Assessing compounding climate-related stresses and development pathways on the power sector in the central U.S.

There are a wide range of electric generation technologies that could play a major role in producing power in the central USA, sometimes referred to as the US Heartland. For our purposes, we are identifying the “Heartland” as largely the Mississippi River drainage basin. Thermal power generation (coal, gas, nuclear) plays a significant current role in this region (68% in 2020); however, large areas are also notable for having high-quality wind and solar resources, while hydropower is an important resource in other parts of the region. The costs of generation from various sources have changed over the past decades. Low natural gas prices in recent years have favored gas generation over coal. At the same time, there has been an ever greater deployment of wind turbines (38% of generation capacity added from 2010 to 2020). Then, in the past few years, the cost of solar photovoltaics has fallen, providing a greater edge for this technology. State and federal energy policy, through tax incentives and renewable energy requirements, has also contributed to a shift in generation sources, as have tighter restrictions on emissions of conventional air pollutants. Going forward, policy goals of increasing the share of clean energy could have further implications for the power generation mix. Many states have focused on renewable portfolio requirements—favoring wind and solar power. Recent legislation including the Infrastructure Investment and Jobs Act, H.R.3684 (U.S. Congress 2021), the CHIPS and Science Act, H.R. 4346 (U.S. Congress 2022a, 2022b), and the Inflation Reduction Act, H.R. 5376 (U.S. Congress 2022a, 2022b) also created support for developing and demonstrating advanced nuclear power and power generation with carbon capture and utilization or storage (CCUS).

Anticipated climate change across the US Heartland presents multiple stresses on the power system, impacting the different generation technologies as well as demand and infrastructure (US Department of Energy—DOE 2022). Higher summer temperatures, combined with lower river levels and streamflow, can lead to higher water temperatures, and limit thermal power plant operations (Kopytko and Perkins 2011; Cook et al. 2015). McCall et al. (2016) found 43 water-related incidents of thermal power plant curtailment for the period 2000 to 2015 in the USA, 18 involving coal-fired power plants and 25 involving nuclear plants, and further climate-warming trends will increase this risk. A shift away from once-through thermal cooling could reduce curtailments but would add cost. A shift to non-thermal power sources, such as wind and solar, could avoid the risks posed by higher water temperatures altogether (e.g., Baker et al. 2014), but climate change may also alter wind patterns (Schlosser et al. 2022; Jung et al. 2019; Jung and Schindler 2019; Karnauskas et al. 2018) and limit solar radiation from increased cloudiness or smoke from wildfires. In addition, the efficiency of photovoltaic conversion of sunlight to electricity declines with higher temperature (Najafi and Woodbury 2013). At the same time, rising temperatures are also likely to increase summer peak demands as a result of more intense and broad use of air conditioning (e.g., Romitti and Sue-Wing 2022), amplifying already high summer peaks in the US Heartland (Bartos et al. 2016). In addition, high temperatures and high demand pose risks for failure of critical grid infrastructure, such as large power transformers (Gao et al. 2018). Other climate-related risks also affect grid infrastructure, including ice storms, wildfires, floods, and high wind-gust events that accompany severe storms (Allen-Dumas et al. 2019; Fant et al. 2020). Notably, many climate-induced stresses to the power system may be more likely in the summer months, creating the potential for a “perfect storm” that could lead to failure of the power system.

Climate impacts on electricity supply, demand, and infrastructure have implications for investment decisions, power system operations, and the resilience of different power system evolutions. A number of studies have investigated aspects of this issue for a variety of geographic regions (e.g., Ralston Fonseca et al. 2021; Khan et al. 2021; Steinberg et al. 2020; Craig et al. 2018). We contribute to the literature by looking at four alternative electricity sector pathways in the US Heartland that were developed without considering climate impacts and constructing a “potential supply gap” metric—the percentage difference between electricity generation in these projections and potential supply losses from climate impacts and additional demand due to warmer temperatures. As such, the calculation provides a first step toward considering systemic impacts, providing an initial estimate for system planners of changes in dispatch or other operational changes, or if those are inadequate, the need for additional capacity/reserves that can offset projected losses in production or be available to meet added demand.

We investigate how the potential supply gap varies by sub-region within the US Heartland and for different electricity sector pathways and identify which components of the system are more relevant to the gap. Our approach is to use a range of energy scenarios relying more or less on fossil, nuclear, and renewable generation. We review existing work in the area to consider realistic power generation scenarios from energy modeling projections and to assess estimates in the literature of potential impacts on electricity system technologies and demand due to future changes in climate and associated extremes. Energy modeling projections usually ignore the potential climate risks on electricity demand and supply.

The approach takes a multi-sectoral dynamics perspective, considering the interconnected and co-evolving environment-human interactions at varying geographic scales and multiple sectors, requiring interdisciplinary models and efforts to understand the behavior of these systems (Moss et al. 2016).

