Highly energy efficient housing can reduce peak load and increase safety under beneficial electrification

Climate change is driving urgent investments in decarbonization. One core decarbonization strategy is to electrify energy services that currently directly use fossil fuels, because electricity can be generated from zero greenhouse gas energy resources. Shifting fossil-based services to electricity, however, requires a major expansion of electricity supply and increases dependence on electricity for critical services. Home heating is a particular challenge, especially in very cold climates. Unserved heating loads can be fatal. Electrified heating is expected to drive peak loads (and thus overall grid size) due to high coincident and nondeferrable loads. This study shows that highly efficient housing presents an opportunity to simultaneously protect people and structurally reduce peak load, reducing the need for electricity supply infrastructure while increasing people’s resilience to weather extremes. This study uses seven building efficiency scenarios from the National Renewable Energy Laboratory’s End Use Saving Shapes to investigate the impact of residential building efficiency on grid size in 2050, using the example of Pierre, South Dakota as a very cold weather location that might also experience substantial new housing demand due to climate-induced human mobility. We find that the deepest efficiency electrification scenario we investigate reduces peak demand by about half relative to low-efficiency electrification. Costs of about $3900/kilowatt (kW) peak load reduction are competitive with the cost of new decarbonized supplies capable of meeting peak load, though building efficiency costs are usually privatized while supply expansion costs are distributed across ratepayers. Decarbonization scenarios suggest the US grid might need to expand by a factor of 5–8 in the next 25 years: extremely rapid growth will be needed regardless, but targets might not be reachable with inefficient end users. Residential building efficiency presents an urgent opportunity to reduce peak demand and provide safer and more resilient housing.


Introduction
As the world continues to experience the impacts of climate change, the need to eliminate greenhouse gas (GHG) emissions and transition towards clean energy sources is ever more urgent (Intergovernmental Panel on Climate Change 2023).In the United States (US), the Biden-Harris Administration's target is net-zero GHG emissions by 2050 and 100% carbon pollution-free electricity by 2035 (White House 2021), similar to global emissions timeframes to limit temperature increases (Intergovernmental Panel on Climate Change 2023).In the US and elsewhere, electrification is expected to play a pivotal role in achieving comprehensive decarbonization, largely because electricity is the most common energy carrier for zero-carbon energy resources and thus presents an opportunity to replace fossil fuels with zero-carbon energy resources (Williams et al 2021, Berrill et al 2022).
Although electrification is a well-known strategy for the decarbonization of fossil fuel-reliant end uses, the way in which electrification proceeds has major ramifications for resource intensity, disbenefits, the pace of transition, the resilience of electrified end uses, and equity (Riofrancos et al 2023).For context, consider that the US electricity supply system has about 1 terawatt (TW) of installed capacity as of 2021 (EIA 2021); net-zero GHG emission scenarios suggest a need for 5-8 TW of installed capacity by 2050 (Williams et al 2021, Browning et al 2023), or roughly in the next 25 years-a level that would take 80-150 years to reach at current installation rates.Installed capacity needs grow faster than demand (Tarroja et al 2020, Williams et al 2021), in part because electricity is difficult to store and key decarbonized resources like wind and solar have variable availability.As such, peak loads, or maximum simultaneous electricity demands, have an outsized impact on the amount of infrastructure that must be built-especially those that cannot be shifted much, if at all, in time, like safety-critical heating loads that might last for hours to days.This relationship between load growth and system growth is one reason for the focus on what is called 'beneficial electrification,' often defined as electrification that reduces GHG emissions, prioritizes efficiency, enables cost savings, and facilitates grid management under deep decarbonization (Dennis et al 2016, Shipley et al 2018).
Residential space heating is expected to both drive grid size (Keskar et al 2023) and considerably intensify the consequences of failure (Busby et al 2021).Historically, utilities have experienced summer peaking, driven by (electrified) air conditioning (Vaishnav andFatimah 2020, Michelfelder andPilotte 2022).Under full building electrification, however, which is increasingly seen as nonoptional for deep decarbonization, winter peaks are likely because of the much higher temperature differences between outdoor and safe indoor temperatures during cold than hot conditions (Waite and Modi 2020, Keskar et al 2023).The remaking and significant expansion of the electricity system over the next few decades occurs within the context of difficult-to-manage and uncertain dynamics from climate and technology change (Grubert and Hastings-Simon 2022).As such, strategies that structurally promote safety and lower peak load are particularly critical to successful and human-centering decarbonization.This study argues that deep residential building efficiency, coupled with electrification, is a primary strategy critical for both emissions reductions and resilience, potentially delivering significant grid benefits alongside improved safety, health, and comfort.
In this letter we ask: how much can residential building efficiency reduce peak electricity load under essentially complete electrification, including for residential space heating, and what would it cost?We are especially interested in understanding how potentially high-growth areas with extreme winter peak load might understand deeply efficient residential buildings and grid investments as complementary.We identify and evaluate Pierre, South Dakota (SD) as an example of a community that experiences extreme cold and could see substantial demand for new housing, including from climate-induced mobility (Maxim and Grubert 2021).

