Embed Size (px)
Citation preview
Wind energy assessment , wind power forecasting and application of NCMRWF data
K.BOOPATHI
SCIENTIST & UNIT CHIEF
WIND RESOURCE ASSESSMENT
CENTRE FOR WIND ENERGY TECHNOLOGY
CHENNAI-600100
ABOUT C-WET
C-WET, -an autonomous R&D institution
ITCS,R&D, Testing, Standards and certification and WRA
Recognized leader in wind resource assessment -2003-
2004
787 - Wind Monitoring Stations across the country till date
620 - stations are closed down and 167 are operational.
minimum 1 year continuous measurements are essential”
Consultancy Services Offered by CWET for WRA
Site condition assessments for wind monitoring & wind farm development and field visits
Procurement, installation and commissioning of met mast of 50m to 120 m height
providing measurement campaign management, assisting clients in the Installation and monitoring of
meteorological masts, LIDAR and SODAR stations
Data collection, management, quality control and wind energy resource reporting
Analysis of Data with sophisticated software tools and techniques
Long-Term Trend Data Analysis (NCEP/NCAR/MERRA)
Turbine array layout design, optimization, field Micro siting and Produce bankable P50 P75, and P90 yield
predictions.
Investment Grade wind energy resource assessment reports (gross/net Preiections, uncertainty analyses, etc.)
Analysis of existing wind farm operations
Technical due diligence in complying with international standards.
Power curve demonstration guarantee test
DPRs (Detailed Project Report) preparation through State of art software tool to wind farm developers
WRA- Achievements
787 wind monitoring stations Established
5 numbers of 120 m tall masts with multilevel instruments were commissioned
Published Wind Energy Resource Survey Vol.V- VIII containing wind data of various wind monitoring stations across the country.
Conducted Micrositing / due diligence for more than 3500 MW (total) wind farms in the Country.
Indian Wind Atlas has been prepared and published.
Offshore Wind Resource Assessment at Dhanuskodi, Rameshwaram, Ramanathapuram District.
Development of Wind Power Forecasting model with special reference to complex terrain.
Wind turbine Wake Study.
States / UTs Installable Potential (MW) 50 m Level
Andaman & Nicobar 2
Andhra Pradesh 5394
Arunachal Pradesh* 201
Assam* 53
Chhattisgarh* 23
Gujarat 10609
Himachal Pradesh * 20
Jammu & Kashmir * 5311
Karnataka 8591
Kerala 790
Lakshadweep 16
Madhya Pradesh 920
Maharashtra 5439
Manipur* 7
Meghalaya * 44
Nagaland * 3
Orissa 910
Rajasthan 5005
Sikkim * 98
Tamil Nadu 5374
Uttarakhand * 161
Uttar Pradesh * 137
West Bengal* 22
Total 49130
INDIAN WIND ATLAS at 50 & 80 M HEIGHT
State Name Installable Potential MW
Andaman &Nicobar Islands 365
Andhra Pradesh 14497
Arunachal Pradesh* 236
Assam* 112
Bihar 144
Chhattisgarh* 314
Dieu Damn 4
Gujarat 35071
Haryana 93
Himachal Pradesh * 64
Jharkhand 91
Jammu & Kashmir * 5685
Karnataka 13593
Kerala 837
Lakshadweep 16
Madhya Pradesh 2931
Maharashtra 5961
Manipur* 56
Meghalaya * 82
Nagaland * 16
Orissa 1384
Pondicherry 120
Rajasthan 5050
Sikkim * 98
Tamil Nadu 14152
Uttarakhand * 534
Uttar Pradesh * 1260
West Bengal* 22
Total 102788
Wind is the fastest growing renewable energy source in the world.
Electricity generated from wind power will paly an important role
the power need to in the existing electric supply system which was mainly designed for
large units of fossil fuel and nuclear power stations
The fuel is free, but its variability poses challenges to wind power’s continued growth and
effective integration into the power grid.
To compensate for this variability, operators and power marketers must rely on other , more
costly power sources to carry the load when the wind isn’t blowing and scheduling wind
power effectively is a high-stakes enterprise. however , this challenge is not
insurmountable
Today’s wind power forecasting technology is giving operators, developers, power
marketers, and investors better and better information on which they can make these
critical decisions.
Wind power forecasting-Introduction
Why Wind Power Forecasting
plan for the future .
To help managing the power system in a robust and economic manner
To help manage network congestions;
To allow reserves management to compensate for wind fluctuations
To define energy storage strategies;
To help planning maintenance of the wind farms for the next days
Coordination of wind energy with storage (i.e. hydro).
Planning power exchanges/flows with neighbour systems.
Planning maintenance of the wind farms for the next days (offshore).
Planning of fuel consumption
Making bids in an electricity market (reduce penalties for imbalances) etc.
Need for forecasting
• Performance requirements for a forecasting - need of both the grid operator and the
wind energy generators
• Wind energy generators - the priority is to minimize the deviation between forecasted
and actual plant output
• First priority: To anticipate changes in wind production as accurately as possible in
very short time (up to few hours ahead) .To enable the Load Dispatch Centers (LDCs) to
manage the grid operations in an optimal fashion
• Second priority :To forecast the wind generation for the next day - to enable the LDCs to
schedule the reserve capacity as efficiently as possible
• Practiced in Denmark, Germany, Spain, U.K. and U.S.A.
