Integrating High Penetrations of Variable Renewable Generation Lori Bird and Debra Lew, NREL NCSL Webinar March 28, 2012 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.
2 Overview Challenges Solutions.Limits? What challenges does variable generation pose? How well can variable renewables be predicted? How can the variability be managed? Is back-up generation needed? What are the costs of integrating variable RE? How much wind/solar can be handled on the system? What is the international experience with managing high penetrations of renewables?
Output Normalized to Mean 3 What Challenges Do Wind/Solar Pose? Wind and solar are: Variable due to daily or seasonal patterns Uncertain due unknown changes in weather (i.e., no wind, cloud cover) Variability is reduced with more resources spread over a wider area (diversity of weather patterns) 1.6 1.4 1.2 1.0 0.8 0.6 1.6 1.4 1.2 1.0 0.8 0.6 1.6 1.4 1.2 1.0 0.8 0.6 0 5 Source: NREL Wind Plant Data 15 Turbines Stdev = 1.21, Stdev/Mean =.184 200 Turbines Stdev = 14.89, Stdev/Mean =.126 215 Turbines Stdev = 15.63, Stdev/Mean =.125 10 15 Seconds 20 25 30x10 3 (Approximately 8 hours)
Loads and Generation Must Be in Balance Each balancing area (BA) must balance supply and demand loads = generation + imports exports 4
System Load (MW) 5 Power System Operations Load fluctuates constantly System operators must constantly match generation and load Power system operators must: 1) make minute-to-minute adjustments to match generation and load 2) follow changes in daily load patterns (minutes to hours) 3) commit power plants to meet load (day-ahead) seconds to minutes Regulation Time (hour of day) 0 4 8 12 16 20 24 tens of minutes to hours Load Following Days Unit Commitment day Scheduling
6 Variability of Wind and Solar Wind Modest minute-to-minute changes Changes over hours more substantial; Impact on following load and committing units Minimum generation can be issue (wind available at night with low loads) PV May have more significant minute-tominute changes (from cloud cover) Predictable daily pattern of generation; but some impact on following load PV Variability Over 1 Day Sunny day, no clouds Cloud cover
How Well Can Variable Generation Be Predicted? 7 Use of forecasts help manage uncertainty Easy to predict wind over short time periods (~5% error hour ahead) More difficult to predict wind day ahead (~8-15% error day ahead) Forecast errors reduced over larger areas (wind farm to balancing area) o Weather is less correlated over a larger area Source: IEA Task 25 Report Source: John Zack, AWS, UWIG 2011
8 What is the Probability of Extreme Events? How often do extreme events occur? Rarely and they are generally slower to occur than contingency events such as loss of power plant Diversity of location of renewables helps ensure that capacity is not all lost at once (weather is different)
9 What happened in Texas, Feb 2008? A frequency drop resulted in a call upon reserves including interruptible loads that are paid to curtail There were 3 major contributors: o Wind generation dropped from 2000 to 360 MW in 3.5 hours o A conventional unit with 370 MW capacity tripped offline o The load forecast was wrong Lessons learned o Load forecast failed to predict the large ramp up in demand o The accurate wind energy forecast was not used in scheduling (this has been rectified) o Demand response was very effective in economically reducing demand o Wind event was a multi-hour ramp event (500MW/hour), not a contingency event. Less expensive resources, such as a 15-minute non-spinning reserves, can be used to address ramp events.
System Load (MW) How Can the Grid Handle Variability? Is 1-1 Back-up Generation Needed? Individual generators are not backed up: but reserves are provided on a system basis Wind and solar can require additional reserves amount of reserves needed changes (not static amount) Generators that change dispatch as a result of wind may have reduced efficiency, but total fuel burn and emissions will decrease seconds to minutes Regulation Time (hour of day) 0 4 8 12 16 20 24 tens of minutes to hours Load Following Days Unit Commitment day Scheduling 10
11 How Can Grid Flexibility Be Increased? Flexibility of Generation o Combustion turbine gas plants, hydro (if not constrained), reciprocating engines o Plants can be modified to ramp more easily o As inflexible generators retire can replace with more flexible plants; may need incentives to adopt most flexible technologies Institutional flexibility o Fast energy markets can help manage fluctuations more easily o Large balancing areas can access more resources to address variability (net out changes in smaller areas) o Use of advanced forecasting techniques Demand response interruptible loads can respond very rapidly, cost-effective form of flexibility Plug-in electric vehicles in the future o Can use electricity off peak to use extra energy when its available (i.e., wind blowing at night) Energy storage o e.g., pumped hydro, compressed air energy storage, batteries