Methods
This paper uses scenario analysis to evaluate potential peak electricity load, intervention costs, and electricity load shapes driven by residential demand for seven levels of building efficiency and/or electrification interventions, including nonelectrified and electrified baselines (details of specific upgrades for each scenario are given in tables S1 and S2).The analysis is applied to a case study community under assumed historical (2018) and future ( 2050) conditions.As one major goal of this work is to inform decadal infrastructure planning related to climate change, we specifically account for potential population growth associated with climate-induced mobility.This section describes the approach, data (temperature, energy consumption, population growth, and costs), case study selection, and calculations made to conduct the analysis.

Approach
In this analysis we estimate 2050 peak residential electric load under various building efficiency conditions, selecting 2050 due to alignment with the US netzero emissions target as of this writing (White House 2021).Two core assumptions are that (1) under full electrification, peak load will occur on the coldest day, and (2) residential load can be estimated by multiplying the average load per residence by the number of residences for each time step of interest.We require four major inputs: (1) high spatiotemporal resolution temperature data to identify conditions likely to drive a peak; (2) high spatiotemporal resolution residential load shapes for building efficiency conditions of interest; (3) the number of residences in 2050; and (4) intervention-level building efficiency cost estimates.We select a case study based on (1) exposure to extreme cold; (2) expected population growth driving demand for new housing (where efficiency interventions could be particularly effective); and (3) availability of compatible data, particularly residential load profiles for extremely cold days.

Data
Data sources and their purpose are summarized in table S3, and detailed in this section.

Temperature
For the purpose of identifying historical extreme cold events, we use EpiNOAA, a derivation of the National Oceanic and Atmospheric Administration (NOAA) US Climate Gridded Dataset (NClimGrid) that includes daily maximum, minimum, and average temperatures at the county level for the continental United States (CONUS) for 1951 to the present (Vose et al 2022).As EpiNOAA does not include hourly data, after identifying the extreme values and selecting locations of interest, we supplement with local hourly data from high quality meteorological databases with appropriate local coverage in order to match load shapes with temperature information (specifically described in the context of the selected case study, below).

Residential load shapes
We use the National Renewable Energy Laboratory's (NREL) End Use Savings Shapes (EUSS) for residential load profiles (2022.1 release with AMY2018 weather data) (NREL 2022b) to quantify perresidence energy consumption.As this EUSS release uses the 2018 weather year, this analysis adopts 2018 as its historical base year for parameters like population.EUSS is based on NREL's End Use Load Profiles, which model building energy consumption across the US commercial and residential building stock at 15 min intervals, validated using empirical data (NREL 2022a).EUSS data are available at multiple spatial levels, including State, Public Use Microdata Area, and County, and can be filtered to specific counties using Federal Information Processing System (FIPS) county codes.Every model in the study signifies the energy usage of a building or housing unit with distinct characteristics, including the type and efficiency of the heating system, wall insulation, and the behavior of a specific set of occupants, such as their occupancy hours, cooking schedule, and thermostat setpoints (NREL 2022b).The energy consumption variables measured include electricity consumption related to heating; electricity consumption not related to heating; and the site energy total, which includes fuel oil, natural gas, propane, and electricity.Instantaneous non-electricity energy use is reported as thermal megawatts in this study.
One major reason for selecting EUSS as the data source for residential load profiles is that it explicitly includes energy efficiency and electrification package impacts on residential load shapes.For this analysis, we investigate seven residential building efficiency and electrification scenarios from NREL's EUSS (NREL 2022b) using the 2018 building model distribution (tables S1, S2 and 1), including both nonelectrified and electrified baselines ('Baseline' and 'Whole-home electrification, min-efficiency,' respectively).These scenarios include both building envelope and appliance interventions with variable efficiency.Details on the development of, and assumptions underlying, these scenarios and the modeled load shapes can be found in the NREL EUSS documentation (NREL 2022b).

