and use the inlet and throat as kind of wind accelerator, supporting the stream tube system. The contra argument is the bigger diameter and the therewith linked higher financial costs.

Fig. 11: Theoretical streamline tube of a bare turbine [24]

1-9

2 VENTURI

Nowadays the Venturi tube is used in a wide field of engineering applications (jet fuel injection, flow rate measurements, vaccination). With its simplicity and no moving parts it is neither cost nor maintenance intense, but with a great effect.

2-1

VENTURI

2.1

EFFECT

The Venturi effect is an example of Bernoulli's principle, in the case of fluid flow through a tube or pipe with a constriction in it. The fluid velocity must increase through the constriction to satisfy the equation of continuity, while its pressure must decrease due to conservation of energy: the gain in kinetic energy is supplied by a drop in pressure or a pressure gradient force. The effect is named after Giovanni Battista Venturi.
>>Giovanni Battista Venturi (1746 - 1822) was an Italian physicist. He was the discoverer and eponym of Venturi effect. He was also the eponym of the Venturi pump and Venturi tube. Born in Bibbiano, he was a pupil of Lazzaro Spallanzani. He was ordained as a priest in 1769, and in the same year appointed as a teacher of logic at the seminary of Reggio Emilia. In 1774 he is professor of geometry and philosophy at the University of Modena, where in 1776 he became professor of physics. He was a contemporary of Leonhard Euler and Daniel Bernoulli. Pupil of Lazzaro Spallanzani.<< [25]

With the equation of continuity in form of

i * Ai * vi

const.

(1)

while the density is constant for an incompressible flow, Ai is a cross section with vi as its mean velocity, and the Bernoulli equation without losses
pstat _ i + * * vi2 + * g * hi
dyn. pressure hyd. pressure

const.

(2)

with the hydrodynamic pressure equal zero, due to the horizontal axis and the low density of air, and pstat as the static pressure, the static pressure difference, pstat, between two cross sections is obtain as

pstat =

pstat _ 2 pstat _1 =

A 2 * * v2 * 2 1 A 1
2

(3)

As soon as A2/A1 is smaller than one the contraction leads to a decrease in pressure but increase in velocity. Equation (1) sets the following relation between two cross sections

v2

A v1 * 2 A1

(4)

A2/A1 bigger than one means an expansion and results in decreasing velocity coupled to increase in static pressure. Fig. 12 illustrates the velocity situation and streamlines in a Venturi shaped tube.

2-2

VENTURI If a flow is first contracted and than expanded back to the same diameter by a surrounding geometry, assuming friction losses, it is important to mention, that the mean velocity of the outlet will be the same as of the inlet due to the conservation of mass law. The losses are only reflected in the diminishing static pressure.

Fig. 12: Principle of the Venturi effect [26]

The Venturi effect is chosen to be utilised because with the mass flow
dm / dt =

*A *v
i i

const.

the power in a cross section is

Pi

1 dm * * vi2 2 dt

1 * i * Ai * vi3 2

(5)

and so proportional to the velocity cubed.

A1 A2 Co 2

For a contraction for example of

Co P2 P 1

4.0

(6)

the power ratio is then

16.0

(7)

2-3

VENTURI

2.2

VENTURI ISO STANDARD

Computational fluid dynamic simulations demand an immense amount of parameter settings, to tune the model as realistic as possible. Every single problem needs its own unique set of parameter, just to mention a few: turbulence modelling, solution approach and mesh strategy. Anyway, in case of a similar problem, the similarity is also reflected in the set of parameters. A European standard for Venturi shaped flow measurement devices exists. Detailed geometrical descriptions with corresponding pressure distributions, gained through numerous experiments are therefore available. The standard constitutes a similar problem with empirical data. It provides so the possibility to rebuild the geometry within FIDAP and find the right parameter settings. In chapter 2.2.1 is the ISO 5167-1:2003 standard with its general restrictions, limits and basic parameters introduced. Among the ISO 5167-3:2003 nozzles is then one concrete geometry chosen and the relevant pressure values are ascertained.

2.2.1

ISO 5167-1:2003

ISO 5167-1:2003 defines terms and symbols and establishes the general principles for methods of measurement and computation of the flow rate of fluid flowing in a conduit by means of pressure differential devices (orifice plates, nozzles and Venturi tubes) when they are inserted into a circular cross-section conduit running full. It also specifies the general requirements for methods of measurement, installation and determination of the uncertainty of the measurement of flow rate. It also defines the general specified limits of pipe size and Reynolds number for which these pressure differential devices are to be used. ISO 5167 (all parts) is applicable only to flow that remains subsonic throughout the measuring section and where the fluid can be considered as single-phase. It is not applicable to the measurement of pulsating flow.
flow conditions which have to be fulfilled are summarized as:

the flow has to be subsonic all over. the flow has to be physical and thermal homogeneous and consists of only one phase. the flow rate has to be constant in time or is changing in small increments in a very slow manner. The flow has to be not pulsating. the incoming flow has to be without spin. At least the angle of spin needs to be lower than 2. the incoming flow has to be completely developed one diameter before the inlet. At least the local axial over the maximum axial velocity ratios all over a cross section have to be in a 5% range of the corresponding ideal ratio. the pressure ratio has to be 0.75 . 2-4

VENTURI The relevant parameter definitions in the standard ISO 5167-1 and the equations derived particularly out of experiments, but based on theoretically assumptions, are briefly discussed below. All parameters and equations are depending on the inlet diameter of the device, D, and the throat diameter, d. See also Fig. 13.

1: conical diffuser 2: measurement point Q 3: inflow cone

4: measurement point J 5: interface plane b: flow direction

Fig. 13: Classical Venturi, geometry profile [1]

diameter ratio :

with

d D

(8)

d D

: throat diameter : inlet diameter

Reynolds number ReD, relating to D:

Dimensionless parameter, giving the inert over friction force ratio.

ReD with

( v1 * D )

( 4* dm / dt ) ( * 1 * D )

(9)

v1 : axial mean velocity in D 1 : kinematical viscosity dm/dt : mass flow 1 : dynamical viscosity

Reynolds number Red, relating to d:

Red

ReD

(10)

2-5

VENTURI
flow rate coefficient C:

Parameter to implicate the actual to the ideal, frictionless (Bernoulli) flow rate. C = dm / dt * 1- 4 (11)

* d * 2 * p * 1
2

with

p 1

: differential static pressure, measured at J and Q : density in D

expansion number :

Parameter, considering the fluid compressibility.

dm / dt * 1- 4 * d * C * 2* p * 1
2

(12)

The expansion number is equal one for incompressible fluids (liquids). In case of compressible fluid (gas) it is smaller than one. With the help of experiments the independence of the Reynolds number is validated. For given diameter ratio the expansion number is only dependent on the static pressure difference between pJ and pQ and the adiabatic exponent. The values of the expansion number for Venturis in ISO 5167 are based on general thermo dynamical equations for adiabatic expansions.
static pressure ratio :

Static pressures measured at the positions (J&Q) defined in the corresponding standards and shown later on.

pQ pJ

(13)

Because the Venturi device is meant to be integrated in a house structure, without a supporting pipe but a smooth blend over from the inlet to the faade (see Fig. 14), the ISO 5167-3 Venturi nozzle is the most reasonable geometry to start out with.
Fig. 14: Schematic Venturi ISO 5167-3 device integration

2-6

VENTURI

2.2.2

ISO 5167-3:2003

Additionally to the general restrictions, given in the ISO 5167-1 sheets and summarized one chapter earlier, a ISO 5167-3 Venturi nozzle geometry (see Fig. 15) has to comply with the following limitations: 0.065 m 0.050 m 0.316 D d 1.5 * 105 0.775 30 2 * 106 0.3 6.125 * 10-2 m
Fig. 15: Venturi nozzle, geometry profile [2]

0.500 m

ReD Ma

Surface roughness

The geometrical parameters for the model, Dm and dm, are chosen to approach the final geometry as close as possible and the diameter ratio is chosen as small as possible to gain the highest possible acceleration. Dm dm = = 0.500 m 0.375 m

0.750

Consequently the geometryoriginal : geometrymodel scale is 18. In wind tunnel tests are also mostly small scale models used. To ensure the same aerodynamic behaviour for the original, certain equality numbers have to be kept. One of the most important equality numbers is the Reynolds number. Here the small scale model is used to be transferred with the same dimensions in a simulation environment. Afterwards though an up scaled Venturi geometry with Do = 9 m, vo_mean 6 m/s and = 1.50 * 10-5 m2/s is simulated with ReDo = vo _ mean * Do = 3.6 * 106

Because this Reynolds number exceeds the range of the standard and therefore can not be kept, the assurance of equal flow behaviour is not given.

2-7

VENTURI At least the limit of the Reynolds number, given in the standard, should be approached with mean velocity in Dm: fluid density: dynamical viscosity: kinematical viscosity: ReD m = vm_mean = = = = vm _ mean * Dm 30.0 m/s 1.225 kg/m3 1.8375 * 105 Ns/m2 1.50 * 10-5 m2/s = 1* 106 .

The velocity is set to 30 m/s to keep the problem incompressible and with it less complex. Flowing air is considered as incompressible medium as long as the Mach number does not exceed 0.3. Defining the Mach number as Ma = v a a vincompressible = 343 m/s 102.90 m/s. (14)

with the sound propagation velocity is the highest maximum local velocity

And flows in tubes are considered as laminar until a critical Reynolds number, referring to the diameter, of around 2 500 is reached. With a Reynolds number of 1 * 106 the flow is by all means a turbulent one. In [27] the complex turbulent velocity profile is simplified to vr vmax = r n 1 R
1

(15)

while vr is the local velocity, vmax is the maximal velocity and emerges along the axis in a developed velocity profile (see Fig. 16), R stands for the tubes radius and r gives the position in radial direction. At this n is a dimensionless parameter, depending on the Reynolds number.
n = 1.7 + 1.8 * log ( Red )

(16)

And with a result of n = 8.9 for the throat, the velocity profile is even for a turbulent profile very blunt with a maximal velocity of around 55 m/s.

Fig. 16: Laminar and turbulent flow in a constant tube [28]

2-8

VENTURI
equations and coefficients within the ISO standard:

GEOMETRY: All geometrical parameters (see Fig. 15) are depending on the throat diameter, d, and are listed with their definitions in Tab. 1:
Tab. 1: Equations to derive the whole geometry, depending on d

parameter Inlet radius R1 centre x-coord. R1 centre y-coord. trans. radius R2 centre x-coord. R2 centre y-coord. throat length 1 throat length 2 point J point Q

symbol

equation

value

R1 xR1 yR1 R2 xR 2
yR 2
E F xJ

= = = = = =
= = =

0.2* d 0.006* d 0.2 * d 0.75 * d d 0.01 * d 3 12 + 39 d * 60 d d + 2 3 0.3 * d 0.4* d ~ 0.45* d inlet

0.3041* d + 0.3* d

= = = = = =
= = =

0.075 m 0.075 m 0.28125 m

0.125 m 0.11403 m 0.3125 m
0.11225 m 0.15 m 0m

(17) (18) (19)

(20) (21) (22) (23) (24) (25)

xQ

0.22654 m

(26)

In the first passage of the ISO 5167-3 standard the ISA-1932 nozzle is described. The Venturi nozzle is an assemblage of such a nozzle, an additional throat section and a diffuser. Due to the diffuser and the smooth expansion thereby, the losses are reduced remarkable, compared to the bare ISA-1932 nozzle. For the length of this ISA device, the following equation is to be used
LISA = d * 0.4041 + 0.75 0.25 - 0.5225 = 0.2197 m (27)

Although the opening angle of the diffuser could be 30, it is set for the model to = 15. The bigger the angle the shorter is the diffuser, which could be attractive, especially in integrative matters. On the other hand, the smaller the geometrical change, the smaller are the pressure losses, which could be attractive in energy harvesting purpose. With this angle and the assumption of widening the outlet diameter back to the entrance diameter, D, the length of the diffuser results in Ldiffuser = D/2 - d /2 tan ( / 2) = 0.4747 m (28)

Altogether the length of the Venturi device is then LVenturi = Ldiffuser + F + LISA = 0.8444 m (29)

2-9

VENTURI PRESSURE LOSSES: Besides the equations given in the ISO 5167-1, the chapter about Venturi nozzles specifies equations for this special application. The static pressure difference between J and Q is ascertainable and delivers the value, which serves as the analogy between the flow problem in reality and simulation.
flow rate coefficient C:

0.9858 - 0.196 * 4.5

0.932093

(30)

expansion number :

for

0.75

(31)

* 2 / 1 - 4 1 - ( 1)/ * * -1 1 - 4 * 2/ 1-

While the static pressure ratio

pQ pJ

is unknown.

For air 20C tempered, the following value for the adiabatic coefficient, , is stated = 1.4. Known is the constant mass flow

dm / dt with

*A*v
D * 2
2

7.2159 kg/s

A1

0.19635 m 2

By rearranging equation (12), the mass flow is also defined as

dm / dt while p = = pJ pQ C 1
4

**

d 2 2 p 1

(32)

is unknown.

Equations (31) and (32) are leading to an implicit system of equations. With the help of trial and error iterations an expansion number with corresponding differential pressure is found. pJ
p

= = =
=

101 325 Pa 0.98615 0.98805209

1 403.35 Pa

2-10

pQ

9 9921.65 Pa

VENTURI As long as the diffuser angle is smaller than or equal to 15, a relative pressure loss, Venturi device losses over the differential pressure in the device, between 5 to 20 percent is acceptable. Therefore is the equation for the relative pressure loss

p p p

0.05 0.2 , (33)

while p represents the loss due to friction in a tube of the same length, but without velocity amplifier, and p on the contrary represents the pressure loss, including the device. Within turbulent flow the pressure loss along a uniform pipe is prominent, compared to the losses in a laminar pipe flow. The velocity gradient at the tube wall is much larger for turbulent flow than for laminar flow. This change in velocity profile causes the wall shear stress to increase sharply, with the same effect on the friction factor.

* f turbulent *

v 2 L * mean 2 D

(34)

With drawn tubing and therefore a roughness of 0.0015 mm the Moody diagram gives a turbulent friction factor of fturbulent = 0.0095. As in Fig. 17 to be seen, the start reference point lies one D in front of the Venturi inlet and the second reference point lies at least six D behind the Venturi outlet. Adding this two demanded lengths to the length of the device, the analogous pipe length is

Lanal

Ldiffuser + 7 * D pturbulent

= =

4.3444 m 45.5027 Pa

(35)

and leads to

The pressure loss of the tube, including the device, over the same length is

p * + pturbulent

= 115.67 Pa 326.17 Pa .

Due to the device a pressure loss will occur about c =

p - p

= 70.17 Pa ~ 280.67 Pa .

Fig. 17: Static pressure along the tube including the Venturi device [1]

2-11

VENTURI

2.3

APPLIED GEOMETRY

The Venturi standard helps to validate the simulation setting for a constricted tube. It hazards the guess, that the geometry, given in the standard, for a Venturi is most investigated and developed over the years to be optimized in matters of losses, due to friction, but also separation. To use the standard geometry would be recommendable, as long as it is meant to accelerate a fluid. But as soon as an obstacle is implemented in the throat, the application is changed and the standard geometry does not constitute an optimum anymore. The configuration of an Eiffel wind tunnel with closed model section is redolent of a Venturi tube but with an obstacle in the throat (see Fig. 18). Instead of the natural wind, a fan is generating the air flow through the tunnel. Wind tunnel tests demand also low turbulence intensity in the flow and losses might be reduced. Therefore based on wind tunnel designs [29] certain laws for the Venturi geometry meant to encase a turbine are derived, as
1 2 3 4 high curvature should take place in big diameter a constant area duct before the rotor section should be employed the throat length behind the rotor should be sufficient long diffusers opening angle has to be sufficient small

and are discussed in the following chapter. It turns out that the ISO standard fulfils the conditions. How the geometry parameters are finally chosen is disclosed in chapter 2.3.2.

Fig. 18: Drawing of a model wind tunnel, Eiffel type, University of Stuttgart [30]

2-12

VENTURI

2.3.1

DESIGN LAWS

The Venturi design in order to be integrated in a building to direct and accelerate the wind toward a rotor, placed inside the Venturi, is mostly alike an open wind tunnel. Disadvantages compared to a closed wind tunnel and therein also facts to consider for an encased turbine are extensive screening at the inlet to get high-quality flow weather (humidity, temperature) affects the operation the wind through the Venturi is blown out, without being used throughout in general opened tunnels tend to be noisy, causing environmental problems, limit hours of operations, and/ or require extensive noise treatment of the tunnel and surrounding room

In each of the sections it is commonly said that a loss of energy occurs. Actually there is an energy transformation from mechanical form to heat, that results in raising the temperature of the flowing gas and the solid material with which it is in contact. The energy transformation comes about due to the viscous actions between the flowing gas and the solid boundaries. With respect to minimize losses due to friction, implicit minimizing the all over length, the nozzle, throat and diffuser length should be kept short as possible, but without incurring flow separation, resulting in significant higher losses. Due to the law of continuity for an incompressible fluid, the loss will not be reflected in the dynamic pressure, but in the static pressure, while the loss coefficient is defined on the basis of total pressure and dynamic pressure. Hence power losses are proportional to power two of the velocity in the cross section. This conclusion leads to the first design law, concerning the inlet.
inlet:

Curvatures of the structure and therefore deflections of the flow are linked to losses, while the losses are proportional to the static pressure but to the squared velocity.
1 high curvature should take place in big diameter

In the ISO standard geometry the inlet is subdivided in two circle arcs with different radius (R1 and R2).

R1

< R2

The smaller radius is placed at the inlet mouth and is passing over tangential to the bigger radius arc, meaning a high curvature by the small radius, hence in the section with lower velocities and a lower curvature in the section, leading to the maximum velocity (Fig. 19).

v1

<

v2

Fig. 19: Curvature of the inlet geometry, given in ISO standard

2-13

VENTURI
throat:

The following throat, or in wind tunnel designs the according test section, is a tube with usually constant cross section. In wind tunnel design matters different cross section shapes have been used over the years, including round, elliptical, squared, rectangular, hexagonal, octagonal, rectangular cross sections with filleted corners, flat ceiling and floor with half round ends, corresponding to the test models and utilities for measurements. In the case of placing a rotor, a cylindrical cross section is the most obvious shape. While the boundary layer thickens through the throat, the core velocity increases and the effective area of the jet decreases. To keep the flow profile constant a slight increase of the geometrical diameter along the axis could be designed (see Fig. 20), while the required opening angles are depending on the wind velocity.

