An Investigation of River Kinetic Turbines: Performance Enhancements, Turbine Modelling Techniques, and a Critical Assessment of Turbulence Models by David.

An Investigation of River Kinetic An Investigation of River Kinetic Turbines: Performance Turbines: Performance

Enhancements,Enhancements,Turbine Modelling Techniques, Turbine Modelling Techniques,

and and a Critical Assessment of a Critical Assessment of

Turbulence ModelsTurbulence Modelsby

David L. F. Gaden

Department of Mechanical and Manufacturing Engineering

University of Manitoba

Committee MembersCommittee Members

Dr. E. Bibeau (departmental advisor)Dr. E. Bibeau (departmental advisor) Dr. A. Gole (Electrical Engineering)Dr. A. Gole (Electrical Engineering) Tom Molinski (Manitoba Hydro)Tom Molinski (Manitoba Hydro) Dr. S. Ormiston (Mechanical Dr. S. Ormiston (Mechanical

Engineering)Engineering)

External ReviewerExternal Reviewer Mr. P. Vauthier (UEK)Mr. P. Vauthier (UEK)

OutlineOutline

IntroductionIntroduction Technology overviewTechnology overview Recent kinetic hydro developmentsRecent kinetic hydro developments Wind energy literature reviewWind energy literature review

Shroud OptimisationShroud Optimisation Anchor ExperimentAnchor Experiment ValidationValidation ConclusionConclusion Future StudyFuture Study

IntroductionIntroductionTechnology Overview Technology Overview

Geographic location with a natural flow restriction

Shroud (cut away)

Turbine, Huband Generator

Anchoring System

To Power Distribution


Example of a kinetic turbine

≈ 8 ft


AdvantagesAdvantages No reservoir or spillway – minimal No reservoir or spillway – minimal

environmental impactenvironmental impact Site selection far less restrictiveSite selection far less restrictive No dams or powerhouses – low cost No dams or powerhouses – low cost

installationinstallation Fast deployment timesFast deployment times Modular – easily scalable energy outputModular – easily scalable energy output Steady flow rates, steady energy Steady flow rates, steady energy

productionproduction

IntroductionIntroductionTechnology OverviewTechnology Overview

DisadvantagesDisadvantages Possibly dangerous flow conditionsPossibly dangerous flow conditions No control over upstream conditionsNo control over upstream conditions Turbulence, foreign debrisTurbulence, foreign debris Unknown fish mortality rateUnknown fish mortality rate


Little in open literature for river Little in open literature for river kinetic turbines kinetic turbines

Purpose:Purpose: To develop modelling techniques for river To develop modelling techniques for river

kinetic turbineskinetic turbines To understand the reliability of these To understand the reliability of these

modelsmodels Use these models to evaluate performance Use these models to evaluate performance

enhancements for kinetic turbinesenhancements for kinetic turbines

IntroductionIntroductionRecent kinetic hydro developments Recent kinetic hydro developments

1970

1980

1990

2000

Coriolis Program (Gulf Stream) ITDG / IT Power

(Sudan)UEK (Various)Nova Energy, NRC (3

sites)Nihon University (Japan) Scottish Nuclear, IT Power

(Scotland)Northern Territory University (Australia) Marine Current

Turbines (UK)Horizontal axis turbineVertical axis turbineDucted turbine

IntroductionIntroductionRecent kinetic hydro developmentsRecent kinetic hydro developments**

*Adapted from Segergren, 2005

1990

2000

Horizontal axis turbineVertical axis turbineDucted turbine

Ontario Power Generation, UEK (Ontario)Hammerfest Strøm AS (Norway)

Exim & Seapower (Sweden / Scotland)Hydro Venturi (Various)

TidEl Generator (Unspecified)

Stingray Tidal Stream, Eng Business Ltd.

New Energy (Alberta)

Pearson College, et al. (B.C.)

Starkraft Development (Norway)

IntroductionIntroductionRecent kinetic hydro developmentsRecent kinetic hydro developments**

*Adapted from Segergren, 2005

IntroductionIntroductionWind energy literature reviewWind energy literature review

1980

1990

2000

N x 5

N x 2

E x 3E x 1

E x 1.25E x 1.3THEORY THEORY

THEORY

THEORY

THEORY – Paper covers ducted turbine theoryN – Numerical study

x 3 – Results show a power increase by a factor of 3E – Experimental results

N x 4N x 3.2

N x 2

Igra Grassmann et al.Lewis et al.

Helmy

HelmyPhillips et al.

Bet et al.

