Energy efficiency (EE) and renewable energy (RE) can benefit public health and the climate by displacing emissions from fossil-fuelled electrical generating units (EGUs). Benefits can vary substantially by EE/RE installation type and location, due to differing electricity generation or savings by location, characteristics of the electrical grid and displaced power plants, along with population patterns. However, previous studies have not formally examined how these dimensions individually and jointly contribute to variability in benefits across locations or EE/RE types. Here, we develop and demonstrate a high-resolution model to simulate and compare the monetized public health and climate benefits of four different illustrative EE/RE installation types in six different locations within the Mid-Atlantic and Lower Great Lakes of the United States. Annual benefits using central estimates for all pathways ranged from US$5.7–US$210 million (US$14–US$170MWh−1), emphasizing the importance of site-specific information in accurately estimating public health and climate benefits of EE/RE efforts.
Implementing EE/RE technologies can displace the emissions of greenhouse gases, both from EGUs and from upstream processes, thus producing climate benefits1. EE/RE will also have important public health ‘co-benefits’ by displacing air pollutant emissions, such as SO2 and NOx, which impact ambient concentrations of important public health drivers such as fine particulate matter (PM2.5; refs 2, 3, 4). Benefits can vary substantially across EE/RE types and locations, which makes understanding the drivers of variability important for decisions and analyses around EE/RE development and siting, along with energy and environmental policies. Site-specific quantification of climate and public health benefits from EE/RE proposals is challenging because: GHG, SO2, and NOx emissions all contribute appreciably to EGU impacts1, 2, 4, 5; the EGUs that EE/RE displaces depend on EE/RE performance, location and time dynamics, properties of other EGUs on the electrical grid, and the interaction of all these factors given electrical grid economics and transmission capabilities6, 7, 8, 9, 10, 11, 12; emissions of affected EGUs vary owing to fuel type, pollution controls and performance; and the public health impacts of PM2.5 formed from SO2 and NOx emissions vary across EGUs as a result of atmospheric conditions and population distributions downwind2, 3, 6. Ideally, a model that compares benefits of different EE/RE types and locations should include all these dimensions with high geographic resolution2, 3, 7, 8, 9, 10, 11, 12, 13, 14. A variety of studies have evaluated aspects of this question1, 2, 5, 6, 7, 8, 9, 12, 14, 15, 16, 17, 18, 19, 20, 21, 22, 23. However, none have connected all of these key elements into a single simulation framework, and then simulated a set of EE/RE implementation scenarios, each differing by type and location, in a manner that facilitates comparison of benefits across both EE/RE type and location.
Here, we have developed the Environmental Policy Simulation Tool for Electrical grid INterventions (EPSTEIN), for the Eastern Interconnection (the electrical grid for regions of the United States (US) and Canada east of the Rocky Mountains) for 2012. The EPSTEIN model links output from a complex economic simulation electrical dispatch model to a public health impact assessment model that provides EGU-specific monetized health impact estimates for SO2 and NOx emissions6, 7, 8, 9, 10, 11, 12 and monetized estimates of the impacts of CO2 emissions2, 3, 5, 7, 12, 13, 24. We then simulated the effects of four different EE/RE installation types—500MW wind, 500MW solar, 500MW peak demand-side management (DSM; reduction in electricity demand), and 150MW baseload DSM (so it conserves an amount of energy comparable to what our wind and solar scenarios generate)—in six different locations on the PJM Interconnection. We simulated each installation type in each location independently, for a total of 24 simulated scenarios. Each scenario represents one installation type in one location, and the set of 24 scenarios represent all combinations of installation type and location. The PJM Interconnection is the Regional Transmission Organization (RTO, a contiguous electrical transmission area) within the Eastern Interconnection that manages the electrical grid and market in much of the Great Lakes and Mid-Atlantic region of the US. The locations are near Chicago (Illinois), North-central Ohio, Southern New Jersey, Eastern Pennsylvania, Virginia, and near Cincinnati (Ohio). We then compare benefits across EE/RE type and location, and examine differences on the basis of differing electricity generation or savings, economics and constraints of the local electrical grid, displaced EGUs, and populations downwind. This approach, using illustrative scenarios, allows direct comparisons between different EE/RE types and locations.
To explain variability in benefits across the 24 scenarios, we employed a basic theoretical decomposition of the model. Total benefits were modelled as a function of the electricity generation from RE installations or demand reductions from EE installations (termed ‘capacity factor’ herein for simplicity), displaced generation by fuel type, and then for each fuel type, the aggregated emissions rates of displaced EGUs and the aggregated health or climate damages per unit emissions for the displaced EGUs. A simplified heuristic equation for this decomposition is presented below:
We refer to this simplified heuristic equation within our manuscript as ‘Benefits = CFEI’. We used this decomposition framework to examine drivers of variability across the 24 scenarios.
Total annual health and climate benefits varied by a factor of approximately 37 across the 24 scenarios, with central estimates ranging from US$5.7 million for DSM peak in Eastern Pennsylvania to US$210 million for a wind energy installation in the areas of both Cincinnati and Chicago (Table 1 and Fig. 1). With central estimates for all parameters, displaced SO2 from coal generally dominated the total benefits (Fig. 1 and Supplementary Fig. 3). Variability in total benefits is explained in part by large differences in EE/RE capacity factor (the ‘C’ term of Benefits = CFEI), which ranged from 1,431GWh from a 500MW wind energy installation near Chicago to 281GWh for 500MW DSM peak near Cincinnati (Supplementary Table 3). The benefits per unit electricity, related to the ‘FEI’ terms of our heuristic equation, also varied by a factor of 12, with values ranging from US$14 to US$170MWh−1 across all 24 scenarios (Table 2).
