Monday Study – New Energy Needs New Wires
Solar and wind grid system value in the United States: The effect of transmission congestion, generation profiles, and curtailment
Dev Millstein, Ryan Wiser, Andrew D. Mills, Mark Bolinger, Joachim Seel, Seongeun Jeong, July 21, 2021 (Joule)
The value of electricity generated from wind and solar sources declines as supply increases. This decline in value has varied over time and across regions, indicating that strategies to mitigate value decline will need to be carefully targeted. To help guide development of these strategies, we empirically determine wind and solar value at 2,100 plants within United States wholesale markets by using local prices and plant-specific generation profiles. We determine how each plant loses (or gains) value because of its output profile, transmission congestion, and curtailment. In regions where wind or solar account for roughly 20% of electricity generation, its value is 30% to 40% below the regional average value of a flat output profile at all plants. Solar value reductions are most sensitive to output profile and wind value reductions are sensitive to both profile and congestion, region dependent. Curtailment was not a major source of value reduction.
Variable renewable energy (VRE), referring to solar and wind power in this study, is essential to cost-effectively decarbonizing the electricity grid. As VRE provides a greater portion of electricity, the value of electricity during windy or sunny hours declines due to the low marginal cost of VRE. This value decline might be larger than future cost declines because of learning.1,2 Other studies expect substantial VRE cost declines3–6 but highlight the key challenge of value decline (or more generally, system integration questions7–10). The possibility of losing the race between declining cost and value puts decarbonization goals at risk and has led to calls for ambitious research and development efforts to cut the cost of VRE technology11 or for more direct policy support.12 Although there is no consensus as to the exact portion of future electricity generation VRE might provide, it is clear that the relatively low cost of VRE generation means that high penetrations are feasible and that the scope of VRE value decline will heavily influence VRE’s deployment potential.
The mechanics of VRE value decline are complex, producing varying results across regions and time periods. This is evidenced by the variety of results found in empirical studies from Europe,13–25 Australia,26 and the United States.27–31 These studies mostly focus on a concept related to VRE value decline, the ‘‘merit order effect,’’ which quantifies the deflation of overall average wholesale power prices due to VRE output. There are only a limited number of empirical studies in the United States that focus specifically on how wind or solar values differ from average electricity values. Studies in Texas32 and California33,34 evaluate the relatively low value of VRE generation (versus average electricity value), and highlight interesting dynamics between VRE and other types of power plants (for example, solar generation has improved the value of flexible gas combustion turbine plants in relation to combined cycle plants). Brown and Sullivan35 calculated solar value at all electric system price nodes. They also evaluated the solar value decline between 2010 and 2017. These studies provide important insights into value decline in the US but leave a number of questions unanswered. Specifically, existing peer-reviewed empirical studies in the US do not provide a national assessment of wind value decline and do not assess the relative importance of different underlying causes of value decline. These topics are important to study because the first step toward designing effective value decline mitigation strategies is to understand the root causes of value decline and how these causes vary by location and technology.
We address these research gaps by providing an empirical assessment of the gridsystem value of 2,100 existing utility-scale wind and solar plants across the contiguous US. We use the term ‘‘value’’ to reflect the energy and capacity revenue potential for wind and solar generation in a wholesale market environment. We analyze regional differences in value decline over the past decade. We then assess, for each plant, what causes the difference between its value and the overall average value of electricity. To do this, we decompose the observed value differences into three separate causes: output profile, transmission congestion, and curtailment. This decomposition loosely builds on concepts from modeling efforts9,36 and complements evaluations of VRE integration costs37–40 and efforts to identify value decline mitigation strategies.8,41–45 We also compare recent value decline trends with forward-looking modeling studies, providing context for how these trends might evolve in the future. We take care to develop realistic temporal profiles of generation—each wind and solar plant is modeled individually, and the resulting output is then bias corrected with a combination of records from various sources, including hourly records where available. We estimate hourly plant-level curtailment on the basis of plant characteristics, regional curtailment reports, and local pricing patterns. In the following sections, we first present overall trends to VRE value, followed by detailed analysis of value broken into the above-mentioned components. These results are followed by a brief comparison between value and cost trends. We conclude with a discussion of key insights relevant to system planners and policy makers…
In this paper, we have provided insight into the underlying mechanisms that have driven VRE value decline in the US. To do this, we found the difference in value between generation at each plant versus flat profile (or ‘‘flat block’’) generation at all power plants in a region. We then decomposed this value difference into three separate causes: output profile, transmission congestion, and curtailment, and we tracked their trends over time. This empirical analysis shines a light on the geographic heterogeneity of VRE value decline with penetration. This heterogeneity was mostly driven by differences in the levels of transmission congestion facing VRE output. In contrast, the decline in value of VRE output profiles was more closely correlated with penetration levels across regions. To date, curtailment levels were relatively low, and had relatively little impact on VRE value. We summarized mitigation strategies that would be most effective in addressing congestion or profile value reductions, and a simple take-away point is that different, and targeted, value decline mitigation strategies should be pursued in each region, at least in the near term.
There are a number of policy and regulatory changes that could support such mitigation strategies. For regions and VRE types where profile leads to substantial value decline (e.g., CAISO and ISO-NE solar, ERCOT and SPP wind), regulations related to storage and load profile modification might be most useful to target. For example, most proposed battery capacity in the interconnection queues is coupled with new VRE plants (mostly solar) as opposed to being standalone storage.63 Yet, these coupled (or ‘‘hybrid’’) plants often present a challenge with respect to existing regulations because coupled plants are new (and evolving) and existing regulations for standalone VRE or standalone storage are not always easily adaptable to coupled plants. Gorman et al.63 identify a set of eight key challenges, which, if addressed, would facilitate full participation of coupled plants in markets and thus help support the overall storage adoption. The challenges range from defining rules around generation forecasting to refining interconnection processes to enhancing resource planning models.
