Assessing the value of improved variable renewable energy ...

1

Assessing the value of improved variable renewable energy forecasting accuracy in the South African power system

WindAc ConferenceCape Town. 5-6 November 2018

Jarrad Wright Greg Landwehr Erol Chartan

- [email protected] CSIR (ZA)

- [email protected] CSIR (ZA)

- [email protected] NREL (USA)

2

Conclusions7

Outcomes6

Some selected scenarios5

System value of VRE forecasting4

Impacts of weather systems on VRE forecasts in South Africa3

Future plans for VRE deployment2

The South African power system and recent VRE deployment1

Overview

3

Conclusions7

Outcomes6






Overview

4

South African power system is coal dominated but has recently begun supplementing this with variable renewables (wind and solar PV)

Sources: CSIR analyses (energy estimated based on production-cost modelling outcomes, 2017)

5

South Africa has world class wind & solar resource and can leverage these for future electricity requirementsRenewable energy resource maps for South Africa

Interim (5 km) High-Resolution Wind Resource Map for South Africa, Metadata and further information, SANEDI, Oct 2017; Global Horizontal Irradiation Map, Solargis,

http://www.sapvia.co.za/sa-solar-irradiation-maps/, accessed May 2018;

Wind Speed (m/s) Irradiance (kWh/m2)

7

257560

1 075

300 300210

960

965

1 460

2 078 2 078

1 474

1 474 1 474

2017 2020

3 852

20162013 20192014 H1 2018

3 134

2015

200

467

1 520

2 040

3 852

+1 053

+520

+1 094

+718 +0

Solar PV

Wind

CSP

Supply Sources

Notes: RSA = Republic of South Africa. Solar PV capacity = capacity at point of common coupling. Wind includes Eskom’s Sere wind farm (100 MW). Sources: Eskom; DoE IPP Office

From 1 November 2013 to 30 Jun 2018, 2 078 MW of wind, 1 474 MW of large-scale solar PV and 300 MW of CSP became operational in RSA

Capacityoperational in MW

(end of year)

8

1.1

0.01 0.05

20142013

2.2

2.6

2.5

2015

1.6

3.7

0.5

3.1

2016

3.3

2018

5.0

1.1 0.7

2017 2019

9.0

2020

0.1

2.2

4.7

6.9

9.0

0.5

3.9

Wind

Solar PV

CSP

Estimated2018 H2production

Notes: Wind includes Eskom’s Sere wind farm (100 MW). CSP energy measured from date when more than two CSP plant were commissioned. Wind and solar PV energy excludes curtailment and is thus lower than actual wind and solar PV generation

Sources: Eskom; DoE IPP Office

Annual energy produced in TWh

In 2017, 9.0 TWh from wind, PV and CSP whilst in H1 2018, 5.2 TWh of wind, PV and CSP energy produced in RSA

Supply Sources

10

Conclusions7

Outcomes6






Overview

11

Energy mix is planned to be 17-21% variable renewable energy across all scenarios in Draft IRP 2018 by 2030 and up to >60% by 2050

Energy production [TWh]

22 (7%)

44 (14%)

25 (8%)

42 (13%)

22 (7%)

IRP1

IRP3

14 (4%)

22 (7%)

44 (14%)

IRP5

44 (14%)

IRP6

23 (7%)

35 (11%)

44 (14%)

Rec.

IRP7

Demand: 313 TWh

OtherWind Solar PV

2030

Sources: DoE Draft IRP 2018; CSIR analysis


78 (20%)

IRP5

164 (42%)

IRP6

110 (28%)

IRP1

63 (16%)IRP3

64 (16%)

107 (27%)

67 (17%)

IRP7

106 (27%)

64 (16%)

106 (27%)

Rec.

2040 2050

Demand: 392 TWhDemand: 353 TWh


54 (15%)

47 (13%)

Rec.

IRP5

129 (36%)

90 (25%)

IRP1

IRP3

47 (13%)

47 (13%)

89 (25%)

47 (13%)

90 (25%)

IRP6

IRP7

90 (25%)

13

Conclusions7

Outcomes6






Overview

14 16 July 2017

0

200

400

600

800

1000

1200

0

5

10

15

20

00

:00

00

:30

01

:00

01

:30

02

:00

02

:30

03

:00

03

:30

04

:00

04

:30

05

:00

05

:30

06

:00

06

:30

07

:00

07

:30

08

:00

08

:30

09

:00

09

:30

10

:00

10

:30

11

:00

11

:30

12

:00

12

:30

13

:00

13

:30

14

:00

14

:30

15

:00

15

:30

16

:00

16

:30

17

:00

17

:30

18

:00

18

:30

19

:00

19

:30

20

:00

20

:30

21

:00

21

:30

22

:00

22

:30

23

:00

23

:30 Ir

rad

ian

ce (

W/m

2)

Win

d S

pee

d (

m/s

)