12 Flexibility Options - Cost is a Consideration Source: DOE
What are the Costs of Integrating RE? Integration cost estimates modest ~ ½ cent/kwh Costs can be reduced with forecasting and operational changes 1% improvement in forecast error saved $800k (PSCO) Most generation sources have costs with integrating them on the grid (e.g., nuclear ramping limitations) Year Study Wind Capacity Penetration Regulation Integration Cost ($/MWh) Load Following Unit Commit. Gas Supply TOTAL 2003 Xcel-UWIG 3.5% 0 0.41 1.44-1.85 2003 We Energies 29% 1.02 0.15 1.75-2.92 2004 Xcel-MNDOC 15% 0.23-4.37-4.60 2005 PacifiCorp-2004 11% 0 1.48 3.16-4.64 2006 Calif. (multi-year)* 4% 0.45 trace trace - 0.45 2006 Xcel-PSCo 15% 0.20-3.32 1.45 4.97 2006 MN-MISO** 31% - - - - 4.41 2007 Puget Sound Energy 12% - - - - 6.94 2007 Arizona Pub. Service 15% 0.37 2.65 1.06-4.08 2007 Avista Utilities 30% 1.43 4.40 3.00-8.84 2007 Idaho Power 20% - - - - 7.92 2007 PacifiCorp-2007 18% - 1.10 4.00-5.10 2008 Xcel-PSCo*** 20% - - - - 8.56 2009 Bonneville (BPA) + 36% 0.22 1.14 - - 5.70 2010 EWITS ++ 48% - - 1.61-4.54 2010 Nebraska +++ 63% - - - - 1.75 * Regulation costs represent 3-year average. Source: ** Highest over LBNL 3-year Wind evaluation Market period. Report *** This integration cost reflects a $10/MMBtu natural gas price scenario. This cost is much higher than the integration cost calculated for Xcel-PSCo in 2006, in large measure due to the higher natural gas price: had the gas price from the 2006 study been used in the 2008 study, the integration cost would drop to $5.13/MWh. + Costs in $/MWh assume 31% capacity factor. Aside from regulation and following reserves, the costs of BPA s imbalance reserves are $4.33/MWh. ++ The unit commitment costs listed in EWITS are the cost of day-ahead wind forecast error; the remaining integration costs included in the total are for shorter term variable reserves that account for regulation and short-term forecast errors (energy imbalance). +++ These integration costs only capture regulating reserves and day-ahead forecast error. A sensitivity case in this study shows that integration costs increase if the differences between the actual hourly deliveries of wind energy are compared to daily flat block of power. The increased costs are shown in Figure 39. Sources: Brooks et al. (2003) [Xcel-UWIG]; Electrotek Concepts, Inc. (2003) [We Energies]; EnerNex Corp. and Wind Logics, Inc. (2004) [Xcel-MNDOC]; PacifiCorp (2005) [Pacificorp-2004]; Shiu et al. (2006) [Calif. (multiyear)]; EnerNex Corp. (2006) [Xcel-PSCo]; EnerNex Corp. and Windlogics Inc. (2006) [MN-MISO]; Puget Sound Energy (2007) [Puget Sound Energy]; Acker (2007) [Arizona Pub. Service]; EnerNex Corp. (2007) [Avista 13
How much wind/solar can be accommodated? No physical limit found in studies it depends on system, operations and economics Large regional integration studies show up to 30% (and 5% solar in the west) can be integrated if there is adequate transmission and if operational changes can provide additional flexibility Eastern Wind Integration and Transmission Study, Jan 2010 Western Wind and Solar Integration Study, Mar 2010 14
System Operations and Generation Mix Matter 15 How much can be integrated depends on the system operation and generation mix How flexible is the generating fleet (fast ramp, quick start-up)? Are there fast markets? How large is the area for system balancing? How much demand response? Markets cover only part of U.S.
16 What Levels of RE Have Been Reached? Xcel Energy (CO) wind penetration record: Hourly record: 55.6% October 9, 2011 Daily Record: 37.0% October 8, 2011
What Levels Have Been Achieved Internationally? 17 Lessons from Denmark Participation in broader Nordic power market has enabled balancing of wind; access to hydro and reserves Integrated wind forecasting in system operations Large amounts of flexible combined heat power for district heating Source: Holttinen, 2011 and EIA 2011
18 Summary Wind and solar energy adds additional variability and uncertainty to power system operations Specific back-up generation is not required, but additional reserves may be necessary System flexibility is important for integrating higher penetrations of renewables; a variety of ways to achieve flexibility New operating practices may be needed to achieve higher RE penetrations (larger, faster markets) Much analysis is ongoing to address operational and planning issues
Contact: Lori.bird@nrel.gov
20 Additional Resources Western Wind and Solar Integration Study Eastern Wind Integration and Transmission Study Utility Variable Generation Integration Group www.uwig.org International Energy Agency Task 25 Report: Design and operation of power systems with large amounts of wind power o http://www.vtt.fi/inf/pdf/tiedotteet/2009/t249 3.pdf NREL Transmission and Grid Integration Group o http://www.nrel.gov/wind/systemsintegration o http://www.nrel.gov/publications