Number of residences
We account for both typical population changes and climate-induced mobility to estimate the number of residences by 2050.Population estimates are taken from Maxim and Grubert (2021), which is based on the Environmental Protection Agency's Integrated Climate and Land Use Scenarios (ICLUS) v2 model.We use 2020 Census values for people per household to estimate the number of residences from population (US Census Bureau 2022).

Residential cost data
We use cost data from the Advanced Building Consortium (ABC) (Advanced Building Construction Collaborative 2023), with guidance from data owners to match upgrade measures to EUSS efficiency and electrification scenarios.For dryers and cooking ranges (where building characteristics do not influence sizing, unlike for water heaters and heat pumps), we use costs for purchase, delivery, and installation to the appropriate location from the Home Depot website (data obtained August 2023).Costs from the ABC database are adjusted to the appropriate location using RSMeans' City Cost Index (2019; most recent publicly available data) (RSMeans 2020) and inflated to June 2023 values using the Bureau of Labor Statistics inflation calculator (US Bureau of Labor Statistics 2023).

Case study selection
With the goal of identifying how residential building interventions might affect grid needs associated with electrification and decarbonization in contexts where both grid size and number of new residences are expected to grow substantially over the next several decades, we select a case study location based on three primary factors: (1) the location experiences extreme cold, which is expected to drive large grid growth under electrification; (2) the location experienced an all-time (or near all-time) record low temperature in weather year 2018 (the year for which we have information about grid response in EUSS), allowing us to investigate load and load shape for likely peak conditions under electrification; and (3) the location is expected to experience high population growth by 2050, driving demand for new residences (and thus opportunity for building-level interventions).
Using the datasets described above, we identify the microarea of Pierre, SD as a case study location  S4).Pierre could experience sustained growth through the end of century with or without climate changeinduced in-migration (Maxim and Grubert 2021), indicating a large need for new housing.Figure 1 shows microareas that could experience large population growth by 2050 under an extreme climate scenario (RCP 8.5, SSP5, GISS-E2-R), overlaid on CONUS minimum temperatures for 1 January 2018.

Calculations
Using Pierre, SD as a case study, we use the approach and data described above to estimate peak load, intervention costs, and daily residential load curves for seven building efficiency scenarios (table S1) for the extreme winter day (matching 1 January 2018) under 2018 and 2050 population conditions.Daily hourly temperatures are taken from South Dakota State University's Mesonet database, using the Dakota Lakes Station as the nearest station with hourly temperature data (South Dakota Mesonet, South Dakota State University 2020).
NREL EUSS 15 min time series data were downloaded for the dates of interest for the customized aggregation of Stanley and Hughes counties that make up the Pierre Microarea.Data are presented as (1) This day-specific single residence load curve is then multiplied by the number of housing units of that building type projected for 2050 to estimate total residential load for each time step, which we use to derive both peak load (by finding the maximum value for any 15 min interval) and load shape for the 24 h period of interest.For Pierre, we estimate a population of 50 000 in 2050 (Maxim and Grubert 2021) and use the 2020 Census value of 2.12 people per household weighted across both counties comprising the Pierre Microarea (US Census Bureau 2022) to estimate a total of 23 600 housing units in 2050, compared to 9000 in 2018.We assume the distribution of building types remains static.
We estimate building efficiency and electrification costs using metadata at the individual building model level from the NREL EUSS dataset, filtered to the Pierre Microarea (county codes G4600650 and G4601170 in EUSS), and multiplying relevant indicators by per-unit costs from the ABC database.For example, to estimate the cost of drill and fill wall insulation (used for both basic and enhanced enclosure upgrade packages), we combine estimates for the total above grade conditioned wall area with cost estimates per unit of treatment area (table S2).We use the 2019 city cost index combined material and installation location adjustment factor of 1.036 for Pierre relative to the national average (RSMeans 2020).