We consider the effects of multiple climate stressors from both gradual climate change and changes in extreme events. For example, a combination of extreme heat, drought, and stagnant meteorological conditions could have significant negative effects on all generation technologies, while simultaneously increasing peak power demands across the region. While we review multiple types of climate events and discuss potential implications for the power sector, in this initial assessment, we are unable to quantify all potential risks. Energy modeling projections usually ignore the potential climate risks on electricity demand and supply. This preliminary analysis, based on existing literature, lays a foundation for follow-on work to assess what these risks might mean for capacity expansion in the region, what strategies could be employed to make the power sector more resilient, and how to incorporate climate change into electricity dispatch and capacity expansion models.

As many have noted, the challenge with wind and solar power is that they are not dispatchable, and that alone may limit their potential contribution to the power mix—to examine whether their potential contribution requires a high time resolution representation of both demand and resource availability to assure that supply is matching demand over the course of a day, and over the year, either directly or through addition of power storage or extra dispatchable back-up capacity. Uncertainty in technology costs and federal and state energy policy creates uncertainty in the power mix, and as discussed above, every major power generation technology appears somewhat vulnerable to climate change. Given that assessment of the vulnerabilities of renewables require high resolution in time, we have chosen to examine scenarios developed using EleMod, an hourly dispatch and capacity expansion model of the US electric sector. Tapia-Ahumada et al. (2019) developed several scenarios under different technology and energy policy assumptions. Since those scenarios were completed, the cost of solar photovoltaics has continued to fall significantly (National Renewable Energy Laboratory—NREL 2022) and there are goals for the USA to achieve net zero emissions by 2050. We supplemented the above scenarios with one from the Solar Future Study developed by NREL and the Office of Energy Efficiency & Renewable Energy (SETO) from the US Department of Energy (SETO-NREL 2021) that relies more heavily on solar (SETO-NREL 2021) and achieves net zero emissions from the power sector by 2050.

We reviewed existing literature to gather insights and estimates of potential impacts of future changes in climate and extreme events on the electricity system and different technologies, which we can then apply to the electricity projections from our four scenarios. While several recent papers have focused on climate impacts on electricity in other regions of the USA, such as the West or Texas (e.g., Cohen et al. 2022a,b; Webster et al. 2022; Dyreson et al. 2022; Turner et al. 2021, 2019; Voisin et al. 2020, 2016) or on climate impacts averaged across the USA (e.g. Craig et al. 2018), here we focus on literature with estimates of impacts for the Heartland regions. A summary of this review is provided in Table (4).

Projections of electricity generation in the Heartland of the USA show alternative pathways for expansion and deployment of several technologies. Given falling costs, and policies to support their deployment, renewables appear likely to achieve a major role in most regions, especially under scenarios focused on decarbonization of the energy mix. However, thermo-electric generation from either fossil fuels or nuclear energy may remain relevant in some regions, depending on their cost and how fast the expected energy transition will proceed. Whatever the future development of the power system in the US Heartland, future climate changes can present multiple stresses on the power system. While there are many studies highlighting these risks, these climate feedbacks on the power sector are usually neglected in most projections of capacity expansion, generation, and demand. As discussed in the previous section, several studies have estimated potential impacts of changing climate conditions on overall electricity demand and alternative power sources. None of these, however, have considered these impacts all together, failing to capture the potential compounding effects and spatial teleconnections among them and the true risks faced by the electricity system in the US Heartland.

As a preliminary attempt to highlight the importance of considering such systemic impacts, we combine climate impact estimates from the literature with electricity mix projections from our scenarios to calculate preliminary estimates of potential supply gaps that could appear due to climate impacts on supply and demand as reviewed in Section 3. These potential supply gap estimates vary given the possible reliance of the power sector on different generating technologies in the four future power sector scenarios we considered in Fig. 3. We multiplied the annual energy projection by region from models discussed in Section 2 by a specific climate impact chosen from the literature review. Table 3 presents the overall assumptions and climate impacts used to calculate the potential supply gaps, which were formulated given the summary of the literature review (Table 6 in the Appendix). We avoided combining projections from different papers in some sort of median or max, since papers in the literature have too many differences in approach and assumptions, including modeling strategies, time and spatial dimensions, and climate scenario considered. For some energy sources, the number of papers is too narrow. These aspects make it virtually impossible to assign probability distribution functions and assess uncertainties. The potential supply gaps assume that the planned power system and its operations were not modified to eliminate or reduce the gap. Simple changes in operations, bringing more capacity on line to meet demand, may be sufficient to meet such potential gaps in many cases, while in other cases, more or different capacity might be required. The scenarios used in this paper were developed by a standalone version of EleMod (Tapia-Ahumada et al. 2019) which treats electricity demand as exogenous taken from the EIA’a Annual Energy Outlook (AEO) 2017. The scenario simulated by ReEDs similarly uses an exogenous demand scenario, based on EIA AEO 2020 (SETO-NREL 2021).