Time scale of forecasting
• Very short-term ( 0-6 hrs) is related to the prediction of small scale atmospheric features in
the vicinity of wind farm
• Short-term (6-72 hrs) is related to the prediction of regional atmospheric features
• Medium range (3-10 days) is related to the prediction of continental, hemispheric and
global systems
• Today the number of tools are available internationally for wind power forecasting
considering these time scales
• Most of the existing power prediction systems are based on the results of numerical
weather prediction (NWP) systems
• There are two approaches to transform the wind prediction given by NWP into a power
prediction namely- statistical and physical
DIFFERENT APPROACHES TO THE POWER
OUTPUT FORECAST
Physical Approach: models try to represent the physical laws (i.e. modelling of wind flow on the terrain Parameterization based on the
physical description of atmosphere).
Statistical Approach: models try to represent the information in the available data (i.e. artificial intelligence techniques) .Based on the measurement data and uses difference between the predicted and actual wind speeds in immediate past to tune the model parameters.
Hybrid Approach: The combination of different approaches
Artificial intelligence (AI) methods to learn the relationship between
predicted wind and power output from time series of the past.
A complex problem
Wind is highly fluctuant by nature – wind speed or wind power time-series can be seen as stochaotic time-series
Example of the power output of a wind farm for a period of 1000 hours.
0
0.1
0.2
0.3
0.4
0.5
0.6
0.7
0.8
0.9
1
12
65
17
61
01
12
61
51
17
62
01
22
62
51
27
63
01
32
63
51
37
64
01
42
64
51
47
65
01
52
65
51
57
66
01
62
66
51
67
67
01
72
67
51
77
68
01
82
68
51
87
69
01
92
69
51
97
61
00
1
Time [x 1 hour]
Power/Pnominal [p.u.]
0 500 1000
– steep slope in the usual
wind speed range
– cut-off effect
0
100
200
300
400
500
600
0 5 10 15 20 25 Wind speed [m/s]
P[i]
v[i]
Power [kW]
The wind turbine characteristic curve introduces nonlinearities
General flow chart for physical models-
wind power forecasting
Physical downscaling (1)
15 km
WEATHER FORECASTS
15 km
blk
90000 95000 100000 105000 110000
315000
320000
325000
330000
OROGRAPHY MAP
• Aims to relate the NWP output to the
local wind by using a meso- or micro-
scale model
• Can be run as once-and-for-all
parameterisation or online
• Examples: Prediktor, Previento,
TrueWind, LocalPred, …
Use of the geostrophic wind for
downscaling
Advantages: rather easy to implement, fast
Drawbacks: weak assumption (bad
performance for complex terrain)
(geostrophic wind: wind in the free
atmosphere, balanced by the pressure
gradient and Coriolis forces)
Mesoscale approach
Physical downscaling (2)
Conversion of the wind
speed to the hub height
Monin-Obuhkov theory (1954):
logarithmic wind speed profile
315000
320000
325000
330000
blk
90000 95000 100000 105000 110000
ROUGHNESS MAP
u(z) = u*.k-1.ln(z/z0)
u(z) : wind speed at height z
U* : inertial velocity
k : von Karman constant (= 0.41)
z0 : roughness length
This theory is not valid for complex
terrains and for offshore winds…
Physical downscaling (3)
Park lay out & Power curve
-200
0
200
400
600
800
1000
1200
1400
0 2 4 6 8 10 12 14 16 18 20 22 24
Po
we
r kW
Wind Speed m/s
Power Curve
Std air density Power kW
Air density corrected power kW
Case study GFS data
The GFS forecasts for the windfarm for 10m a.g.l. Every forecast is 196 hours long and presented in its own colour. There is a new forecast every 12 hours.
The power production at the wind farm
0
10
20
30
40
50
1
16
31
46
61
76
91
10
6
12
1
13
6
15
1
16
6
18
1
19
6
21
1
22
6
Po
we
r (M
W)
3 Hrs Data
Observed
Predicted
0
20
40
60
1.0
0
9.0
0
17
.00
2
5.0
0
33
.00
4
1.0
0
49
.00
5
7.0
0
65
.00
7
3.0
0
81
.00
8
9.0
0
97
.00
1
05
.00
1
13
.00
1
21
.00
1
29
.00
1
37
.00
1
45
.00
1
53
.00
1
61
.00
1
69
.00
1
77
.00
1
85
.00
1
93
.00
2
01
.00
2
09
.00
2
17
.00
2
25
.00
2
33
.00
2
41
.00
Po
we
r (M
W)
Forecast Hours- December 2010
Predicted Vs Observed power
Observed Predicted
Case study of NCMRWF data
Wind speed predictions based on NCMRWF and upscaled measurements from the Khandke wind farm.
zoomed in to the overlapping period
NCMRWF predicts a strong diurnal variation,
NCMRWF wind data has
been utilized to predict power in the
wind farm , as a part of the case study
CONCLUSION
In order to ensure that the grid operates safely and to take appropriate measures, it is very important for the System Operator to foresee what is expected wind power generation in a few hours ahead.
Wind power forecasting reduces cost of integrating wind on the grid and so reduces energy costs, both financial and environmental, for everyone.
As India has tropical weather , GFS data some times could not provide correct wind forecasting. Hence, Indian forecast data of NCMRWF data can be utilised and fine tune the system.
