The Use of Solar and Wind as a Physical Hedge against Price Variability within a Generation Portfolio Thomas Jenkin, Victor Diakov, Easan Drury, Brian Bush, Paul Denholm, James Milford, Doug Arent, and Robert Margolis National Renewable Energy Laboratory Ray Byrne Sandia National Laboratories NREL is a national laboratory of the U.S. Department of Energy Office of Energy Efficiency & Renewable Energy Operated by the Alliance for Sustainable Energy, LLC This report is available at no cost from the National Renewable Energy Laboratory (NREL) at www.nrel.gov/publications. Technical Report NREL/TP-6A20-59065 August 2013 Contract No. DE-AC36-08GO28308 The Use of Solar and Wind as a Physical Hedge against Price Variability within a Generation Portfolio Thomas Jenkin, Victor Diakov, Easan Drury, Brian Bush, Paul Denholm, James Milford, Doug Arent, and Robert Margolis National Renewable Energy Laboratory Ray Byrne Sandia National Laboratories Prepared under Task No. SS13.1030 NREL is a national laboratory of the U.S. Department of Energy Office of Energy Efficiency & Renewable Energy Operated by the Alliance for Sustainable Energy, LLC This report is available at no cost from the National Renewable Energy Laboratory (NREL) at www.nrel.gov/publications. National Renewable Energy Laboratory Technical Report 15013 Denver West Parkway NREL/TP-6A20-59065 Golden, CO 80401 August 2013 303-275-3000 • www.nrel.gov Contract No. DE-AC36-08GO28308 NOTICE This report was prepared as an account of work sponsored by an agency of the United States government. Neither the United States government nor any agency thereof, nor any of their employees, makes any warranty, express or implied, or assumes any legal liability or responsibility for the accuracy, completeness, or usefulness of any information, apparatus, product, or process disclosed, or represents that its use would not infringe privately owned rights. Reference herein to any specific commercial product, process, or service by trade name, trademark, manufacturer, or otherwise does not necessarily constitute or imply its endorsement, recommendation, or favoring by the United States government or any agency thereof. The views and opinions of authors expressed herein do not necessarily state or reflect those of the United States government or any agency thereof. This report is available at no cost from the National Renewable Energy Laboratory (NREL) at www.nrel.gov/publications. Available electronically at http://www.osti.gov/bridge Available for a processing fee to U.S. Department of Energy and its contractors, in paper, from: U.S. Department of Energy Office of Scientific and Technical Information P.O. Box 62 Oak Ridge, TN 37831-0062 phone: 865.576.8401 fax: 865.576.5728 email: mailto:[email protected] Available for sale to the public, in paper, from: U.S. Department of Commerce National Technical Information Service 5285 Port Royal Road Springfield, VA 22161 phone: 800.553.6847 fax: 703.605.6900 email: [email protected] online ordering: http://www.ntis.gov/help/ordermethods.aspx Cover Photos: (left to right) photo by Pat Corkery, NREL 16416, photo from SunEdison, NREL 17423, photo by Pat Corkery, NREL 16560, photo by Dennis Schroeder, NREL 17613, photo by Dean Armstrong, NREL 17436, photo by Pat Corkery, NREL 17721. Printed on paper containing at least 50% wastepaper, including 10% post consumer waste. Table of Contents Executive Summary ................................................................................................................................... vi 1 Introduction ........................................................................................................................................... 1 2 Study Methodology .............................................................................................................................. 5 2.1 Model Scenarios ............................................................................................................................ 5 2.2 Study Region and Modeling Tools ................................................................................................ 6 2.3 Natural Gas Price Distribution Modeling ...................................................................................... 9 3 The Impact of RE Generation in Reducing the Uncertainty of Future Electricity Prices ............ 13 3.1 Impact of Increasing RE Generation on Annualized Variable Wholesale Electricity Cost and Cost Uncertainty—50:50 Contribution of Solar and Wind Energy ............................................ 13 3.2 Impact of Increasing RE generation on Variable Wholesale Electricity Cost and Cost Uncertainty—Variable Solar and Wind Energy Contributions ................................................... 16 3.3 Impact of Coal-to-natural Gas Ratios on Annualized Variable Electricity Cost and Cost Uncertainty .................................................................................................................................. 18 3.4 Regulated vs. Restructured Markets and the Impact of Market Structure Assumptions on Cost and Cost Variance with Increasing RE ........................................................................................ 