Fig. 20: Thickening boundary layer, to the left, and influence of a slight opening angle, to the right

On one hand it could be nice to start the diffuser within this design and therefore smooth the geometrical transition from throat to diffuser and/or enable a bigger diffuser opening angle (see Fig. 21). On the other hand it efforts higher constructional complexity and therefore higher financial costs.

D/2 d/2 d/2

D/2

Fig. 21: Straight throat diffuser interface, left, and throat section as gradual diffuser initiation, right

Borger [29] recommended a slight expansion near the nozzle exit (see Fig. 22) to improve the exit flow uniformity. Indeed, resulting velocity profiles for a geometry without this expansion show significant suction peaks close to the wall, but which can be advantageous as mentioned later.

Fig. 22: Slight expansion near the nozzle exit

2-14

VENTURI It is in general desirable to reduce the length of the whole Venturi device, thus an idea is to skip the whole throat section and join the nozzle directly to the diffuser (see Fig. 23). Also, the power losses in the throat are sizable. Power can be saved by keeping it short. But design law number two and three tell us: Contractions do not deliver a uniform velocity distribution to the beginning of the throat, so
2 a constant area duct before the rotor section should be employed

Behind the rotor the flow might be separated and all separated flow zones shall be close before the beginning of the diffuser, therefore
3 the throat length behind the rotor should be sufficient long

the ISO standard describes a throat, subdivided into two constant area duct sections. The interface of these two sections is the position where one of the reference pressures is measured, according to the standard. This position is chosen for measurement, because the flow here is suspected to be uniform again.

rotor non- uniform separation

rotor
Fig. 23: Throat subdivisions with constant cross sections

diffuser:

Since the power losses at any point in the tunnel are expected to vary as the speed cubed, the purpose of the diffuser is to reduce the speed with as little energy loss as possible. Minimum energy loss corresponds to maximum pressure recovery. It is generally desirable to reduce the speed in the shortest possible distance without incurring flow separation. Diffusers are sensitive to design errors that may cause either intermittent separation or steady separation. Such separations can be hard to localize but can cause vibrations, oscillating rotor loads and increased losses in the tunnel downstream of their origin. 2-15

VENTURI The primary parameters of a diffuser (see Fig. 24) are the diffuser opening angle and the area ratio (contraction, Co).

/2

D/2d /2 arctan Ldiffuser

(36)

Diffusers are common elements in fluid flow devices, and in many applications the equivalent opening angle can be quite large. The principal constraint on the angle is that it is sufficient small so that the turbulent boundary layer does not separate.
4 diffusers opening angle has to be sufficient small

Relatively large angles are possible for the case of uniform flow entering the upstream end of the diffuser. The pressure recovery and pressure gradients and therefore the risk of separation are dependent on both, the opening angle and the area ratio. Thick boundary layers at the diffuser entrance as well as the use of a very long diffuser to obtain a large contraction ratio will increase the risk of separation. Space limitations for the Venturi or/ and the cost of construction may dictate that a slightly larger divergence be employed to an increase in cost of operation. A diffuser is classed as wide angle when its cross section increases so quickly with axial distance that separation can be avoid only by boundary layer control (suction, blowing, insertion of fine mesh screens). These possibilities are not traced within this basic work. With additional moving parts, the error rate of the system would increase sharply. Because of the throat segment behind the rotor plane, the separated flow zones are expected to be closed again. Therefore a bigger opening angle is applicable. In the standard the diffuser opening angle is limited to max. = 30 .

/2 vthroat d/2 vout

D/2

Ldiffuser

Fig. 24: Diffusers primary parameters

2-16

VENTURI

2.3.2

PARAMETER MODELS

In chapter 1.2 the coin model is introduced, which constitutes the constant dimensions. Orientated on the architectural design, the inner diameter, D, and the outer diameter, two times b, are determined and kept constant. The length is the third constant parameter and has to be decided (D = 9.00 m; b = 6.00 m; length = const.). Consistence of the ISO standard geometry with the design laws discussed earlier advices the use of the standardized Venturi nozzle as backing. Four different contractions are chosen Co = 1.00; Co = 2.25; Co = 4.00 and Co = 9.00. The contraction gives the ratio of the inlet over throat area and can be expressed as Co = 1

(37)

A contraction of one is a bare tube with no constriction in it and therefore represents the bare coin the first contraction. The three remaining contractions are inserted into the coin model as inlays, while the coin face passes over tangentially to the inlay inlet (see Fig. 25).
f a c e tangential transition
wind

COIN INLAY

Fig. 25: Tangential transitions at the inlet

Based on the way the standard derives the inlet curvature for a contraction of 2.25, an up scale is constructed for the contraction of 4.00 and 9.00. How the curvature is constructed is shown in Fig. 26 and explained in detail in the next few lines.

R1 R* M2 lc o M1 O l1 s

D/2

y x Fig. 26: Inlet curvature construction

l2

d/2

2-17

VENTURI Starting out with parameters fixed by the contraction, D and d, the coordinates of the circle centre points are found in the following way: Two parallel lines, l1 and l2, starting at x = 0 and y1 = D/2 and y2 = d/2, respectively, with the distance s are drawn. s = D d 2 2 (38)

The radius R1 for the first section of the inlet has due to the first design law to be the smaller one. R1 is smaller than s and R2 is bigger than s. With the requisite of a tangential transition from the face to the inlet, the circle with the radius R1 has clear defined centre point coordinates M1[R1; D/2], while the second circle has with the requisite of a tangential transition from the inlet to the throat a fixed y coordinate. The x value of the centre point has to be identified M2[xR2; d/2+R2] to exchange the inlet curvature tangentially from R1 into R2. Another parameter, o, has to be declared, which is set to 0.1 m for all of the three geometries. R1 R2 = = s o s + o (39) (40)

Additionally a construction circle around M1, with a radius of R* = R2 R1 = 2 *o (41)

and a parallel construction line, lc, at yc = D/2 + o are needed. There are two intersection points of the construction circle and the construction line. The intersection point with the bigger x value is the centre point of R2. Circle one and circle two have got one osculation point, lying on the extended connecting line of M2 to M1. The equation for the x value of M2 is therefore
xR 2

R1 + o *

(42)

and the coordinates of the osculation point are xo yo = = R1 * (1 cos 30) D 1 * R1 2 2 (43) (44)

To overcome higher contraction, a bigger inlet radius has to be used and therefore the higher the contraction, the longer is the inlet section. Because higher Reynolds numbers are connected to faster developing velocity profiles, the throat length dependency on the throat diameter given in the standard is reasonable and directly applied. The higher the contraction, the more is the air accelerated and results in higher velocities in the throat and hence shorter throat sections are necessary. In order to keep the diffuser opening angle in a useful range and stick to the limitation given in the ISO standard, the maximum angle applied is 30 degrees and therefore results a length of 16.30 meter for the over all device. Corresponding to the constant kept length for all four shapes, the diffuser lengths and angles respectively are concluded. In Tab. 2 the geometrical characters are summarized. 2-18

VENTURI
Tab. 2: Geometrical characters for the contractions Co1.00, Co2.25, Co4.00, Co9.00

contraction, Co diameter ratio, throat diameter, d inlet length throat length diffuser length half diffuser angle, /2

[-] [-] [m] [m] [m] [m] []

1.00 1 9.00 0.00

2.25 6.00 1.57 4.20 10.28 8.21

4.00 4.50 2.32 3.15 10.87 12.04

9.00 3.00 3.07 2.10 11.20 15.00

The three coin inlay geometries are shown in Fig. 28 to Fig. 30. Fig. 30 a) - c) shows how the contraction nine inlay is inserted in the coin to obtain the model Co9.00. Model Co4.00 and Co2.25 are obtained accordingly. The technical drawings are attached in the appendix.

Fig. 27: Co1.00 model

Fig. 28: Co2.25 inlay

Fig. 29: Co4.00 inlay

Fig. 30: a) Co9.00 inlay, left;

b) Co1.00 model, middle;

c) Co9.00 model, right

2-19

VENTURI

2-20

3 FIDAP

FIDAP is a single integrated environment for the simulation of fluid flow problems. The program can be regarded as an integrated set of components and program modules designed to perform all aspects of the model generation, problem setup, post processing and solution phases of a flow analysis. FIDAP can be executed with or without a graphical user interface (GUI, see Fig. 31). The program also provides extensive capabilities for accessing operating system functions such as editing and printing files, as well as launching background jobs. All these features and capabilities can be accessed directly from FIDAP.

Fig.

31:

Graphical

user

interface,

GUI,

FIG-GEN

module

opened

3-1

FIDAP A brief description of each FIDAP program module and its function follows. FI-GEN: Mesh Generation Module

Interactive geometry-based mesh generator Automatic generation of meshes for 2-D or 3-D surfaces Mapped mesh generation for complex 3-D geometries

FI-BC: Initial and Boundary Condition Processor Module Interactive graphical specification of initial and boundary conditions Graphical creation of boundary element groups Problem Data Specification

FI-PREP:

Specification of class of problem and equations to be solved Specification of solution procedures and other control information Specification of fluid and material properties Specification of initial and boundary conditions Mesh Generation Module

FI-MESH:

Mesh generation based on an I,J,K integer space Specification of initial and boundary conditions Postprocessor Module

FI-POST:

Graphical post processing of computed results Computation and graphical display of derived variables and quantities Computational Simulation

FI-SOLV:

Transformation of governing partial differential equations into algebraic equations Solution of algebraic equations and time-integration

To validate the right approach in matters of grid definition, boundary conditions, turbulence modelling and usage of the so called near-wall function, the ISO 5167-3 standard geometry is constructed within FIDAP and the computed results are compared with the values derived from the equations, based on elementary fluid dynamical laws and adjusted by empirical parameters from experiments, given in the standard introduced in chapter 2.2. In the next chapter the modelling of the standard Venturi tube is explained throughout to give a more detailed description of the software and its theory.

3-2

FIDAP

3.1

INTRODUCTION AND VALIDATION

Turbulent flow is a highly complex phenomenon. Although researchers have studied the phenomenon for many years, it is not yet possible to characterize turbulence from a purely theoretical standpoint. Many important characteristics of turbulence are well known, however, including the following: Turbulence is time-dependent, three-dimensional, and highly non-linear. Fully-developed turbulent motion is characterized by entangled eddies of various sizes. The largest eddies arise from hydrodynamic instabilities in the mean flow fieldfor example, shearing between a flowing stream and a solid boundary or unstable stratification produced by heating the fluid from below. The largest eddies break down into smaller eddies which, in turn, break down into even smaller eddies. This process of eddy break-down transfers kinetic energy from the mean flow to progressively smaller scales of motion. At the smallest scales of turbulent motion, the kinetic energy is converted to heat by means of viscous dissipation. The dynamic and geometrical properties of the largest eddies are closely related to the corresponding properties of the mean flow field. For example, large, unstable vortices that form on the perimeter of a turbulent jet tend to possess well defined toroidal structures. The time and length scales of the smallest turbulent eddies are many orders of magnitude greater than the time scales and free paths of molecular motion. As a result, the processes of viscous dissipation are statistically independent of molecular motion. Turbulent motion is not a random phenomenon. As a consequence, turbulent fields possess definite spatial and temporal structures.

To create a usable numerical model of a turbulent flow field, it is necessary to describe turbulent motion in terms of averaged quantities. Although it is theoretically possible to directly apply the conservation equations to the entire flow field, it is unreasonably difficult to do so in practice. As noted above, turbulent flows involve entangled eddies, which encompass a wide range of length scales. To resolve an entire turbulent flow field by direct application of the conservation equations, it is necessary to employ a computational mesh, which elements sizes are smaller than the smallest eddies. Such meshes are extremely dense and - because computational cost increases dramatically with mesh density - result in models that are prohibitively expensive to solve. By contrast, models that are based on averaged quantities characterize turbulent flows using meshes of reasonable density; therefore, they result in reasonable computational times and costs. To model turbulent flow, FIDAP employs a statistical approach, first suggested by Osborne Reynolds [31], that averages the flow equations over a time scale that is long compared to that of the turbulent motion.

3-3

FIDAP REQUIRED DATA:

O O O O O O O DIAGRAM OF THE FLOW DOMAIN, INCLUDING DIMENSIONS BOUNDARY CONDITIONS MESH STRATEGY FLUID PROPERTIES INITIAL CONDITIONS EQUATIONS TO SOLVE SOLUTION APPROACH

3.1.1
O DIAGRAM OF THE FLOW DOMAIN, INCLUDING DIMENSIONS

FI-GEN

The geometry of the standard (see Fig. 32) is modelled in the module FI-GEN. It is possible to work entirely with the graphical user interface (GUI). The history will be saved in different data files, corresponding to the modules. Detailed information about the various data files are given in [32].

Fig. 32: Venturi device in supporting tube, geometry plot

To keep an overview and to change parameters quick and especially more comfortable, it is highly recommended to write a READ in file. However, to create READ files, the graphical user interface is very helpful and for the first steps using FIDAP a good start. Every now and then parts of READ files will be listed with the explanation below to give an idea how FIDAP READ files can be operated. Complete versions of the most important input files are attached in the appendix. It is important to mention that if the mesh in FI-MESH is being created for an AXISYMMETRIC or CYLINDRICAL analysis in FIDAP (refer to the FIPREP PROBLEM command), then the x axis in FI-MESH must correspond to the z direction in the physical domain, the y axis to the radial direction and the z direction to the azimuthal direction in the physical domain. This is quite independent of the use of CYLINDRICAL or SPHERICAL coordinates in the FIMESH mesh generation phase. These options are provided for convenience in defining points in geometric space and have no relation to the AXISYMMETRIC or CYLINDRICAL options in FIDAP. 3-4

FIDAP One of the demands in the standard is a fully developed velocity profile, one diameter in front of the inlet of the Venturi device. Because of the high Reynolds number it has to be a fully developed turbulent velocity profile. The modelling of the wall near region of a turbulent profile is complex and difficult to model. To circumvent this problem, a rectangular velocity profile is set on the entrance of the tube, which is elongated (see Fig. 33) to allow the development of the profile until one diameter in front of the Venturi device.

z
Fig. 33: Computational domain, elongated tube

The tube supporting the device starts 12 m in front of the Venturi device inlet, resulting in a velocity profile development shown in Fig. 34.

Fig. 34: FI-POST plot: velocity profiles in a tube at 0m; 2m; 7m; 11,5m; vmean= 30 m/s; D = 0.5 m

3-5

FIDAP
O BOUNDARY CONDITIONS

Before going on with meshing the computational region, the boundary conditions have to be clarified. Assuming that the problem can be handled as an axis-symmetrically problem, four different boundaries are found (see Fig. 35).

wall

wall wall wall

wall

wall o u t l e t

i n l e t

symmetry axis

Fig. 35: Boundary elements

The inlet with following conditions:

ENTITY( PLOT, NAME = "infl" ) BCNODE( URC, ZERO, ENTI = "infl" ) BCNODE( UZC, CONS = 30, ENTI = "infl" ) BCNODE( KINE, CONS = 0.003, ENTI = "infl" ) BCNODE( DISS, CONS = 0.00045, ENTI = "infl" ) URC is the abbreviation for the velocity components in radial direction. In the inlet, ENTITY labelled infl, the flow is supposed not to be skewed. Accordingly URC is set to ZERO. Correspondingly UZC represents the velocity components in the cylindrical direction. As mentioned earlier, a rectangular velocity profile is set on the inlet, developing into a fully developed turbulent profile in the elongated tube to serve the demands. The mean velocity is set to 30 m/s, resulting in a mass flux of 7.2158 kg/s. The values for the turbulent kinetic energy and the viscous dissipation are set to k = 0.003 and = 0.00045. NOTE: It is important to mention at this point, that it is not possible to set a pressure and velocity condition on the boundary. If so, the equation system would be over-determined and the computational results would fail.
The outlet with following conditions:

ENTITY( PLOT, NAME = "outfl" ) There are no restrictions to the outlet. The flow is free to develop. 3-6

FIDAP
The axis with following conditions:

ENTITY( PLOT, NAME = "axis" ) BCNODE( URC, ZERO, ENTI = "axis" ) The axis- ENTITY is the symmetry axis for the problem. Because there are no sinks or sources along the axis, the radial velocity components, URC, need to be ZERO.
The wall with following conditions:

ENTITY( WALL, NAME = "wall" ) BCNODE( VELO, ZERO, ENTI = "wall" ) To declare the wall- ENTITY as WALL, instead of PLOT, imposes certain conditions, implemented by FIDAP. By applying the WALL ENTITY command, use is made of universal near-wall flow profiles to construct the shape function for the special near-wall elements. The general characteristics of the flow profiles in the near-wall region follow from a consideration of the mean flow equations in these regions and from the application of some simplifying assumptions to these equations. If the following assumptions are made near the wall, the flow is predominantly parallel to the wall and in local equilibrium, there is no transpiration (injection and/or suction) at the wall and the effects of body forces are small,

then the shear stress , the heat flux, q, and the mass flux, dm/dt, across the flow are very nearly constant and equal to the corresponding values of these quantities at the wall (that is, w, qw , and dm/dtw, respectively). Given the above assumptions, the mean flow equations can be manipulated to show that the dimensionless velocity becomes universal functions of y+, the characteristic non-dimensioned distance from the wall. That is, u+ = u uw u* (45) y+ =

* u* *

(46)

with the so called friction velocity u* =

w ,

(47)

where is the normal distance from the wall. The wall stress, w, in a pipe, with respect to the common 1/7- power profile in a pipe, can be derived with

0.0233 * * vi 2 * v *

1/ 4

(48)

3-7

FIDAP The near-wall region is subdivided in three sections (see Fig. 36), the fully turbulent region or core flow, representing a high Reynolds number flow, characterized by the occurrence of the largest eddies, produced by hydrodynamic instabilities in the mean flow field (e.g. shearing between a stream and a solid boundary). These large eddies extract kinetic energy from the mean flow and by doing so provide the kinetic energy input which is necessary to maintain turbulent motion. the transitional region, where kinetic energy cascading takes place. The largest eddies in the core flow become unstable and break down into progressively smaller eddies and transfer their energy, extracted from the core flow, to smaller scales of motion. the laminar viscous sub layer, representing a low Reynolds number region, where finally the cascaded down kinetic energy of turbulence is destroyed by viscous dissipation (i.e. it is converted into heat).

sub layer trans. layer core flow

Fig. 36: Turbulent flow regions

Viscous sub layer: u+ = y+ 5 < y+ < 30 for y+ < 5 (49)

The transitional region, for u+ =

, can be described by the so called Reichards law

y+ y+ 1 *ln (1 + K * y + ) + 7.8* 1 exp *exp ( 0,33* y + ) , (50) K 11 11

with = 0.41 as the Karman constant. The fully turbulent region is modelled with u+ = 1

* ln ( G * y + )

for

y+ >

30,

(51)

while G is an empirical constant assuming a value of 9.0 for aerodynamically smooth walls. 3-8

FIDAP It is very important to note that the near-wall methodology described only functions (see Fig. 37) satisfactorily if the viscous and transitional sub layers are fully contained within the special near-wall elements. If this condition is violated, then the k and equations, which are designed for high Reynolds number flow regions, will be pushed into the low Reynolds number near-wall regions and could result in inaccurate model predictions. Thus, in order for the near-wall model to perform properly, the element Reynolds number y+ in any special element must not be significantly less that 30.

