A R EVIEW OF L ARGE -S CALE R ENEWABLE E LECTRICITY I NTEGRATION S TUDIES Paulina Jaramillo, Carnegie Mellon University And Paul Hines, University of Vermont.

A REVIEW OF LARGE-SCALE RENEWABLE ELECTRICITY

INTEGRATION STUDIES

Paulina Jaramillo, Carnegie Mellon University

And

Paul Hines, University of Vermont

Introduction• 33 States have developed

Renewable Portfolio Standards

• Many RPS call for large percentages (~20%) of Renewable electricity

• Wind is the fastest growing renewable source

• Wind: Intermittent and Variable

2

0

10

20

30

40

50

60

Bill

ion

kWh

USA Wind Production

www.renewelec.org

Integration Studies

• Several recent studies evaluate the impacts of renewables on grid operations, and identify strategies to mitigate these impacts.

• We performed a systematic review of recent integration studies, focusing on wind

Goals of our review• What grid, wind data

were used?• Evaluate methodology for

estimating– Wind power variation– Reserves requirements– Regulation requirements

• Identify research gaps

3

NYSERDA 2005• 3,300 MW of Wind in New York State.• The analysis separates among different time

scales.• Brief analysis of forecast value.• Main recommendation: Wind farms build voltage

controls and low voltage ride through capability. • Major Concern: Use of Gaussian methods for

reserve calculations– Conclusions based largely on measured standard

deviation and mean

4

5

Real wind data, 31% CF, Std. Dev. ΔP = 21 MW

Gaussian data, 31% CF, Std. Dev. ΔP = 21 MW

Empirical comparison of real wind data and “Normal” wind data

The Gaussian assumption dramatically underestimates the probability of multiple sequential large changes in the same direction.

6

2006 Minnesota Wind Integration Study

• 15%, 20%, and 25% wind integration in MISO for the year 2020.

• Conclusion:– Penalty for variability between $2 and $4 per MWh.– Increasing spatial diversity reduces the number of

“no-wind power” events, reserves requirements.

• Concerns:– Gaussian methods for reserves calculations.– Analysis gap for short term modeling.

7

2007 CAISO Wind Integration Study

• Modeled theoretical wind plants in California and identified transmission requirements.

• Conclusion:– Using Types 3 and 4 turbines will allow for reliable

wind integration.

• Concern:– Use of Gaussian methods for reserve calculations.

2008 NREL’s 20% Wind by 2020

• Not really an integration study, but a projection of technology and economic requirements to achieve 20% wind by 2030.

• Good comparison of available wind power at various wind speed class levels.

• Recommendation: Build transmission

• Concern: Transmission system modeling not based on Kirchhoff’s & Ohm’s laws

8

2008 ERCOT Wind Integration

• Analysis of impact of wind generation on net load.

• Conclusions:– Wind AND load are variable and out-of-phase.– Seasonal variations exist.– Reserve and regulation requirements increase with

increased wind power.

• Concern:– Use of Gaussian methods for reserve calculations.– No grid model.

9

2009 Trade Wind Integration Study - Europe

• Study focused on transmission flows to identify transmission needs.

• Assumes that regional diversity is sufficient to deal with the variability of wind power.

• No discussion about reliability and reserves.

• Potentially erroneous finding: “Wind and Load are positively correlated.”

10

2010 Eastern Wind Integration and Transmission Study

• 4 different scenarios with different percentage of wind generation and different wind production locations.

• Use of DC power flow model allows them to identify transmission investment that will be needed at larger wind generation percentages.

• Estimated reserved requirements, forecast error, curtailment and impacts of geographic diversity.

11

2010 SW Power Pool (CRA)

• Study 10%, 20% and (limited) 40% wind penetration.

• Detailed contingency study.

• Based on hourly and limited high-resolution data.

• Conclude that no additional contingency reserves needed.

12

2010 CEC/KEMA study of reserves and regulation

• Analyze 20% and 33% renewable scenarios.

• First large-scale study to include dynamic generator models.

• Conclude that fast-ramping storage is needed to manage ACE and frequency deviations.

13

2010 studies by NERC, CAISONERC analysis of renewables & reliability

• Qualitative study of reliability risks, given renewables, DSM, storage

• Emphasize the need for more load-following during morning and evening ramps

• New technology will require changes to operating policies.

CAISO analysis of 20% renewable in 2012 (PNNL)

• 1-minute wind data data

• Monte-Carlo model to model forecasts

• Regulation estimates based on 1-minute data

• One of the most careful studies reviewed (but still use standard deviations)

• Emphasize need to better understand load-following in morning/evening.

14

Research gaps

• Gaussian statistical methods Frequently conclusions are drawn from the mean and standard deviation of sampled wind data. – Need better models.

• Larger control areas Several studies conclude that aggregating control areas reduces costs. – Further analysis needed.

15

Research gaps

• Meteorological vs. anemometer data – Need empirical research to find the appropriate role

for each.

• Estimation of regulation requirements– Need new methods, for estimating regulation needs,

given accurate wind and solar data.

• Morning and evening ramping– Wind and load are generally anti-correlated during the

morning and evening. Need new operating policies and technology

16