19 4 Characterizing the Impact of RE Generation on the Annualized Variable Cost of Electricity and Cost Uncertainty using MC Simulations .......................................................................................... 22 4.1 The Stability of Optimal Production Cost Model Dispatch to a Range of Natural Gas Prices ... 22 4.2 Annualized Electricity Cost Distribution .................................................................................... 25 5 The Potential Use and Value of Solar and Wind as a Physical Hedge against Cost Risk .......... 28 5.1 Economic Utility of Risk and Loss Aversion for Consumers ..................................................... 28 5.2 The Cost of Hedging and Alternative Methods ........................................................................... 31 6 Summary and Conclusions ............................................................................................................... 34 References ................................................................................................................................................. 37 List of Tables Table 1. Generation and Capacity by Technology for the 35% 50:50 Solar-and-wind Penetration Scenario ......................................................................................................................................... 8 Table 2. Stochastic Model Parameters for Henry Hub, Colorado Utility, and Arizona Utility Prices . 12 Table 3. Impact of Alternative Dispatch Assumptions on the Annualized Cost of Electricity ............ 25 Table 4. Characteristics of the Natural Gas Price and Variable Electricity Cost Distributions ........... 27 List of Figures Figure 1. Schematic of main elements of analysis ................................................................................. 6 Figure 2. The Rocky Mountain Power Authority (RMPA) (FERC 2012) ............................................. 7 Figure 4. EIA forecasts of U.S. wellhead natural gas prices in various years (blue lines) compared with actual prices (red line) ......................................................................................................... 10 Figure 5. Historical natural gas price data: Henry Hub (1997–2011), Colorado utility prices (2002– 2011), and Arizona utility prices (2002–2011) (EIA 2013b) ...................................................... 11 Figure 6. 20 time-series simulations of 30 years of monthly Colorado utility prices ........................... 12 Figure 7. Annualized variable cost of energy ($/MWh) for different natural gas prices under a range of 50:50 solar-wind generation penetration scenarios in RMPP (isolated) ................................. 14 Figure 8. Daily average generation (GW) by technology for an equal mix of solar and wind generation (50:50) for different RE penetration (by percent of generation) scenarios in RMPP isolated ........................................................................................................................................ 15 Figure 9. Percentage of generation curtailed with increasing RE penetration in RMPP (isolated) ...... 16 This report is available at no cost from the iii National Renewable Energy Laboratory (NREL) at www.nrel.gov/publications. Figure 10. Annual cost variance and cost variance reduction with RE penetration under high ($9.2/MMBtu) and low ($3.9/MMBtu) natural gas prices for different solar-to-wind ratios in RMPP (isolated) .......................................................................................................................... 17 Figure 11. Daily average generation (MW) by technology for 35% overall annual RE penetration for three cases with different solar-wind ratios in RMPP (isolated) ................................................. 17 Figure 12. Annualized variable cost of electricity with RE penetration for the 50:50 solar-wind case and coal- and natural gas-dominated fossil scenarios ................................................................. 18 Figure 13. Annualized sensitivity of electricity prices (for restructured market) and annualized average variable cost (for regulated market) under different natural gas price scenarios for different RE penetration levels in RMPP (isolated) .................................................................... 20 Figure 15. Hourly gas use over one year (2020) under different natural gas price and RE penetration scenarios, 50:50 solar-wind scenario in RMPP (isolated) ........................................................... 23 Figure 16. Comparison of daily average dispatch for coal, CCGT, and CT technologies for different pairs of natural gas prices in RMPP (isolated) ............................................................................ 24 Figure 17. Annualized variable electricity system costs for different RE penetrations based on MC simulations using underlying natural gas price distribution in RMPP (isolated) ........................ 