Cost-Causation and Integration Cost Analysis for Variable...

NREL is a national laboratory of the U.S. Department of Energy, Office of Energy Efficiency and Renewable Energy, operated by the Alliance for Sustainable Energy, LLC.

Cost-Causation and Integration Cost Analysis for Variable Generation

Michael Milligan, Ph.D. National Renewable Energy Laboratory

National Renewable Energy Laboratory Innovation for Our Energy Future

About this presentation

•  Information in this presentation is taken from “Cost-Causation and Integration Cost Analysis for Variable Generation,” Milligan, M.; Ela, E.; Hodge, B. M.; Kirby, B.; Lew, D.; Clark, C.; DeCesaro, J.; Lynn, K.

•  http://www.nrel.gov/docs/fy11osti/51860.pdf

2


Outline

•  Power system operation: variability and uncertainty

•  Cost-causation and integration tariffs

•  Thought experiments: testing tariffs

•  Conclusions

3


Time scale for power system operation

4


Additional ramping/rangemore flexibility

10000

8000

6000

4000

2000

0

MW

160140120100806040200Hours (1 week)

800

600

400

200

0

-200

-400

-600

-800

Ram

p (M

W/h

our)

Net load (load-wind) Additional ramping needs with wind Maximum/Minimum Load Peak Load

5


Integration costs: wind and solar

•  Wind and solar generation increase the variability and uncertainty in power systems operation

•  Solar and wind integration issues are similar –  Wind is becoming reasonably well

understood –  Solar

•  PV has high potential for short-term variability from cloud variations, but the impact of large PV plants is largely unknown because of limited experience with small plants

•  CSP without storage has some thermal inertia and is likely less variable than PV

•  CSP with storage is thought to be much less of an integration challenge but still unknown

•  Cycling efficiency •  Are not unique to wind or solar

6


Variability and Uncertainty Variability: Wind and solar generator outputs vary on different 3me scales as the intensity of their energy sources (wind and sun) Uncertainty: Wind and solar genera3on cannot be predicted with perfect accuracy

0

5000

10000

15000

20000

25000

0 6 12 18 24 30 36 42 48Hours

MW

Wind ActualForecasted Wind

0

500

1000

1500

2000

2500

0 6 12 18 24 30 36 42 48Hours

MW

CSP ActualCSP Forecasted

Variability Uncertainty

Variability: load varies throughout the day, conven3onal genera3on can o=en stray from schedules Uncertainty: Con3ngencies are unexpected, load forecast errors are unexpected

7


Integration cost of wind and solar •  Can it be measured? •  If so, how is it defined? •  What is the proper

benchmark unit? •  How are cost and value

untangled? •  What about units in one

region that economically respond to needs in another region?

•  Are there integration costs for other units? –  Do all AGC units follow the

signal? –  Are there efficiency costs of

adding conventional generators?

100x103

50

0

-50Del

ta M

arke

t Val

ue ($

)

160140120100806040200

-80x103

-60

-40

-20

0

20

40

Del

ta M

arke

t Val

ue ($

)

-10

-5

0

5

10

Over or U

nder-estimate of W

ind Value ($/MW

h)

-10

-5

0

5

10

(Flat block value) - (wind value)

Daily flat energy block ($52.33/MWh) Daily flat block difference $/MWh (right) 6-Hour flat energy block ($48.59/MWh) 6_Hour flat energy block difference $/MWh (right)

(Wind: $48.98/MWh)

Related reports: Milligan, M.; Kirby, B. (2009). Calcula3ng Wind Integra3on Costs: Separa3ng Wind Energy Value from Integra3on Cost Impacts. 28 pp.; NREL Report No. TP-‐550-‐46275. hXp://www.nrel.gov/docs/fy09os3/46275.pdf Milligan, M.; Ela, E.; Lew, D.; Corbus, D.; Wan, Y. H. (2010). Advancing Wind Integra3on Study Methodologies: Implica3ons of Higher Levels of Wind. 50 pp.; NREL Report No. CP-‐550-‐48944. hXp://www.nrel.gov/docs/fy10os3/48944.pdf

8


How are integration costs calculated?

•  Compare two (or more) alternative simulations of the power system using production simulation/cost models –  With wind/solar –  Without wind/solar

•  To provide an energy-equivalent basis, a hypothetical unit is often chosen for the “without wind/solar” case

•  This proxy resource may introduce unintended consequences

•  It is natural to ask about integration costs, but extremely difficult, if not impossible, to measure them accurately

9


The flat-block proxy resource distorts the value of the energy

-300x103

-200

-100

0

100

200

300

Val

ue (

$)

160140120100806040200

-150x103

-100

-50

0

50

100

150

Del

ta M

arke

t

-10

-5

0

5

10

Average O

ver or Under-estim

ate of Wind V

alue ($/MW

h)-10

-5

0

5

10

6000

5000

4000

3000

2000

1000

0

Win

d/D

aily

Blo

ck (

MW

)

