1 | Future wind power forecast errors and associated costs in the Swedish power system | Fredrik Carlsson | 2012.02.07

Future wind power forecast errors and associated costs in the Swedish power system

Presentation at Winterwind 2012 Fredrik Carlsson/Viktoria Neimane

Confidentiality class: None (C1)

2012.02.07

2 | Future wind power forecast errors and associated costs in the Swedish power system | Fredrik Carlsson | 2012.02.07

Future wind power forecast errors and associated costs in the Swedish power system • Nordic electricity market • Day ahead forecast errors • Model with different actors 10 and 30 TWh wind power scenarios

• Imbalance costs • Are there any possibilities to reduce imbalance costs? • Conclusions

3 | Future wind power forecast errors and associated costs in the Swedish power system | Fredrik Carlsson | 2012.02.07

Nordic Electricity market

•  Bids to the day-ahead market are provided 12-36 hours in advance

4 | Future wind power forecast errors and associated costs in the Swedish power system | Fredrik Carlsson | 2012.02.07

Nordic Electricity market

•  Bids to the day-ahead market are provided 12-36 hours in advance

•  It is possible to trade on the intra-day market in order to compensate for forecast errors

5 | Future wind power forecast errors and associated costs in the Swedish power system | Fredrik Carlsson | 2012.02.07

Nordic Electricity market

•  Bids to the day-ahead market are provided 12-36 hours in advance •  It is possible to trade on the intra-day market in order to compensate for forecast errors

•  The forecast errors are handled by the TSO during the hour of delivery •  Those who have caused to for forecast errors then have to pay

6 | Future wind power forecast errors and associated costs in the Swedish power system | Fredrik Carlsson | 2012.02.07

Future wind power forecast errors and associated costs in the Swedish power system • Nordic electricity market • Day ahead forecast errors • Model with different actors 10 and 30 TWh wind power scenarios

• Imbalance costs • Are there any possibilities to reduce imbalance costs? • Conclusions

7 | Future wind power forecast errors and associated costs in the Swedish power system | Fredrik Carlsson | 2012.02.07

Example of day-ahead forecast from Lillgrund wind farm

-20

0

20

40

60

80

100

120

2009-02-27

2009-02-28

2009-03-01

2009-03-02

2009-03-03

2009-03-04

2009-03-05

2009-03-06

2009-03-07

2009-03-08

2009-03-09

2009-03-10

2009-03-11

2009-03-12

2009-03-13

2009-03-14

2009-03-15

2009-03-16

2009-03-17

2009-03-18

MW

Day-ahead (12-36h) forecast Real production

8 | Future wind power forecast errors and associated costs in the Swedish power system | Fredrik Carlsson | 2012.02.07

Example of day-ahead forecast from Lillgrund wind farm

-20

0

20

40

60

80

100

120

2009-02-27

2009-02-28

2009-03-01

2009-03-02

2009-03-03

2009-03-04

2009-03-05

2009-03-06

2009-03-07

2009-03-08

2009-03-09

2009-03-10

2009-03-11

2009-03-12

2009-03-13

2009-03-14

2009-03-15

2009-03-16

2009-03-17

2009-03-18

MW

Day-ahead (12-36h) forecast Real production

Weather front arrived later than forecasted

9 | Future wind power forecast errors and associated costs in the Swedish power system | Fredrik Carlsson | 2012.02.07

Example of day-ahead forecast from Lillgrund wind farm

-20

0

20

40

60

80

100

120

2009-02-27

2009-02-28

2009-03-01

2009-03-02

2009-03-03

2009-03-04

2009-03-05

2009-03-06

2009-03-07

2009-03-08

2009-03-09

2009-03-10

2009-03-11

2009-03-12

2009-03-13

2009-03-14

2009-03-15

2009-03-16

2009-03-17

2009-03-18

MW

Day-ahead (12-36h) forecast Real production

Wind speed less than forecasted

10 | Future wind power forecast errors and associated costs in the Swedish power system | Fredrik Carlsson | 2012.02.07

Analysis of Lillgrund forecast errors (day-ahead, 12-36h)