15 16 July 2017

0

5

10

15

20

00

:00

00

:30

01

:00

01

:30

02

:00

02

:30

03

:00

03:3

00

4:0

00

4:3

005

:00

05

:30

06

:00

06

:30

07

:00

07

:30

08

:00

08

:30

09

:00

09

:30

10

:00

10

:30

11

:00

11

:30

12

:00

12

:30

13

:00

13

:30

14

:00

14

:30

15

:00

15

:30

16

:00

16

:30

17

:00

17

:30

18

:00

18

:30

19

:00

19

:30

20

:00

20:3

02

1:0

02

1:3

022

:00

22

:30

23

:00

23

:30W

ind

Sp

eed

(m

/s)

16

Several Major Climate InfluencersSouth African Climate

Warren, M., Climatology – a South African Perspective, http://slideplayer.com/slide/10180352/, accessed May 2018

Summer – October to MarchWinter – April to September

South Africa terrain elevation (SRTM+, NASA version 3)

Dynamics of late Cenozoic aeolian deposition along the South African coast: a record of evolving climate and ecosystems, http://sp.lyellcollection.org/content/388/1/353

17

Several Major Climate InfluencersSouth African Climate

Warren, M., Climatology – a South African Perspective, http://slideplayer.com/slide/10180352/, accessed May 2018

1.

Escarpment height

2000 m to 3482 m

Escarpment

South Africa terrain elevation (SRTM+, NASA version 3)

Dynamics of late Cenozoic aeolian deposition along the South African coast: a record of evolving climate and ecosystems, http://sp.lyellcollection.org/content/388/1/353

Ocean Currents

18

Mid Latitude Cyclones (MLC’s)Challenging Weather Systems to Forecast

SA Weather and Disaster Observation Service, http://sawdis1.blogspot.co.za/2012/10/images-mid-latitude-cyclone-offshore.htmlvan Wyk, E., van Tonder, GJ., Vermeulen, D., Characteristics of local groundwater recharge cycles in South African semi-arid hard rock terrains -rainwater input, Water SA vol. 37 n.2 Pretoria Apr 2011

Mid Latitude Cyclone

19de Villiers, M., Roll cloud on the South African east coast, Weather vol. 66, Issue 2, Jan 2011Lange, M., Evaluation of forecasts, meteorological characteristics in South Africa and lessons learned, Workshop presentation, Eskom, Nov 2017

Berg WindsChallenging Weather Systems to Forecast

Change in wind directionSynoptic ‘Berg Wind’

20

Challenging Weather Systems to ForecastCloud build up on the West coast and interior and Low Level Jet formation (LLJ)

Lange, M., Evaluation of forecasts, meteorological characteristics in South Africa and lessons learned, Workshop presentation, Eskom, Nov 2017

21

16.815.9

16.8

15.1

12.511.5

12.3

14.6

16.9

19.0 19.318.1

15.7

Febru

ary

Janu

ary

Jun

e

Decem

ber

Ap

ril

March

May

July

Au

gust

Septem

ber

Octo

ber

No

vemb

er

Average

Uncertainty with VRE forecasts in South Africa highly dependent on localised weather systems – we have started to understand thisSemi-Operational VRE forecasting model

Sources: CSIR Short Term VRE forecasting model initial results; Energy and Meteo Systems intra-day and day-ahead high level forecasts for existing wind and solar PV generators

(LANGE, 2018)

13.1 13.4 13.7 13.6

12.311.4

13.3 13.013.7 14.0

13.1 12.8 13.1

Ap

ril

Janu

ary

March

Febru

ary

Au

gust

May

July

Jun

e

Decem

ber

Septem

ber

Octo

ber

No

vemb

er

Average

Wind(MAE forecast, %)

Solar PV(MAE forecast, %)

22

Conclusions7

Outcomes6






Overview

25

15

0

10

5

20

25

30

35

VRE Uncertainty results in different residual demand – worsened at high penetration levels

Monday Tuesday Wednesday Thursday Friday Saturday Sunday

Sources: CSIR analysis

Demand

GW e.g. 10 GW installed wind and solar PV capacity

26

20

15

0

5

10

25

30

35


Demand

Residual demand (DA)

GW

Subtracting forecasted wind & solar PV = residual demand


e.g. 10 GW installed wind and solar PV capacity


27

20

0

5

30

15

10

25

35

0

5

25

15

30

10

20

35

2.7 GW(+7%)


Demand

Residual demand (DA)

Residual demand (RT)


Solar PV (DA)

Wind (DA)

GW e.g. 10 GW installed wind and solar PV capacity


28

Day 0

Day 0

DAUCED

(1 step of 1 day)

RTUpdated UCED

(24 steps of 1 hour)

Improved forecastUnit commitmentDispatchReserves

Day 1

DAUCED

(1 step of 1 day)

Day 1

RTUpdated UCED


Note: UCED = Unit Commitment and Economic Dispatch; DA = Day-ahead, RT = Real-time

Day 2

DAUCED

(1 step of 1 day)

Day 2

RTUpdated UCED


. . .

