TODAY’S STUDY: Impacts Of New Energy On Power Prices
Impact of Wind, Solar, and Other Factors on Wholesale Power Prices; An Historical Analysis—2008 through 2017
Andrew D. Mills, Dev Millstein, Ryan Wiser, Joachim Seel, Juan Pablo Carvallo, Seongeun Jeong, Will Gorman, November 2019 (Lawrence Berkeley National Laboratory)
Wholesale power markets in the United States have evolved over time. Some of the more notable changes over the last decade include growth in wind and solar, a steep reduction in the price of natural gas, limited growth in electrical load, and an increase in the retirement of thermal power plants. This report assesses the impact of these changes on wholesale electricity prices using two approaches. First, a supply curve model is used to quantify impacts to annual average wholesale prices at each centrally organized wholesale power market between 2008 and 2017. Second, hourly wholesale prices at all of the more than 60,000 pricing nodes are used to highlight the impacts of wind, solar, and other factors on trends in geographic and temporal pricing patterns.
In most markets, growth in wind and solar reduced average wholesale prices by less than $1.3/MWh. California is an exception, where growth in solar reduced prices by $2.2/MWh—perhaps foreshadowing greater impacts from solar in other regions as solar penetrations grow. Falling natural gas prices over this same period were the dominant driver of average market-wide wholesale prices, reducing average annual wholesale prices by $7–$53/MWh. The impact of wind and solar was secondary compared to the impact of natural gas, but among the biggest drivers in a second tier of factors with similar magnitudes, Figure ES-1. The second tier includes expansion and retirement of thermal generation, changes in demand, generator efficiency, coal prices, variations in hydropower, and emissions prices.
Beyond the impacts to market-wide average annual wholesale prices, growth in wind and solar had a more consequential impact on prices in some locations and in altering how prices change based on the hour of the day and season. Specifically, growth in wind and solar impacted time-of-day and seasonal pricing patterns, growth in the frequency of negative prices was correlated geographically with deployment of wind and solar (Figure ES-2), and negative prices in high-wind and high-solar regions occurred most frequently in hours with high wind and solar output.
Despite the recent increase in frequency of negative prices, annual average prices at most locations have not been heavily impacted by these negative-price hours because negative prices were mostly small in magnitude. However, some regions have seen significant declines in annual average prices owing to negative hourly prices, specifically parts of the Midwest in the Southwest Power Pool, California, and northern areas of New York, New Hampshire, and Maine.
The regional clustering of negative prices means that not all generation has been equally impacted. In 2017, negative prices decreased the average annual real-time energy price at nodes near wind plants by about 6%, at nodes near solar plants by about 3%, and nodes near hydropower plants by about 3%. Pricing nodes near coal, gas, and nuclear plants saw a smaller reduction of about 1.5%, though those (modest) impacts have slightly increased over time.
Numerous factors beyond wind and solar influence local pricing patterns. Attempts to assess the impacts of wind and solar must carefully consider the full regional context.
Centrally organized wholesale power markets in the United States have evolved over time. Some of the more notable recent trends include growth in wind and solar, a steep reduction in the price of natural gas, limited growth in electrical load, and an increase in the retirement of thermal power plants. Building on recent related work (Wiser et al. 2017), this report has assessed the degree to which growth in VRE has influenced wholesale power energy prices in the United States, not in isolation but in comparison to other possible drivers and focused on regions of the country that feature ISOs/RTOs.
Across all U.S. ISO/RTO markets, the dominant driver of the decline in average wholesale prices between 2008 and 2017 was the fall in natural gas prices. Even after the shale-gas boom caused a sustained reduction in natural gas prices, variability of natural gas prices continued to be the largest driver of changes in average wholesale prices—albeit sometimes increasing and sometimes decreasing prices.