Related to load profile modification, dynamic retail rates, i.e., rates that allow for greater representation of hourly variability in wholesale prices, such as ‘‘time-ofuse’’ pricing, are an important regulatory tool to shift a portion of load across hours of the day (for example, reducing demand by 5% in peak times).64–66 Carefully developed retail rate structures might also effectively guide electric vehicle charging toward times of greatest benefit.67 Other long-term options exist for modifying load profile, including scheduling high-load industrial processes for low-cost hours, such as hydrogen production or desalinization.66
For regions where transmission congestion leads to substantial value decline (e.g., wind in SPP, MISO, and NYISO), reforms to the transmission planning process are essential. Analyses suggest that existing transmission planning processes were not designed to adapt to the speed of VRE deployment, and that, in particular, the rules surrounding who pays for new transmission and when new transmission builds are triggered, are particularly detrimental to VRE deployment and slow infrastructure project development needed to alleviate congestion.68,69
Although the above mitigation strategies and policy and regulatory topics are appropriate responses to the value decline observed to date, achieving long-term decarbonization goals requires that we keep in mind a larger perspective that compares future VRE value and cost. To achieve this perspective, we must combine the empirical results described in this paper with studies that simulate value decline into the future.
Some models indicate, as highlighted in the previous sections, that value decline might soon get worse. Our empirical analysis provides little solace on that front, that is, there is no evidence yet that the real world is behaving substantially differently from what was predicted years ago. Yes, transmission congestion is more geographically heterogeneous than is often described in system simulations, but overall, the decline we have observed in profile value, and even overall value, largely confirms the shape and magnitude of modeled predictions of value decline.
From one perspective, the trends described above leave VRE in a precarious position. Value decline to date has been matched and beaten by cost declines. But forward-looking models, which have been roughly correct to date, suggest that we will soon enter a regime of accelerating value decline. If the rate of cost declines levels out, we could soon see a situation where value decline outpaces cost decline, slowing VRE deployment. To avoid this slow down, value decline mitigation strategies would need to be employed, but the policy and regulatory changes required of these strategies take time. Thus, value decline mitigation efforts would need to begin before the value curve crosses back below the cost curve (see Figure 8), but it is more challenging to motivate action when costs are currently below value. This perspective is perhaps more relevant to wind, which compared with solar, faces greater congestion value decline, has reached relatively high penetration in multiple regions, and whose drivers of value decline are not addressed as well by short-duration (4 to 8 h) storage.
The substantial time needed to implement transmission value decline mitigation measures is one reason why multiple research efforts call for work to begin on more far ranging policy actions than described above.68,70,71 From this perspective, the key action needed is expanded inter-regional transmission. Even though inter-regional transmission is often part of longer-term strategies for deep decarbonization,72,73 there is a push to begin the process now. This push is driven by the long lead times needed for transmission construction and because existing high-VRE systems (e.g., SPP and MISO) would already benefit from multiregional transmission planning.68–71 In our analysis, we tested the impact of intra-regional transmission on the congestion component of value reductions, but inter-regional transmission might address all three of the value decline components we have examined. The important concept here is that long-distance, high-voltage transmission can link regions with different demand and VRE profiles to each other, and that these profile differences occur over various time frames ranging from diurnal to seasonal.70
An alternate, more optimistic perspective can also be presented. Optimism, in this case, is based on the recent decline in battery storage costs.74 Multiple modeling studies have shown that low-cost energy storage can effectively mitigate VRE curtailment, or value decline, out to VRE penetration levels of 50% (for combined wind and solar).8,72,75–77 In CAISO, for example, simulations of low-cost storage at high VRE penetration were shown to increase prices during relatively low priced sunny hours, significantly increasing the value of solar at high penetrations.8,77 In ERCOT, simulations show that low-cost, short-duration (4 to 8 h) storage can cut curtailment of VRE roughly in half at penetrations near 50%, even when total VRE is weighted toward wind.76 Finally, a simulation of high VRE penetration in 2050 indicated that multiple pathways exist to achieve the needed grid flexibility with high VRE penetration and specifically that storage can be part of an effective solution even when inter-regional transmission is limited.72
We therefore conclude on a mixed note of positivity and caution. We have found that cost decline has taken a recent lead in the race versus value decline. VRE deployment thus continues at a healthy rate. However, some modeling analyses point to the threat of accelerated value decline and the resulting slowing of VRE deployment, and our empirical analysis does not contradict this assessment. Yet, recent declines in battery storage costs and the trend toward low specific power wind plants suggests that options to slow value decline might be more cost effective than previously modeled. This leaves open the question of how aggressively policy makers should pursue value decline mitigation strategies. Are targeted regional strategies sufficient to ensure decarbonization goals, or should stronger efforts be made in the near term, such as to develop inter-regional transmission or more aggressively subsidize storage? Research might help to illuminate the factors needed to make these decisions. Simulations of value decline could be updated to reflect design changes to new wind, solar, and coupled storage plants. Continued empirical analyses are needed to highlight evolving regional dynamics. Stakeholders can then evaluate policy tradeoffs given the evolving information…