Table 1: Total benefits from CO2, NOx, and SO2 emissions reductions for each of four EE/RE types in six locations, in 2012 US$ millions.
There was substantial variability in benefits across all 24 scenarios, and generation displaced, proportion of fuel types displaced, emissions rates of displaced EGUs, and impacts per unit emission of the displaced EGUs equation (1) were all important in determining the total benefits of each EE/RE scenario. Although there were significant differences in the percentage of each fuel displaced between scenarios, most EE/RE types displaced substantial amounts of coal, especially wind and DSM base. A likely contributor to this is the recent decrease in natural gas prices and increase in coal prices15, 25, 26, consequently making coal more likely to be displaced on the margin than it would have been in previous years.
There were some important differences in displaced fuel types due to both EE/RE type and location. Solar PV and DSM peak operate at times of relatively high electricity demand, so they tend to displace more expensive EGUs operating during times of peak electricity demand, generally natural gas. In contrast, wind and DSM base both operate at off-peak times, including spring, fall, and nights, resulting in more displacement of coal EGUs because they are closer to the margin during these times. Although there were small changes in generation across the entire Eastern Interconnection (Supplementary Fig. 4), the majority of the displaced EGUs were located near where EE/RE was implemented, even where coal generation is the major generation source. As a result, locations with greater amounts of coal generation generally had greater coal displacement, and higher benefits. This also occurred in areas where nuclear contributes substantially to baseload generation. These patterns indicate that EE/RE is capable of displacing coal generation in areas where peaking sources such as natural gas have small shares of generation. EE/RE can put coal in price competition with natural gas, which is much more flexible, and sufficient price pressure can force coal to reduce operation.
Our results demonstrate the importance of dynamics that our model and scenario design can simulate. EE/RE can relieve constraints on transmission capacity, can force some EGUs to operate below minimum operating capacity (thus forcing them to shut off completely and others to turn on), or can force some EGUs to operate at lower capacity and higher heat rates of fuel, resulting in higher emissions rates. The variability in emission rates of displaced EGUs is also predicated on underlying variability across all EGUs, owing to differences in EGU efficiency, emissions control technologies, and fuel mix. The variability in emissions and impact rates of displaced EGUs is also partially due to EE/RE decreasing utilization of some EGUs but increasing utilization of others, which can result in net rates changing beyond the range of individual EGUs.
An example of these complex dynamics is that EE/RE displaces coal in areas with other baseload generation sources, for instance nuclear, such as the Chicago area (Supplementary Table 2). This may indicate that coal is on the margin frequently, owing to transmission constraints and the need for natural gas generation as a peaking source, emphasizing the importance of detailed dispatch modelling reflecting contemporary fuel prices. Another example of these dynamics is in Eastern Pennsylvania, where DSM peak has a small total benefit, including some induced impact from SO2 emissions. Here, DSM peak displaces a substantial amount of natural gas while inducing use of coal-fired EGUs with high SO2 emissions rates. This indicates that natural gas is the dominant peak generation type in this area, and that DSM peak installations can induce emissions by shifting the optimal dispatch order to higher-emitting sources, based on unit-specific minimum operation levels and commitment decisions.
In general, EE/RE has a net displacement of EGUs with high emissions rates, because EGUs operating in EE/RE scenarios have slightly lower emissions rates for both NOx and SO2 than EGUs operating in the baseline scenario (Supplementary Fig. 3). This is more pronounced for natural gas EGUs than for coal. The displaced coal EGUs are probably older, less efficient, run without SO2 controls, and use higher-sulphur coal. For natural gas EGUs, these are probably older, less efficient EGUs or EGUs that co-fire with other fuels. These patterns indicate that the incremental costs to operate pollution control technologies may be generally small relative to other operating costs, or that higher-emitting EGUs are generally less efficient and more expensive to operate; otherwise, lower-emitting EGUs would probably be first to be displaced.
The impacts per unit emissions from displaced EGUs also varied across EE/RE scenarios. In general, impact rates of NOx had higher variability than SO2, probably owing in part to the complex chemistry of secondary particulate matter formation from NOx emissions reflected in the health impact model2, 5, 6. Impact rates per ton of emissions for coal-fired EGUs tended to be higher than for natural gas-fired EGUs (Fig. 4), indicating that coal-fired EGUs tend to be located in places with larger exposed populations than natural gas plants.
Although our estimates are interpretable and reflect factors that influence site-specific benefits, our model has some limitations. We included only a subset of pollutants and impact pathways that dominated prior analyses2, 3, 5, 7, 12, 24, 27, 28 and for which we could construct detailed site-specific estimates. We did not include stack emissions of primary PM2.5, methane, mercury, or other pollutants, nor morbidity effects of PM2.5 or secondary formation of ozone. We also did not account for possible differences in impact/ton depending on emission timing, or slight changes in emissions rates due to power plants cycling in response to higher variability in both electricity supply and demand13, URL:
Jonathan J. Buonocore. Health and climate benefits of different energy-efficiency and renewable energy choices[J]. Nature Climate Change,2015-08-31,Volume:6:Pages:100;105 (2016).