27 Figure 18. Utilities of various RE mixes for different attitudes towards risk and impact of time horizon ......................................................................................................................................... 30 Figure 19. Premium for natural gas spot prices in the Northeast compared to other locations during winter 2012/13 (EIA 2013a)........................................................................................................ 33 This report is available at no cost from the iv National Renewable Energy Laboratory (NREL) at www.nrel.gov/publications. Acknowledgments The authors would like to thank the following individuals for their input and comments: Venkat Banunarayanan, Morgan Bazilian, Nate Blair, Mark Bolinger, Karlynn Cory, Hugh Li, Jeff Logan, Maggie Mann, Richard Tusing, and Jurgen Weiss. We also spoke informally to individuals from a number of utilities who provided many helpful comments. We would also like to thank Jarrett Zuboy and Kendra Palmer for their editorial support. This work was funded by the U.S. Department of Energy’s (DOE’s) Office of Energy Efficiency and Renewable Energy (EERE). This report is available at no cost from the v National Renewable Energy Laboratory (NREL) at www.nrel.gov/publications. Executive Summary This study provides a framework to explore the potential use and incremental value of small- to large-scale penetration of solar and wind technologies as a physical hedge against the risk and uncertainty of electricity cost. The idea that adding renewable energy (RE) to a conventional fossil portfolio generates diversity-related benefits is not new and has been discussed by many others (e.g., Bolinger et al. 2002; Awerbuch and Berger 2003; Bazilian and Roques 2008; Roques et al. 2010). Similarly, there may be related benefits from combining RE and natural gas generation (Lee et al. 2012; Weiss et al. 2013) as well as from combining wind and solar within the RE component of the larger portfolio. The core idea behind the value of diversification, of not putting all the “eggs into one basket”--or in this case electric generation technologies--has widespread acceptance. In finance applications the value of diversification forms the foundation behind the application of mean-variance portfolio (MVP) theory to choose “efficient” portfolios of stocks and bonds (Markowitz 1952).1 The related concept of quantifying the value of diversity in the electric sector that may result from reducing the risk and uncertainty of the overall system costs over multi-year to multi-decade time horizons is less well understood or accepted (Stirling 1994; Awerbuch and Berger 2003). Adding RE can be expected to reduce the variability of the overall electric system costs over a variety of timescales as natural gas-fired generation is displaced. However, the direct application of MVP theory to “optimize” the mix of generation assets within a generation portfolio is problematic for a number of reasons, including the fact that the operational characteristics of some types of generation assets are dissimilar. Earlier studies characterizing the impacts of adding RE to portfolios of electricity generators have often used a levelized cost of energy (LCOE) or simplified net cash flow approach. In this study, we expand on previous work by using an hourly production cost model (PLEXOS) to analyze the incremental impact of solar and wind penetration under a wide range of deployment scenarios for a region in the western U.S. We do not attempt to “optimize” the portfolio in any of these cases. Rather, we consider different RE penetration scenarios that might, for example, result from the implementation of a Renewable Portfolio Standard (RPS) to explore the dynamics, risk mitigation characteristics, and incremental value that RE might add to the system.2 For our reference case, in which solar and wind make equal contributions (1:1) to total generation on an annual basis, we varied the annual RE generation from about 10% to more than 50% under a range of natural gas price scenarios. We then explored the impact of altering the annual solar-to-wind generation ratio to 3:1 and 1:3 and also varied the ratio of natural gas to coal generation in the fossil generation mix for the 1:1 reference case. We also simulated the variation in electricity costs using a Monte Carlo (MC) simulation approach. This allowed us to characterize the value of variance reduction for customers with different levels of risk and loss aversion and to compare this, at least in the near term, with the use of alternative mechanisms for 1 “Efficient” refers to portfolios of assets that lie on a curve (the “efficient frontier”) where each point represents a portfolio with lowest risk for a given return (over a range of returns). It is the lack of correlation of outcomes (returns in the case of financial assets) that reduces the risk (as measured by the variance of returns) for a given expected portfolio return. 