Results
Figure 2 shows estimated peak load for Pierre, SD in 2018 and 2050 by building efficiency electrification scenario for the extreme cold day (matching 1 January 2018: see figures S1-S4 for estimates based on arbitrarily selected typical winter and summer days).For both 2018 (top) and 2050 (bottom), figure 2 depicts demand on the electric grid from electric heating equipment alone ('Electricity-Heating'); demand on the electric grid from all aggregated electricity use ('Total-Electricity'); and lastly, for reference, all site energy requirements (bottom: 'Total-Site Energy').Note that the site total includes propane, fuel oil, and natural gas (presented as thermal megawatts) in addition to electricity for scenarios that are not completely electrified.
Full electrification without attention to efficiency, which we define as the electrified baseline (Wholehome electrification, min-efficiency), increases electricity demand by a factor of 2.5, but reduces peak site energy by 20% due to the higher efficiency of some electrified equipment relative to fuel-based equipment.The most efficient electrified scenario (Wholehome electrification, high efficiency + enhanced enclosure) reduces peak load by 45% relative to the electrified baseline.Relative to the nonelectrified baseline, the most efficient electrified scenario increases peak electricity load by 35% and reduces peak site energy demand by 55%.The most efficient electrified scenario accommodates a factor of 2.6 population growth by 2050 while increasing peak site energy demand by only about 10% relative to the 2018 baseline.
Table 1 shows estimated costs for each residential building efficiency intervention package, alongside an estimate of capital cost per kW peak reduction relative to either the nonelectrified or electrified baseline for an extreme winter day, which we assume is the overall peak under electrification (figure 2).
Deeper efficiency interventions are more costly per average residence, but also more cost effective for peak load reduction.
Figure 3 presents the estimated residential grid load for Pierre, SD for an extreme winter day at 15 min intervals assuming 2018 and 2050 populations across seven scenarios for residential building efficiency and electrification.For reference, the hourly temperature graph for 1 January 2018 is plotted on the second y-axis in light blue, illustrating the relationship between temperature and energy demand.

