Energy Storage and Renewables: A Cautionary Tale Presentation to the Harvard Electricity Policy Group Paul Denholm March 24, 2014 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 My Opinions* Storage is undervalued in today s market place But at today s prices, storage isn't a nobrainer Renewables increase the value for storage, but other options are probably more cost effective in the near term There appear to be substantial market challenges for storage revenue capture Price-suppression effects Need for ISO optimization of dispatch *Some of which are actually fact-based
Capacity (MW) 3 Ancient History Pumped Storage CAES (Macintosh) 20,000 18,000 16,000 14,000 12,000 10,000 8,000 6,000 4,000 2,000 0 1950 1960 1970 1980 1990 2000 Year
Historical Planning of Storage Storage as part of the integrated resource planning process Compares a new storage plant to an alternative generation resource (oil or gas fired steam plant) Assume approximately equivalent performance (capacity factor, grid services etc) Assume low cost charging from coal or nuclear power Assume increasing cost of natural gas and oil Restrictions on use of oil and natural gas (Power Plant and Industrial Fuel Use Act) Low-efficiency oil and steam gas plants as opposed to today s efficient gas turbines 4 4
Revised Interest in Energy Storage Advances in storage technologies Volatility in fossil fuel prices T&D siting challenges Perceived need for storage with renewables Emergence of electricity markets Puts value on operating reserves Mandates 5 5
Does wind and solar need storage? The wind doesn t always blow and the sun doesn t always shine? Increased reserves requirements? Curtailment and negative LMPs? Can the duck pay for energy storage? 6 6
So what is storage worth? 1) Identify market opportunities 2) Apply valuation approaches 7 7
Applications of Energy Storage Application Load Leveling/ Arbitrage Description Purchasing low-cost off-peak energy and selling it during periods of high prices. Timescale of Operation Response in minutes to hours. Discharge time of hours. Firm Capacity Operating Reserves Provide reliable capacity to meet peak system demand. Must be able to discharge continuously for several hours or more. Regulation Contingency Spinning Reserve Replacement/ Supplemental Fast responding increase or decrease in generation (or load) to respond to random, unpredictable variations in demand. Fast response increase in generation (or decrease load) to respond to a contingency such as a generator failure. Units brought on-line to replace spinning units. Unit must be able to respond in seconds to minutes. Discharge time is typically minutes. Service is theoretically net zero energy over extended time periods. Unit must begin responding immediately and be fully responsive within 10 minutes. Must be able to hold output for 30 minutes to 2 hours depending on the market. Service is infrequently called. [2] Typical response time requirement of 30-60 minutes depending on market minutes. Discharge time may be several hours. Ramping/Load Following T&D Replacement and Deferral Black-Start End-Use Applications TOU Rates Demand Charge Reduction Backup Power/ UPS/Power Quality Follow longer term (hourly) changes in electricity demand. Reduce loading on T&D system during peak times. Units brought online to start system after a system-wide failure (blackout). Functionally the same as arbitrage, just at the customer site. Functionally the same as firm capacity, just at the customer site. Functionally the same as contingency reserve, just at the customer site. Response time in minutes to hours. Discharge time may be minutes to hours. Response in minutes to hours. Discharge time of hours. Response time requirement is several minutes to over an hour. Discharge time requirement may be several to many hours. [3] Same as arbitrage. Same as firm capacity. I nstantaneous response. Discharge time depends on level of reliability needed by customer. National Renewable Energy Laboratory 8 Innovation for Our Energy Future 8