Flat block value - wind value

Wind Generation Daily wind-equivalent energy block Daily flat energy block $43.12/MWh Daily flat block difference $1.06/MWh 6-Hour flat energy block $42.18/MWh 6-Hour flat energy block difference $0.11/MWh

(Wind: $42.06/MWh)

10


Total system costs or integration costs

•  Total operating costs are relatively easy to calculate

•  Integration costs are difficult to calculate correctly

•  Both of these are sensitive to assumptions about the other parts of the power system –  What is the mix of conventional generation? –  What is the transmission build-out (if any)? –  What are the institutional constraints? –  Electrical footprint? –  Do markets allow access to physical capability that

exists, or is this access constrained? –  What will the power system look like in 20xx?

11


Are there other sources of integration costs?

•  Contingency reserves •  Conventional units may impose additional

variability and uncertainty that must be managed

•  Interaction between generators in the economic dispatch process can result in generator A imposing a cost on generator B, even if both units are “conventional”

•  Gas purchase/nomination requirements

12


Contingency reserves

•  Specific rules vary, but the contingency reserve is typically set by the largest unit in the pool.

•  Often the specific reserve allocation is based on load ratio share or other similar metric

•  When the largest unit is replaced by a still larger unit, contingency reserve obligations increase

•  à if I am a generation owner/operator, I will find my contingency reserve obligation may increase independently of any action I have taken (or not taken)

13


Contingency reserve costs could be allocated based on generators’ contribution to contingency reserve activation…but this is not done

14


Conventional units may impose regulation costs

Two similar coal fired generators: both are trying to provide regula;on but the upper generator is following dispatch instruc;ons fairly well providing regula;on while the lower generator is not and is imposing a regula;on burden on the power system.

15


New, low-cost base-load may cause integration costs

1. Coal is operated as base-load unit

2. With new wind generation added, gas and coal cycling increase and capacity factors decline

3. Instead of adding wind, a new, cheap base-load technology is introduced. Coal cycling increases; gas is nearly pushed out. Both coal and gas have lower capacity factors.

16


Gas nominations

•  Day-ahead nominations •  Week-end (or holiday

weekends) can pose challenges because of long forecast horizons and uncertainty, and can increase costs and/or limit flexible use of gas generation

•  This is an institutional issue and is unrelated to the capability of the gas generation

17


Principles of Cost-causation: 1

•  Maintaining reliability is critical •  If tariffs are based on costs, they provide

transparency and can induce desired behavior

18



•  Individuals who cause costs should pay •  Individuals who mitigate (reduce, eliminate)

costs should either incur a lower cost, or be paid for helpful actions

•  Complex systems like electric grids product both joint produces and joint costs that must be allocated among the users of the system

•  Joint costs can be recovered base on the principle of “relative use.”

19



•  Tariffs should not collect revenue if no cost is incurred

20



•  Tariffs should be based on the physical characteristics of the power system

•  Aggregate load and generation must be balanced

•  It is un-necessary and usually quite costly to balance individual loads or resources, and this is inconsistent with the way the power system is operated

21



•  Tariffs should result in an efficient allocation of resources

•  This can be tested: is there another way that the required services can be supplied at less cost? Or is there another way that the system can be planned or operated at less cost?

•  If either of these are true, resources are not efficiently allocated

22


Other characteristics of tariffs

•  Vertical consistency: individuals who impose higher costs should be assessed more than an individual who imposes lower (or no) cost

•  Horizontal consistency: individuals who cause similar (identical) costs should be assessed similar (identical) costs

A: High cost

B: Low cost

A and B have similar cost contributions

A B

23


The sum of all parts physically cannot exceed the whole •  Methods that separate

regulation, following, uncertainty for the analysis must follow the principle of re-composition.

•  à The sum of –  Regulation –  Following –  Uncertainty

•  Components must combine so that they do not exceed the total variability + uncertainty…

•  Sum of all parts of the tariff revenue cannot exceed total costs

24


Thought experiments: How can tariffs be tested to see how they behave?

25


Thought experiment #1: Block schedules and regulation •  Useful to test the behavior of proposed, or actual, tariffs •  How does the tariff treat perfect following of a volatile

schedule?

17000

19000

21000

23000

25000

27000

29000

31000

00:00 03:00 06:00 09:00 12:00 15:00 18:00 21:00 00:00

Time

To

tal L

oad

(M

W)

200

400

600

800

1000

1200

1400

1600

Sch

edu

led G

eneratio

n (M

W)

LoadPure block schedule

Block schedule

Load

200

400

600

800

1000

1200

1400

1600

00:00 03:00 06:00 09:00 12:00 15:00 18:00 21:00 00:00Time

Gen

erat

ion

sch

edu

le (

MW

)

-350

-250

-150

-50

50

150

250

350

Gen

eration

ramp

rate (MW

/min

)

Pure Block Schedule

Generation ramp rate

26


Thought experiment #2: Ramping

•  Should a tariff quantify peak-to-peak movements of generator or load?

