aprescribed+wakevortexlinemethodfor...

16
A PrescribedWake Vortex Line Method for Aerodynamic Analysis and Op=miza=on of Mul=Rotor Wind Turbines A. Rosenberg and A. Sharma Department of Aerospace Engineering Iowa State University NAWEA 2015 Symposium Virginia Tech

Upload: others

Post on 13-Jun-2020

1 views

Category:

Documents


0 download

TRANSCRIPT

Page 1: APrescribed+WakeVortexLineMethodfor ...APrescribed+Wake"Vortex"Line"Method"for" Aerodynamic"Analysis"and"Op=miza=on"of" Mul=+Rotor"Wind"Turbines" A."Rosenberg"and"A."Sharma" Departmentof)Aerospace)Engineering)

A  Prescribed-­‐Wake  Vortex  Line  Method  for  Aerodynamic  Analysis  and  Op=miza=on  of  Mul=-­‐Rotor  Wind  Turbines  A.  Rosenberg  and  A.  Sharma  Department  of  Aerospace  Engineering  Iowa  State  University            NAWEA  2015  Symposium  Virginia  Tech  

Page 2: APrescribed+WakeVortexLineMethodfor ...APrescribed+Wake"Vortex"Line"Method"for" Aerodynamic"Analysis"and"Op=miza=on"of" Mul=+Rotor"Wind"Turbines" A."Rosenberg"and"A."Sharma" Departmentof)Aerospace)Engineering)

Mo=va=on  

•  Betz  limit:    –  59.3%  of  energy  capture  –  Modern  HAWTs  operate  well  

below  this  

•  Root  loss  (≈5%)  –  Inner  25%  of  rotor  …      …  designed  for  structural  integrity  à  Poor  root  aerodynamics  

•  Wake  loss  (≈8−40%)  –  Not  operaWng  in  isolaWon  

NAWEA  2015  Rosenberg  &  Sharma   2  

Motivation

• Betz limit: – 59.3% of energy capture– Modern HAWTs operate well

below this

• Root loss  (≈5%)– Inner 25% of rotor  …  …  designed  for structural integrity Poor root aerodynamics

• Wake  loss  (≈8−40%)– Not operating in isolation

NAWEA 2015Rosenberg & Sharma 2

Page 3: APrescribed+WakeVortexLineMethodfor ...APrescribed+Wake"Vortex"Line"Method"for" Aerodynamic"Analysis"and"Op=miza=on"of" Mul=+Rotor"Wind"Turbines" A."Rosenberg"and"A."Sharma" Departmentof)Aerospace)Engineering)

Novel  Turbine  Concept  

•  Dual  Rotor  Wind  Turbine  (DRWT)*  –  Add  a  secondary,  co-­‐axial  rotor  –  Rotors  rotate  independently  –  Co-­‐  or  counter-­‐rotaWng  

•  Aims:  –  Reduce  root  loss  –  Enhance  wake  mixing  

•  Secondary  rotor  “inversely”  designed  using  BEM  

NAWEA  2015  Rosenberg  &  Sharma   3  

Dual  rotor  wind  turbine  (DRWT)  concept  

Lab  models  (courtesy  Dr.  Hui  Hu,  ISU)  *A.  Rosenberg,  S.  Selvaraj,  A.  Sharma,  A  novel  dual-­‐rotor  turbine  for  increased  wind  energy  capture,  Journal  of  Physics:  Conference  Series,  vol.  524,  2014.    

Page 4: APrescribed+WakeVortexLineMethodfor ...APrescribed+Wake"Vortex"Line"Method"for" Aerodynamic"Analysis"and"Op=miza=on"of" Mul=+Rotor"Wind"Turbines" A."Rosenberg"and"A."Sharma" Departmentof)Aerospace)Engineering)

Vortex  Line  Method  

Rosenberg  &  Sharma   NAWEA  2015   4  

•  Prescribed-­‐  or  free-­‐wake  •  Prandtl’s  Liding  Line  Theory  

–  Blade  bound  vorWcity  –  Helmhotlz  →  trailing  vortex  helix  

•  InducWon  from  Biot-­‐Savart  law  

–  Compute  local  inflow,  AoA  –  A/f  polars  →  secWonal  lid  &  drag  à  

turbine  torque/power  

•  Goal:  Develop  computaWonally  inexpensive  and  robust  method  to  analyze  DRWTs  

dv = ��dl⇥ r

4⇡|r|3

Page 5: APrescribed+WakeVortexLineMethodfor ...APrescribed+Wake"Vortex"Line"Method"for" Aerodynamic"Analysis"and"Op=miza=on"of" Mul=+Rotor"Wind"Turbines" A."Rosenberg"and"A."Sharma" Departmentof)Aerospace)Engineering)