•  Installed capacity 110 MW •  Yearly production: 330 GWh •  Mean power: 37 MW = 33% of installed capacity •  Forecast error: 130 GWh = 41% of yearly production •  Mean deviation: 12 MW = 10% of installed capacity •  Standard deviation: 21 MW = 19% of installed capacity Prognosfel per timme

0

100

200

300

400500

600

700

800

900

-80 -70 -60 -50 -40 -30 -20 -10 0 10 20 30 40 50 60 70 80

Prognosfel [MWh/h]Tim

mar [

h]

Lillgrund prognosfel

-100,0

-60,0

-20,0

20,0

60,0

100,0

2009-01-07 2009-02-06 2009-03-08 2009-04-07 2009-05-07 2009-06-06

MW

11 | Future wind power forecast errors and associated costs in the Swedish power system | Fredrik Carlsson | 2012.02.07

Future wind power forecast errors and associated costs in the Swedish power system • Nordic electricity market • Day ahead forecast errors • Model with different actors 10 and 30 TWh wind power scenarios

• Imbalance costs • Are there any possibilities to reduce imbalance costs? • Conclusions

12 | Future wind power forecast errors and associated costs in the Swedish power system | Fredrik Carlsson | 2012.02.07

Swedish wind power scenarios with 10 and 30 TWh

Kapacitet per area

0

500

1 000

1 500

2 000

2 500

3 000

3 500

4 000

4 3 2 1 Total

Area

Instal

lerad

effek

t [MW

]

Kapacitet per area

0

2000

4000

6000

8000

10000

12000

4 3 2 1 Total

Area

Inst

aller

ad ef

fekt

[MW

]

+ = Under construction

With permission

Scenarios are based on: •  Projects under construction •  Projects which have been

granted permission

13 | Future wind power forecast errors and associated costs in the Swedish power system | Fredrik Carlsson | 2012.02.07

Eight actors

0

1

2

3

4

5

6

1 2 3 4 5 6 7 8

Aktör

Årlig

prod

uktio

n [TW

h]

Different combinations of: •  Amount of installed wind power •  Number of power farms •  Geographical location and spread

Ann

ual p

rodu

ctio

n

Actor

14 | Future wind power forecast errors and associated costs in the Swedish power system | Fredrik Carlsson | 2012.02.07

Model of forecast error for 12 – 36h forecasts •  The forecast error, per farm, is assumed to

have a standard deviation of 13% of installed capacity (20% in Lillgrund)

•  The relative forecast error decreases when an actor owns more parks in the same area.

•  The relative forecast error decreases when an actor owns farms in several areas.

Correlation data are adopted from Germany

15 | Future wind power forecast errors and associated costs in the Swedish power system | Fredrik Carlsson | 2012.02.07

Model of forecast error for 12 – 36h forecasts •  The forecast error, per farm, is assumed to

have a standard deviation of 13% of installed capacity (20% in Lillgrund)

•  The relative forecast error decreases when an actor owns more parks in the same area.

•  The relative forecast error decreases when an actor owns farms in several areas.

Correlation data are adopted from Germany

16 | Future wind power forecast errors and associated costs in the Swedish power system | Fredrik Carlsson | 2012.02.07

Model of forecast error for 12 – 36h forecasts •  The forecast error, per farm, is assumed to

have a standard deviation of 13% of installed capacity (20% in Lillgrund)

•  The relative forecast error decreases when an actor owns more parks in the same area.

•  The relative forecast error decreases when an actor owns farms in several areas.

Correlation data are adopted from Germany

17 | Future wind power forecast errors and associated costs in the Swedish power system | Fredrik Carlsson | 2012.02.07

Future wind power forecast errors and associated costs in the Swedish power system • Nordic electricity market • Day ahead forecast errors • Model with different actors 10 and 30 TWh wind power scenarios

• Imbalance costs • Are there any possibilities to reduce imbalance costs? • Conclusions

18 | Future wind power forecast errors and associated costs in the Swedish power system | Fredrik Carlsson | 2012.02.07

