aerodynamic design of a high solidity canted vertical axis ... · pdf filefirst symposium on...

PR

OD

-F-0

15-0

1 Ariane FrèreEnergy & Buildings Team

Contacts:[email protected]. ref.: D4WIND-NS-010-00

First Symposium on OpenFOAMⓇ in Wind Energy20th & 21st of March 2013 - Oldenburg

Aerodynamic design of a high solidity canted Vertical Axis Wind Turbine with OpenFOAM

PR

OD

-F-0

15-0

1

© 2013 Cenaero – All rights reserved

1/ Project Introduction

OpenFOAM interest for VAWT design

Symposium on OpenFOAM in Wind Energy 21/03/2013 - Oldenburg

Implicit Large Eddy SimulationCenaero in-house tool:

Argo DGM

Dynamic polars: CL for various TSR [3]

D4Wind Project :

• Walloon Region funds

• 1 SME, 1 university & 1 research center

• Task 1: Aerodynamics rotor design

Vertical Axis Wind Turbines:

• High & varying AOAs

• Highly unsteady flow, more complex than HAWT

• Difficulty to find accurate polar (dynamic stall)

Strong interest in unsteady CFD:

• Airfoils � static & dynamic polars

• Full rotor structure � arms, attachment… effects on performance & acoustic

PR

OD

-F-0

15-0

11/ Project Introduction

Presentation of the 2D validation case


Experimental set-up for the straight (left VAWT) and canted (right) VAWT [1] (naca0015, chord 0.420m, rotor diameter 2.8m, height 2.9m)


Project steps:

• Validate CFD approach using OpenFOAMtools on 2D cases

• Compare 2D to 3D (straight blades)

• Study the canted effect

• Use developed methodology for D4Wind

Set-up:

• Meshing Gambit/ANSYS

• Transient, incompressible, unsteady flows with dynamic mesh: pimpleDyMFoam

• 2D URANS with turbulence models: k-ε & k-ω SST

• Δt as Courant number ≤ 0.9

• Naca0015, 0.420m chord, 2.8 rotor diameter, 0.1m axis diameter, no pitch

PR

OD

-F-0

15-0

1


2/ Modelling Approach

Mesh & boundary conditions

Mesh specification:

• Unstructured triangular mesh

• Specification:

• Domain: 14φ*21φ (φ=2.8m)

• Rotating disk: 4φ

• 2D mesh extruded

• Blades first cell as y+ ≤200

• Max aspect ratio 1.2

Operating conditions:

• Inlet velocity: U=10m/s

• Turbulence intensity 10%

• AMI rotation 30:140 RPM (Re 4:8e5)


Inletv=constant

dp/dy=0

Symmetry

cyclicAMI

movingWallWallFunction

Outletp=constant

dv/dy=0

Symmetry

movingWallWallFunction

Top&Bottom: empty

PR

OD

-F-0

15-0

1



Mesh dependence study

• 50k nodes mesh chosen (vs. 150k in reference [2])

• Further analysis: 220k nodes


Simulations & ref. data [2] at TSR 1.32, U=10m/s, k- ω SST (k=1.5, ω=5.6), third revolution of blade n°2

Nodes Run time(3*360°)

Cp (exp:0.23)

20k 40 CPUh 0.09

50k 420 CPUh 0.20

110k 1200 CPUh 0.20

PR

OD

-F-0

15-0

1



Turbulence sensitivity study


Simulations at TSR 1.2, U=10m/s, 50knodes mesh, third revolution of blade n°2

• k- ω SST chosen for the present study (as in reference [2])

• Turbulence level doesn’t alter much the results

PR

OD

-F-0

15-0

1


3/ Result & Discussion

Vorticity TSR 1.32

TSR 1.32, U=10m/s, k- ω SST (k=1.5, ω=5.6) 50k nodes mesh

PR

OD

-F-0

15-0

1


3/ Result & Discussion

Vorticity TSR 0.4

TSR 0.4, U=10m/s, k- ω SST (k=1.5, ω=5.6) 50k nodes mesh

PR

OD

-F-0

15-0

1


4/ Results & Discussion

Thrust force coefficient variation with θ for 4 TSR


TSR 0.4

TSR 1.32

TSR 0.7

TSR 1.81

Simulations at U=10m/s, k- ω SST (k=1.5, ω=5.6), third revolution of blade n°2, 50k nodes mesh

PR

OD

-F-0

15-0

1


4/ Results & Discussion

Cp curve compared to reference numerical & experimental results


Red dots=openFoam results with U=10m/s, k- ω SST (k=1.5, ω=5.6) 50k nodes mesh: TSR=0.4 Cp=0.03,

TSR=0.7 Cp=0.05, TSR=1.2 Cp= 0.16, TSR=1.32 Cp=0.20, TSR=1.81 Cp=0.29

Fig

ure

6.11

from

[2]

v

v

• 2D analysis shows good agreement with the reference

• Further analysis required in [1.5:2] TSR region

v

PR

OD

-F-0

15-0

1

Conclusions:

• 2D comparison to benchmark shows very encouraging results

• Mesh 50k converges in reasonable time

• k-ω SST well adapted for VAWT highly separated flow

Perspectives:

• Mesh dependency & numerical schemes need to be further analysed

• Methodology to reduce computational time for 3D cases:

• Change mesh: tetrahedral to hexahedral

• Improve parallelisation approach

• Steps with different numerical schemes

• … ?


5/ Conclusion and Perspectives

Accuracy vs. computational time


3D simulations: extrusion (top) & full 3D (bottom) (Naca0015 chord

0.420, V10m/s, TSR 1.2 )

PR

OD

-F-0

15-0

1

Thank you for your attention!

Symposium on OpenFOAM in Wind Energy 21/03/2013 - Oldenburg © 2013 Cenaero – All rights reserved

References:

[1] Armstrong S., Fiedler A., Tullis S., Flow separation on a high Reynolds number, high solidity vertical axis wind turbine with straight and canted

blades and canted blades with fences, Renewable Energy, Elsevier, May 2012

[2] McLaren K., A numerical and experimental study of unstedy loading of high solidity vertical axis win turbines, Open Access Dissertations and

Theses McMaster University, Paper 6091, http://digitalcommons.mcmaster.ca/opendissertations/6091

[3] Mukinovic M., Brenner G., Rahimi A., Analysis of Vertical Axis Wind Turbines, New Results in Numerical and Experimental Fluid Mechanics VII

Notes on Numerical Fluid Mechanics and Multidisciplinary Design, Volume 112, 2010, pp 587-594

Questions?

Comments?

TSR 0.4, U=10m/s, k- ω SST (k=1.5, ω=5.6) 50k nodes mesh(video at 1/10 rate)

aerodynamic design of a high solidity canted vertical axis ... · pdf filefirst symposium on...

Documents