Algorithm  •  Prescribed  wake  –  No  mutual  inducWon  in  trailing  vorWces  

–  Trailing  vortex  stays  intact  except:  •  Helix  pitch  &  Treiz  plane  locaWon  are  updated  

–  Chajot*  (2011)    

•  IteraWve  soluWon  –  Converges  when  trailing  vortex  

sheet  in  equilibrium  with  blade  bound  circulaWon  

Rosenberg  &  Sharma   NAWEA  2015   5  

*J.J.  Chajot,  Wind  turbine  aerodynamics:  analysis  and  design,  InternaGonal  Journal  of  Aerodynamics,  vol.  1,  2011  

dv

��i=

X

j

dlj ⇥ rij4⇡|rij|3

Under-­‐relax  

Page 6: APrescribed+WakeVortexLineMethodfor ...APrescribed+Wake"Vortex"Line"Method"for" Aerodynamic"Analysis"and"Op=miza=on"of" Mul=+Rotor"Wind"Turbines" A."Rosenberg"and"A."Sharma" Departmentof)Aerospace)Engineering)

Algorithm  -­‐  con=nued  

Rosenberg  &  Sharma   NAWEA  2015   6  

•  Vortex  pitch  determined  by  area-­‐averaged  axial  inducWon  -­‐  1-­‐D  momentum  theory:  

-­‐  f,  g  are  mulW-­‐valued!  

•  Buhl’s  (2005)*correcWon  

 makes  the  funcWon  single-­‐valued  

CT (a) =

8<

:

4aF (1� a) if 0 a 0.4

8

9+

✓4F � 40

9

◆a+

✓50

9� 4F

◆a2 if 0.4 < a 1

*M.  Buhl,  A  new  empirical  relaWonship  between  thrust  coefficient  and  inducWon  factor  for  the  turbulent  windmill  state,  NREL/TP-­‐500-­‐36834,  2005  

CT = 4a(1� a) :: a = f(CP )

CP = 4a(1� a)2 :: a = g(CT )

Page 7: APrescribed+WakeVortexLineMethodfor ...APrescribed+Wake"Vortex"Line"Method"for" Aerodynamic"Analysis"and"Op=miza=on"of" Mul=+Rotor"Wind"Turbines" A."Rosenberg"and"A."Sharma" Departmentof)Aerospace)Engineering)

VLM  –  Single  Rotor  Valida=on    

•  Tellus  T-­‐1995  –  95  kW  –  Stall  controlled  –  19  m  diameter  –  Chord/twist  distribuWon:  

•  Good  agreement  with:  –  Measured  data  –  BEM  (blade  element  method)  

Rosenberg  &  Sharma   NAWEA  2015   7  

����

����

����

����

���� ���� ���� ��� ��

��

��

���

���

��������������

�� �������

�� ������ ���� ��

�� ������

CTCP

� �

Page 8: APrescribed+WakeVortexLineMethodfor ...APrescribed+Wake"Vortex"Line"Method"for" Aerodynamic"Analysis"and"Op=miza=on"of" Mul=+Rotor"Wind"Turbines" A."Rosenberg"and"A."Sharma" Departmentof)Aerospace)Engineering)

Extension  to  Dual-­‐Rotor  Wind  Turbines  

Rosenberg  &  Sharma   NAWEA  2015   8  

•  MulWple  trailing  helices  –  Treat  singulariWes  in  Biot-­‐Savart  (r  >  rmin)  

•  PotenWal  flow  effects  ignored  •  Both  helices  convect  with  

same  flow  speed  –  Equiv.  CT  à  inducWon    

•  Inherently  unsteady  –  Pseudo-­‐steady  assumpWon  –  Phase  average  for  final  results  

(a) c⌧F (b) cT

Figure 7: Comparisons of radial variation of force coe�cients at � = 7.0 against BEMresults for the Tellus T-1995 turbine

2. If the two rotors rotate independently, then the problem becomes in-herently unsteady; approximations need to be made to solve it as asteady problem.