. . .

Improved forecastUnit commitmentDispatchReserves

Using a production cost model of a representative South African system – establish the change in system costs as the forecast improves

29

Conclusions7

Outcomes6






Overview

30

Some scenarios explored to understand the value of VRE forecast improvements

31

Conclusions7

Two US jurisdictions6.2

South Africa6.1

Outcomes6






Overview

32

Conclusions7


South Africa6.1

Outcomes6






Overview

33

0.0% 10.0%5.0% 15.0% 20.0%

0.00%

0.10%

0.20%

0.30%

0.40%

VRE penetration level[% of annual energy demand]

Production cost saving[%]

Current RSApower system

Value of an improved VRE forecast is already notable at low VRE penetration – increases with increasing VRE penetrationRelative cost difference with perfect foresight and forecast uncertainty

Relative to the state-of-the-artforecast available in South Africafor VRE (solar PV and wind), animproved forecast could result innotables savings

20% improvement:

≈ 0.02 - 0.12% of production costs≈ 1.4 – 5.8 USD-million/yr1

40% improvement:

≈ 0.04 - 0.21% of production costs≈ 2.3 – 10.1 USD-million/yr1

1 USD:ZAR = 13.32 (2017-average)


20% improvement

40% improvement

34

Conclusions7


South Africa6.1

Outcomes6






Overview

35

Two very different jurisdictions in the U.S. – previous studies quantified the value of VRE forecast

T e c h n o lo g y I n s t a l le d

C a p a c it y [M W ]

E n e r g y s h a r e

e s t . [ % ]

C o a l 5 7 5 0 0 4 8

N u c le a r 1 3 0 0 0 1 6

G a s 5 9 9 0 0 2 4

O i l 4 1 5 0 < 0 .1

H y d r o 3 3 0 0 2

W in d 1 4 7 0 0 9

S o la r P V 4 0 0 < 1

B io m a s s 2 2 4 < 1

P u m p e d S to r a g e 2 5 1 8 -




e s t . [ % ]

C o a l 5 2 0 .2

N u c le a r 2 6 9 4 9

G a s 4 2 2 2 7 4 3

O i l 3 5 2 < 0 .1

H y d r o 1 4 0 0 2 2 1

W in d 5 6 3 2 6

S o la r P V 9 5 8 8 1 1

B io m a s s 1 3 1 4 2

P u m p e d S to r a g e - -

G e o th e r m a l 2 6 9 4 6

MISO CAISO

Sources: NREL; U.S. DoE; CAISO; MISO

Approach for both jurisdictions

Varied VRE penetration: 12-56%VRE forecast improvement: 0%, 20%, 40%

36

System costs savings range as a result of the differing energy mixes –MISO quite similar to South Africa

≈ 0.20% of production costs

≈ 0.25-0.35% of production costs






e s t . [ % ]

C o a l 5 7 5 0 0 4 8

N u c le a r 1 3 0 0 0 1 6

G a s 5 9 9 0 0 2 4

O i l 4 1 5 0 < 0 .1

H y d r o 3 3 0 0 2

W in d 1 4 7 0 0 9

S o la r P V 4 0 0 < 1

B io m a s s 2 2 4 < 1

P u m p e d S to r a g e 2 5 1 8 -




e s t . [ % ]

C o a l 5 2 0 .2

N u c le a r 2 6 9 4 9

G a s 4 2 2 2 7 4 3

O i l 3 5 2 < 0 .1

H y d r o 1 4 0 0 2 2 1

W in d 5 6 3 2 6

S o la r P V 9 5 8 8 1 1

B io m a s s 1 3 1 4 2

P u m p e d S to r a g e - -

G e o th e r m a l 2 6 9 4 6

MISO CAISO

20% improvement (10-20% VRE penetration)

40% improvement (10-20% VRE penetration)

Sources: NREL; U.S. DoE; CAISO; MISO

37

Conclusions7

Outcomes6






Overview

38

Conclusions

South Africa has started deploying VRE (solar PV and wind) with plans for significant further expansion

Value of improved VRE forecast in representative South African system at a national scale has been quantified

Value of improved VRE forecast in South Africa increases with increasing VRE penetration levels (as expected)

Value is ≈0.02-0.12%1 and ≈0.04-0.21%2 of production costs for VRE ranges considered (0-20% - by energy)

When comparing to representative U.S power systems (MISO, CAISO) - outcomes show more correlation andalignment with the coal-dominated MISO region than that of gas-dominated CAISO region (as expected)

Future research is necessary for higher variable renewable energy penetration levels

A number of unique weather systems exist in South Africa – a better understanding these can facilitate thepresented improvements to VRE forecasts

1 For 20% VRE forecast improvement; 2 For 40% VRE forecast improvement

39

Thank you

Assessing the value of improved variable renewable energy ...

Documents

Transcript of Assessing the value of improved variable renewable energy ...