Renewables, Solar, Wind

Study Explores Impacts of High Wind and Solar Futures on Wholesale Power Markets

Issue 7 and Volume 122.

Study Explores Impacts of High Wind and Solar Futures on Wholesale Power Markets

Increasing penetrations of variable renewable energy (VRE) can affect wholesale electricity price patterns and make them meaningfully different from past price patterns.

A new study from the Lawrence Berkeley National Laboratory specifically assesses wind and solar shares of up to 40 to 5 percent in California, the Midwest, Texas and New York, and contrasts those results to a low-VRE scenario.

The study’s motivating question is whether certain electric-sector decisions that are made based on assumptions reflecting low VRE levels will still achieve their intended objective in a high VRE future.

Some of the study’s key findings include:

• High shares of wind and solar lead to several profound changes in wholesale electricity price patterns. 

• Average annual hourly wholesale energy prices decrease with more VRE penetration, by $5-$16/MWh depending on the region and mix of wind and solar.

• Perhaps the most fundamental changes relate to the timing of when electricity is cheap or expensive and the degree of regularity in those patterns.

• Diurnal price patterns change significantly, especially in scenarios with large solar shares. With high solar shares in ERCOT, for example, prices slump in the middle of the day to an average of $10/MWh and then rise in the evening to an average of $80/MWh.

• The frequency of periods with low prices (below $5/MWh) increases to between 3 percent and 19 percent of hours in the high VRE scenarios depending on the region and mix of renewables. High solar in ERCOT, with its limited interconnection capacity to neighboring regions, experiences the highest frequency of periods with near-zero prices.

• Price volatility increases with higher VRE shares, particularly in the high wind scenarios. Morning prices in the spring in CAISO can vary between $0 and $50/MWh with high wind, but fall in a much narrower range in the low VRE scenario.

Ancillary service prices rise substantially with high VRE, especially for regulation and spinning reserves. Higher ancillary service prices could attract new market entrants such as batteries or incent wind and solar to offer these services themselves.

Peak net-load hours are shifted over shorter time periods into the evening, yet are distributed over more days of the year.

Researchers then highlighted qualitatively the possible impact of these altered price patterns on various demand- and supply-side electric sector decisions, from energy efficiency and demand-response programs to electric-vehicle charging locations, to the value of generation flexibility. The study also provides a foundation for later planned quantitative evaluations of these decisions in low and high VRE futures.

The study found that high shares of variable energy resources lead to several profound changes in the characteristics of electric power systems. 

The most fundamental changes relate to the timing of when electricity is cheap or expensive and the degree of regularity in those patterns. Diurnal price patterns change significantly, especially in scenarios with large solar shares. With high solar shares in ERCOT, for example, prices slump in the middle of the day to an average of $10/MWh and then rise in the evening to an average of $80/MWh. These price dynamics may support electric-vehicle charging infrastructure at commercial sites that can be accessed during the day instead of residential charging that would occur at night—opposite of the situation with low shares of VRE.

The frequency of periods with low prices (below $5/MWh) increases to between 3 percent and 19 percent of hours in the high VRE scenarios depending on the region and mix of renewables. High solar in ERCOT, with its limited interconnection capacity to neighboring regions, experiences the highest frequency of periods with near-zero prices.

“The study found that obtaining high shares of energy from variable energy resources leads to several profound changes in the characteristics of power systems.”

Average annual hourly wholesale energy prices also decrease with more VRE penetration, by $5-$16/MWh depending on the region and mix of wind and solar. Decreases in average wholesale prices and common occurrences of periods with very low prices will affect the profitability of VRE and inflexible generators that operate in these hours, but also presents an opportunity to shift or increase demand at very low cost.

Evidence of VRE-induced Price Changes

Many of these developments can be observed through changes in the patterns of wholesale prices – although it would be wrong to attribute all price changes exclusively to VRE growth, especially in an environment with dynamic natural gas pricing or stagnant load growth. A broad body of literature has discussed these empirical effects both internationally2 and in the United States3.

For example, analyses of wholesale prices in Australia (Gilmore, Rose, Vanderwaal, & Riesz, 2015) show that the deployment of photovoltaic capacity can lead to price changes: historical capacity additions by 2013 had already eroded a mid-day peak in prices in comparison to 2009 and caused the diurnal price profile to flatten significantly. Forward-looking modeling projections for the year 2030 exhibit a further reversal in price peaks to non-solar hours in the early morning and late evening. Keay (2016) summarizes recent European developments and demonstrates a substantial flattening in German diurnal price profiles between 2000 and 2012 that coincided with strong deployment in solar capacity.