Imbalance costs: scenario with 10 TWh wind power

10 TWh

0,00

5,00

10,00

15,00

20,00

25,00

30,00

35,00

1 2 3 4 5 6 7 8

Aktör

kr/M

Wh

Dagens priserNya priserNya prisområden

Assumptions: •  Today’s prices are based on statistical data

•  New prices consider higher price level due to higher demand

•  New bidding zones assume that cut 4 is congested 30% of time

Actor

19 | Future wind power forecast errors and associated costs in the Swedish power system | Fredrik Carlsson | 2012.02.07

Imbalance costs: scenario with 10 TWh wind power

10 TWh

0,00

5,00

10,00

15,00

20,00

25,00

30,00

35,00

1 2 3 4 5 6 7 8

Aktör

kr/M

Wh

Dagens priserNya priserNya prisområden

Assumptions: •  Today’s prices are based on statistical data

•  New prices consider higher price level due to higher demand

•  New bidding zones assume that cut 4 is congested 30% of time

Actor

20 | Future wind power forecast errors and associated costs in the Swedish power system | Fredrik Carlsson | 2012.02.07

Imbalance costs: scenario with 10 TWh wind power

10 TWh

0,00

5,00

10,00

15,00

20,00

25,00

30,00

35,00

1 2 3 4 5 6 7 8

Aktör

kr/M

Wh

Dagens priserNya priserNya prisområden

Assumptions: •  Today’s prices are based on statistical data

•  New prices consider higher price level due to higher demand

•  New bidding zones assume that cut 4 is congested 30% of time

Actor

10 TWh

0,00

5,00

10,00

15,00

20,00

25,00

30,00

35,00

1 2 3 4 5 6 7 8

Aktör

kr/M

Wh

Dagens priserNya priserNya prisområden

21 | Future wind power forecast errors and associated costs in the Swedish power system | Fredrik Carlsson | 2012.02.07

Imbalance costs: scenario with 10 TWh wind power

Assumptions: •  Today’s prices are based on statistical data

•  New prices consider higher price level due to higher demand

•  New bidding zones assume that cut 4 is congested 30% of time

Actor

22 | Future wind power forecast errors and associated costs in the Swedish power system | Fredrik Carlsson | 2012.02.07

Why does it become more expensive? •  Wind power dominates forecast errors – fewer hours in “correct” direction.

50% error (= independent) changes to 75 % error •  Higher regulating prices due to higher volumes •  Number of hour without activation of regulating power decreases (to 50%). •  Price areas è imbalances in different areas are not added.

Scenarier

0,00

5,00

10,00

15,00

20,00

25,00

30,00

35,00

40,00

45,00

50,00

1 2 3 4 5 6 7 8

Aktör

kr/MWh

Today5 GW = 10 - 15 TWh12 GW = 30 TWh

Actor

23 | Future wind power forecast errors and associated costs in the Swedish power system | Fredrik Carlsson | 2012.02.07

Why does it become more expensive? •  Wind power dominates forecast errors – fewer hours in “correct” direction.

50% error (= independent) changes to 75 % error •  Higher regulating prices due to higher volumes •  Number of hour without activation of regulating power decreases (to 50%). •  Price areas è imbalances in different areas are not added.

Scenarier

0,00

5,00

10,00

15,00

20,00

25,00

30,00

35,00

40,00

45,00

50,00

1 2 3 4 5 6 7 8

Aktör

kr/MWh

Today5 GW = 10 - 15 TWh12 GW = 30 TWh

Actor

24 | Future wind power forecast errors and associated costs in the Swedish power system | Fredrik Carlsson | 2012.02.07

Why does it become more expensive? •  Wind power dominates forecast errors – fewer hours in “correct” direction.

50% error (= independent) changes to 75 % error •  Higher regulating prices due to higher volumes •  Number of hour without activation of regulating power decreases (to 50%). •  Price areas è imbalances in different areas are not added.