(51) (50) (49)

Fig. 37: Function graphs of the velocity profiles for the viscous sub layer, transitional region, core flow [31]

FIDAP allows plotting the element Reynolds number along wall boundaries by means of the FI-POST module (see Fig. 38). With this feature, chosen values for can be controlled. A first estimation for the required normal distance between wall and first node is made by inserting equation (48) in (47) and then in (46) to

* *

0.0233 * vi

* v *

1/ 4

30

(52)

In the biggest diameter the mean velocity is 30 m/s, therefore the first node has to be posed at least 1 > 0.0002 m away from the wall boundary.

In the throat a mean velocity of around 53 m/s leads to a necessary distance of 2 > 0.0001183 m . 3-9

FIDAP

Fig. 38: FI-POST plot: y+ values along the ENTITY wall, device starts at ARC LENGTH 12.00.

MESH STRATEGY

FI-GEN introduces a mesh generation technology based on the paving automatic mesh generation algorithms developed at Sandia National Laboratories. The paving algorithm allows automated generation of quadrilateral meshes on arbitrary 3-D surfaces (of which 2-D models are a subject). The algorithm simply requires the geometrical definition of the boundary after which the mesh generation procedure is entirely automatic. It is indeed a time saving method to mesh the domain, but with the drawback of random placed nodes (see Fig. 39). In particular close to the wall or in regions with high gradients in for example pressure, more control about node positions is desired. Besides this, FIDAP comes, as mentioned earlier, with a special near wall- function for turbulent modelling problems.

Fig. 39: Paved mesh, detail from the inlet region

3-10

FIDAP For the near-wall element a distance from the wall in the inlet of 0.0008 m and in the throat of 0.0003 m is chosen. The results for y+ along the wall are shown in Fig. 38. Four significant peaks can be figured out, arising from sharp changes in the geometry (see Fig. 40).

Fig. 40: Significant geometrical deflections

However, for the set near wall element the values are in a tolerable range. Furthermore the core flow is subdivided into two regions; the high gradients region and the blunt core region (see Fig. 41). The high gradient region needs a higher resolution, than the blunt core region.

near-wall-element high gradients region

blunt core region

Fig. 41: Characteristic meshing regions

To control the mesh, a mapped meshing strategy is applied to all of the three regions shown in Fig. 41. To see how the paved meshing strategy works, the same region is shown in Fig. 39. The nodes along the boundaries are kept, but do not influence the generation in the desired way.

3-11

FIDAP

3.1.2

FI-PREP

The model can be defined throughout in the module FI-PREP. In different categories, described below, the nodes and entities, generated in FI-GEN, are labelled with the boundary conditions for the single entities or element groups, the fluid properties are chosen, the problem is described in detail and the solution strategy is settled. ENTITY: The entities chosen in the FI-GEN module are attached here with their properties. The single entity groups, infl, outfl, axis, wall , y+1, trans1, y+2, trans2, y+3, trans3, y+4, trans4, y+5, trans5, core1, core2 and core3 are attributed with ENTITY TYPES according to the boundary conditions applied to them later on. The entity type PLOT gives the opportunity to specify the boundary conditions manually. The FLUID entity gets its information through the PROPERTIES. The WALL entity with its implemented near-wall qualities is discussed detailed in chapter 3.1.1. BOUNDARY CONDITIONS: In this feature it is possible to define the boundary conditions for each of the entities. The single entities are discussed above and corresponding boundary conditions are noted in the FIGEN chapter. SIMULATION: Herein is the description of the PROBLEM: axis-symmetric, incompressible steady simulation type turbulent flow type nonlinear convective term Newtonian fluid type single phase

The DATAPRINT OUTPUT, the EXECUTION of a NEW JOB and the PRINTOUT STEPS are to be settled, if it is wanted to change the default settings. Furthermore the turbulence model has to be defined here. Generally it consists of two parts, the eddy-viscosity constitutive relation (Boussinesq, Speziale, Launder) and the eddyviscosity model (zero-equation or two-equation). The Boussinesq relation combined with a RNG two-equation eddy viscosity model is applied and the FIDAP revised model coefficients for the RNG k- model are used instead of the original coefficients.

3-12

FIDAP
O FLUID PROPERTIES

Numerical entries of the fluid character are constant density with 1.225 kg/m3 and a constant value of 1.8375 * 10-5 Ns/m2 for the dynamical viscosity.
O INITIAL CONDITIONS

The initial nodes are defined in position by the mesh. Conditions are set for the turbulent kinetic energy and the viscous dissipation. A constant value for k = 0.003 and a constant value for = 0.00045 are chosen and turn out to deliver the best results in terms to fit to the standard.
O SOLUTION APPROACH

Beside a successive solution algorithm in FIDAP a second so called segregated solution algorithm exists for solving the nonlinear system of matrix equations arising from the finite element method discretisation of the flow equations. With it each conservation equation (velocity component, pressure, temperature, etc.) is solved separately in a sequential manner. The velocity convergence tolerance, velc, is set to 10-5 and the residual vector convergence tolerance, resc, is set to 10-5. With the segregated solution method the parameter U= cylindrical velocity, V= radial velocity, P= pressure, K= kinetic energy and E= viscous dissipation are decoupled solved. The solution converges tolerable after 208 iteration steps (Fig. 42).

Fig. 42: FI-POST plot: Convergence history for the uncoupled quantities, velc=10-5, resc=10-5

Again, it is recommended to write a READ file for the FI-PREP module. Changes can be made rapidly and it is more secure to not forget settings beside the defaults. 3-13

FIDAP The READ in file for the problem preparation is in the form of:
FIPREP( ) PROBLEM( NONL, TURB, AXI- ) EXECUTION( NEWJ ) DENSITY( CONS = 1.225 ) ROUGHNESS( ADD, SET = 1, CONS = 1.5e-006, MOOD ) ICNODE( KINE, CONS = 0.003, ALL ) ICNODE( DISS, CONS = 0.00045, ALL ) VISCOSITY( TWO-, CONS = 1.8375e-005 ) EDDYVISCOSITY( BOUS ) TURBOPTIONS( RNG ) SOLUTION( SEGR = 350, VELC = 1e-005, RESC = 1e-005 ) OPTIONS( UPWI ) DATAPRINT( CONT ) ENTITY( FLUI, NAME = "y+1" ) ENTITY( FLUI, NAME = "trans1" ) ENTITY( FLUI, NAME = "y+2" ) ENTITY( FLUI, NAME = "trans2" ) ENTITY( FLUI, NAME = "y+3" ) ENTITY( FLUI, NAME = "trans3" ) ENTITY( FLUI, NAME = "y+4" ) ENTITY( FLUI, NAME = "trans4" ) ENTITY( FLUI, NAME = "y+5" ) ENTITY( FLUI, NAME = "trans5" ) ENTITY( FLUI, NAME = "core1" ) ENTITY( FLUI, NAME = "core2" ) ENTITY( FLUI, NAME = "core3" ) ENTITY( PLOT, NAME = "infl" ) ENTITY( PLOT, NAME = "outfl" ) ENTITY( PLOT, NAME = "axis" ) ENTITY( WALL, NAME = "wall" ) BCNODE( URC, ZERO, ENTI = "infl" ) BCNODE( UZC, CONS = 30, ENTI = "infl" ) BCNODE( KINE, CONS = 0.003, ENTI = "infl" ) BCNODE( DISS, CONS = 0.00045, ENTI = "infl" ) BCNODE( URC, ZERO, ENTI = "axis" ) BCNODE( VELO, ZERO, ENTI = "wall" ) RENUMBER( ) END( )

CREATE:

After setting all conditions in the FI-PREP section the CREATE section will check the input data correctness. As soon as the input data is complete and checked, FI-SOLV can start the computation. If the solution converged, diverged or the limit of set iteration steps go to To save the various result data to the identifier session, go to IDENT and accept. In the command history window has to be written UNCOMPRESSING DATABASE. Now it is possible to go on with the post-processing.

RUN:

IDENT:

3-14

FIDAP

3.1.3

FI-POST

The computational domain includes 40440 points. The solution converged in 208 iterations. Different plots and calculation options are given in this module. In the former chapters use was already made of some of the plots generated by FI-POST. Most interesting is the differential pressure between reference point J and reference point Q (see Fig. 43). To keep in mind, this geometry is generally used to validate the right approach of the simulation. The standard ISO 5167-3 gives empirical data, which also should be gained by CFD.
J Q

plane1

plane2

Fig. 43: ISO measurement positions vs. FIDAP measurement positions

The differential pressure between J and Q derived by the ISO standard is

p

1403.35 Pa.

Plotting along a line through J (Q) perpendicular to the mean flow direction gives the numerical values along the radius, plane 1 (plane 2). The graphs can be plotted out on the GUI or as a Postscript file (Fig. 44). At the same time, the numerical values are written in the FIOUT file. Because no pressure value is defined at the inlet, the pressure values are relative ones.

plane 1

plane 2

Fig. 44: Superimposed pressure plots of plane 1 and plane 2

3-15

FIDAP Qualitative develops the pressure as expected (Fig. 45). In the corner in front of the inlet the flow stagnates sharply and results in high pressure. Interesting are the two suction peaks; one on the convex inlet throat interface and another one at the convex throat diffuser interface.

Fig. 45: Relative pressure distribution, above: isobars in the device, below: pressure solid patterned

3-16

FIDAP Also the development of the velocity behaves qualitative as expected (Fig. 46). In the throat the highest velocities occur, while the velocity towards the wall diminishes. According to the suction peaks high velocities emerge close due to the convex shapes. The velocity profiles at these cross sections are strongly deformed compared to the velocity profiles at J and Q.

Fig. 46: Plot of the velocity distribution in the standard Venturi device

Fig. 47 shows the cylindrical velocity profiles in plane1 and plane2. At the inlet, plane1, is the velocity profile slightly deformed, whereas the profile in the throat, where the second pressure is measured, is fully developed.

Fig. 47: Cylindrical velocity profiles at plane1 and plane2

3-17

FIDAP As examinable in the pressure plots along the wall- entity and axis- entity (Fig. 48), the device causes an irreversible decreased pressure level. Unfortunately the mean pressure in a cross section is not computable within FIDAP. The pressure plot along the wall shows three distinctive peaks. With the device starting at 12 m the first peak is generated by the strongly decelerated air in the corner. The two minimums are the already observed suction peaks. Here takes the highest acceleration place, at the outer radius of the throat inlet and at the outer radius of the throat outlet.

Fig. 48: a) Static pressure plot along the wall

b) Static pressure plot along the axis

To compare the simulation with the standard, the pressure values at the wall are necessary. The static pressure in J is and the static pressure in Q is therefore is the differential pressure pJ pQ = = 364.86 Pa -1104.12 Pa 1468.98 Pa.

pFIDAP_discrete =

In the entrance of the Venturi device, along its length and exit, big changes in pressure are to be seen and due to the fact, that measurement point J is just positioned were high gradients occur, the periphery is explored (Fig. 49). Additionally it has to be considered, that the measurement method, described in the ISO sheets, measures the punctual pressure with boreholes having a certain diameter and not discrete surface-less nodes like in a computation (see Fig. 50). The measurement holes in the ISO standard have a defined diameter, which means, that the measurement is an average of the diameter values. The standard gives also a limitation for the bore diameter which is 0.01 m. Averaging with the arithmetic means of Jmin and Jmax and Qmin and Qmax leads to the following pressure values pJmin pJmax = = 354.482 Pa 364.860 Pa -1104.340 Pa -1103.127 Pa pJm pQm = = 359.671 Pa -1103.734 Pa

pQmin = pQmax =

3-18

FIDAP
500 12,006821

0 11,6

11,8

12

12,2

12,4

12,6

12,8

12,8592917 13 13,2

13,4

-500

-1000 12,2316476 12,3891469 -1500 12,1197994

-2000

Fig. 49: Pressure along the device wall with geometrical important points emphasized

pISO

averaged

pFIDAP

discrete

Jmin Jmax

Qmin Qmax

Fig. 50: Discrete measurement points and averaging measurement holes

With this way of comparison the result pISO = 1403.35 Pa pFIDAP_averaged = 1460.41 Pa

a derivation of 60.06 Pa gives a tolerable inaccuracy of around 4.07%. Variations of the k and parameters have not lead to significant changes of the differential pressure. Instead of applying the RNG model the Standard model was applied, also without significant changes. In the simulation the walls are considered as smooth, but the loss due to roughness is, as calculated earlier, only about 2.30 Pa. 3-19

FIDAP

3-20

4 MODEL WITHOUT TURBINE

Simulating the ISO 5167-3 standard Venturi nozzle and comparing in chapter 3.1.3 the computed results from FIDAP with the calculated ones from the standard validates the right model approach. Geometrical design aspects are examined in chapter 2.3.1 and qualified shapes are found and described in chapter 2.3.2. Here the modified geometries, up scaled in dimensions, without supporting tube but with redesigned inlet shapes are simulated in a flow, conditioned by the validated parameter values. Changed is the inlet velocity with now vw = 6 m/s to embrace the actual wind conditions. Adjustment concerning the computational domain is necessary as well. Since the supporting tube is removed, the air is free to flow around the structure. Consequently the computational domain is enlarged in radial direction, while the outer cylindrical domain boundaries are no walls anymore. In this chapter the turbine is not implemented yet, to find out about the blocking and separation behaviour of the single contractions structures themselves and to choose later the most reasonable models to insert turbines at different positions.

4-1

MODEL WITHOUT TURBINE

4.1

SETTINGS

Following the applied domain, the boundary conditions and the problem properties are described. The coin model with the corresponding inlays is situated at the first third of the computational domain. So the flow structure is free to develop behind the structure.

4.1.1

DOMAIN

coin with and without corresponding contraction inlay

14 * b

14 * b

Ldevice

24 * b

Fig. 51: Diagram of the computational domain, including dimensions, without turbine

4.1.2

BOUNDARY CONDITIONS

radial outlet

inlet

fluid

cylindrical outlet

structure

symmetry axis

Fig. 52: Boundary groups identification, without turbine

4-2

MODEL WITHOUT TURBINE

4.1.3
problem viscosity model solution accuracy FLUID initial condition INLET : : : : : : : nonlinear, turbulent, axis-symmetric Boussinesq-RNG (two equation) segregated solver, upwinding velc= 1 * 10-5; resc = 1 * 10-5 k : : : : const. const. const. const. const. const. const. const. = 1.225 kg/m3 = 1.8375 10-5 Ns/m2 = 0.003 = 0.00045 = 0 m/s = 6 m/s = 0.003 = 0.00045

INPUT

radial velocity : cyl. velocity : k : : radial velocity : wall velocity :

SYMMETRY AXIS : STRUCTURE :

const. = 0 m/s const. = 0 m/s

4.2

RESULTS

The kinetic energy increasing potential of the bare structures in general is examined, without implementing the turbine. To declare a mean velocity in a cross section, the cylindrical velocity values computed for discrete points along the radius (Fig. 53) are averaged by
structure

vi+1 vi
dAi

(53)

vm ,i

vi + vi +1 2

and taken into account in area weighted manner (54)

vm =

(vm ,i * dAi )
A
Fig. 53: Velocity values at discrete points along the radius

Qualities and quantities of the flow behaviour around and through the models, Co1.00, Co2.25, Co4.00 and Co9.00 are discussed and accompanied by diverse plots. 4-3

MODEL WITHOUT TURBINE

4.2.1

RESULTS CO1.00

Fig. 54: Streamline plot; Co1.00

The streamline plot (see Fig. 54) shows a separation bubble inside the device, which forms the so called Vena Contracta and therefore works like a Venturi constriction. A separation bubble contains velocities in, but also towards the flow direction. In the velocity profile plot (Fig. 55) is to be seen, that negative values occur at least until a device depth of 1.5 m, while the profile in 2.5 m depth still shows an inflexion point. A blade, attacked at the root and along most of the length by positive wind velocities, but at the tips by the reversed velocity, would be difficult to optimize. Therefore the Venturi effect could be used but with regards to keep a certain distance to the tubing wall, to prevent the tips exposure to the vortex. Finding this distance is definitely a more complex problem, due to the separation dependency on the incoming flow velocity and then by the influence of the turbines thrust plus the impact of the blade tip vortex itself.
Co_1.0 INLET velocity profiles
8 7 6 5 cylindrical velocity [m/s 4 3 2 1 0 0 -1 -2 radius [m] 0,5 1 1,5 2 2,5 3 3,5 4 4,5 5 0.0 m 0.5 m 1.0 m 1.5 m 2.0 m 2.5 m

Fig. 55: Velocity profile development inside Co1.00 at 0 m, 0.5 m, 1.0 m, 1.5 m, 2.0 m and 2.5 m

4-4

MODEL WITHOUT TURBINE The stagnation point is obviously on the front face of the device, while the streamline through the stagnation point represents the separating streamline between flow going through the Venturi structure and flow going around the structure. The static pressure at the separation point is expected to be maximal compared to the other points at the wall due to the maximal deceleration. With the relative pressure plot along the face wall (see Fig. 56), a maximum can be figured out at node number 1692 with the coordinates xst_Co1.00 = 0 m; yst_Co1.00 = 5.023 m. The velocity vector plot identifies and so confirms the same point.
Co1.00; relative static pressure along face wall 30 20 10 0 -10 -20 -30 curve length [m] 0 0,2 0,4 max: y = 5.023 m; 21.8629 Pa; node: 1692 0,6 0,8 1 1,2 1,4 1,6

Fig. 56: Relative static pressure plot along the face wall; Co1.00

In the streamline plot is also to be seen, that the flow separates inside the device, but reattaches again. To determine the point of reattachment the shear rate plot along this wall section is used (Fig. 57).
Co1.00; shear rate along "wall1"
1200 1000 800 600 400 200 0 0 1 2 3 4 5 6 7

shear rate [m/s]

relative static pressure [Pa]

min: x = 1.56 m; 9.32 m/s; node: 147

curve length [m]

Fig. 57: Shear rate plot along the inner device wall; Co1.00

Because the shear rate definition, depending on the shear stress at the wall, is given by u =

u ( y ) y

w ,

(55)

separation points and reattachment points as well should be defined by points with a shear rate of zero (Fig. 58). The minimum value is found at node 147, but it is not equal zero. A closer look on the velocity vector plot clarifies the applicability of declaring the shear rate minimum the separation and reattachment respectively.

