Sin ttulo de diapositiva

1
A METHODOLOGY FOR ESTIMATING WIND FARM PRODUCTION
THROUGH CFD CODES. DESCRIPTION AND VALIDATION
Daniel Cabezón, Ignacio Martí CENER, National
Renewable Energy Centre (Spain) Wind Energy
Department dcabezon_at_cener.com
2
INDEX

• Introduction
• Description of the methodology
• 3.1 Numerical model
• 3.2 Estimation of wind farm power
• Alaiz wind farm
• Experimental validation
• Conclusions

3
1. INTRODUCTION

• Complex terrain sites
• Increasing uncertainty when estimating power
  production with linear models
• Higher uncertainties for larger wind farms and
  for larger distances to meteorological mast
• The proposed analysis
• Presents a methodology for estimating power
  production of large wind farms through a CFD
  (Computational Fluid Dynamics) code
• Compiles power measurements of a real wind farm
  during a 4 years period
• Validates the methodology in terms of power
  production for each wind turbine and compares it
  with conventional tools

4
2. DESCRIPTION OF THE METHODOLOGY

• 2.1 NUMERICAL MODEL

5
2. DESCRIPTION OF THE METHODOLOGY

• 2.2 ESTIMATION OF WIND FARM POWER

• How wind speed estimation is transformed into
  power estimation?

• The CFD model solves instantaneous situations
  for every direction
• 1 simulation for sector f Field of
  V,TI,P when wind comes from f

CFD - Output
Ratios Wind Turbine velocity Mast velocity for
sector f and WTi
6
2. DESCRIPTION OF THE METHODOLOGY

• 2.2 ESTIMATION OF WIND FARM POWER

INPUTS
OUTPUTS
Wind speed and direction at MAST
Net Annual Energy Production / Wind Turbine
RATIOS Wind Turbine velocity Mast velocity for
sector f and WTi
Net Annual Energy Production at Wind Farm
WAKE EFFECTS for sector f and WTi
Normalized POWER CURVE for WTi
7
2. DESCRIPTION OF THE METHODOLOGY

• 2.2 ESTIMATION OF WIND FARM POWER

WT50
WT1
WT2
WT3
WT1
WT2
WT3
WT50
bin_1
bin_2
bin_3
. . .
. . . . . . . . . . . . . . .
. . . . . . . . . . . . . . . . . . . . . .
. . . . . . . . .
. . . . . . . . . . . . . . . . . . .
INPUT
. . .
bin_30
WT1
WT2
ANNUAL FREQUENCY (HRS) FOR EACH WIND TURBINE
FOR SECTOR f
WT3
. . .
WIND SPEED FOR EACH WIND TURBINE FOR SECTOR f
. .
WT50
INPUT
8
2. DESCRIPTION OF THE METHODOLOGY

• 2.2 ESTIMATION OF WIND FARM POWER

WT1
WT2
WT3
WT50
WT1
WT2
WT3
WT50
WT1
WT2
WT50
bin_1
bin_1
20º
bin_2
bin_2
40º
bin_3
bin_3
60º
. . .
. . .
. . .
bin_30
bin_30
340º
INPUT
INPUT
ANNUAL FREQUENCY (HRS)
POWER CURVES
WAKE EFFECTS
PARK
WT1
WT2
WT3
WT4
WT5
WT50
. . .
PRODUCTION / WT (GWh)
9
3. ALAIZ WIND FARM

• Alaiz hill site
• Complex terrain (global RIX 16 )
• 4 kilometers hill, E-W orientation
• Prevailing wind direction N
• Highly roughed on the hilltop

Measurement campaign met masts
Wind farm met mast

• Alaiz wind farm
• Installed power 33.09 MW
• 49 WTs (660 kW) 1 WT (750 kW)
• Measurement campaign 1996-1997
• Wind farm measurements 2000 (40 WTs)
  2001-2003 (50 WTs)

10
4. EXPERIMENTAL VALIDATION
AEP (Anual Energy Production)
COMPARATIVE
AEP 2000
WT1_WT40
(AEP_WTi / AEP_ref) MODELLED
WAsP
AEP 2001
WT1_WT50
vs
MEASUREMENTS
AEP 2002
(AEP_WTi / AEP_ref) REAL
CFD
WT1_WT50
AEP 2003
WT1_WT50
AEP Average
AEP_ref AEP corresponding to the nearest WT to
the met mast

• AEP for just north direction at Alaiz_9 (20º
  sector)
• Production filtering WT availability gt 70
• Modeling from Alaiz3_55 and Alaiz 6_40

11
4. EXPERIMENTAL VALIDATION
I. AEP modelling from ALAIZ 3 55 m
WT_refWT_28
1
3
2
4
Underestimation on WT 13 to 19
20º degrees turning clockwise!
12
4. EXPERIMENTAL VALIDATION
I. AEP modelling from ALAIZ 3 55 m
20º degrees turning clockwise around Alaiz 2
23.2
ALAIZ 2
9.7
ALAIZ 6
28.2
ALAIZ 3

• Turning caused by an upstream hill
• Production moved to sector 2 (10º-30º)

13
4. EXPERIMENTAL VALIDATION
II. AEP modelling from ALAIZ 6 40 m
WT_refWT_35

• Similar trend for Alaiz3_55 y Alaiz 6_40
• Underestimation for alignement 1(WT1_WT11) and 2
  (WT12_WT19)

14
4. EXPERIMENTAL VALIDATION
III. Global Error. Wind Farm AEP

• Results using one met. mast

AEP Error from ALAIZ 3_55
AEP Error from ALAIZ 6_40

• CFD errors below 18. WAsP errors up to 39.5

15
4. EXPERIMENTAL VALIDATION
WT segregation according to similar RIX indexes
IV. Combined WAsP simulation with 2 masts
ALAIZ 6
ALAIZ 3

• Results using two met. masts

16
4. EXPERIMENTAL VALIDATION
IV. Combined WAsP simulation with 2 masts

• Global Production Error with Alaiz6_40 -31
• Global Production Error with Alaiz6_40
  Alaiz3_55 -29.2

• Using two met. masts with WAsP the error
  decreases in a 1.8

17
5. CONCLUSIONS

• A specific methodology for the estimation of
  wind farm output power from CFD codes has been
  developped and validated in a complex terrain
  wind farm.
• Only conventional inputs needed (mast data,
  power curve).
• Uncertainty decrease of 25 at the test site
  based on power measurements
• CFD annual absolute error variation in AEP are
  in the range 0.46 - 17.41 for a wind farm in
  complex terrain while with WAsP the error range
  is 16.64 - 39.5.
• The reduction of errors with WAsP using two
  meteorological masts in this case was only 1.8.
• A CFD simulation with CENER methodology can help
  to increase accuracy in AEP estimation reducing
  the number of meteorological masts.

18
(No Transcript)