Shroud OptimisationShroud OptimisationTheoryTheory

Conventional turbine

Small power available

Pa < 60% P∞ Betz limit (Betz, 1926)

Shroud OptimisationShroud OptimisationTheoryTheory

Shrouded turbine

Greater power available

(Lewis et al., 1977)

Shroud OptimisationShroud OptimisationTurbine ModellingTurbine Modelling

Four turbine modelling strategies:1.1. No model2.2. Momentum source3.3. Averaging rotating reference frame4.4. Sliding mesh rotating reference frame

• Does not capture pressure drop, swirl• Non-linear response to pressure not modelled• Not used

• Open passage


Four turbine modelling strategies:1.1. No model

2.2. Momentum source3.3. Averaging rotating reference frame4.4. Sliding mesh rotating reference frame

• Does not capture pressure drop, swirl• Non-linear response to pressure not modelled• Not used

• Open passage

k – Momentum source factor

• Models turbine as block of momentum

• Captures pressure drop

• Avoids complex geometry


Four turbine modelling strategies:1.1. No model2.2. Momentum source

3.3. Averaging rotating reference frame4.4. Sliding mesh rotating reference frame

k – Momentum source factor

• Models turbine as block of momentum

• Captures pressure drop

• Avoids complex geometry

0

0.1

0.2

0.3

0.4

0.5

0.6

0.7

-0.6 -0.1 0.4 0.9

Uw / U∞

P /

P∞

TheoryExperiment

• Does not account for power curves, mechanical losses

• Close to Betz theory

• ≈ 5% over-prediction of power


Four turbine modelling strategies:1.1. No model2.2. Momentum source

3.3. Averaging rotating reference frame4.4. Sliding mesh rotating reference frame

0

0.1

0.2

0.3

0.4

0.5

0.6

0.7

-0.6 -0.1 0.4 0.9

Uw / U∞

P /

P∞

TheoryExperiment

• Does not account for power curves, mechanical losses

• Close to Betz theory

• ≈ 5% over-prediction of power

• Models rotor geometry

• Averages along circumference of rotation for pseudo steady-state

• Streamwise axis-symmetric only


Four turbine modelling strategies:1.1. No model2.2. Momentum source3.3. Averaging rotating reference frame

4.4. Sliding mesh rotating reference frame

• Models rotor geometry

• Averages along circumference of rotation for pseudo steady-state

• Streamwise axis-symmetric only

• Rotates and interpolates mesh at each time step

• Computationally intensive; large output

• Fully transient solution

Shroud OptimisationShroud OptimisationMomentum SourceMomentum Source

Design variables:

1.1. Diffuser Angle


Design variables:

1.1. Diffuser Angle

2.2. Area ratio

Shroud Shroud OptimisationOptimisationMomentum SourceMomentum Source

Model dimensions

Flow domain

Surface mesh

0

10

20

30

40

50

60

0 2 4 6 8

Area ratio

Po

we

r [k

W]

0

20

40

60

80

100

120

140

160

0 2 4 6 8

Area ratio

Dra

g [

kN

]

Shroud Shroud OptimisatioOptimisatio

nnMomentum SourceMomentum Source

Variable: Area ratio

15

Variable: Angle0

10

20

30

40

50

60

0 20 40 60

Diffuser Angle [degrees]

Po

wer

[kW

]

No diffuser

0

5

10

15

20

25

30

35

0 20 40 60

Diffuser angle [°]

Dra

g [k

N]

Total dragShroud dragTurbine drag

No diffuser

■■ Power increase by a factor of 3.1

■■ Drag increase by a factor of 3.9

Shroud Shroud OptimisatioOptimisatio

nnMomentum SourceMomentum Source

Streamlines for45° diffuser

Streamlines for20° diffuser

Shroud Shroud OptimisationOptimisationMomentum SourceMomentum Source

No diffuser versus diffuser


3.8703.3612.8512.3421.324 1.8330.814-0.204 0.305-0.714

Axial ve locity [m /s]



If area is limited, shroud will reduce If area is limited, shroud will reduce turbine sizeturbine size

Shroud is still beneficialShroud is still beneficial

Output: 25.6 kW Output: 51.3 kW

Diameter: 3.0 m Diameter: 2.4 m

Shroud OptimisationShroud OptimisationRotating Reference FrameRotating Reference Frame

Tetrahedral mesh

Flow domain

Hexahedral mesh


A.A. B.B.

C.C. D.D.


100%46.4 kW

95.8%44.4 kW

84.7%39.3 kW

105.5%48.9 kW

Relative power output

(standard)A.A.B.B. C.C.D.D.


Anchor ExperimentAnchor Experiment

Boundary-layer causes power lossBoundary-layer causes power loss

U/U∞

y/δ

Velocity

P/P∞

Power

Shroud (cut away)

Turbine, Huband Generator

Anchoring System

To Power Distribution



Four anchor modelsFour anchor models

A.A. B.B.

C.C. D.D.