2 This approach was suggested in Bush et al. (2012). This report is available at no cost from the vi National Renewable Energy Laboratory (NREL) at www.nrel.gov/publications. partially hedging against future price uncertainty. Some market structure issues were also considered. Some key findings of our analysis include the following: • Solar and wind generation significantly reduce the exposure of electricity costs to natural gas price uncertainty in fossil-based generation portfolios on a multi-year to multi-decade time horizon. The incremental impact, and any associated marginal value of RE in decreasing o electricity cost volatility, declines with increasing RE penetration. The reduction in volatility of electricity costs with increased RE penetration is o greater for natural gas-dominated portfolios than for coal-dominated portfolios. • At low RE penetrations (e.g., 10%–15% annual RE generation) the annualized variable system costs vary widely with the price of natural gas in both our coal-dominated and natural gas-dominated fossil portfolios. For the modified region studied in this report: At 15% RE penetration in the coal-dominated system,3 a $5/MMBtu variation in o natural gas prices (between $4/MMBtu and $9/MMBtu) translates to approximately a $8/MWh range in the variable cost of electricity. For similar RE penetration (15%) in the gas-dominated portfolio, a $5/MMBtu o variation in natural gas prices changes the variable cost of electricity by about $35/MWh--a more than three-fold difference compared to the coal-dominant portfolio.4 In the coal-dominated fossil portfolio the incremental impact of further solar and o wind penetration decreases with increasing RE penetration with only small incremental benefits achievable beyond 35% penetration. This is largely because, at these higher levels of RE penetration, very little natural gas generation remains to be displaced. In contrast to the natural gas-dominated portfolio, the saturation effect in o electricity cost variance reduction is not observed even at higher RE penetration levels (of over 40%) because a large amount of natural gas generation remains to be displaced. • In the region studied, a mix of wind and solar provides a better physical hedge against uncertain fuel prices than either wind or solar alone because of the observed anti- correlation in solar and wind generation profiles at time scales ranging from intra-day to seasons. 3 Where the ratio of coal thermal to natural gas combined cycle gas turbine (CCGT) capacity was approximately 2:1. For the natural gas-dominated portfolio, all coal thermal units were switched out with CCGTs. 4 The relative ratio of price variation depends not only on the ratio of coal thermal to natural gas plants but also on the cost of coal. Coal prices, even on an energy equivalent basis, vary significantly by location. The cost of coal per MMBtu for Colorado used in the study is amongst the lowest in the U.S. This report is available at no cost from the vii National Renewable Energy Laboratory (NREL) at www.nrel.gov/publications. • Market structure choices are important. Adding RE reduces uncertainty in cost to consumers5 much more in restructured markets than in regulated markets since natural gas often sets the marginal price in a given hour in restructured markets (particularly during higher-priced peak periods), and this price is then paid to all generators dispatched. • MC analysis of the impact of natural gas price variations over multi-decade time horizons complements scenario analysis by generating electricity cost distributions that show the likelihood (or “density”) of outcomes. These distributions also show that the electricity costs are positively skewed. While the upside risk (lower electricity prices) is largely capped by physical o constraints on fuel costs, the downside risk (higher electricity prices) is not. RE may be important to both reduce the overall variance of system costs, as well as provide insurance future price increases, which may be particularly important given the current low natural gas prices (Bolinger 2013). Inter-annual variability in generation is also important since it can lead to o deviations from average annual generation of ±10% or more in any year for solar and wind generation. However, while year to year variation in RE generation was not explicitly integrated into the production cost runs used in this study, the impact of such resource variation may be expected to be mitigated over long time horizons as year to year variations will tend to offset each other (Drury et al. 2013). • We find that much of the MC analysis of natural gas price uncertainty impacts can be done outside of the production cost model by recognizing the stability of the simulated hourly system dispatch for a wide range of natural gas prices. This greatly enhances our ability to perform many simulations which otherwise would be limited by model run times.6 The potential benefits of diversified portfolios containing significant solar and wind generation will depend on two main factors. One factor is how much consumers’ values lower price uncertainty due to risk aversion, loss aversion, scarcity, or other characteristics. The second factor is the potential cost and effectiveness of alternative financial or physical hedging methods, such as forward contracts, swaps, or physical supply contracts7, and the timeframe over which these are available; this includes the degree to which price uncertainty risks are mitigated and the 5 Bilateral contracts within a restructured market, which are common for solar and wind, may mitigate this leverage and have an asymmetrical effect on consumers. This and other market structure-related issues are a focus of our follow-on research. 