•  Ramping the block schedule does not impact the energy delivery or forecast accuracy but reduces regulation requirements.

17000

19000

21000

23000

25000

27000

29000

31000

00:00 03:00 06:00 09:00 12:00 15:00 18:00 21:00 00:00

Time

Tota

l Loa

d (M

W)

200

400

600

800

1000

1200

1400

1600

Scheduled G

eneration (MW

)

LoadRamped block scheduleBlock schedule

Load

200

400

600

800

1000

1200

1400

1600

00:00 03:00 06:00 09:00 12:00 15:00 18:00 21:00 00:00Time

Gen

erat

ion

sche

dule

(MW

)

-35

-25

-15

-5

5

15

25

35

Generation ram

p rate (MW

/min)

Ramped Block Schedule

Generation ramp rate

27


Thought experiment #3: Ramp metric •  Some approaches to assessing ramping needs (or supply) may not

produce desired result •  Red: regulation, lots of small movements •  Green: longer time interval but essentially energy-neutral •  Blue: likely the most challenging

•  If considered in isolation, does not capture what the system must do

0

10

20

30

40

50

60

0 10 20 30 40 50 60Minutes

MW

28


Thought experiment #4

•  Equal but opposite behavior is benign to the power system operator

•  Even though #4 may not be realistic, it can identify tariffs that over-charge based on this principle

10000

15000

20000

25000

30000

00:00 03:00 06:00 09:00 12:00 15:00 18:00 21:00 00:00

Time

Tota

l Loa

d (M

W)

0

500

1000

1500

2000

Generation (MW)

System LoadWind GeneratorMirror Wind Generator

29


Thought experiment #5

•  How does the tariff assess beneficial movement? •  For example, would both coal plants be paid the

same amount?

30


Other considerations

•  Does the tariff recognize all cost-causers? •  Does the tariff recognize all helpful actions

(intended or otherwise)?

31


Common errors in integration analysis

•  Double-counting •  Assuming fixed schedules/resources that may

be variables in the long-run •  Balancing individual actors •  Scaling •  How are wind/solar forecasts simulated? •  Excessive or unknown implied CPS

performance •  Assumptions regarding replacement power

sources and costs

32


Common errors in integration analysis

•  Constant reserves (wind and solar generation cannot be less than zero, nor greater than rated)

•  Failure to release following (or related reserves) when they are called on

•  Excessive lead times prior to the dispatch period

•  Assuming specific fleet characteristics (limited turn-down, for example) for future scenarios

•  Generally – nearly any aspect of the system may change in the future. Assuming all else constant may drive results

33


Large BA Geographically Dispersed Wind and Solar

Wind/Solar Forecasting Effectively Integrated Into System Operations Sub-Hourly Energy Markets

Fast Access to Neighboring Markets NonSpinning and 30 Minute Reserves for Wind/Solar Event Response

Regional Transmission Planning For Economics and Reliability Robust Electrical Grid

More Flexible Transmission Service Flexibility in Generation

Responsive Load Overall

Example Utility Structures10 8 7 10 7 2 7 6 7 7 3 7 Large RTO with spot markets

6 6 6 3 3 2 6 4 7 2 2 4 Smaller ISO

1 3 2 1 2 1 2 3 2 2 2 2 Interior west & upper Midwest (non-MISO)

7 6 6 2 2 2 5 4 2 5 2 4 Large vertically integrated utility

1 3 2 1 2 1 2 4 2 2 2 2 Smaller Vertically Integrated Local Utility

8 Unconstrained hydro system

3 Heavily fish constrained hydro system1 1 1 1 1 1 1 1 1 1 1 11 Weightings Factors

Accommodating Wind and Solar Integration

Adapted from Milligan, M.; Kirby, B.; Gramlich, R.; Goggin, M. (2009). Impact of Electric Industry Structure on High Wind Penetra3on Poten3al. 31 pp.; NREL Report No. TP-‐550-‐46273. hXp://www.nrel.gov/docs/fy09os3/46273.pdf

34


Conclusions

•  There is no universal agreement on integration cost methods, or whether these costs are measureable

•  Integration costs are part of normal power systems operation, beyond wind/solar –  Conventional units may impose integration costs –  Performance-based tariffs are more appropriate than

technology-based tariffs, assuming other factors are properly considered

•  There are many potential non-(wind/solar) cases that may be good base cases

•  High penetrations of wind/solar will have an impact on the conventional plant mix and institutional practice

35

Questions?

36

Date post:	23-Apr-2018
Category:	Documents
Upload:	vuongduong
View:	218 times
Download:	3 times

Cost-Causation and Integration Cost Analysis for Variable...

Documents