Figure 4 shows the potential supply gaps estimated based on climate impacts on each of the power sources and electricity demand (Fig. 5 in the Appendix decomposes these supply gaps by source in TWh, while Fig. 6 shows total supply gaps relative to model results in Fig. 3). The “medium-risk” calculation assumes climate scenarios compatible with RCP4.5 or similar, while “high risk” assumes RCP8.5 climate impacts. Under medium risks, potential supply gaps range from 5 to 6% in the NCENT region in comparison to power generation displayed in Fig. 3, 6 to 9% in the Lakes-Mid Atlantic, 8 to 11% in the SCENT, and 8 to 10% in the SEAST. Under high risks, potential supply gaps can reach between 11 and 12% in the NCENT, 14% and 21% in the Lakes-Mid Atlantic, 14% and 16% in the SCENT, and 15% and 19% in the SEAST. The additional electricity demand from climate impacts has a greater effect on the potential supply gaps than the losses suffered in energy supply in all scenarios, except in the Lakes-Mid Atlantic region under the Current Policies scenario, where its reliance on thermal power generation is higher than in other regions. Scenarios relying more on renewables are subject to lower potential supply gaps, since overall climate impacts on renewables mentioned in the literature are lower compared to those on thermal and hydropower. It may indicate higher resilience of renewables to climate impacts, but may also reflect a less-developed literature on climate impacts on renewables given these sources are relatively new in the power system and much less investigated than thermal and hydro power. It is also important to note that these potential supply gaps are based on estimated changes in annual supply and demand, whereas more serious threats with compounding impacts could arise with extreme events. For example, large-scale meteorological blocking events during summer months would support higher heat stress, reduce wind power, threaten thermal power generation, greatly increase peak power demand when there was little or now available excess capacity, and might involve periods of increased pollution/aerosol haze that reduce solar radiation reaching the surface.

The current study set out to determine potential supply gaps between the amount of electricity produced and used in US Heartland regions in scenarios that optimized dispatch and deployment of various supply technologies, under various policy assumptions, assuming climate was unchanged. Based on a review of the literature, we identified how demand and different supply technologies could be affected by changes in climate by mid-century. In general, most of these impacts resulted in reduced electricity production, given technology deployment and dispatch in the various scenarios we examined, and an increase in demand. We described the resulting difference between electricity production and use as potential supply gaps, which are meant to provide a useful first-order metric of the potential magnitude of the compounding effects of climate change on the electricity system (e.g., the impacts that could result if climate change is not appropriately factored into electricity system planning and/or adaptive responses are not properly pursued). However, they are only potential supply gaps as we recognize that utilities serving these regions have a variety of options at their disposal that they can adopt to ensure that electricity demand can be met each year and continuously over the course of a year. These options include short-term responses such as changing dispatch to make use of extra capacity, reserves and existing storage, options such as demand management and load shifting that may require somewhat more time to implement, and, because we are looking forward to mid-century, investing in extra reserve capacity, more storage, a different mix of generations options, or efforts to limit climate impacts on various production technologies (e.g., moving away from once-through cooling for thermal power plants, or different location decisions for wind turbines).

A clear limitation of our approach is that it does not identify the mix of options that could (would likely) be used to close this potential gap. Hence, it would be useful for future work to go a next step to produce new scenarios with an hourly dispatch and capacity expansion model, such as EleMod or a similar model, that would consider, first, whether changing dispatch and use of existing extra production and storage capacities (for 2050 as simulated with the model without climate change) were alone adequate to eliminate these supply gaps. A second step could then look at the potential role of load management via further improved efficiency beyond that simulated in the base scenarios and load shifting to help close potential supply gaps. A third step could then assess how changes in the investment mix of technologies, additional storage, and/or overall production capacity over the years leading up to mid-century could result in a system better optimized for the changed climate conditions. To optimize the system on the basis of expectations of future conditions would require a stochastic, forward-looking approach: forward looking because investment decisions in years leading up to mid-century would need to anticipate future conditions and stochastic because the details of future conditions are uncertain. Climate projections at relevant regional levels remain highly uncertain as do technology costs and the direction of GHG and conventional pollutant mitigation efforts that would affect the mix of production technologies. Electricity demand is also uncertain because the overall path of development (economic and population growth) is uncertain as is electricity demand growth because of competing influences (e.g., improved efficiency and increased electrification) and the impacts of climate change. Various decision-making under uncertainty approaches should be investigated, including robust decision making (e.g., Lempert et al. 2013), stochastic dynamic programming (e.g., Bellman 1957; Puterman 2005), stochastic programming (Birge & Louveaux 1997), approximate dynamic programming (e.g., Powell 2011), stochastic dual dynamic programming (e.g., Pereira and Pinto 1991), robust optimization (e.g., Ben-Tal et al. 2009), and others.