3. If the two rotors are very close, potential flow e↵ects (due to finite bladethickness) also come into play.

To account for mutual induction between the bound and trailing vorticesof the two rotors of a DRWT, we first need to setup the associated vortexlines/lattices. Figure 8 shows the vortex lattice structure of a DRWT. Twosets of helices are now present; one set each for the two rotors. The pitch ofthe two helices is set to be the same since the trailing vorticity is convectedby the same flow speed. Equation 1 is used to determine the area-averageinduction, a, which then determines the pitch of the helices. For DRWTs,an equivalent area-weighted C

T,eq

is used with Eq. 1, where

C

T,eq

= C

T,m

+A

s

A

m

C

T,s

, (2)

and subscripts ‘m’ and ‘s’ refer to the main rotor and the secondary rotorrespectively; ‘A’ is area of rotor disk. Once the vorticity structure is set,computation of induction coe�cients using the Biot-Savart law is straight-forward. It should be emphasized again that induction from bound vorticity

8

Page 9: APrescribed+WakeVortexLineMethodfor ...APrescribed+Wake"Vortex"Line"Method"for" Aerodynamic"Analysis"and"Op=miza=on"of" Mul=+Rotor"Wind"Turbines" A."Rosenberg"and"A."Sharma" Departmentof)Aerospace)Engineering)

RANS/AD  Method  for  DRWT  Aero  

Rosenberg  &  Sharma   NAWEA  2015   9  

RANS/AD:  •  Actuator  disk  to  model  circ.  avg.  forces  •  Axisymmetric  grids;  SIMPLE;  k-­‐omega  RANS  model  •  Previously  validated*  

Figure 9: Predicted variation in % CP of a DRWT with relative angular position betweenthe primary and the secondary rotors.

Rosenberg et al. [1] suggest that the optimum axial separation for enhancedisolated rotor aerodynamic performance is about 0.2 ⇥ main rotor diameter,which is much greater than the maximum blade chord of the main rotor.Thus, the approximation to neglect potential e↵ects due to blade thicknessshould be valid for such turbines.

3.1. Verification with CFD

The proposed extended vortex line method to analyze dual-rotor windturbines is verified against results obtained using the Reynolds AveragedNavier-Stokes + actuator disk (RANS/AD) method described in Rosenberget al. [1] and Selvaraj [12]. Subtle aspects of this RANS/AD methodologyare summarized here for completeness.

3.1.1. RANS/AD Method

The RANS/AD method [1, 12] solves the incompressible RANS equations(Eq. 3) with the rotor blades modeled as body forces (actuator disk). Thegoverning equations are

@u

i

@x

i

= 0, and,

u

j

@u

i

@x

j

= �1

@p

@x

i

+ ⌫

@

2u

i

@x

j

2�

@u

0i

u

0j

@x

j

+f

i

. (3)

10

*A.  Rosenberg,  S.  Selvaraj,  A.  Sharma,  A  novel  dual-­‐rotor  turbine  for  increased  wind  energy  capture,  Journal  of  Physics:  Conference  Series,  vol.  524,  2014.    

Page 10: APrescribed+WakeVortexLineMethodfor ...APrescribed+Wake"Vortex"Line"Method"for" Aerodynamic"Analysis"and"Op=miza=on"of" Mul=+Rotor"Wind"Turbines" A."Rosenberg"and"A."Sharma" Departmentof)Aerospace)Engineering)

DRWT:    Code-­‐to-­‐Code  Comparisons  

Rosenberg  &  Sharma   NAWEA  2015   10  

Verify  VLM  against  RANS/AD  predicWons:  2  DRWT  designs:    rWp  sec  =  0.25  and  0.4  

Good  overall  agreement  Discrepancy  at  r  =  rWp  sec  due  to  circ  avg.  in  AD  method  

-0.50

-0.25

0.00

0.25

0.50

0.75

1.00

0.00 0.25 0.50 0.75 1.00To

rque

forc

e co

e�.

radius / tip radius of main rotor

CFDVLM

-0.50

-0.25

0.00

0.25

0.50

0.75

1.00

0.00 0.25 0.50 0.75 1.00

Torq

ue fo

rce

coe�

.