4-5

MODEL WITHOUT TURBINE The shear rate plot detects node 147, while the velocity vector plot determines node 148. So the assumption is not correct, but points out the adjacencies. Outside the device, along the cylindrical direction, the flow is also detached, but after a certain distance reattaches and the velocity vector plot identifies this at node number 1879 located 5.93 m downstream the model face.

Fig. 58: Separating velocity profile [33]

In the inlet is with the averaging method mentioned above a mean velocity of 5.35 m/s calculated and in a depth of 2.726 m, 6.326 m and also in the outlet at 16.36 m a mean cylindrical velocity of 5.33 m/s (Fig. 59). Wind tunnel measurements carried out within the Castle House feasibility study found an increase of the undisturbed wind velocity taking place through the tubing, while the house model was additionally blocking the wind. The strong upwind, generated at the house facade concentrated the wind towards the upper part and therefore forced the building blockage more flow going through the holes. This parameter study here decouples the effect of the building and therefore the skewed inflow velocity components and concentrates on the inlay shaping only. Fig. 59 shows the cylindrical velocity distribution around the structure, while the profiles at certain positions are plotted as well. Fig. 60 shows the static pressure distribution with the maximum at the stagnation point.

cylindrical velocity [m/s] -5 -4 -5 -4

cylindrical velocity [m/s] -5 -4

cylindrical velocity [m/s] -5 -4

cylindrical velocity [m/s]

cylindrical velocity profile; 2.726 m

cylindrical velocity profile; 6.326 m

cylindrical velocity profile; 16.36 m

cylindrical velocity profile; 0.00 m

-3 -2 -1 radius [m] 0 1 2 3 4 10 9 8 7 6 5 4 3 2 1 0 -1 5

-3 -2 -1 radius [m] 0 1 2 3 4 10 9 8 7 6 5 4 3 2 1 0 -1 5

-3 -2 -1 radius [m] 0 1 2 3 4 10 9 8 7 6 5 4 3 2 1 0 -1 5

-3 -2 -1 radius [m] 0 1 2 3 4 10 9 8 7 6 5 4 3 2 1 0 -1 5

Fig. 59: Cyl. velocity distribution plot & velocity profile inside the device at 0.0, 2.7, 6.3, 16.36 m; Co1.00

Fig. 60: Relative pressure distribution; Co1.00

4-6

MODEL WITHOUT TURBINE

4.2.2

RESULTS CO2.25

Fig. 61: Streamline plot; Co2.25

In comparison to the tube without contraction, no separation takes place in the inlet for the contraction 2.25 (see Fig. 61). Especially the diffuser in a Venturi shape turns out to be the most critical part, but also here the flow keeps attached. Remarkable is the all over shape of the structure added up with the separation bubbles. It is redolent of a common profile shape. Fig. 62 shows the cylindrical velocity distribution around the structure Co2.25. Negative values are identifying the back flow in the separation areas. The field of interest is the throat section where the highest velocities occur. At the throat inlet and outlet close to the wall the flow is deflected most and therefore must be accelerated most as well, resulting in suction and velocity peaks, respectively. The cylindrical velocity profiles are plotted at the significant positions along the device axis; device inlet at 0 m, throat inlet at 1.6 m, position where the velocity profile is expected to be developed, 3.4 m, throat outlet at 5.8 m and the device outlet at 16.3 m.

cylindrical velocity [m/s] -5 -4 -5 -4

cylindrical velocity [m/s]

-5 -4

cylindrical velocity [m/s] -5 -4

cylindrical velocity [m/s]

cylindrical velocity [m/s] -5 -4

cylindrical velocity profile; 1.573 m

cylindrical velocity profile; 3.373 m

cylindrical velocity profile; 5.773 m

cylindrical velocity profile; 16.30 m

cylindrical velocity profile; 0.00 m

-3 -2 -1 radius [m] 0 1 2 3 4 11 10 9 8 7 6 5 4 3 2 1 0 -1 5

-3 -2 -1 radius [m] 0 1 2 3 4 11 10 9 8 7 6 5 4 3 2 1 0 -1 5

-3 -2 -1 radius [m] 0 1 2 3 4 11 10 9 8 7 6 5 4 3 2 1 0 -1 5

-3 -2 -1 radius [m] 0 1 2 3 4 11 10 9 8 7 6 5 4 3 2 1 0 -1 5

-3 -2 -1 radius [m] 0 1 2 3 4 11 10 9 8 7 6 5 4 3 2 1 0 -1 5

Fig. 62: Cyl. velocity distribution & velocity profile inside the device at 0.0, 1.6, 3.4, 5.8 m, 16.3 m; Co2.25

4-7

MODEL WITHOUT TURBINE Usually a turbine blade is exposed to a regular wind velocity profile. By rotation the blade experiences a combined velocity. The more far out, towards the tip, the smaller gets the angle of attack. To gain the best lift over drag ratio geometrical twist is applied. An argument to position a turbine in the cross sections with these distinctive peaks, throat inlet and throat outlet, is that the blade twist could be reduced. Additionally would the resulting velocity at the tip be very high and with this a high lift force could be obtained at a position generating moment most effective. With the same methods as for Co1.00 the stagnation point is determined on the front face. At node 1763 a pressure maximum is found (Fig. 63) and a closer look on the velocity vector plot identified as well the same point as stagnation point with the coordinates xst_Co2.25 = 0.00 m; yst_Co2.25 = 4.623 m.

Co2.25; relative static pressure along "wall 6"

25 20 reltive static pressure [Pa] 15 10 5 0 -5 -10 -15 -20 -25 curve length [m] st.pt.: 0.123 m; 21.9566510 Pa; node: 1763 0 0,2 0,4 0,6 0,8 1 1,2 1,4 1,6

Fig. 63: Relative static pressure plot along the face wall; Co2.25

For the inlet and the outlet mean velocity, found with the same averaging method again, a value of 4.23 m/s is stated. Through the throat a mean velocity of 9.52 m/s is determined, while the velocity profiles differs from each other in significant way. No point along the diffuser wall got a minimum shear rate. The values are even three-digit. Also the look into the vector plot shows that the flow keeps attached. Fig. 64 shows the high pressure area around the stagnation point and its spacious influence.

Fig. 64: Relative pressure distribution; Co2.25

4-8

MODEL WITHOUT TURBINE

4.2.3

RESULTS CO4.00

Fig. 65: Streamline plot; Co4.00

On the outer surface of the device the separation bubble is grown compared to the contraction 2.25, but it still reattaches to the rear (see Fig. 65). In contrast to the flow through the diffuser in Co2.25, flow separation is detected in this case, which leads to higher pressure losses through the diffuser but also to reduced efficiency of the whole device. While the vector plot identifies node 788 as separation point in the diffuser with the coordinates xsep_Co4.00 = 11.236 m; ysep_Co4.00 = 3.443, the global shear rate minimum is given at node 803 (see Fig. 66).

Co4.00; DIFFUSER shear rate

3000 2500 shear rate [m/s] 2000 1500 1000 500 0 0 2 4 6 curve length [m] 8 10 12 min: 6,61 m; 5,41720 m/s; node: 803

Fig. 66: Shear rate plot along the diffuser wall; Co4.00

The stagnation point of the contraction of 4.00 is positioned already on the curvature of the inlet (xst_Co4.00 = 0.019 m; yst_Co4.00 = 4.187 m), which reflects the blocking effect of the higher contraction. Negative cylindrical velocity components exist in the very entrance of the device, resulting in a smaller effective inlet and thus contraction (Fig. 67).

SP

Ainlet Aeff

Aoutlet

Fig. 67: Stagnation point determines effective inlet

4-9

MODEL WITHOUT TURBINE

Co4.00; INLET relative static pressure along wall 40 20 0 -20 -40 -60 -80 st. pt.: 0.314 m; 22.045238 Pa; node: 139 curve length [m] 0 0,5 1 1,5 2 2,5 3 3,5 4

Fig. 68: Relative static pressure plot along the inlet wall; Co4.00

The mean velocity in the inlet and the outlet cross section is 2.67 m/s, while the averaged velocity in the throat is 10.67 m/s. This is already far below the undisturbed wind velocity. Fig. 69 gives the velocity profile development along the device, while at the diffuser outlet the negative values indicate the separation.

relative static pressure [Pa]

cylindrical velocity [m/s] -5 -4 -5 -4

cylindrical velocity [m/s] -5 -4

cylindrical velocity [m/s] -5 -4

cylindrical velocity [m/s] -5 -4

cylindrical velocity [m/s]

cylindrical velocity profile; 2.323 m

cylindrical velocity profile; 3.673 m

cylindrical velocity profile; 5.473 m

cylindrical velocity profile; 16.34 m

cylindrical velocity profile; 0.00 m

-3 -2 -1 radius [m] 0 1 2 3 4 12 11 10 9 8 7 6 5 4 3 2 1 0 -1 5

-3 -2 -1 radius [m] 0 1 2 3 4 12 11 10 9 8 7 6 5 4 3 2 1 0 -1 5

-3 -2 -1 radius [m] 0 1 2 3 4 12 11 10 9 8 7 6 5 4 3 2 1 0 -1 5

-3 -2 -1 radius [m] 0 1 2 3 4 12 11 10 9 8 7 6 5 4 3 2 1 0 -1 5

-3 -2 -1 radius [m] 0 1 2 3 4 12 11 10 9 8 7 6 5 4 3 2 1 0 -1 5

Fig. 69: Cyl. velocity distribution & velocity profile inside the device at 0.0, 2.3, 3.7, 5.5 m, 16.3 m; Co4.00

Such a strong contraction of the flow increases the drag intensely. Directly linked to that is a reduced mass flow through the device. The high pressure region around the stagnation point is shutting down the inlet (Fig. 70).

Fig. 70: Relative pressure distribution; Co4.00

4-10

MODEL WITHOUT TURBINE

4.2.4

RESULTS CO9.00

Fig. 71: Streamline plot; Co9.00

For the contraction 9.00 geometry the streamline plot shows big separation areas (Fig. 71). The bubble on the outer surface has grown over the rear edge and mingles with the outflow of the device. Inside the diffuser the flow is almost not attached at all. While the diffuser starts at 5.17 m the flow separates at xsep_Co9.00 = 6.39 m; ysep_Co9.00 = 1.83. The stagnation point lies with xst_Co9.00 = 0.19 m; yst_Co9.00.= 3.45 m deep in the nozzle and hazards the guess that the high amplification failed. Indeed, the throat velocity lies with 8.95 m/s even below the contraction 4.00 throat velocity, although in both cases the flow separates in the diffuser. The low velocity area behind the structure is deflected towards the axis and generates therein a kind of shielding (Fig. 72). Already for contraction 4.00 this deflecting trend can be observed. At the same time the pressure shield at the front face is getting more rigorous (see Fig. 73).

cylindrical velocity [m/s] -5 -4 -5

cylindrical velocity [m/s] -5 -4

cylindrical velocity [m/s] -5 -4

cylindrical velocity [m/s] -5 -4

cylindrical velocity [m/s]

cylindrical velocity profile; 16.3700 m

cylindrical velocity profile; 0.000 m

cylindrical velocity profile; 3.973 m

cylindrical velocity profile; 5.173 m

cylindrical velocity profile; 3.073 m

-3 -2 -1 radius [m] 0 1 2 3 4 10 9 8 7 6 5 4 3 2 1 0 -1 5

-3

-3

-3

-3 -1 radius [m] 1 3 10 9 8 7 6 5 4 3 2 1 0 -1 5

-2 -1 radius [m] 0 1 2 3 4 10 9 8 7 6 5 4 3 2 1 0 -1 5

-2 -1 radius [m] 0 1 2 3 4 10 9 8 7 6 5 4 3 2 1 0 -1 5

-2 -1 radius [m] 0 1 2 3 4 10 9 8 7 6 5 4 3 2 1 0 -1 5

Fig. 72: Cyl. velocity distribution & velocity profile inside the device at 0.0, 3.1, 4.0, 5.2 m, 16.4 m; Co9.00

Fig. 73: Relative pressure distribution; Co9.00

4-11

MODEL WITHOUT TURBINE

4.3

CONCLUSION

Two properties are influencing the structures in their quality to increase the achievable energy in the throat. One aspect to look at is the blockage by the contraction. Another one is the velocity amplification. With increasing contraction the velocity amplification increases, but the resistance increases as well, so that mass surrounds the structure instead of flowing through it to get accelerated. The optimum between mass reduction and mass acceleration has to be found. Fig. 74 illustrates the accelerated mass flow through the structure, while the rest, surrounding the structure is not contributing. The blockage is defined by the velocity in the throat, which would be achieved with an inlet velocity equal to the wind velocity, over the actual lower throat velocity, due to mass surrounding the structure. vthroat _ tubed blockage = (56) vthroat _ CFD The velocity amplification is defined by the actual throat velocity over the undisturbed wind velocity. This character does not take the mass flow into account. velocity amplification = vthroat _ CFD vwind (57)

mass loss by BLOCKAGE

Ainlet

Athroat

kinetic energy by AMPLIFICATION

mass loss by BLOCKAGE

Fig. 74: Aspects influencing the efficiency of contraction and therefore accelerating, but blocking the flow

In Fig. 75 the trends for the blockage and velocity amplification are plotted against each other.
BLOCKAGE vs. ACCELERATION
7 6 5 4 [-] 3 2 1 0 0 1 2 3 4 5 6 7 8 9 10 contraction [A_inlet/A_throat] acceleration blockage
[-] 7 6 5 4 3 2 1 0 1 2,25 4 blockage acceleration energy amplification 9

comparison

contraction [A_inlet/A_throat]

Fig. 75: Comparison of the four geometries; blockage, acceleration, energy amplification

4-12

MODEL WITHOUT TURBINE In terms of taking into account the two aspects mentioned above, a contraction of approximately 2.25 constitutes the maximum in effectiveness, where velocity amplification over blockage ratio is biggest. A contraction of one is then more effective than a contraction of four. A closer look on the energy flux shows something else. Air flowing through a cross section in the size of the inlet, with the velocity of the undisturbed wind has an energy flux potential of dEwind = dt 1 dm 2 * * vwind 2 dt = 1 3 * Ainlet * * vw 2 (58)

The actual energy fluxes in the single throat sections are calculated corresponding and listed in Tab. 3. To respect the two properties, blockage and velocity amplification, in a reasonable manner, the energy flux amplification is used to compare the quality of the four tubing shapes. With this analysis and among the four investigated geometries the contraction of 2.25 still states the maximum in matters of the energy flux amplification with the factor of 1.77, while contracting the flow with a value of four is indeed amplifying in a positive way, whereas a contraction of one does the amplification with a factor smaller one and therefore reduces the energy flux and so is less effective than the contraction four structure. To keep the application of the geometries feasible, focus is taken on the contraction 2.25. With a throat diameter of six meters this structure offers space for a small but still reasonable turbine. Besides this more practical aspect the fact of the highest energy amplification and no separation through the whole Co2.25 device makes it most attractive.
Tab. 3: Bare structure simulation results overview

definition d Athroat vthroat vinlet vthroat_tubed

Co1.00

Co2.25

Co4.00

Co9.00

[m] [m2] [m/s] [m/s] [m/s]

throat diameter * (d/2)2 CFD CFD Ainlet/Athroat * vwind

9.00 63.62 5.35 5.35 6.00

6.00 28.27 9.52 4.23 13.50

4.50 15.90
10.67

3.00 7.07 8.95 1.00 54.00 77.51 1.49 6.03 3.104 0.37

2.67 24.00

[kg/s] 329.68 207.82 Athroat * * vthroat dm/dt 416.95 velocity [-] 0.89 1.59 vthroat/vwind 1.78 amplification [-] 1.42 2.25 vthroat_tubed/vthroat blockage 1.12 energy [kJ/s] * dm/dt * vthroat2 5.967 11.830 14.940 fluxthroat (dm/dt*vthroat2)/ energy flux 0.71 [-] 1.41 1.77 (dm/dtwind*vwind2) amplification Ainlet = 63.62 m2; vwind = 6 m/s; = 1.225 kg/m3; dm/dtwind = 467.50 kg/s

By inserting a turbine, the blockage will be increased even more and therefore the mass flow through the throat reduced. In the following chapter this circumstance is examined for the contraction 2.25, while the turbine is positioned at three different significant throat depths. Various thrust coefficients are applied to find a power coefficient maximum for a turbine at the throat inlet (pos1), directly placed at the interface nozzle-throat, the throat mid section (pos2) and finally the throat outlet (pos3), placed at the interface throat-diffuser. 4-13

MODEL WITHOUT TURBINE

4-14

5 MODEL WITH TURBINE

In chapter four the bare contraction structures are examined, while a contraction of 2.25 turns out to be most attractive in matters of energy flux amplification, attached flow and applicability with a throat diameter of six meters. In this chapter a turbine modeled as an actuator disc (AD) is implemented. First, a validation of the right settings for an actuator disc is carried out with the help of the well examined bare turbine in a free field. Afterwards actuator discs are inserted into the Venturi models with the contraction 2.25, Co2.25AD, and 1.00, Co1.00AD, while three different significant positions, pos1, pos2, pos3, for the turbines are chosen. With variations of the thrust coefficient the power coefficient optimums are found for the configuration Co2.25pos1, Co2.25pos2 and Co2.25pos3. Finally, comparison for all situations, including Co1.00AD, is performed to figure out the optimum configuration. 5-1

MODEL WITH TURBINE

5.1

ACTUATOR DISC SIMULATION IN FREE FIELD/ VALIDATION

With the help of body forces implementation (Fig. 76) a turbine modelled as an actuator disc can be simulated. To do so it is necessary to define a volume domain. In former applications at Ris satisfying results were obtained by dimensioning the actuator disc volume, VAD, as VAD while =

x * Arotor x
= 0.05 * Rrotor

(59) (60)

and applying a force density distribution, fx, with the definition fx = force unity mass = T VAD * . (61)

pos1

pos2

pos3 Rrotor

x = 0.05 * Rrotor

For the free standing turbine, the thrust coefficient is defined [34] as
CT = T 1 2 * * vw * Arotor 2 , (62)

R
referring to the undisturbed wind velocity flowing through an area in the size of the rotor, while T is the turbines thrust. The power coefficient is the ratio of harvested power over available power in a cross section equal to the area swept by the rotor and can also be written as
x

CP

T * v AD 1 3 * * vw * Arotor 2

(63)

Fig. 76: Sketch: AD implementation

with vAD as the retarded wind velocity in the rotor plane. Assuming the ideal conditions from Froude- Rankine theorem a Betz maximum CP_f =16/27 and a CT_f = 8/9 are expected for a bare turbine.