The other major limitation of the current study is the coarse resolution in time and space. While we reviewed the potential changes in extreme events that could affect the system, these extremes typically occur over short time periods, and may occur over very limited space. While extreme heat may spread over a large region, even over the entire mid-section of the country, it has different implications in different regions given the infrastructure in place in different regions. And, the effects of extremes such as floodin, depend on whether specific infrastructure is located in the flooded area or not. Clearly, extreme heat in mid-summer, when air conditioning demand tends to result in an annual peak in electricity demand even without climate change, that could also affect thermal power plant production and may be associated with low wind turbine output, would create greater stresses on the electricity grid than unusually warm weather in spring or fall, when wind turbine output may be strong, and there exists excess thermal power capacity because demand is relatively low in these periods. Further, changes in climate and in electrification will likely affect the seasonal and diurnal pattern of electricity demand. Warmer winters could reduce power demand then because of less heating and less use of fans and blowers in forced air heating systems, while warmer summers could increase demand then due to more air conditioning use. This could lead to a more extreme seasonal peaking profile. On the other hand, greater electrification through use of heat pumps could increase winter electricity demand, perhaps reducing the difference between winter and summer demand. Further electrification of the vehicle fleet is considered in the base scenarios we investigated, but electric vehicles recharging could shift diurnal patterns of demand, or by controlling when vehicles are recharged, perhaps through time of day pricing, could create greater load shifting capabilities.

The literature indicates a variety of impacts on power production and demand in the US Heartland from changing climate, which affects all power generation technologies. In general, this literature suggests increases in demand and a reduction in power production. We reviewed these estimates and made some preliminary calculations of how these various effects might combine to affect the power sector, developing a potential supply gap metric. The largest potential supply gap achieves 21% of total power generation projected by models for the 2040–2050 period, and is observed in the Lakes-Mid Atlantic region, due to potential large climate impacts both on energy demand and thermal power generation. While the impact on demand is expected in all the US Heartland, those sub-regions and scenarios counting on larger share of renewables in the power system tend to experience lower impacts on total power generation from the supply side, since the scientific literature on climate impacts on power technologies has indicated that thermal power will be more negatively affected than renewables.

These preliminary estimates are an indication of possible stresses on the power sector that may compound and can reach, on an annual basis, a potential supply gap of at least 5% in the NCENT region under a mild change in climate (consistent with RCP4.5). This will require operators to develop strategies such as changes in dispatch or other operations or, if those are insufficient, changes in the mix of generation, or more generation or storage capacity. Our calculation does not factor in all of the potential changes identified in the literature because in many cases, there is not yet solid quantification of some of them. Also, we have not attempted to quantify the risks from extreme events to the power sector infrastructure. We recognize that there are many uncertainties and a lack of consensus in the literature about the climate impacts on the various power technologies, as also as there is a relatively low number of climate impact estimates for some technologies, such as wind and solar. As a result, our supply gap measure should be taken as a first approximation of the potential risks climate may pose to the power system.

There are important aspects of our assessment that warrant further refinement, as mentioned in Section 5. First, we are looking at a regionally coarse representation. While this study provides initial insights on the potential impacts of compounding stressors that lead to potential supply gaps, future work should consider more disaggregated resolution and check if the electricity mix at state (or local) level is as vulnerable as at the US Heartland regional level. Second, while we are addressing annual potential supply gaps, it is expected that climate change will pose salient seasonal threats, especially in summer, when prolonged higher water temperatures, stagnant air, and dry conditions could reduce thermal power plant efficiency or require scaled back operations to meet regulations on water discharge. A key challenge will be the ability to faithfully predict changes in the extent, intensity, and occurrence of meteorological “blocking” events that cause these severe and compounding conditions. The ability to predict atmospheric blocking events under human-forced climate change remains a challenge (e.g., Lupo 2021), but recent evidence indicates that a warmer climate will cause blocking events to become larger in spatial extent (e.g., Nabizadeh et al. 2019). Third, while trends in the improvement in technologies such as solar and wind as well as efficiency improvements in end use and continuing trends in electrification were incorporated in baseline electricity demands, potential further responses due to climate change are not considered in most studies. Future work needs to consider technological improvements in the energy system in combination with climate shocks impacting demand. Overall, such improvements may allow a better integration of climate risks and electricity pathway options. It may also help to answer relevant questions, related to the best generation portfolio choice (more diverse or reliance on less vulnerable sources), the need for backup and storage capacity, water cooling options, chances of grid failure, or how costs may change to deal with these challenges.