≈ 3 m


y / δ

P /

P∞

At 7.5 m downstream from Anchor

Anchor ExperimentAnchor ExperimentMidstream velocity contours

ValidationValidation

Particle Image Velocimetry (PIV) usedParticle Image Velocimetry (PIV) used Six experimental runs:Six experimental runs:

2 configurations (nozzle & diffuser)2 configurations (nozzle & diffuser) 3 flow speeds (0.5 m/s, 0.8 m/s and 1.0 m/s)3 flow speeds (0.5 m/s, 0.8 m/s and 1.0 m/s)

For each, four CFD simulations For each, four CFD simulations performed:performed: 2 Eddy-viscosity turbulence models (2 Eddy-viscosity turbulence models (k-k-εε & &

SST)SST) 2 Reynolds stress transport models (SSG & 2 Reynolds stress transport models (SSG &

BSL)BSL)


Laser Mirror Camera

Water tunnel test section

Model

Ruler (for alignment)


FLUID WITH SEEDING PARTICLES

CAMERA LASER AND OPTICS

DATA ACQUISITION AND CONTROL SYSTEM

COMPUTER AND SOFTWARE

TEST SECTION AND MODEL

PIV Apparatus


d1

d3

d2

Frame 1 Frame 2

Both framesRaw Image


Diffuser, 1 m/s Nozzle, 1 m/s

PIV Streamlines & velocity contours

ValidationValidationk-ε streamlines & velocity contours


ValidationValidationSSG streamlines & velocity contours


ValidationValidationk-ε velocity error


ValidationValidationSSG velocity error



■ ■ Root mean square error (RMSE) used to evaluate each model across the entire field:

Full-field validation results:


PIV Experimental errorPIV Experimental error Seeding particle density too lowSeeding particle density too low

5 particles / IA recommended (Dantec 2000)5 particles / IA recommended (Dantec 2000) ≈ ≈ 3 particles / IA3 particles / IA Velocity up to 55% under-read (Keane et al. Velocity up to 55% under-read (Keane et al.

1992)1992) Field of view too largeField of view too large

Poor handling of high velocity gradientsPoor handling of high velocity gradients 60% probability of valid detection (Keane et al. 60% probability of valid detection (Keane et al.

1992)1992) Regions with high gradients cannot be trustedRegions with high gradients cannot be trusted

ValidationValidation CFD inlet conditions inadequateCFD inlet conditions inadequate Modelled as uniform flow, but it Modelled as uniform flow, but it

was not:was not:

0

0.2

0.4

0.6

0.8

1

1.2

-0.12 -0.06 0 0.06 0.12y [m]

Up

iv /

Ucf

d

D60D45D30N60N45N30

ConclusionsConclusions

River kinetic turbines are studiedRiver kinetic turbines are studied Shroud optimisation (momentum source Shroud optimisation (momentum source

model):model): Power increase by a factor of 3.1Power increase by a factor of 3.1 Sacrificing turbine area for duct can double power Sacrificing turbine area for duct can double power

outputoutput Shroud optimisation (rotating reference frame):Shroud optimisation (rotating reference frame):

Cylindrical shroud can cause 30% power lossCylindrical shroud can cause 30% power loss Power increase of 4% with a diffuserPower increase of 4% with a diffuser Power increase of 25% comparing against shrouded Power increase of 25% comparing against shrouded

turbineturbine


Anchor experimentAnchor experiment Up to 90% power loss due to boundary Up to 90% power loss due to boundary

layerlayer Upstream flow obstruction can increase Upstream flow obstruction can increase

power availablepower available 30% power increase seen 12 meters 30% power increase seen 12 meters

downstreamdownstream Geometries designed to maximize Geometries designed to maximize

vertical disturbance were most vertical disturbance were most successfulsuccessful


ValidationValidation Full field velocity RMSE of between 21.2% to Full field velocity RMSE of between 21.2% to

47.4%47.4% PIV experimental errors:PIV experimental errors:

Low seeding particle density Low seeding particle density velocity under-read velocity under-read Small field of view Small field of view lower probability of valid lower probability of valid

detectiondetection CFD modelling errors:CFD modelling errors:

Inlet velocity assumed to be uniformInlet velocity assumed to be uniform Eddy-viscosity based turbulence models Eddy-viscosity based turbulence models

performed superior than Reynolds stress performed superior than Reynolds stress turbulence modelsturbulence models

Future StudyFuture Study Turbine rotor geometryTurbine rotor geometry Study of cavitationStudy of cavitation Mechanical and electrical lossesMechanical and electrical losses Additional shroud optimisation studyAdditional shroud optimisation study Further performance enhancements:Further performance enhancements:

Wing designWing design Inlet statorsInlet stators

Improve the shroud validation; validate Improve the shroud validation; validate the turbine modelthe turbine model

Study interactions with array installationsStudy interactions with array installations Fish mortality and damage susceptibilityFish mortality and damage susceptibility

AcknowledgmentsAcknowledgments

Dr. Eric Bibeau

Dr. A Gole

Andrea Kraj

Jeremy Langner

Manitoba Hydro

Mr. T. Molinsky

NSERC

Dr. S. Ormiston

Dr. M. Tachie

Mr. P. Vauthier

Dr. Eric Bibeau

Dr. A Gole

Andrea Kraj

Jeremy Langner

Manitoba Hydro

Mr. T. Molinsky

NSERC

Dr. S. Ormiston

Dr. M. Tachie

Mr. P. Vauthier

An Investigation of River Kinetic Turbines: Performance Enhancements, Turbine Modelling Techniques, and a Critical Assessment of Turbulence Models by David.

Documents

river kinetic turbines

vauthier uek slide

turbine modelling techniques

natural flow restriction

dangerous flow conditions

scalable energy output

open literature

upstream conditions