6 The wide range that this stability effect was due in part is due to the low coal prices found in the region studied (on a $/MMBtu basis), and so the effect is likely to be less pronounced in many other regions of the U.S. with higher coal prices. 7 A buyer (or seller) of natural gas (or electricity) can protect itself, or hedge against future price uncertainty by agreeing to an over the counter (OTC) forward contract to buy (or sell) a commodity at some time. The price to be paid at delivery is specified in advance when the contract is made. An alternative way for a buyer to hedge is to buy gas at spot market prices, but also have an arrangement where the buyer pays the third party a fixed price for natural gas and in return receives (or swaps) payments linked to the market price of natural gas (Eydeland and Wolyniec 2003). This report is available at no cost from the viii National Renewable Energy Laboratory (NREL) at www.nrel.gov/publications. extent to which new risks may be introduced (e.g., associated with natural gas transportation constraints, counterparty risks, market liquidity, and others). The cost of using financial instruments to hedge against future price uncertainty depends in part on whether long-term forward contracts (for natural gas or electricity) contain a premium over expected future prices. Electricity sellers and buyers may both be risk averse, and there is no consensus about the net impact of this on the existence of a forward premium for eliminating price volatility in the United States. Some studies that suggest, at least in the short-term, it may be more cost effective to use financial-hedging instruments often assume (either implicitly or explicitly) there is no risk aversion or other premium in the forward price over the expected futures price. On the other hand, some studies have suggested there may a positive premium over the expected future price due to risk aversion (Bolinger et al. 2002) or due to scarcity or other factors (Borenstein et al. 2007),8 while others suggest a negative premium (Modjtahedi and Movassagh 2005). The answer may be “all of the above”, with the existence and magnitude of a premium (positive or negative) likely to vary with location, commodity, and timescale, while changing over time. Of particular relevance to RE, it is difficult and rare to be able to lock in financial or physical supply contracts of 10 years or more for natural gas. Such contracts may include premiums that reflect lack of liquidity and counterparty risk (Bolinger 2013).9 Because of these and other issues, in the longer term solar and wind may be able to provide a physical hedge that is not easily replicated in the financial and physical commodity markets.10 It also provides insurance value against rising electricity prices in futures where natural gas prices rise or carbon emissions are priced via a tax or some other mechanism. Even in the shorter term, RE may be the better choice for some consumers. While most of this report deals with the system wide effect on the average consumer at a multi-utility level, the preference for cost mitigation and over what timeframes may vary widely by customer type. Size also matters where some residential and commercial customers may decide to install distributed RE in part if their ability to hedge using financial or physical instruments is limited by a lack of knowledge, high transaction costs, or a lack of availability of such instruments. 8 Graves and Levine (2010) make the interesting observation about how the positively skewed nature of the price distribution for natural gas could explain observed differences between the expected forward price and the observed prices--even if there is no meaningful premium simply due to the expected value of the distribution lying above the mostly likely and the median values. 9 “Passive” hedging with RE could also provide benefits by affecting a wide range of buyers in a similar manner. This may be helpful because many firms have trouble knowing how to hedge appropriately (possibly overreacting to a crisis and locking in high prices), and this can bring business risks. Alternatively, a firm could hedge in a smart way—while many of its competitors do not—and get “unlucky” if, for example, the prices of inputs fall sharply for the industry. Passive or natural hedging with RE in this way may provide a “cushioning” effect to help mitigate these types of business risks. 10 The use of rolling, short-term hedging over longer time horizons provides a hedge against evolving market conditions and prices. It does not provide a long-term hedge against future price changes (as might a hedge due to RE). This report is available at no cost from the ix National Renewable Energy Laboratory (NREL) at www.nrel.gov/publications.