Energy Storage in Restructured Markets Application Load Leveling/ Arbitrage Firm Capacity Regulation Reserves Spinning Reserves Valued in Restructured Markets? Yes Via scarcity pricing or combined scarcity plus capacity markets. Suffers from missing money problem. Yes, with potentially increased compensation for fast response through FERC 755 initiated market reforms Yes Replacement/Supplemental/ Non-Spinning Primary Frequency Response / Inertia Ramping/Load Following Transmission Replacement and Deferral All Distribution Specific Applications Renewable Integration End-Use Applications National Renewable Energy Laboratory Yes but values are very low No. Early stage proposals No. Proposed in several markets Only partially via congestion prices No. Will likely remain cost of service through regulated entities Captured through other services. Only via rate structure, perhaps combined with aggregated wholesale services (adds transaction costs) 9 Innovation for Our Energy Future 9
Valuation Approaches Value using historical market data (merchant price-taker) Full system value using production cost models (operational value to a utility or value to society, with some insight into market value) 10 10
11 Price-Taker Value in Restructured Markets Use historical market data to estimate what a storage plant would have received if optimally dispatched (big caveat) Typically Single Unit Optimal Dispatch Simulations (Price Taker) Based on historical price and load patterns Use a mixed-integer linear program or other optimization routine Typically assumes perfect foresight Can evaluate some price-suppression impacts (using price load relationships)
Wholesale Energy Price ($/MWh) 12 Example - Load Leveling & Arbitrage 200 180 160 140 120 100 80 60 40 20 0 Summer Winter Spring 0 6 12 18 24 Hour 12
Example: Storage in PJM 75% AC-AC Efficiency Perfect foresight of prices for 1 week 13 13
Net Energy Sold (% of Power Capacity) Electricity Price ($/MWh) 14 Optimal Dispatch 200 90 150 80 100 70 50 60 0 50-50 40-100 30-150 -200 Net Energy Sold Electricity Price Thu Fri Sat Sun Mon Tue Wed Day 20 10 Energy Arbitrage in PJM
Value of Storage ($/kw-year) 15 Value 140 120 100 80 60 40 20 0 2 4 6 8 10 12 14 16 18 20 22 24 26 28 30 32 34 36 38 40 Hours of Storage Energy Arbitrage in PJM 2002 2003 2004 2005 2006 2007
Value of Storage (% of Maximum) Marginal Value of Storage ($/kwh-year) 16 Sizing Optimization 100 95 Total Value 90 85 80 75 70 65 60 55 50 2002 2003 2004 2005 2006 2007 2 4 6 8 10 12 14 16 18 20 22 24 26 28 30 Hours of Storage 25 20 15 10 2002 2003 2004 2005 2006 2007 5 Marginal Value 0 2 4 6 8 10 12 14 16 18 20 Hours of Storage
Locational Variation Arbitrage Value 2004 2005 2006 2007 Sioshansi, Denholm & Jenkin (2009) The Value of Electricity Storage in PJM: Arbitrage and Some Welfare Effects. 17
Value of Storage ($/kw-year) % of Theoretical Value Captured 18 Forecasting & Uncertainty Analysis 120 100 Backcasting Perfect Foresight Accuracy 100 80 95 60 40 90 20 0 2002 2003 2004 2005 2006 2007 Year 85
Arbitrage Estimates Location Years Evaluated Annual Value ($/kw) Assumptions PJM a 2002-2007 $60-$115 12 hour, 80% efficient device. Range of efficiencies and sizes evaluated NYISO b 2001-2005 $87-$240 (NYC) $29-$84 (rest) 10 hour, 83% efficient device. Range of efficiencies and sizes evaluated. USA c 1997-2001 $37-$45 80% efficient device, Covers NE, No Cal, PJM CA d 2003 $49 10 hour, 90% efficient device. CA f 2010-2011 $25-41 4 hour, 90% efficient device a Sioshansi et al. 2009 b Walawalkar et al. 2007 c Figueiredo et al. 2006 d Eyer et al. 2004 f Byrne and Silva-Monroy 2012 19 19