5-2

MODEL WITH TURBINE Inserting equation (59) and (62) into (61), assuming a homogeneous thrust distribution, an ideal power extraction (CT_f = 8/9) of the 6 m/s wind, the force density to employ for a turbine with a rotor radius Rrotor = 3.0 m results in CT * 1 2 * vw 2 x

fx

106.67 m/s 2 . (64)

A simulation of such a rotor in a free field is accomplished in the following section to ensure the right usage of the body forces for the encased turbine later on. Prerequisite of an ideal power harvesting is an axial induction factor of one third, which shall be reflected in a velocity in the actuator disc of two third of the undisturbed wind velocity, hence vAD = 4 m/s.

5.1.1

DOMAIN

10 * Rrotor

Rrotor 10 * Rrotor
x

20 * Rrotor

Fig. 77: Diagram of the computational domain, including dimensions, bare turbine

Setting up the computational domain as in Fig. 77 ensures a free development of the flow.

5.1.2
radial outlet

BOUNDARY CONDITIONS

inlet

fluid

cylindrical outlet

actuator disc

symmetry axis

Fig. 78: Boundary groups identification, bare turbine

5-3

MODEL WITH TURBINE

5.1.3
problem viscosity model solution accuracy FLUID initial condition INLET : : : : : : : nonlinear, turbulent, axis-symmetric Boussinesq-RNG (two equation) segregated solver, upwinding velc= 10-5; resc = 10-5 k : : : : const. = 1.225 kg/m3 const. = 1.8375 * 10-5 Ns/m2 const. = 0.003 const. = 0.00045 const. const. const. const. = 0 m/s = 6 m/s = 0.003 = 0.00045

INPUT

radial velocity : cyl. velocity : k : : radial velocity : cyl. body force:

SYMMETRY AXIS : ACTUATOR DISC :

const. = 0 m/s const. = -106.67 m/s2

5.1.4
fx=-106.67 for CT=8/9; values along symmetrical axis 15

RESULTS AD FREE

10 pressure [Pa] / velocity [m/s]

5 velocity [m/s] pressure [Pa] 0 0 -5 10 20 30 40 50 60 70 80 90

-10 5 diameter before and 10 diameter behind AD (at 30) [m]

Fig. 79: Velocity and static pressure plot along the symmetrical axis, AD in free field, Betz optimum

Fig. 79 shows the resulting values for the pressure and the velocity along the symmetry axis. The plot shows the expected characteristics, vAD = 4.0 m/s, and therein confirms the right implementation and so allows the adoption of turbine modelling by body forces for the following cases. 5-4

MODEL WITH TURBINE

5.2

VENTURI ENCASED ACTUATOR DISC SIMULATION

As soon as the rotor is encased it must be clarified to which area the dimensionless coefficients refer to. For the first consideration the coefficients, CP and CT, shall refer to the actual rotor sweeping area (see Fig. 80 and Fig. 81). Rearranging equation (62), T = CT * 1 2 * * vw * Arotor , 2 1 2 CT * * * v w * Arotor * v AD 2 = = 1 3 * * v w * Arotor 2 (65)

and inserting into (63) leads to CP = T * v AD 1 3 * * vw * Arotor 2

vw

CT *

v AD . vw

(66)

Arotor

AD

vAD

Fig. 80: Ainlet is 2.25 times bigger than Arotor for a contraction of 2.25 vw

AD Arotor vAD

Fig. 81: Ainlet corresponds to Arotor for a contraction of one

With is then

T fx

= =

p * Arotor T m =

(67)

p * Arotor = * Arotor * x
5-5

p . * x

(68)

MODEL WITH TURBINE Values for the thrust coefficient, corresponding thrust and pressure drop which are investigated and analysed in the next chapter for the contraction of 1.00 and 2.25 are listed below in Tab. 4.
Tab. 4: Applied thrust coefficients, referring to the rotor area, for the model Co2.25AD and Co1.00AD

pi [Pa] p0 = 5.210 p1 = 6.615 p2 = 8.004 p3 = 9.685 p4 = 11.718 p5 = 13.020 p6 = 13.744 p7 = 14.105 p8 = 14.467 p9 = 15.190

CTi (referred to rotor) CT0 = 0.236 CT1 = 0.300 CT2 = 0.363 CT3 = 0.439 CT4 = 0.531 CT5 = 0.591 CT6 = 0.623 CT7 = 0.640 CT8 = 0.656 CT9 = 0.689

Co2.25AD fxi [m/s2] T [N] X 36.000 43.560 52.708 63.773 70.859 74.795 76.764 78.732 82.669 X 187.04 226.31 273.84 331.33 368.14 388.59 398.82 409.05 429.50

Co1.00AD fxi [m/s2] T [N]

18.904 24.000 29.040 35.138 42.515 X X X X X

331.47 420.83 509.20 616.13 745.48 X X X X X

This is the conventional way of looking at it, but makes a comparison difficult. Mostly, in investigations done before, the increased mass flow through the turbine caused by a diffuser or shroud was referred to the rotor cross section and compared to a bare rotor of the same size. Here is the inlet area kept constant and given is the same initial situation, but the rotor diameter varies corresponding to the contractions. Which inner geometry combined with a turbine makes most out of the offered wind, entering the same inlet, is ascertained. In general the power coefficient is the ratio of harvested power over power contained in the wind. That is why in a second investigation the power coefficient refers to the inlet area, Cp*.
* CP

p * v AD * Arotor
1 3 * * vw * Ainlet 2

CP Co

(69)

5.3

SETTINGS

The settings considering the geometry, turbulence modeling, wind properties and boundary conditions are not changed compared to the bare structure simulations. But the settings here are added up with an actuator disc, already introduced in 5.1. The thrust of this actuator disc is modeled in form of a body force density in the actuator disc volume, while the distribution is constant and pointing against the wind direction. For the contraction 2.25 and 1.00 an actuator disc is positioned at three different significant locations inside the device. The systems, casing and turbine, are named in the following way Co2.25pos1, Co2.25pos2, Co2.25pos3 and Co1.00pos1, Co1.00pos2, Co1.00pos3, respectively. In case of the contraction 1.00 is pos1 positioned directly at the device inlet, pos2 at a location where the velocity profile is expected to be developed without the influence of the applied thrust and pos3 directly at the outlet of the device. Pos1 is located directly at the throat inlet for the contraction 2.25, while pos2 is located in the throat where the velocity profile is expected to be developed in case of a bare structure and pos3 is positioned at the outlet of the throat, before the diffuser starts. 5-6

MODEL WITH TURBINE

5.3.1

DOMAIN

AD at pos1, pos2, pos3, respectively, for Co2.25 and Co1.00

14 * b

b 14 * b Ldevice 24 * b

Fig. 82: Diagram of the computational domain, including dimensions, device with turbine

In Fig. 83 and Fig. 84 the grids are to be seen. Obviously the grid around the structure is denser than the skewed surrounding domains. Changes in pressure, velocity and direction are higher close to the structure and so the finite elements need to be defined finer. Detailed information about the mesh strategy and the software FIDAP can be found in chapter 3.

Fig. 83: Grid plot of the Co1.00 domain; 51742 nodes

Fig. 84: Grid plot of the Co 2.25 domain, 46918 nodes

5-7

MODEL WITH TURBINE

5.3.2
radial outlet

BOUNDARY CONDITIONS

inlet

fluid

cylindrical outlet

structure

AD

symmetry axis

Fig. 85: Boundary groups identification, device with turbine

5.3.3
problem viscosity model solution accuracy FLUID initial condition INLET : : : : : : : nonlinear, turbulent, axis-symmetric Boussinesq-RNG (two equation) segregated solver, upwinding velc= 10-5; resc = 10-5 k : : : : const. = 1.225 kg/m3 const. = 1.8375 * 10-5 Ns/m2 const. = 0.003 const. = 0.00045 const. const. const. const. = 0 m/s = 6 m/s = 0.003 = 0.00045

INPUT

radial velocity : cyl. velocity : k : : radial velocity : wall velocity :

SYMMETRY AXIS : STRUCTURE : CO2.25AD : (POS1) (POS2) (POS3) fx1 : fx2 : fx3 : fx4 : fx5 : fx6 : fx7 : fx8 : fx9 :

const. = 0 m/s const. = 0 m/s CO1.00AD : (POS1) (POS2) (POS3) fx0 : fx1 : fx2 : fx3 : fx4 : const = 18.904 m/s2 const = 24.000 m/s2 const = 29.040 m/s2 const = 35.138 m/s2 const = 42.515 m/s2

const = 36.000 m/s2 const = 43.560 m/s2 const = 52.708 m/s2 const = 63.773 m/s2 const = 70.859 m/s2 const = 74.795 m/s2 const = 76.764 m/s2 const = 78.732 m/s2 const = 82.669 m/s2 5-8

MODEL WITH TURBINE

5.4

RESULTS

In this section a detailed examination of the simulation results for a Co2.25AD is performed. First a qualitative observation and description of the influence of the turbine position is carried out. With the help of computed values and graphs, characteristic changes, peaks or failure are determined. For these peculiar values static pressure, cylindrical velocity and/or streamline plots around the structure are chosen to explain the e.g. derived velocity values in the actuator disc cross section. Case Co1.00AD is run only until a thrust coefficient of CT = 0.53. Streamline plots visualise the thrust increase influence.

5.4.1

RESULTS CO2.25AD

In the same way as for the bare structures the mean velocity in the actuator disc cross section is determined and the results at position one, position two and position three are plotted over the applied CT in Fig. 86 while the values are listed in Tab. 5, corresponding to the thrust coefficients, related to the rotor cross section.

Co2.25_v_AD over CT
9 8 7 6 5 v_A D 4 3 2 1 0,2 0,25 0,3 0,35 0,4 0,45 0,5 0,55 0,6 0,65 0,7 0 0,75 -1 -2 CT pos1 pos2 pos3

Fig. 86: vAD plotted over the corresponding CT for pos1, pos2 and pos3; Co2.25AD Tab. 5: Actuator disc mean velocities depending on the position and thrust of the actuator disc; Co2.25AD

vthroat = 9.52 m/s pos1 pos2 pos3

vAD1 [m/s] 8.05 8.00 7.97

vAD2 [m/s] 7.76 7.84 7.76

vAD3 [m/s] 7.55 5.75 5.69

vAD4 [m/s] 7.81 6.17 5.24

vAD5 [m/s] 6.08 5.17 5.43

vAD6 [m/s] 5.40 X 5.36

vAD7 [m/s] 5.21 X 4.99

vAD8 [m/s] X X 4.87

vAD9 [m/s] X X X

Along the graphs plotted in Fig. 86, the characteristics of the three different positions are qualitatively examined, starting out at the low thrust implementation. For the lower pressure sinks due to the actuator disc the behaviour of the flow does not differ distinctively. At CT3 = 0.44 the first characteristic branching occurs. The turbine at position one is less blocking the flow, than the turbines placed at position two and three, although the same thrust is affecting the flow. A look on the static pressure distribution for the single cases will explain this phenomenon (Fig. 87 to Fig. 89). 5-9

MODEL WITH TURBINE The static pressure sink due to the turbine has the typical pressure maximum in front of the AD and the minimum behind, approaching asymptotic the atmospheric pressure again. To recall this pressure development, please see Fig. 79 in chapter 5.1.4. This pressure maximum upstream the turbine is reaching more out of the structure inlet in case one, than in two or three. So the pressure peak has more space compared to the pressure peaks in front of the turbines positioned at pos2 or pos3. Because of the more restricted space for the pressure peak the pressure density is increasing and therefore the blockage is increasing as well.

Fig. 87: Static pressure, large picture, and cyl. velocity plot, small picture; Co2.25pos1, CT = 0.44

Fig. 88: Static pressure, large picture, and cyl. velocity plot, small picture; Co2.25pos2, CT = 0.44

Fig. 89: Static pressure, large picture, and cyl. velocity plot, small picture; Co2.25pos3, CT = 0.44

This circumstance is reflected by the decreasing inlet velocity and with it the smaller mass flow through the structure (see Fig. 90). In this figure it can also be seen, that the static pressure maximum along the axis for the AD at pos1 is located 3.14 meters, for the AD at pos2 2.42 meters and for the AD at pos3 1.97 meters in front of the structure inlet. 5-10

MODEL WITH TURBINE

static pressure along the axis in device Co2.25AD; Ct = 0.53
20 [-1.97; 13.40] [-2.42; 12.19] [-3.14; 9.54] 10 [0.00; 11.02] [0.00; 8.12] [0.00; 1.42] [-0.00; -7.52] 0 static pressure [Pa] -10 -5 0 5 10 15 20 bare pos1 pos2 pos3 v_inlet = 2.33 m/s v_inlet = 2.74 m/s v_inlet = 3.47 m/s v_inlet = 4.23 m/s

-10

-20

-30

-40 device depth [m]

Fig. 90: Static pressure plotted along the axis; structure inlet at 0m; CT = 0.53&0.59: Co2.25pos2&3

For the bare structure the global maximum pressure was found at the stagnation point, with a spacious influence on the device inlet area (see Fig. 91).This pressure field is now additionally influenced by the turbine. A kind of pressure shield is developing at the face of the device, which is less dense for pos1. The deeper the turbine, and with it the retardation, is inside the casing, the more is the shield affected by the steep pressure increase in front of the turbine and closes therein the device (see Fig 92).
20,00-30,00 10,00-20,00 0,00-10,00 -10,00-0,00 -20,00--10,00 -30,00--20,00 -40,00--30,00
30 static pressure [Pa] 20 10 0 -10 -20 -30 -40 -6.0 m -5.0 m -4.0 m -3.0 m -2.0 m 14 rad i 12 al c oo -1.0 m 9 rdin
at e
th [m ] ep v ic de ed

wind

5
[m]

inlet 4 3 3 1.0 m 2 1 1.6 m 0

Fig. 91: Static pressure distribution, 3-D plot: pressure shield in front of the device, bare structure

5-11

MODEL WITH TURBINE

wind

30 static pressure [Pa] 20 10 0 -10 -20 -30 -40 -4.0 m -3.0 m

h[ de pt de vic e

-6.0 m -5.0 m

-2.0 m 14 12 -1.0 m 9 5 inlet 4 3 3 1.0 m 2 1 AD at pos1 0

20,00-30,00 10,00-20,00 0,00-10,00 -10,00-0,00 -20,00--10,00 -30,00--20,00 -40,00--30,00

wind

30 static pressure [Pa] 20 10 0 -10 -20 -30 -40 -0.6 m -0.5 m -0.4 m -0.3 m
de pt m]

-0.2 m 14 12 -0.1 m 9 5 inlet 4 3 3 1.0 m 2 1 AD at pos3 0

Fig. 92: Pressure shield in front of the device, CT = 0.53: upper pic.: Co2.25pos1; lower pic.: Co2.25pos3

5-12

de vic e

h[

m]

MODEL WITH TURBINE By implementing CT5 = 0.59 a new effect starts to take over. Since the velocity is decelerated by the turbine, but in contrast to the pressure change in a continuous manner, the lowest speed for the turbine wake at position two is closer situated to the diffuser end, than for the turbine at position three (compare Fig. 93 with Fig. 94). What happens is that the turbine wake with very low velocities grows and marches upstream until it acts at the diffuser outlet like a blocking area. In the mixing area of the two flow streams behind the structure, the outer stream is deflected towards the axis due to the over expanded jet situation there.

Fig. 93: Cylindrical velocity plot; Co2.25pos2, CT = 0.59

Fig. 94: Cylindrical velocity plot; Co2.25pos3, CT = 0.59

At this point, additionally, the wake of the structure with its separation bubble on the outer surface and the rear face starts to interact. The outflow has to find its way between the turbine wake and the structure wake (see Fig. 95). A barrier from behind is generated vice versa to the one at the front face. An advantage of the turbine wake could be the deflection of the devices inner flow towards the diffuser wall and therein it contributes to the prevention of separation. Because the wake is oscillating and therefore this effect is unstable, an idea could be to place a solid structure behind the device to create an artificial but constant deflection in the wake area. Further investigations in this field are necessary, but not done within this study.