Scenarier

0,00

5,00

10,00

15,00

20,00

25,00

30,00

35,00

40,00

45,00

50,00

1 2 3 4 5 6 7 8

Aktör

kr/MWh

Today5 GW = 10 - 15 TWh12 GW = 30 TWh

Actor

25 | Future wind power forecast errors and associated costs in the Swedish power system | Fredrik Carlsson | 2012.02.07

Future wind power forecast errors and associated costs in the Swedish power system • Nordic electricity market • Day ahead forecast errors • Model with different actors 10 and 30 TWh wind power scenarios

• Imbalance costs • Are there any possibilities to reduce imbalance costs? • Conclusions

26 | Future wind power forecast errors and associated costs in the Swedish power system | Fredrik Carlsson | 2012.02.07

Possible improvement 1: Use intraday market Elbas

•  Trading forecast errors on Elbas reduces the volumes with 50% •  Assumption: Forecast error (RMS) is reduced to 6% instead of 13%.

0,00

5,00

10,00

15,00

20,00

25,00

30,00

35,00

40,00

1 2 3 4 5 6 7 8

Actor

kr/MWh

New PricesNew Bidding zonesNew Prices and ElbasNew Bidding zones and Elbas

27 | Future wind power forecast errors and associated costs in the Swedish power system | Fredrik Carlsson | 2012.02.07

•  Change spot market to trade every 6 hour •  Standard deviation 10%

Possible improvement 2: Assuming 6h spot market

0

5

10

15

20

25

30

35

40

1 2 3 4 5 6 7 8

Aktör

kr/MWh

3 - 9 h new prices12 - 36 h new prices3 - 9 h new biddning zones12 - 36 h new biddings zones

Actor

28 | Future wind power forecast errors and associated costs in the Swedish power system | Fredrik Carlsson | 2012.02.07

Future wind power forecast errors and associated costs in the Swedish power system • Nordic electricity market • Day ahead forecast errors • Model with different actors 10 and 30 TWh wind power scenarios

• Imbalance costs • Are there any possibilities to reduce imbalance costs? • Conclusions

29 | Future wind power forecast errors and associated costs in the Swedish power system | Fredrik Carlsson | 2012.02.07

Conclusions

•  Larger volumes of wind power lead to higher forecast errors in the system which results in:

-  higher regulating prices -  more hours with regulation -  more hours with regulation correlated to wind power forecast error -  thereby higher costs for balance responsible wind power owners.

•  Geographical spread of the wind power leads to lower forecast error which implies lower costs.

•  The introduction of price areas leads to lower possibility to counterbalance imbalances between different areas.

•  Participation on intra-day market can reduce the imbalance costs with about 20%.

30 | Future wind power forecast errors and associated costs in the Swedish power system | Fredrik Carlsson | 2012.02.07

Conclusions

•  Larger volumes of wind power lead to higher forecast errors in the system which results in:

-  higher regulating prices -  more hours with regulation -  more hours with regulation correlated to wind power forecast error -  thereby higher costs for balance responsible wind power owners.

•  Geographical spread of the wind power leads to lower forecast error which implies lower costs.

•  The introduction of price areas leads to lower possibility to counterbalance imbalances between different areas.

•  Participation on intra-day market can reduce the imbalance costs with about 20%.

31 | Future wind power forecast errors and associated costs in the Swedish power system | Fredrik Carlsson | 2012.02.07

Conclusions

•  Larger volumes of wind power lead to higher forecast errors in the system which results in:

-  higher regulating prices -  more hours with regulation -  more hours with regulation correlated to wind power forecast error -  thereby higher costs for balance responsible wind power owners.

•  Geographical spread of the wind power leads to lower forecast error which implies lower costs.

•  The introduction of price areas leads to lower possibility to counterbalance imbalances between different areas.

•  Participation on intra-day market can reduce the imbalance costs with about 20%.

32 | Future wind power forecast errors and associated costs in the Swedish power system | Fredrik Carlsson | 2012.02.07

Conclusions

•  Larger volumes of wind power lead to higher forecast errors in the system which results in:

-  higher regulating prices -  more hours with regulation -  more hours with regulation correlated to wind power forecast error -  thereby higher costs for balance responsible wind power owners.

•  Geographical spread of the wind power leads to lower forecast error which implies lower costs.

•  The introduction of price areas leads to lower possibility to counterbalance imbalances between different areas.

•  Participation on intra-day market can reduce the imbalance costs with about 20%.

33 | Future wind power forecast errors and associated costs in the Swedish power system | Fredrik Carlsson | 2012.02.07

Thank you for your attention!

33