Required Storage Capital Cost ($/kw) Supported Capital Cost 1600 1400 1200 1000 Fixed Charge Rate 9.8% 12.0% 13.9% 800 600 400 200 0 0 50 100 150 Annual Benefit of Storage ($/kw) 20 20
Current focus on regulation applications Higher Value Capacity factor closer to 100% Cheaper device (shorter duration) But very small market 21 21
22 Example Market Prices for Spin and Regulation Average Market Clearing Price $/MW-hour 2005 2006 2007 2008 2009 2010 2011 2012 California ISO Regulation (Up + Down) 35.2 38.5 26.1 33.4 12.6 10.6 16.1 10.0 Spinning 9.9 8.4 4.5 6.0 3.9 4.1 7.2 3.3 Electric Reliability Council of Texas Regulation (Up + Down) 38.6 25.2 21.4 43.1 17.0 18.1 31.3 9.2 Responsive 16.6 14.6 12.6 27.2 10.0 9.1 22.9 9.1 New York ISO (east) Regulation 39.6 55.7 56.3 59.5 37.2 28.8 11.8 10.4 Spinning 7.6 8.4 6.8 10.1 5.1 6.2 7.4 6.0 30 Minute 0.4 0.6 0.9 1.1 0.5 0.1 0.1 0.3 Midwest ISO (day ahead) Regulation 12.3 12.2 10.8 7.8 Spinning 4.0 4.0 2.8 2.3 ISO New England Regulation + mileage 30.2 22.7 12.7 13.8 9.3 7.1 7.2 6.7 Spinning 0.3 0.4 1.7 0.7 1.8 1.0 1.7 10 Minute 0.1 0.3 1.2 0.5 1.6 0.4 1.0 30 Minute 0.0 0.1 0.1 0.1 0.4 0.3 1.0
Required Storage Capital Cost ($/kw) Historical Value of Energy Storage in U.S. Markets 4000 3500 3000 2500 Capital Charge Rate 9.8% 12.0% 13.9% 2000 1500 Arbitrage Only 1000 Regulation Only 500 0 Contingency Only 0 50 100 150 200 250 300 350 Annual Benefit of Storage ($/kw) 23 23
Limits of Price-Taker Analysis 1. Examines a static historic system Difficult to examine impact of fuel prices, renewables, extract impact of scarcity prices etc. 2. Depth of Market 3. Doesn t consider values not captured in today s markets Capacity payments, avoided start costs. 24 24
System Value Analysis Example PLEXOS SCUC Simulation Can examine production costs as well as simulated market revenue Example Energy Only Device 75% AC-AC efficiency No minimum generation, ramp constraints 8 hours Base device is 300 MW 25 25
Storage Generation (MW) System Marginal Price ($/MWh) Storage Generation (MW) System Marginal Price ($/MWh) Results Storage Dispatch (Energy) 350 60 300 50 250 200 150 40 30 Storage Generation System Marginal Price February 3-5 100 20 50 10 0 0 0 12 24 36 48 60 72 Hour 350 60 What s going on here? July 19-21 300 250 200 150 50 40 30 Storage Generation System Marginal Price 100 20 50 10 0 0 0 12 24 36 48 60 72 Hour 26
Value of Avoided Starts 27
28 Results Fuel/Generation Generation (GWh) Base Case With Storage (300 MW) Increase with Storage Coal 46,134 46,375 241 Hydro 3,792 3,792 - Gas CC 14,761 14,947 186 Gas CT 1,024 763-260 Other 103 89-14 Existing Pumped Storage 1,054 1,050-4 New Storage - 465 465 PV 1,834 1,834 0 Wind 10,705 10,705 0 Total Generation (GWh) 79,407 80,020 613 Fuel Use (1,000 MMBtu) Coal 488,140 490,930 2,790 Gas 126,651 124,728-1,923 Total Fuel 614,719 615,658 867
29 Results Change in Production Costs (300 MW Device) Base Case With Storage (300 MW) Increase with Storage Total Fuel Cost (M$) 1,210.5 1,204.7-5.8 Total Variable Operations and Maintenance Cost (M$) 152.1 152.8 0.7 Total Start Cost (M$) 58.2 52.8-5.5 Total Regulation Adder Cost (M$) 4.7 4.8 0.1 Total Production Cost (M$) 1,425.6 1,415.1-10.5 ~50% of value from avoided starts Price taker value - $8.5 million Market value - $5.2 million
30 Market Challenge 1: Starts and Scheduling Large fraction of storage value may come from avoided starts, which are not captured in LMPs You can t self-schedule to avoid starts I don t see any way to maximize value without complete ISO optimization of storage
31 Capacity Value Storage is a Peaker A storage plant with 6-8 hours of capacity is functionally equivalent to a CT for capacity planning purposes. But should short duration A/S devices receive capacity payments?