Fig. 95: Superposition of cyl. velocity and streamline plot; Co2.25pos3, CT = 0.59

5-13

MODEL WITH TURBINE In areas with low velocities, the static pressure is high. How the cylindrical velocity develops along the axis for CT = 0.53 is shown in Fig. 96 for the turbine positioned at pos2 and pos3. Also the velocity development for the higher thrust implementation at pos3 is plotted in the same figure. Here it can be seen, that the CT = 0.53 velocity minimum of 1 m/s has decreased to almost 0 m/s for the CT = 0.59 implementation. So the velocity 1 m/s is 2 meters closer to the device outlet and shows almost the same behaviour as for CT = 0.53, pos2 .
cylindrical velocity along the axis Ct=0.53 and Ct=0.59
10,0 9,5 9,0 8,5 8,0 7,5 7,0 cylindrical velocity [m/s] 6,5 6,0 5,5 5,0 4,5 4,0 3,5 3,0 2,5 2,0 1,5 1,0 0,5 0,0 -8 -4 0 4 8 device depth [m] 12 16 20 24 pos2_Ct=0.53 pos3_Ct=0.53 pos3_Ct=0.59

DEVICE INLET

DEVICE OUTLET

Fig. 96: Cylindrical velocity along axis plot: CT = 0.53, pos2, pos3; CT = 0.59, pos3

At 16.0 m device depth the structure ends with the outlet diameter of 4.5 m. The static pressure distribution behind the device is plotted in 3-D for the bare structure in Fig. 97. How the reduced static pressure by the turbines, pos2&3, influences this area is visible in Fig. 98.
0,5-1,0 0,0-0,5 -0,5-0,0 -1,0--0,5 -1,5--1,0 -2,0--1,5 -2,5--2,0 -3,0--2,5

1,0 0,5 static pressure [Pa] 0,0 -0,5 -1,0 -1,5 -2,0 -2,5 -3,0
rad ia l co o

24.0 m 23.0 m 22.0 m

e [m ]

20.0 m 4,4 4,2 19.0 m 3,1

de v

3,9

3,5

18.0 m 2,6 17.0 m 2,0 1,4 0,7 16.0 m 0,0

wind

Fig. 97: Static pressure distribution, 3-D plot: pressure shield behind the device, bare structure

5-14

ice

de pt h[ m]

rd i n at

21.0 m

MODEL WITH TURBINE

1,0 0,5 static pressure [Pa] 0,0 -0,5 -1,0 -1,5 -2,0 -2,5 -3,0
rad ial coo rd

24.0 m 23.0 m 22.0 m

ina t

21.0 m
h[ ep t
e [m ]

4,4

0,5-1,0 0,0-0,5 -0,5-0,0 -1,0--0,5 -1,5--1,0 -2,0--1,5 -2,5--2,0 -3,0--2,5

4,0

3,6

18.0 m 3,2 2,6 17.0 m 2,1 1,4 0,7 16.0 m 0,0

wind

1,0 0,5 static pressure [Pa] 0,0 -0,5 -1,0 -1,5 -2,0 -2,5 -3,0
rad ial coo r

de vic ed

4,3

19.0 m

22.0 m
m]

3,2

wind

2,6

17.0 m 2,1 1,4 0,7 16.0 m 0,0

Fig. 98: Static pressure field behind device, CT = 0.53: upper pic.: Co2.25pos2; lower pic.: Co2.25pos3

5-15

de v

3,6

18.0 m

ic e

4,3

4,0

de

4,4

19.0 m

pt

h[

20.0 m

m]

din

ate [

21.0 m

m]

20.0 m

24.0 m 23.0 m

MODEL WITH TURBINE If more energy is extracted, the flow through the Venturi device is decelerated more, while the flow around the structure is increasing in velocity, mass and deflection angle. With increasing thrust system Co2.25pos collapses first. One may expect system Co2.25pos3 to breakdown next, but system Co2.25pos1 fails earlier. Thus the advantage of the less dense inlet shield does not pay until the end, but far. The turbine at position three with the biggest disadvantage concerning the blockage at the inlet operates with the highest thrust coefficient. From another point of view, the power harvesting side, the turbine at position one reaches the highest power coefficients among the three positions (see Fig. 99).
Co2.25_Cp over Ct
0,8 0,7 0,6 0,5 C p 0,4 0,3 0,2 0,1 0 0,75 pos1 pos2 pos3

0,2

0,25

0,3

0,35

0,4

0,45 Ct

0,5

0,55

0,6

0,65

0,7

Fig. 99: CP over CT for pos1, pos2, pos3; Co2.25AD

Fig. 100 to Fig. 105 show a storyboard for the flow development with the turbine implemented at position one.

Fig. 100: Streamline, AD velocity profile, pressure and cylindrical velocity plot; Co2.25pos1, CT = 0.44

Fig. 101: Streamline, AD velocity profile, pressure and cylindrical velocity plot; Co2.25pos1, CT = 0.53

5-16

MODEL WITH TURBINE

Fig. 102: Streamline, AD velocity profile, pressure and cylindrical velocity plot; Co2.25pos1, CT = 0.59

Fig. 103: Streamline, AD velocity profile, pressure and cylindrical velocity plot; Co2.25pos1, CT = 0.62

Fig. 104: Streamline, AD velocity profile, pressure and cylindrical velocity plot; Co2.25pos1, CT = 0.64

Fig. 105: Streamline, AD velocity profile, pressure and cylindrical velocity plot; Co2.25pos1, CT = 0.66

5-17

MODEL WITH TURBINE

5.4.2

RESULTS CO1.00AD

As soon as the thrust is applied, the velocity inside the device decreases. The intensity of this effect depends on the interaction between the structure and the turbine. Fig. 106 shows the cylindrical mean velocity in the actuator disc cross section, depending on the CT and the positioning. The exact values are summarized in Tab. 6. No distinctive differences occur between the three positions and the retardation is almost linear. The maximum applied CT is 0.53. For higher thrust coefficients the solution does not converge and need another adjustment of parameters setting, which is not carried out in this work.
Co1.00_vAD over Ct
9 8 7 6 5 v_A D 4 3 2 1 0 0,2 0,25 0,3 0,35 0,4 0,45 0,5 0,55 -1 0,6 -2 Ct pos 1 pos 2 pos3

Fig. 106: vAD plotted over the corresponding CT for pos1, pos2 and pos3; Co1.00AD Tab. 6: Actuator disc mean velocities depending on position and thrust of the actuator disc; Co1.00AD

vthroat = 5.35 m/s pos1 pos2 pos3

vAD0 [m/s] X 4.69 X

vAD1 [m/s] 4.64 4.53 4.55

vAD2 [m/s] 4.15 4.32 4.30

vAD3 [m/s] 4.04 3.97 4.00

vAD4 [m/s] 3.48 3.69 3.76

Interesting are the velocity profiles in the actuator disc positioned at position one (Fig. 107). As for the bare Co1.00 model examined, separation takes place at the inlet and forms the so called Vena Contracta, acting like a Venturi constriction. For lower thrust coefficients these negative velocity components still exist. This recirculation zone might interact with the blade tips in a disadvantageous way. The velocity peaks close to the structure are characteristic inside a contraction and are discussed for a turbine positioned at a Venturi throat inlet.

Fig. 107: Cylindrical velocity profile plot at pos1; CT = 0.3000, CT = 0.4392, CT = 0.5310; Co1.00pos1

5-18

MODEL WITH TURBINE With the streamline plots for Co1.00pos1 (Fig. 108) the growing separation bubble on the outer surface of the model is visible. Remarkable is the streamline tube widening behind the structure, resembling the expected streamline tube for a bare turbine. The typical streamline tube seems to be shortly interrupted by the structure and continues afterwards. The behaviour that can be observed is, the higher the deceleration of the wind velocity is the more narrows the supporting streamline tube. For CT = 0.3, for example, the inlet cross section of the streamline tube far in front of the structure is around 9 m2 larger than for the case of CT = 0.53. This also reflects the decreasing mass flow through the structure and actuator disc, respectively.

Fig. 108: Streamline plot around structure: CT = 0.30, CT = 0.36, CT = 0.44, CT = 0.53; Co1.00pos1

Fig. 109 shows the power coefficient plotted over the thrust coefficient, while both dimensionless parameters refer to the rotor diameter. CP for position three lies slightly above position twos, although the flow will lose more energy through the device before passing the actuator disc.
Co1.00_Cp over Ct 0,35 0,3 0,25 Cp 0,2 0,15 0,1 0,05 0
0,2 0,25 0,3 0,35 0,4 0,45 0,5 0,55 0,6 pos 1 pos2 pos3

Ct

Fig. 109: CP over CT for pos1, pos2, pos3; Co1.00AD

5-19

MODEL WITH TURBINE

5.5

CONCLUSION

The qualitative description of the untypical system Co2.25AD flow behaviour and the explanation by the shield at the front face and the blocking at the rear face is not analytically defined. It could be the influence of the more or less deep located turbine with its pressure peaks. In contrast to the contraction of 2.25, the model with no Venturi shaping inside shows a very regular behaviour with no significant differences among the three turbine positions. Probably higher thrust coefficients would enforce more distinctive characters. Nonetheless, Fig. 110 visualises how the pressure peak influence in front of the actuator disc lies more inside and is restricted by the structure for pos2 and pos3, respectively. For the Co2.25AD model, position one turns out to be the most effective position, therefore its results are compared in the following with the Co1.00pos1 results.
Fig. 110: Pressure distribution, pos1, pos2, pos3, Co1.00AD

As mentioned earlier, to compare the two cases with the help of the power coefficient is difficult, because the rotor diameters are not the same, but the inlet diameter. In case of referring the power coefficient to the rotor area, the Betz maximum of 16/27 is exceeded with CP4(Co2.25pos1) = 0.69 by 16.68 percent, while the thrust coefficient, referring to the rotor area as well, lies with CT4(Co2.25pos1) = 0.53 far below 8/9. Which value can be directly used for comparison is the power itself. With P = CP * 1 3 * * vw * Arotor 2 (70)

the following values listed in Tab. 7 are derived.

Tab. 7: Power values for the Co2.25pos1 and Co1.00pos1 models

power(CT1) [W] Co2.25pos1 Co1.00pos1 1506.34 1952.79

Power(CT2) [W] 1756.04 2115.19

power(CT3) [W] 2066.94 2486.21

power(CT4) [W]
2586.37 2591.65

5-20

MODEL WITH TURBINE Although the Co1.00AD model is not run for higher thrust coefficients, higher power values could be achieved certainly. But comparing the power maximum for the contraction 2.25, turbine positioned at pos1, with the power value achieved by the model without constriction but same pressure drop, shows the potential of the shaping combined with the right turbine positioning. An ideal bare turbine and its operation, one rotor with inlet diameter and one rotor with throat diameter, is analytically calculated below to see the advantages or disadvantages compared to encased ones. The retarded velocity over wind velocity is commonly [34] defined as (1 a ) = v AD vw (71)

and the thrust coefficient and power coefficient referred to the rotor diameter are then CT CP = = 4 * a * (1 a) 4 * a * (1 a) 2 (72) (73)

while the power coefficient depending on the thrust coefficient can also be written as CP = CT * v AD vw (74)

With these equation and the initial condition of the same thrust coefficients, CT1 = 0.30, CT2 = 0.36, CT3 = 0.44, CT4 = 0.53 following results, summarized in Tab. 8, are derived.
Tab. 8: Power coefficients and power values for bare turbines the same size of encased ones, comparison

a [-]
CT1 CT2 CT3 CT4

big bare rotor power CP [-] [W]

Co1.00pos1 CP power [-] [W]

small bare rotor CP power CP [-] [W] [-]

Co2.25pos1 CP* power [-] [W]

0.08 0.10 0.13 0.16

0.276 0.326 0.384 0.448

2318.76 2746.82 3232.57 3767.34

0.232 0.251 0.295 0.308

1952.79 2115.19 2486.21 2591.65

0.276 0.326 0.384 0.448

1030.56 1220.81 1436.70 1674.37

0.403 0.469 0.553 0.691

0.179 0.209 0.246 0.307

1506.34 1756.04 2066.94 2586.37

To compare the turbine encased by the constricted tubing with the one encased by a constant diameter tubing the Cp* seems to be a convenient coefficient, while the fact of the power harvesting by a smaller turbine is not reflected with it. Comparing the big bare rotor power with the power gained by the Co1.00 encased turbine, it performs better. Of course, it has to be remarked, that effects like tip loss reduction, varying thrust distribution etc. are not considered here and might change the conclusion. Definitely the comparison between a small bare rotor and the C02.25 encased rotor is significant. The power amplification is 54.24 %. 5-21

MODEL WITH TURBINE

5-22

6 SUMMARY

In this chapter the difficulties and results of the parameter study and especially ideas for further investigations are summarized. 6-1

SUMMARY

6.1

STATE OF THE ART

At the beginning was the Castle House with its straight ducts, reaching from the faade and breaking through to the rear at the upper sloped section. Wind tunnel tests and rough CFD simulations for a simplified three dimensional building were carried out. Based on the buildings geometry the task was to find an advantageous shaping for the ducts inlay. To decouple the effect of the building itself, focus was taken on the upper part of the building and was simplified to an axis symmetrical parameter model and problem. On the way finding geometries for a constricted tube with a reasonable diffuser angle and respecting at the same time the given inlet diameter to keep the faade layout unchanged, the length turned out to be far out of range to be integrated (Fig. 111). Anyway, a more general parameter study developed, regarding the contractions 1.00, 2.25, 4.00 and 9.00, while the inlet diameter, the outer diameter and the length were kept constant.

Fig. 111: Venturi tube integrated

Simulations were done within the computational fluid dynamical tool FIDAP. Most difficult was the right definition of the first cell around the structure to maintain the laminar sub layer and transitional layer, reflected by the dimensionless cell Reynolds number y+. After approximately 50 iteration steps it was for example still bigger than 30 at the first node, but 50 iteration steps later, the velocity close to the structure had changed and with it the value for y+, depending on the normal distances from the wall and the velocity at that distance. Fine tuning the distance was quite time consuming, especially for the higher contracting geometries. And as soon as the actuator discs were implemented with different thrust coefficients and so velocities were changing again, numerous adjustments were necessary. In general computer simulations do not correspond to the reality. Already Newton only tried to describe the nature. However, to achieve an idea of trends and qualitative properties, computer aided calculations and visualisations are very convenient. One of the Venturi geometries was chosen to observe the influence of the positioning of a turbine. The contraction 2.25 was the one to be examined throughout, because of its qualities in matters of separation, blockage and velocity amplification. Compared to the other shapes no separation took place through the whole device, the energy flux amplification with 1.77 was higher than 1.41 for the contraction of four and the throat diameter with six meters states a reasonable dimension for small turbines. Interesting characteristics, depending on the turbine position, were found. All three positions were located in the Venturi throat, while position one was at the entrance, position two at a position where the velocity profile was expected to be developed as long as no thrust was applied and position three was located at the throat end, in front of the diffuser. With the turbine at position one the highest power coefficients were obtained. Depending on the reference area the Betz limit was with Cp = 0.69 exceeded by 16.68 percent, with a thrust coefficient of CT = 0.53. For the same pressure drop the turbine, 6-2

SUMMARY encased with straight tubing, gained the same power, although the rotor area was 2.25 times larger. For the straight duct model the position of the turbine had no significant impact. But the separation inside the inlet area has to be considered. Turbines should not be located there. Within a comparison with bare turbines, the encased turbine, nine meters in diameter, produced less power than the bare turbine the same size. Reviewing the bare Co1.00 structure, already blocking the wind, a mean velocity of 5.35 m/s was found in the device, which lies below the 6 m/s for the approaching wind velocity. In contrast to that the six meters in diameter turbine, placed inside the Venturi shaped duct, seemed to be able to produce 54.24 percent more power than a bare turbine the same size. Regards to the different tip loss situations might lead to another result, but were not considered in this work. Comparing the Castle House feasibility study results with the results derived within this work is difficult, because neither horizontal skewed wind was regarded nor the blockage of the building, generating an upwind with accelerated components and vertical skewed angles. Thus comes a Venturi shaped casing compared to a straight duct with the advantages of a smaller rotor and with it lower production, transportation etc. costs and especially installation on high buildings with corresponding erection equipment might be more feasible or less complicated.

Also important is due to the contracting inlet the lowered turbulence intensity in the flow approaching the turbine. Furthermore is the Venturi effect amplifying the wind velocity and so the air attacking the turbine in the throat might be sufficient high to operate the turbine, while the big rotor, attacked by velocities even below wind velocity still stands still.

The new application of turbines integrated in buildings cancels out the always returning augmented turbine contra argument - additional structure does not pay gained power - because the building structure is at the same time encasing structure, and even better no tower is necessary anymore and therein the conventional structure is even reduced.

On the other hand is it very important to consider the changed loads situation on the building structure, which might needs to be strengthened, planned stiffer or may more flexible at certain locations. In general is an encasing for turbines in urban areas recommendable, e.g. in case of a blade loss the centrifugal energy will be first caught by the casing. Also frequently moving shadows, cast on buildings, could be reduced, due to the more hidden turbine.

Lower adequate noise levels, demanded in towns, could be fulfilled by the possibility to integrate damping material in the encasing structure and probably directing of the sound waves.

But these are already some topics which have to be investigated in further studies. 6-3

SUMMARY

6.2

FUTURE PROSPECTS

It was a challenge to restrict the parameter study and find constant parameter, respectively. Therefore a wide field for further investigations already exists only for the parameter study itself. So it would be interesting to know more about the flow behaviour with different wind velocities influence of skewed inflow consideration of rotational velocities in the turbine wake regards to the tip loss situation more realistic thrust distribution implementation varying coin parameter b the same contraction, but varying diffuser opening angles and therein device length a bigger diffuser outlet area than Venturi inlet area different outer shapes of the coin, e.g. smooth edges, flange etc. no throat section at all, only constriction and directly widening effect of honey combs, screening the flow structure to prevent human hazard in case of blade loss considering of turbine fixing and nacelle structure a well shaped structure in the turbine wake area ADs in the diffuser ADs in the contraction 4.00 to observe the separation behaviour in the diffuser

The next step could be to go from the decoupled coin model back to the Venturi, integrated in a building, and take the whole structure geometry and its three turbines, exposed to a wind velocity profile in urban areas, into account. Probably a vertical skewed device build in angle could catch the upwind better to prevent the separation, which was observed in horizontal arranged Venturi encased turbines[10]. At the same time noise would be directed upwards. Apropos noise, investigations in matters of noise propagation, noise guidance and damping opportunities,

depending on blade design, diffuser shape and material (anechoic structure) are desirable. Furthermore, wind tunnel tests should be carried out to verify simulation tendencies and moreover to obtain exacting values. To realise shorter diffusers with larger opening angles, boundary layer control could be applied, just as suction holes. With it integration in smaller structures could be possible, although the KISS = keep it simple system, especially at difficult maintenance places, rules. 6-4

SUMMARY

6.3 ACKNOWLEDGEMENTS
Thanks to Ris for the friendly and free work environment. The accessibility to all facilities and opportunities were the fundament of this work. Thanks a lot for the competent support onsite Juan- Jos Trujillo and thanks Herr Prof. Martin Khn for the rules, which are reasonable and guiding through the autonomous work. Special thanks for the candid backing and the recuperative motivation Christian Bak. Mange tak for the t(h)rust in me Ole Sangill and also for the close cooperation. Your project brought me to Denmark. Thanks for the qualified answers to my questions Helge Madsen Aagaard. And thank you dear Peter Bjrn Andersen.