The impacts of wind and solar on market-wide average annual wholesale prices were secondary compared to the impacts of natural gas, but they were among the biggest drivers in a second tier of factors that also included expansion and retirement of other generation capacity, changes in demand, generator efficiency, variations in hydropower, and emissions prices. The impact of wind and solar on average wholesale prices increased with their share of total generation. Building on near-term projections from EIA, the impact of additional wind and solar on average wholesale prices will be similar to the impact of thermal generation additions, except in the case of additional solar in California. The projected doubling of solar in California by 2022 is expected to have substantial impacts on average wholesale prices—perhaps foreshadowing larger impacts in other regions on a longer-term basis as solar penetrations grow. Storage and other forms of flexibility could affect these results, but impact of storage on prices was not to captured in the simple supply-curve model.
Beyond the impacts to market-wide average annual prices, VRE has had a more substantial impact on prices in some locations and in altering the temporal patterns of prices. In particular, VRE impacts timeof-day and seasonal pricing patterns, often depressing prices when VRE supply is high but, in some cases, inflating prices at other times.
The analysis demonstrates that the frequency of negative prices is correlated geographically with VRE deployment, and that negative prices in high-VRE regions occur most frequently in those hours with high VRE output. Despite the recent increase in frequency of negative prices, annual average LMPs at most locations have not been heavily impacted by these negative-price hours (i.e., negative prices were mostly small in magnitude). However, some regions have seen significant declines in annual average LMPs owing to negative hourly prices, specifically regions in SPP, regions in CAISO, and northern areas of New York, New Hampshire, and Maine.
Along with the limited regional impacts of negative prices, negative prices reduced the prices near wind, solar, and hydropower generators significantly more than near natural gas, nuclear, and coal generators.
Finally, through a series of in-depth regional analyses, this analysis shows how numerous factors beyond VRE have interacted with VRE to influence local pricing patterns. For example, given the backdrop of expanding VRE, annual changes in hydropower output drove negative pricing events in the Northwest; nuclear retirements, changes in load, and solar expansion led to markedly different diurnal patterns of pricing in California; and the expansion of transmission reduced negative-price hours near wind in Texas and near nuclear in Illinois. The conclusion to draw from all of this is that, while expansion of wind and solar is leading to significant changes in pricing patterns in some regions (by reducing prices, increasing the frequency of negative-price hours, and changing the diurnal patterns of pricing), other factors are also influencing pricing patterns, and attempts to assess the impacts of VRE must carefully consider the full regional context.
A number of important additional areas of research are not covered in this analysis.
• VRE and other factors are likely to impact other grid services priced in wholesale markets, including capacity and ancillary services. Similar to wholesale energy prices, the price of these services varies by region and has changed over time. While the analysis presented in this paper focuses exclusively on energy prices, additional assessments might usefully also address capacity and ancillary service markets, including uplift payments associated with generation that is directed by system operators to operate in ways that differ from their schedule.
• Price changes have differential impacts on the revenue earned by different resources depending on whether the resource operates at a near-constant output irrespective of grid conditions (e.g., nuclear), the resource flexibly responds to changing grid conditions as signaled by changing prices (e.g., combustion turbines), or the resource dispatch is variable and largely driven by weather (e.g., wind and solar). Future research might therefore explore the implications of price changes on the net revenue of different generation assets, depending on their typical dispatch patterns.
• Storage and flexible demand can mitigate some of the price variability associated with growing shares of VRE. While storage was not accounted for in the simple supply-curve model, other approaches are available to integrate storage and other more-complicated features of electricity markets into fundamental models of wholesale prices. Incorporating storage into the analysis appears to be particularly important for assessing near-future wholesale prices in the solar-dominated California market.
• Exploring longer-term power-sector transformation scenarios and related impacts on pricing and market design will require more sophisticated tools than employed in the present paper. Use of such tools can enable a more thorough investigation of future temporal and geographic pricing patterns under a range of future assumptions and conditions. Of particular interest for an investigation with such tools will be the impact of VRE on price volatility and the subsequent impact on revenues of flexible resources.