Similar developments can be found increasingly in the United States as well. Wiser et al (2017) comprehensively review wholesale electricity price data of U.S. ISOs and find evidence of changed temporal and geographic price patterns in areas with high VRE penetrations. Growth in PV in the California market drove down net-load levels during the mid-day in 2017 relative to 2012 resulting in an associated change in price patterns (U.S. Energy Information Administration (EIA), 2017). In contrast to more even prices over the course of the day in the first half of 2015, the more recent price profile resembled a “duck” in the first half of 2017. In particular, prices have a local maximum around 7am at slightly under $40/MWh followed by a mid-day price slump of about $15/MWh and an evening price peak of nearly $60/MWh at 8pm. Another example of VRE-induced price changes are low power prices at night in wind-rich areas in Texas that have caused some electricity retailers to offer “free” electricity at night.

Study Explores Impacts of High Wind and Solar Futures on Wholesale Power Markets

Summary

The study found that obtaining high shares of energy from variable energy resources leads to several profound changes in the characteristics of electric power systems.

The most fundamental changes relate to the timing of when electricity is cheap or expensive and the degree of regularity in those patterns. The frequency of periods with low prices (below $5/MWh) increases from zero hours in the low VRE scenarios to between 3% and 19% of hours in the high VRE scenarios depending on the region and mix of renewables. High solar in ERCOT, with its limited interconnection capacity to neighboring regions, experiences the highest frequency of periods with near zero prices. Common occurrences of periods with very low prices will affect the profitability of VRE and inflexible generators that operate in these hours, but also presents an opportunity to shift or increase demand at very low cost.

Across all of the regions, high solar scenarios lead to the largest change in the diurnal profile of prices and the greatest overall variation in prices. High wind scenarios, on the other hand, lead to the greatest increase in irregularity of pricing patterns. As a result, electricity suppliers or various electric-sector programs may need to be more flexible and adaptable in a high wind future than in a low VRE or even a high solar future.

High VRE scenarios enable some reduction in the capacity of thermal generation, yet energy from nonVRE generators decreases more significantly, particularly for natural gas and coal. Furthermore, average annual hourly energy prices decline in high VRE scenarios relative to low VRE. For many generators, this reduction in average energy prices will increase the relative importance of ancillary service and capacity market products.

In all regions, the study found that high VRE scenarios result in higher ancillary service prices, absent the ability of VRE to provide ancillary services or the entry of new emerging providers of ancillary services, such as batteries. Capacity prices on the other hand remain relatively steady. Nonetheless, the high VRE scenarios consistently spread peak net-load hours over more days of the year and push the timing of such hours into the early evening, indicating a potential shift in the resource portfolio that can contribute to meeting resource adequacy requirements.

It is crucial to note however, that the portrayed price changes will elicit responses by other market participants which in turn will affect prices. While the capacity expansion model that researchers used has optimized the non-VRE supply portfolio by selecting among traditional generator types, it has not considered investments into demand-side assets that would change the aggregate load profiles (certain energy efficiency measures or demand-response programs) or investments into electro-chemical battery storage. Very high energy prices during scarcity hours or sustained high ancillary service prices would likely motivate investments into these technologies, which subsequently would moderate prices again.

The price results are further a consequence of our modeling assumptions: The expansion of intra-regional transmission masks price variability related to local congestion, while the assumed high VRE penetrations in neighboring regions limit price mitigation due to exports and imports. Changes in our fuel price assumptions (e.g., natural gas relative to coal) would impact the merit order curve and could lead to a different optimal generator portfolio with different flexibility and ramping characteristics. Altered load profiles (such as mass deployment of electric vehicles with price-responsive charge management) would affect our diurnal price profiles. Differences in the absolute load level forecast that do not affect the load shape (e.g., due to better energy efficiency performance or less energy-intensive economic growth) would likely have less of an impact, as the generator portfolio would adjust with the retirement of some marginal plants. Because researchers only considered a single exemplary year of 2030, inter-annual variation (that may include stronger cold-spells with high heating demands, droughts with less hydro-power availability, or heat waves with large additional cooling loads) and a further evolution of the electric system beyond 2030 are not captured by our analysis.

Despite these limitations, the study found that electric systems with large shares of VRE penetration will see profound changes in average electricity prices, diurnal price patterns, and price volatility that should be considered in decisions related to long-lasting assets. This paper qualitatively highlighted some of the possible impacts on other demand- and supply-side decisions.

While the decision-making processes and considerations may differ between regulated and de-regulated regions of the country, analysis of the marginal value of different resources can be informative in either case. As such, these simulated wholesale prices can provide a foundation for quantitative evaluations to explore how various demand- and supply-side decisions might be affected by changes in the future electricity supply mix.