6-5

REFERENCES
[1] DIN EN ISO 5167-3:2003; Durchflussmessung von Fluiden mit Drosselgerten in voll durchstrmten Leitungen mit Kreisquerschnitt- Teil3: Dsen und Venturidsen; DIN Deutsches Institut fr Normung e. V., Berlin, Deutschland DIN EN ISO 5167-1:2003; Durchflussmessung von Fluiden mit Drosselgerten in voll durchstrmten Leitungen mit Kreisquerschnitt- Teil1: Allgemeine Grundlagen und Anforderungen; DIN Deutsches Institut fr Normung e. V., Berlin, Deutschland Jewel B. Barlow; William H. Rae, Jr.; Alan Pope; Low- Speed Wind Tunnel Testing; Wiley- Interscience, third edition, 1999 Yongtao Han; Shufang Han; Ling-an Xu; The practice on DN = 1400 mm Venturi tubes, Tianjin University, China; Elsevier 1998 R. A. Oman; K. M. Foreman; Cost effective diffuser augmentation of wind turbine power generators; Proc. Second Workshop on Wind Energy Conversion systems, Washington, 1975 A. Betz; Windenergie und ihre Ausnutzung durch Windmhlen; Vandenhoeck und Ruprecht, Gttingen, 1926 K. Abe; M. Nishida; A. Sakurai; Y. Ohya; H. Kihara; E. Wada; K. Sato; Experimental and numerical investigations of flow fields behind a small turbine with flanged diffuser; Journal of Wind Engineering and Industrial Aerodynamics 93, 2005 Toshio Matsusima, Shinya Takagi, Seiichi Muroyama; Characteristics of a highly efficient propeller type small wind turbine with a diffuser; Renewable Energy 31, 2006 Gerald J. W. Van Bussel; The science of making more torque from wind: Diffuser experiments and theory revisited; Journal of Physics: Conference Series 75; 2007 S.J. Watson, D. G. Infield, J. P. Barton, S. J. Wylie; Modelling of the performance of a building-mounted ducted wind turbine; Journal of Physics: Conference Series 75, 2007 M. A. Heath; J. D. Walshe; Estimating the potential yield of small building- mounted wind turbines; John Wiley & Sons; 2007 www.enflo-windtec.ch www.energieprojecten.nl www.loopwing.co.jp www.aeroliftpatent.com www.peswiki.com/index.php/Directory:O%27Conner_Hush_Turbine A

[2]

[3] [4] [5]

[6] [7]

[8] [9] [10] [11] [12] [13] [14] [15] [16]

REFERENCES

[17] [18] [19] [20] [21] [22] [23] [24] [25] [26] [27] [28] [29] [30] [31] [32] [33] [34]

www.krystal-planet.com www.avinc.com www.aerotecture.com www.architecture.myninjaplease.com www.dac.dk www.elephantandcastle.org.uk www.brighton.ac.uk M. Khn; Windenergienutzung I; lecture notes, Universitt Stuttgart, 2006 www.en.wikipedia.org/wiki/Giovanni_Battista_Venturi www.centennialofflight.gov Robert W. Fox; Alan T. McDonald; Introduction to Fluid Mechanics; John Wiley & Sons, Inc.; fifth edition www.me.queensu.ca/people/sellens/teaching/fluids/power_law.php J. B. Barlow; W. H. Jr. Rae; A. Pope; Low-Speed Wind Tunnel Testing; John Wiley & Sons, Inc., third edition, 1999 www.iag.uni-stuttgart.de/laminarwindkanal/modellwindkanal FIDAP 8- Volume 10: FIDAP Theory Manual; Fluent, Incorporated; 1998 FIDAP 8- Volume 1: FIDAP User`s Manual; Fluent, Incorporated; 1998 S. Wagner; Vorlesung Strmungslehre; lecture notes, Universitt Stuttgart, 2006 Martin O. L. Hansen; Aerodynamics of Wind Turbines; James & James (Science Publishers) Ltd, London; 2001

APPENDIX
A. FI-GEN READ in file, geometry and mesh Co2.25

Tab. 9: FI-GEN READ in file to generate the geometry, computational domain and mesh; Co2.25
FI-GEN ELEM = 1, POIN = 1, CURV = 1, SURF = 1, NODE = 0, MEDG = 1, MLOO = 1, MFAC = 1, BEDG = 1, SPAV = 1, MSHE = 1, MSOL = 1, COOR = 1, TOLE = 1e-005 ) WINDOW(CHANGE= 1, MATRIX ) 1.000000 0.000000 0.000000 0.000000 0.000000 1.000000 0.000000 0.000000 0.000000 0.000000 1.000000 0.000000 0.000000 0.000000 0.000000 1.000000 -10.00000 10.00000 -7.50000 7.50000 -7.50000 7.50000 PGRID( LINE ) PGRID( COLO = 13 ) PGRID( ON ) POINT( ADD, COOR, X = 0, Y = 4.5 ) POINT( ADD, COOR, X = 0.187564, Y = 3.8 ) POINT( ADD, COOR, X = 1.573205, Y = 3 ) POINT( ADD, COOR, X = 3.373205, Y = 3 ) POINT( ADD, COOR, X = 5.773205, Y = 3 ) POINT( ADD, COOR, X = 16.3127543, Y = 4.5 ) POINT( ADD, COOR, X = -0.009, Y = 4.494804 ) POINT( ADD, COOR, X = 0.181564, Y = 3.79400 ) POINT( ADD, COOR, X = 1.573205, Y = 2.9925 ) POINT( ADD, COOR, X = 5.773025, Y = 2.9925 ) POINT( ADD, COOR, X = 16.318754, Y = 4.496536 ) POINT( ADD, COOR, X = -0.209, Y = 4.37933379 ) POINT( ADD, COOR, X = 1.573205, Y = 2.7925 ) POINT( ADD, COOR, X = 5.773205, Y = 2.7925 ) POINT( ADD, COOR, X = 16.51875434, Y = 4.38106584 ) POINT( ADD, COOR, X = -7.79422863, Y = 0 ) POINT( ADD, COOR, X = 1.573205, Y = 0 ) POINT( ADD, COOR, X = 5.773205, Y = 0 ) POINT( ADD, COOR, X = 24.10698297, Y = 0 ) POINT( ADD, COOR, X = 1.4, Y = 5.8 ) POINT( ADD, COOR, X = 1.8244998, Y = 6.2 ) POINT( ADD, COOR, X = 2.227332, Y = 4.72442 ) POINT( ADD, COOR, X = 0, Y = 6 ) POINT( ADD, COOR, X = 16.312754, Y = 6 ) POINT( ADD, COOR, X = -0.004800, Y = 6.0048 ) POINT( ADD, COOR, X = 16.324754, Y = 6.012 ) POINT( ADD, COOR, X = -0.2048, Y = 6.2048 ) POINT( ADD, COOR, X = 16.524754, Y = 6.212 ) POINT( ADD, COOR, X = -85, Y = 85 ) POINT( ADD, COOR, X = 160, Y = 85 ) POINT( ADD, COOR, X = -85, Y = 4.37933379 ) POINT( ADD, COOR, X = 160, Y = 4.381066 ) POINT( ADD, COOR, X = -85, Y = 0 ) POINT( ADD, COOR, X = 160, Y = 0 ) POINT( ADD, COOR, X = 2.50809, Y = 4.911978 ) POINT( ADD, COOR, X = 1.723205, Y = 3 ) POINT( ADD, COOR, X = 1.723205, Y = 2.9925 ) POINT( ADD, COOR, X = 1.723205, Y = 2.7925 ) POINT( ADD, COOR, X = 1.723205, Y = 0 ) CURVE( SELE, ID ) 8 POINT( SELE, ID ) 7 CURVE( SPLIT ) POINT( ADD, COOR, X = 3.373205, Y = 2.9925 ) POINT( ADD, COOR, X = 3.373205, Y = 2.7925 ) POINT( ADD, COOR, X = 3.523205, Y = 2.7925 ) POINT( ADD, COOR, X = 3.523205, Y = 0 ) POINT( ADD, COOR, X = 5.623205, Y = 3 ) POINT( ADD, COOR, X = 5.623205, Y = 2.9925 ) POINT( ADD, COOR, X = 5.623205, Y = 2.7925 ) POINT( ADD, COOR, X = 5.623205, Y = 0 ) POINT( SELE, ID ) 2 1 20 CURVE( ADD, CIRC ) CURVE( SELE, ID ) 1 POINT( SELE, ID ) 1 CURVE( SPLIT ) CURVE( SELE, ID ) 3 CURVE( DELE ) POINT( SELE, ID ) 3 2 21 CURVE( ADD, CIRC ) CURVE( SELE, ID ) 3 POINT( SELE, ID ) 2 CURVE( SPLIT ) CURVE( SELE, ID ) 5 CURVE( DELE ) CURVE( SELE, ID ) 2 4 CURVE( JOIN ) POINT( SELE, ID ) 3 5 CURVE( ADD, LINE ) POINT( SELE, ID ) 5 6 CURVE( ADD, LINE ) POINT( SELE, ID ) 8 7 35 CURVE( ADD, CIRC )

APPENDIX

CURVE( SELE, ID ) 10 CURVE( DELE ) POINT( SELE, ID ) 9 8 21 CURVE( ADD, CIRC ) CURVE( SELE,ID ) 10 POINT( SELE, ID ) 8 CURVE( SPLIT ) CURVE( SELE, ID ) 12 CURVE( DELE ) CURVE( SELE, ID ) 9 11 CURVE( JOIN ) POINT( SELE, ID ) 9 10 CURVE( ADD, LINE) POINT( SELE, ID ) 10 11 CURVE( ADD, LINE ) POINT( SELE, ID ) 13 12 21 CURVE( ADD, CIRC ) CURVE( SELE, ID ) 15 POINT( SELE, ID ) 12 CURVE( SPLIT ) CURVE( SELE, ID ) 17 CURVE( DELE ) POINT( SELE, ID ) 13 14 CURVE( ADD, LINE ) POINT( SELE, ID ) 14 15 CURVE( ADD, LINE ) POINT( SELE, ID ) 16 17 CURVE( ADD, LINE ) POINT( SELE, ID ) 17 18 CURVE( ADD, LINE ) POINT( SELE, ID ) 18 19 CURVE( ADD, LINE ) POINT( SELE, ID ) 1 7 CURVE( ADD, LINE )

POINT( SELE, ID ) 7 12 CURVE( ADD, LINE ) POINT( SELE, ID ) 12 16 CURVE( ADD, LINE ) POINT( SELE, ID ) 3 9 CURVE( ADD, LINE ) POINT( SELE, ID ) 9 13 CURVE( ADD, LINE ) POINT( SELE, ID ) 13 17 CURVE( ADD, LINE ) POINT( SELE, ID ) 5 10 CURVE( ADD, LINE ) POINT( SELE, ID ) 10 14 CURVE( ADD, LINE ) POINT( SELE, ID ) 14 18 CURVE( ADD, LINE ) POINT( SELE, ID ) 6 11 CURVE( ADD, LINE ) POINT( SELE, ID ) 11 15 CURVE( ADD, LINE ) POINT( SELE, ID ) 15 19 CURVE( ADD, LINE ) POINT( SELE, ID ) 33 16 CURVE( ADD, LINE ) POINT( SELE, ID ) 19 34 CURVE( ADD, LINE ) POINT( SELE, ID ) 12 31 CURVE( ADD, LINE ) POINT( SELE, ID ) 15 32 CURVE( ADD, LINE )

POINT( SELE, ID ) 23 25 CURVE( ADD, LINE ) POINT( SELE, ID ) 25 27 CURVE( ADD, LINE ) POINT( SELE, ID ) 27 29 CURVE( ADD, LINE ) POINT( SELE, ID ) 24 26 CURVE( ADD, LINE ) POINT( SELE, ID ) 26 28 CURVE( ADD, LINE ) POINT( SELE, ID ) 28 30 CURVE( ADD, LINE ) POINT( SELE, ID ) 29 30 CURVE( ADD, LINE ) POINT( SELE, ID ) 27 28 CURVE( ADD, LINE ) POINT( SELE, ID ) 25 26 CURVE( ADD, LINE ) POINT( SELE, ID ) 23 24 CURVE( ADD, LINE ) POINT( SELE, ID ) 27 12 CURVE( ADD, LINE ) POINT( SELE, ID ) 25 7 CURVE( ADD, LINE ) POINT( SELE, ID ) 23 1 CURVE( ADD, LINE ) POINT( SELE, ID ) 24 6 CURVE( ADD, LINE ) POINT( SELE, ID ) 26 11 CURVE( ADD, LINE )

APPENDIX
POINT( SELE, ID ) 28 15 CURVE( ADD, LINE ) POINT( SELE, ID ) 29 31 CURVE( ADD, LINE ) POINT( SELE, ID ) 31 33 CURVE( ADD, LINE ) POINT( SELE, ID ) 30 32 CURVE( ADD, LINE ) POINT( SELE, ID ) 32 34 CURVE( ADD, LINE ) CURVE( SELE, ID) 6 POINT( SELE, ID) 36 CURVE( SPLIT ) CURVE( SELE, ID ) 59 POINT( SELE, ID ) 4 CURVE( SPLIT ) CURVE( SELE, ID ) 61 POINT( SELE, ID ) 43 CURVE( SPLIT ) CURVE( SELE, ID ) 63 POINT( SELE, ID ) 47 CURVE( SPLIT ) CURVE( SELE, ID ) 13 POINT( SELE, ID) 37 CURVE( SPLIT ) CURVE( SELE, ID ) 67 POINT( SELE, ID ) 40 CURVE( SPLIT ) CURVE( SELE, ID ) 69 POINT( SELE, ID ) 44 CURVE( SPLIT ) CURVE( SELE, ID ) 71 POINT( SELE, ID ) 48 CURVE( SPLIT ) CURVE( SELE, ID ) 17 POINT( SELE, ID ) 38 CURVE( SPLIT ) CURVE( SELE, ID ) 75 POINT( SELE, ID ) 41 CURVE( SPLIT ) CURVE( SELE, ID ) 77 POINT( SELE, ID ) 45 CURVE( SPLIT ) CURVE( SELE, ID ) 79 POINT( SELE, ID ) 49 CURVE( SPLIT ) CURVE( SELE, ID ) 20 POINT( SELE, ID ) 39 CURVE( SPLIT ) CURVE( SELE, ID ) 83 POINT( SELE, ID ) 42 CURVE( SPLIT ) CURVE( SELE, ID ) 85 POINT( SELE, ID ) 46 CURVE( SPLIT ) CURVE( SELE, ID ) 87 POINT( SELE, ID ) 50 CURVE( SPLIT ) POINT( SELE, ID ) 36 37 CURVE( ADD, LINE ) POINT( SELE, ID ) 37 38 CURVE( ADD, LINE ) POINT( SELE, ID ) 38 39 CURVE( ADD, LINE ) POINT( SELE, ID ) 4 40 CURVE( ADD, LINE ) POINT( SELE, ID ) 40 41 CURVE( ADD, LINE ) POINT( SELE, ID ) 41 42 CURVE( ADD, LINE ) POINT( SELE, ID ) 43 44 CURVE( ADD, LINE ) POINT( SELE, ID ) 44 45 CURVE( ADD, LINE ) POINT( SELE, ID ) 45 46 CURVE( ADD, LINE ) POINT( SELE, ID ) 47 48 CURVE( ADD, LINE ) POINT( SELE, ID ) 48 49 CURVE( ADD, LINE ) POINT( SELE, ID ) 49 50 CURVE( ADD, LINE ) CURVE( SELE, ID ) 5 25 12 22 MFACE( WIREFRAME, EDG1CNT=1, EDG2CNT=1, EDG3CNT=1, EDG4CNT=1 ) CURVE( SELE, ID ) 58 90 66 25 MFACE( WIREFRAME, EDG1CNT=1, EDG2CNT=1, EDG3CNT=1, EDG4CNT=1 ) CURVE( SELE, ID ) 7 31 14 28 MFACE( WIREFRAME, EDG1CNT=1, EDG2CNT=1, EDG3CNT=1, EDG4CNT=1 ) CURVE( SELE, ID ) 12 26 16 23 MFACE( WIREFRAME, EDG1CNT=1, EDG2CNT=1, EDG3CNT=1, EDG4CNT=1 )

APPENDIX

CURVE( SELE, ID ) 66 91 74 26 MFACE( WIREFRAME, EDG1CNT=1, EDG2CNT=1, EDG3CNT=1, EDG4CNT=1 ) CURVE( SELE, ID ) 14 32 18 29 MFACE( WIREFRAME, EDG1CNT=1, EDG2CNT=1, EDG3CNT=1, EDG4CNT=1 ) CURVE( SELE, ID ) 16 27 19 24 MFACE( WIREFRAME, EDG1CNT=1, EDG2CNT=1, EDG3CNT=1, EDG4CNT=1 ) CURVE( SELE, ID ) 74 92 82 27 MFACE( WIREFRAME, EDG1CNT=1, EDG2CNT=1, EDG3CNT=1, EDG4CNT=1 ) CURVE( SELE, ID ) 18 33 21 30 MFACE( WIREFRAME, EDG1CNT=1, EDG2CNT=1, EDG3CNT=1, EDG4CNT=1 ) CURVE( SELE, ID ) 50 22 49 38 MFACE( WIREFRAME, EDG1CNT=1, EDG2CNT=1, EDG3CNT=1, EDG4CNT=1 ) CURVE( SELE, ID ) 47 38 46 41 MFACE( WIREFRAME, EDG1CNT=1, EDG2CNT=1, EDG3CNT=1, EDG4CNT=1 ) CURVE( SELE, ID ) 51 41 52 31 MFACE( WIREFRAME, EDG1CNT=1, EDG2CNT=1, EDG3CNT=1, EDG4CNT=1 ) CURVE( SELE, ID ) 49 23 48 39 MFACE( WIREFRAME, EDG1CNT=1, EDG2CNT=1, EDG3CNT=1, EDG4CNT=1 )