32 Market Challenge 2: Capacity Insert standard discussion of missing money problem Scarcity pricing? Perhaps, but how does a storage plant effect on-peak prices?
33 Impact of RE Penetration Increased number of starts Decrease in both on-peak and off-peak prices Increased reserve requirements
Value as a Function of RE Penetration 34
35 Source of Value Difference $4.1/MMBTU $8.2/MMBTU
36 Market Challenge 3: Price Suppression Storage is uniquely exposed to price suppression in BOTH directions
System Net Load System Marginal Price ($/MWh) 37 Storage is uniquely exposed to price-suppression effects 10000 9000 8000 7000 6000 5000 4000 3000 2000 1000 0 Curtailed wind 0 6 12 18 24 Curtailed solar Hour Net Load Marginal Price 60 50 40 30 20 10 0 A day sometime in the future with a significant penetration of wind and solar.. What would the value of storage be on this day?
System Net Load 38 Storage Impact on Net Load and Value 10000 9000 Avoided curtailment Avoided peaking generation 8000 7000 6000 5000 4000 3000 0 6 12 18 24 Hour Net Load with storage Net Load Before Storage Added 300 MW storage device. Absorbs 2,400 MWh of otherwise curtailed solar and wind energy Avoids 1,800 MWh of combustion turbine generation. At 9,500 BTU/kWh the storage device saves about 17,000 MMBTU of gas. At $5/MMBTU this is about $85,500 of system value
System Maringal Price ($/MWh) 39 Impact on Storage Revenue 60 50 40 30 20 10 0 0 6 12 18 24 Hour Marginal Prices with Storage Marginal Price Before Storage Storage device must purchase 2,400 MWh of energy at marginal cost of a CCGT (even though it absorbed zero cost wind and solar) at $75,900 Sells 1,800 MWh at marginal costs of CT/CCGT at $75,950 Net revenue is $50, compared to system value of $85,500
Storage Value ($/kw of Storage Capacity) 40 System Value vs. Market Capture 80 70 60 50 40 30 System Value System Value (Curve Fit) Incremental System Value Market Value Market Value (Curve Fit) Incremental Market Value 20 10 0 0 200 400 600 800 1000 Storage Size (MW) Storage is uniquely exposed to price suppression effects in market environments!
Value Capture? 41
42 Conclusions / Opinions 1. The value of storage for energy arbitrage is relatively low and will never pay for almost any storage device in existence. o Arbitrage devices are peaking resources and suffer from the missing money problem. 2. Ancillary services (regulating reserves) appears to be cost effective for a few storage devices, but how deep is the market? o Capacity payments would help, but should short duration devices get them? 3. Storage is undervalued in existing markets and it is still difficult to assess the true value and opportunities for energy storage in the current and future grid 4. Renewables appear to increase the value of storage, but significant challenges to capture this revenue o Need ISO optimized scheduling in both day ahead and real time o Value of avoided starts o Impact of price suppression o Large benefits do not appear until significant curtailment occurs
43 Questions? Paul Denholm paul.denholm@nrel.gov
Contact Paul Denholm Paul.denholm@nrel.gov
Dedicated Renewable Storage? Dedicated renewable storage is generally a non-optimal use Could have scenarios where one storage device is charging while another is discharging simultaneously in the same system Renewable specific applications are already typically captured in grid operations RE Specific Application Transmission Curtailment Time Shifting Forecast Hedging Frequency Support Fluctuation Suppression Whole Grid Application Transmission Deferral Load Leveling/Arbitrage Forecast Error Frequency Regulation Transient Stability 45 45
Storage Caveats Efficiency Not uniformly defined (should be AC-AC, but sometimes stated in terms of DC-DC, which doesn t capture conversion) May not include parasitics CAES (which uses natural gas) and thermal storage cannot be easily compared to pure electricity storage devices such as pumped hydro Cost Many technologies have not been deployed as large scale, so costs are largely unknown Commodity prices affect estimates from different years Difficult to compare devices that offer different services (power vs. energy) 46 46
Charging from Curtailed Wind and Solar 47
Carbon Emissions 48