radius / tip radius of main rotor

CFDVLM

Page 11: APrescribed+WakeVortexLineMethodfor ...APrescribed+Wake"Vortex"Line"Method"for" Aerodynamic"Analysis"and"Op=miza=on"of" Mul=+Rotor"Wind"Turbines" A."Rosenberg"and"A."Sharma" Departmentof)Aerospace)Engineering)

Poor  Results  at  High  Loading  

Rosenberg  &  Sharma   NAWEA  2015   11  

OpWmizaWon:    rudimentary  approach  –  parametric  sweeps  •  Vary  secondary  rotor  radius  &  TSR  •  Main  rotor  geometry  &  TSR  held  fixed  

RANS/AD  

Stability  &  accuracy  issues  with  VLM  @  high  loading  

VLM  

Page 12: APrescribed+WakeVortexLineMethodfor ...APrescribed+Wake"Vortex"Line"Method"for" Aerodynamic"Analysis"and"Op=miza=on"of" Mul=+Rotor"Wind"Turbines" A."Rosenberg"and"A."Sharma" Departmentof)Aerospace)Engineering)

Enhancement  to  VLM  Model  

Rosenberg  &  Sharma   NAWEA  2015   12  

•  Previous  approaches  assumed:  •  axial  inducWon,  a(x)  increases  linearly  from  a0  to  2a0  •  a(x)  à  wake  helix  pitch  

•  Calibrate  wake  pitch  using  RANS  results  Old  model  –  linear  approx.  

Page 13: APrescribed+WakeVortexLineMethodfor ...APrescribed+Wake"Vortex"Line"Method"for" Aerodynamic"Analysis"and"Op=miza=on"of" Mul=+Rotor"Wind"Turbines" A."Rosenberg"and"A."Sharma" Departmentof)Aerospace)Engineering)

New  Wake  Model  

Rosenberg  &  Sharma   NAWEA  2015   13  

a(x) = a0

✓✓a

xr3

a0� 1

◆tanh(B ⇥ x/x

r3)

tanh(B)+ 1

◆axr3 = 0.47 C

T

+ 0.03

with  B=3  

solid  lines  à  RANS  dashed  à  new  model  

Page 14: APrescribed+WakeVortexLineMethodfor ...APrescribed+Wake"Vortex"Line"Method"for" Aerodynamic"Analysis"and"Op=miza=on"of" Mul=+Rotor"Wind"Turbines" A."Rosenberg"and"A."Sharma" Departmentof)Aerospace)Engineering)

Improved  Results  @  High  Loading  

Rosenberg  &  Sharma   NAWEA  2015   14  

RANS/AD   VLM  

Repeat  parametric  sweep  

•  Stability  issues  resolved  with  the  proposed  wake  correcWon  •  QualitaWve  agreement  à  useful  for  opWmizaWon  •  Parameter  tuning  to  improve  results  ongoing  

Page 15: APrescribed+WakeVortexLineMethodfor ...APrescribed+Wake"Vortex"Line"Method"for" Aerodynamic"Analysis"and"Op=miza=on"of" Mul=+Rotor"Wind"Turbines" A."Rosenberg"and"A."Sharma" Departmentof)Aerospace)Engineering)

Conclusion  

•  Prescribed-­‐wake  VLM  method  proposed  for  DRWTs  –  Useful  for  prelim  design  &  opWmizaWon  

•  Valida=on:    experiments  &  BEMT  for  SRWTs  

•  Verifica=on:    Code-­‐to-­‐code  agreement  -­‐  VLM  &  RANS/AD  for  DRWTs  

 •  RANS-­‐inspired  wake  model  alleviates  stability  issues  at  high  loading  for  DRWTs  

Rosenberg  &  Sharma   NAWEA  2015   15  

Page 16: APrescribed+WakeVortexLineMethodfor ...APrescribed+Wake"Vortex"Line"Method"for" Aerodynamic"Analysis"and"Op=miza=on"of" Mul=+Rotor"Wind"Turbines" A."Rosenberg"and"A."Sharma" Departmentof)Aerospace)Engineering)

QUESTIONS?  

Rosenberg  &  Sharma   NAWEA  2015   16  

ACKNOWLEDGEMENT  •  NSF  CBET:  #1438099    •  IEC:  #14-­‐008-­‐OG    •  ISGC:  #475-­‐20-­‐5    •  NSF  XSEDE:  #TG-­‐CTS130004