CURVE( SELE, ID ) 46 39 45 42 MFACE( WIREFRAME, EDG1CNT=1, EDG2CNT=1, EDG3CNT=1, EDG4CNT=1 ) CURVE( SELE, ID ) 52 42 53 32 MFACE( WIREFRAME, EDG1CNT=1, EDG2CNT=1, EDG3CNT=1, EDG4CNT=1 ) CURVE( SELE, ID ) 24 34 55 36 MFACE( WIREFRAME, EDG1CNT=1, EDG2CNT=1, EDG3CNT=1, EDG4CNT=1 ) CURVE( SELE, ID ) 48 36 54 40 MFACE( WIREFRAME, EDG1CNT=1, EDG2CNT=1, EDG3CNT=1, EDG4CNT=1 ) CURVE( SELE, ID ) 45 40 44 43 MFACE( WIREFRAME, EDG1CNT=1, EDG2CNT=1, EDG3CNT=1, EDG4CNT=1 ) CURVE( SELE, ID ) 53 43 56 37 MFACE( WIREFRAME, EDG1CNT=1, EDG2CNT=1, EDG3CNT=1, EDG4CNT=1 ) CURVE( SELE, ID ) 33 37 57 35 MFACE( WIREFRAME, EDG1CNT=1, EDG2CNT=1, EDG3CNT=1, EDG4CNT=1 ) CURVE( SELE, ID ) 60 93 68 90 MFACE( WIREFRAME, EDG1CNT=1, EDG2CNT=1, EDG3CNT=1, EDG4CNT=1 )

CURVE( SELE, ID ) 62 96 70 93 MFACE( WIREFRAME, EDG1CNT=1, EDG2CNT=1, EDG3CNT=1, EDG4CNT=1 ) CURVE( SELE, ID ) 64 99 72 96 MFACE( WIREFRAME, EDG1CNT=1, EDG2CNT=1, EDG3CNT=1, EDG4CNT=1 ) CURVE( SELE, ID ) 65 28 73 99 MFACE( WIREFRAME, EDG1CNT=1, EDG2CNT=1, EDG3CNT=1, EDG4CNT=1 ) CURVE( SELE, ID ) 68 94 76 91 MFACE( WIREFRAME, EDG1CNT=1, EDG2CNT=1, EDG3CNT=1, EDG4CNT=1 ) CURVE( SELE, ID ) 70 97 78 94 MFACE( WIREFRAME, EDG1CNT=1, EDG2CNT=1, EDG3CNT=1, EDG4CNT=1 ) CURVE( SELE, ID ) 72 100 80 97 MFACE( WIREFRAME, EDG1CNT=1, EDG2CNT=1, EDG3CNT=1, EDG4CNT=1 ) CURVE( SELE, ID ) 73 29 81 100 MFACE( WIREFRAME, EDG1CNT=1, EDG2CNT=1, EDG3CNT=1, EDG4CNT=1 ) CURVE( SELE, ID ) 76 95 84 92 MFACE( WIREFRAME, EDG1CNT=1, EDG2CNT=1, EDG3CNT=1, EDG4CNT=1 )

APPENDIX
CURVE( SELE, ID ) 78 98 86 95 MFACE( WIREFRAME, EDG1CNT=1, EDG2CNT=1, EDG3CNT=1, EDG4CNT=1 ) CURVE( SELE, ID ) 80 101 88 98 MFACE( WIREFRAME, EDG1CNT=1, EDG2CNT=1, EDG3CNT=1, EDG4CNT=1 ) CURVE( SELE, ID ) 81 30 89 101 MFACE( WIREFRAME, EDG1CNT=1, EDG2CNT=1, EDG3CNT=1, EDG4CNT=1 ) CURVE( SELE, ID ) 22 25 28 31 41 38 MEDGE( ADD, SUCC, INTE=1, RATI = 0.000000, 2RATI=0.000000, PCENTR=0.000000 ) CURVE( SELE, ID ) 23 26 29 32 42 39 MEDGE( ADD, SUCC, INTE=20, RATI = 1.100000, 2RATI=0.000000, PCENTR=0.000000 ) CURVE( SELE, ID ) 5 12 MEDGE( ADD, REVERSE, SUCC, INTE=80, RATI = 1.00000, 2RATI=0.000000, PCENTR=0.000000 ) CURVE( SELE, ID ) 16 MEDGE( ADD, SUCC, INTE=80, RATI = 1.00000, 2RATI=0.000000, PCENTR=0.000000 ) CURVE( SELE, ID ) 58 66 74 MEDGE( ADD, SUCC, INTE=3, RATI = 0.000000, 2RATI=0.000000, PCENTR=0.000000 ) CURVE( SELE, ID ) 7 14 18 MEDGE( ADD, SUCC, INTE=230, RATI = 0.000000, 2RATI=0.000000, PCENTR=0.000000 ) CURVE( SELE, ID ) 51 52 53 MEDGE( ADD, LSTFIRST, INTE=30, RATI = 3.000000, 2RATI=3.000000, PCENTR=0.500000 ) CURVE( SELE, ID ) 47 46 45 MEDGE( ADD, LSTFIRST, INTE=100, RATI = 3.000000, 2RATI=3.000000, PCENTR=0.500000 ) CURVE( SELE, ID ) 50 49 48 MEDGE( ADD, LSTFIRST, INTE=30, RATI = 3.000000, 2RATI=3.000000, PCENTR=0.500000 ) CURVE( SELE, ID ) 19 MEDGE( ADD, REVE, SUCC, INTE=80, RATI = 1.0100000, 2RATI=0.000000, PCENTR=0.000000 ) CURVE( SELE, ID ) 82 MEDGE( ADD, SUCC, INTE=3, RATI = 0.000000, 2RATI=0.000000, PCENTR=0.000000 ) CURVE( SELE, ID ) 21 MEDGE( ADD, SUCC, INTE=230, RATI = 0.000000, 2RATI=0.000000, PCENTR=0.000000 ) CURVE( SELE, ID ) 56 MEDGE( ADD, LSTFIRST, INTE=30, RATI = 3.000000, 2RATI=3.000000, PCENTR=0.500000 ) CURVE( SELE, ID ) 44 MEDGE( ADD, LSTFIRST, INTE=100, RATI = 3.000000, 2RATI=3.000000, PCENTR=0.500000 ) CURVE( SELE, ID ) 54 MEDGE( ADD, LSTFIRST, INTE=30, RATI = 3.000000, 2RATI=3.000000, PCENTR=0.500000 ) CURVE( SELE, ID ) 55 24 MEDGE( ADD, SUCC, INTE=30, RATI = 1.120000, 2RATI=0.000000, PCENTR=0.000000 ) CURVE( SELE, ID ) 27 30 MEDGE( ADD, SUCC, INTE=30, RATI = 1.060000, 2RATI=0.000000, PCENTR=0.000000 ) CURVE( SELE, ID ) 33 57 MEDGE( ADD, SUCC, INTE=30, RATI = 1.110000, 2RATI=0.000000, PCENTR=0.000000 ) CURVE( SELE, ID ) 34 MEDGE( ADD, REVERSE, SUCC, INTE=100, RATI = 1.030000, 2RATI=0.000000, PCENTR=0.000000 ) CURVE( SELE, ID ) 36 40 MEDGE( ADD, SUCC, INTE=100, RATI = 1.0480000, 2RATI=0.000000, PCENTR=0.000000 ) CURVE( SELE, ID ) 43 37 35 MEDGE( ADD, SUCC, INTE=100, RATI = 1.0520000, 2RATI=0.000000, PCENTR=0.000000 ) CURVE( SELE, ID ) 60 68 76 84 MEDGE( ADD, SUCC, INTE=40, RATI = 0.000000, 2RATI=0.000000, PCENTR=0.000000 ) CURVE( SELE, ID ) 62 70 78 86 MEDGE( ADD, SUCC, INTE=3, RATI = 0.000000, 2RATI=0.000000, PCENTR=0.000000 ) CURVE( SELE, ID ) 64 72 80 88 MEDGE( ADD, SUCC, INTE=50, RATI = 0.000000, 2RATI=0.000000, PCENTR=0.000000 ) CURVE( SELE, ID ) 65 73 81 89 MEDGE( ADD, SUCC, INTE=3, RATI = 0.000000, 2RATI=0.000000, PCENTR=0.000000 ) CURVE( SELE, ID ) 90 93 96 99 MEDGE( ADD, SUCC, INTE=1, RATI = 0.000000, 2RATI=0.000000, PCENTR=0.000000 )

APPENDIX
CURVE( SELE, ID ) 91 94 97 100 MEDGE( ADD, SUCC, INTE=20, RATI = 1.100000, 2RATI=0.000000, PCENTR=0.000000 ) CURVE( SELE, ID ) 92 95 98 101 MEDGE( ADD, SUCC, INTE=30, RATI = 1.060000, 2RATI=0.000000, PCENTR=0.000000 ) MFACE(SELE, ID ) 1 MFACE( MESH, MAP, ENTITY="y+1" ) MFACE(SELE, ID ) 2 MFACE( MESH, MAP, ENTITY="AD+1" ) MFACE(SELE, ID ) 21 MFACE( MESH, MAP, ENTITY="T+1" ) MFACE(SELE, ID ) 22 MFACE( MESH, MAP, ENTITY="AD+2" ) MFACE(SELE, ID ) 23 MFACE( MESH, MAP, ENTITY="T+2" ) MFACE(SELE, ID ) 24 MFACE( MESH, MAP, ENTITY="AD+3" ) MFACE(SELE, ID ) 3 MFACE( MESH, MAP, ENTITY="y+2" ) MFACE(SELE, ID ) 4 MFACE( MESH, MAP, ENTITY="trans1" ) MFACE(SELE, ID ) 5 MFACE( MESH, MAP, ENTITY="ADtra1" ) MFACE(SELE, ID ) 25 MFACE( MESH, MAP, ENTITY="Ttra1" ) MFACE(SELE, ID ) 26 MFACE( MESH, MAP, ENTITY="ADtra2" ) MFACE(SELE, ID ) 27 MFACE( MESH, MAP, ENTITY="Ttra2" ) MFACE(SELE, ID ) 28 MFACE( MESH, MAP, ENTITY="ADtra3" ) MFACE(SELE, ID ) 6 MFACE( MESH, MAP, ENTITY="trans2" ) MFACE(SELE, ID ) 7 MFACE( MESH, MAP, ENTITY="core1" ) MFACE(SELE, ID ) 8 MFACE( MESH, MAP, ENTITY="ADco1" ) MFACE(SELE, ID ) 29 MFACE( MESH, MAP, ENTITY="Tco1" ) MFACE(SELE, ID ) 30 MFACE( MESH, MAP, ENTITY="ADco2" ) MFACE(SELE, ID ) 31 MFACE( MESH, MAP, ENTITY="Tco2" ) MFACE(SELE, ID ) 32 MFACE( MESH, MAP, ENTITY="ADco3" ) MFACE(SELE, ID ) 9 MFACE( MESH, MAP, ENTITY="core2" ) MFACE(SELE, ID ) 10 MFACE( MESH, MAP, ENTITY="y+3" ) MFACE(SELE, ID ) 11 MFACE( MESH, MAP, ENTITY="y+4" ) MFACE(SELE, ID ) 12 MFACE( MESH, MAP, ENTITY="y+5" ) MFACE(SELE, ID ) 13 MFACE( MESH, MAP, ENTITY="trans3" ) MFACE(SELE, ID ) 14 MFACE( MESH, MAP, ENTITY="trans4" ) MFACE(SELE, ID ) 15 MFACE( MESH, MAP, ENTITY="trans5" ) MFACE(SELE, ID ) 16 MFACE( MESH, MAP, ENTITY="inlet" ) MFACE(SELE, ID ) 17 MFACE( MESH, MAP, ENTITY="front" ) MFACE(SELE, ID ) 18 MFACE( MESH, MAP, ENTITY="top" ) MFACE(SELE, ID ) 19 MFACE( MESH, MAP, ENTITY="rear" ) MFACE(SELE, ID ) 20 MFACE( MESH, MAP, ENTITY="outlet" ) MEDGE( SELE, ID ) 43 31 32 52 56 60 64 33 48 MEDGE( MESH, MAP, ENTITY="axis" ) MEDGE( SELE, ID ) 37 36 MEDGE( MESH, MAP, ENTITY="inflow" ) MEDGE( SELE, ID ) 42 34 MEDGE( MESH, MAP, ENTITY="outflow" ) MEDGE( SELE, ID ) 35 MEDGE( MESH, MAP, ENTITY="noflow" ) MEDGE( SELE, ID ) 13 MEDGE( MESH, MAP, ENTITY="wall1" ) MEDGE( SELE, ID ) 16 49 53 57 61 MEDGE( MESH, MAP, ENTITY="wall2" ) MEDGE( SELE, ID ) 19 MEDGE( MESH, MAP, ENTITY="wall3" ) MEDGE( SELE, ID ) 22 MEDGE( MESH, MAP, ENTITY="wall4" ) MEDGE( SELE, ID ) 25 MEDGE( MESH, MAP, ENTITY="wall5" ) MEDGE( SELE, ID ) 28 MEDGE( MESH, MAP, ENTITY="wall6" ) MEDGE( SELE, ID ) 2 8 39 MEDGE( MESH, MAP, ENTITY="mass" ) END( )

APPENDIX
B. FI-PREP READ in file, Co2.25pos1, CT = 0.5314410

Tab. 10: FI-PREP READ in file to set fluid properties, boundary conditions, implement AD; Co2.25 pos1
FIPREP PROBLEM( NONLINEAR, TURB, AXI-SYMMETRIC ) EXECUTION( NEWJOB ) DENSITY( CONSTANT=1.225 ) ICNODE( KINETIC, CONST=0.003, ALL ) ICNODE( DISSIPATION, CONST=0.00045, ALL ) VISCOSITY( TWO-EQUATION, CONST=1.8375e-5 ) EDDYVISCOSITY(BOUSSINESQ) TURBOPTION(RNG) SOLUTION( SEGR=400, VELC=1e-005,RESC=1e-005 ) OPTIONS( UPWINDING ) DATAPRINT( CONTROL ) ENTITY( FLUID, NAME="y+1" ) ENTITY( FLUID, NAME="trans1" ) ENTITY( FLUID, NAME="AD+1" ) ENTITY( FLUID, NAME="ADtra1" ) ENTITY( FLUID, NAME="T+1" ) ENTITY( FLUID, NAME="Ttra1" ) ENTITY( FLUID, NAME="AD+2" ) ENTITY( FLUID, NAME="ADtra2" ) ENTITY( FLUID, NAME="T+2" ) ENTITY( FLUID, NAME="Ttra2" ) ENTITY( FLUID, NAME="AD+3" ) ENTITY( FLUID, NAME="ADtra3" ) ENTITY( FLUID, NAME="y+2" ) ENTITY( FLUID, NAME="trans2" ) ENTITY( FLUID, NAME="y+3" ) ENTITY( FLUID, NAME="trans3" ) ENTITY( FLUID, NAME="y+4" ) ENTITY( FLUID, NAME="trans4" ) ENTITY( FLUID, NAME="y+5" ) ENTITY( FLUID, NAME="trans5" ) ENTITY( FLUID, NAME="core1" ) ENTITY( FLUID, NAME="core2" ) ENTITY( FLUID, NAME="ADco1" ) ENTITY( FLUID, NAME="Tco1" ) ENTITY( FLUID, NAME="ADco2" ) ENTITY( FLUID, NAME="Tco2" ) ENTITY( FLUID, NAME="ADco3" ) ENTITY( FLUID, NAME="inlet" ) ENTITY( FLUID, NAME="front" ) ENTITY( FLUID, NAME="top" ) ENTITY( FLUID, NAME="rear" ) ENTITY( FLUID, NAME="outlet" ) ENTITY( PLOT, NAME="inflow" ) ENTITY( PLOT, NAME="outflow" ) ENTITY( PLOT, NAME="axis" ) ENTITY( PLOT, NAME="noflow" ) ENTITY( PLOT, NAME="mass" ) BODYFORCE( ADD, CONSTANT, FZC=-63.77292, FRC=0., FTHETA=0., ENTITY="AD+1" ) BODYFORCE( ADD, CONSTANT, FZC=-63.77292, FRC=0., FTHETA=0., ENTITY="ADtra1" ) BODYFORCE( ADD, CONSTANT, FZC=-63.77292, FRC=0., FTHETA=0., ENTITY="ADco1" ) RENUMBER END ( ) ENTITY( WALL, NAME="wall1" ) ENTITY( WALL, NAME="wall2" ) ENTITY( WALL, NAME="wall3" ) ENTITY( WALL, NAME="wall4" ) ENTITY( WALL, NAME="wall5" ) ENTITY( WALL, NAME="wall6" ) BCNODE( URC, ZERO, ENTITY="inflow" ) BCNODE( UZC, CONS=6, ENTITY="inflow" ) BCNODE( KINETIC, CONST=0.003, ENTITY="inflow" ) BCNODE( DISSIPATION, CONST=0.00045, ENTITY="inflow" ) BCNODE( URC, ZERO, ENTITY="axis" ) BCNODE( VELOCITY, ZERO, ENTITY="wall1") BCNODE( VELOCITY, ZERO, ENTITY="wall2") BCNODE( VELOCITY, ZERO, ENTITY="wall3") BCNODE( VELOCITY, ZERO, ENTITY="wall4") BCNODE( VELOCITY, ZERO, ENTITY="wall5") BCNODE( VELOCITY, ZERO, ENTITY="wall6")

APPENDIX
C. TECHNICAL DRAWINGS: Co2.25; Co4.00; Co9.00

Fig. 112: Technical model drawings: first Co2.25; second Co4.00; third Co9.00

APPENDIX

queen of nothing there is nothing to lose, there is nothing to care, besides my lust for life i