Discussion
As this analysis of Pierre, SD reiterates (see also Williams et al 2021, Browning et al 2023), electrification will almost certainly result in a substantial increase in peak demand and require growth across the entire grid (Vaishnav and Fatimah 2020).Although electrification is likely to drive overall energy efficiency improvements, the extent to which efficiency is prioritized could have a meaningful impact on how much grid buildout is required.As figure 2 shows, low efficiency electrification with no building envelope improvements (WHE-ME) would have raised Pierre's estimated residential heating electricity load on 1 January 2018 from about 50 MW to about 150 MW (total electricity from 64 to 158 MW)-tripling the need for deliverable electricity for residential heat alone.Full electrification with maximum efficiency and enhanced enclosure interventions (WHE-HE-EC), however, cuts peak load for the electrified scenarios approximately in half.Although the high efficiency electrification scenario reduces peak site energy load, the electrical load still exceeds the nonelectrified baseline level by about 35%, reinforcing the point that decarbonization via electrification will require grid expansion even with highly efficient load.Crucially, though, the level of this grid expansion need can be moderated by attention to efficient housing (figure 2).For the 2050, higher population scenario, the buildout of new, fully electrified housing could drive peak loads of between 200 and 400 MW by 2050, depending on efficiency measures-growth to 3-6 times the size of the current system, for a population growing to 2.6 times its current size.Such growth rate projections are consistent with estimates for the country as a whole under net zero GHG emissions scenarios (Browning et al 2023), highlighting the critical importance of attention to end use efficiency during build out.Particularly for places like Pierre where much of the potential housing stock could be new build, focusing on building efficiency could be a crucial and relatively straightforward tool for reducing the total amount of electricity system growth required to achieve decarbonization goals.
Demand-side interventions in the form of deep residential building efficiency might also be cost effective relative to supply-side buildout.For the interventions that include both envelope and appliance efficiency (WHE-HE-BE and WHE-HE-EC), expected cost per kilowatt peak reduction under full electrification is competitive with expected capital costs for decarbonized supply side investments capable of up and down dispatchability, which are necessary for meeting peak loads under full decarbonization.These scenarios produce peak load reduction at costs of $3900 and $4000/kW (table 1), with few if any additional costs beyond capital investments-and likely significant savings on electricity bills.By comparison, NREL's 2023 Annual Technology Baseline (ATB) estimates overnight capital costs of ($2500, $4900) for pumped storage hydroelectricity; ($5400, $7100) for geothermal; and ($8800, $8800) for nuclear ([low, high], adjusted from $2020 to $2023, June: high and low values for nuclear are the same) (NREL 2023).Estimates for residential and utility-scale batteries, which do not include any costs associated with the capital infrastructure required to charge them, are ($4300, $5600) and ($1100, $4200), respectively (NREL 2023).As is commonly observed for building efficiency, the greatest benefits are available for packages of interventions that are collectively cost competitive due to synergies across upgrades (e.g. a more efficient heater can deliver heating services even more effectively in a well-insulated home).Furthermore, non-climate health benefits of replacing home combustion sources with electricity (Gruenwald et al 2022), insulating homes (Howden-Chapman et al 2007), and otherwise improving housing quality can be meaningful, with opportunities to improve health equity (Swope and Hernández 2019).
One major and well known challenge is that high upfront cost can be a significant barrier for building upgrades.In addition to the direct costs we evaluate here, electrical service and distribution grid upgrades can be costly (Elmallah et al 2022), though crucial for most electricity-based decarbonization approaches.As investments in electricity supply measures are more commonly socialized (e.g. through utility rates) than investments in electricity demand measures, increasing energy supply to serve inefficient load is often more straightforward to fund even when demand-side interventions are more cost effective.Electrification-based decarbonization, however, poses a paradigm shift that demands revisiting whether socializing supply costs to overcome inefficient demand can be tolerated: as this case study shows, the difference between efficient and inefficient electrification could be a question of sextupling versus tripling electricity generating capacity that is available in extreme conditions, within 25 years, while managing major technology and climate dynamics.Even tripling the size of the electricity system will be extraordinarily challenging.Demand-side interventions like residential building efficiency simultaneously significantly reduce the need for grid buildout and significantly improve safety and resilience for occupants, likely at lower cost than marginal supply.
As figure 3 shows, the high-efficiency wholehome electrification scenarios that also include more efficient building enclosures (WHE-HE-BC and WHE-HE-EC) both have flattened load shapes, likely reflecting that residences are able to maintain temperature for longer durations and thus can spread heating load over the day relative to the less efficient envelopes.Flatter load shape allows for lower capacity-to-demand ratios on the grid, allowing for more distribution of load across resources (Odukomaiya et al 2021).This has significant implications for how much of the system can go offline in extreme events while keeping life-saving equipment online for the population.Higher grid resilience is particularly relevant under extreme conditions: an extremely cold day makes it harder for equipment to operate, as occurred during winter storm Uri and the cascading system failure issues (Glazer et al 2021).
Electrification without other temperature control methods makes homes entirely reliant on the electric Inefficiency and high peak load needs are costly (Vaishnav and Fatimah 2020) and disproportionately affect vulnerable people (Maxim and Grubert 2022).Low-income people are more likely to live in poorly insulated homes and rely on electric heating systems that are less efficient and more expensive to operate than other heating options (Drehobl and Ross 2016).As a result, they may face higher energy bills and struggle to keep their homes warm during the winter months, which can have negative health impacts and contribute to energy poverty.Low-income households spend a higher percentage of their income on energy bills than higher-income households, and energy costs can represent a significant financial burden for these households (Drehobl and Ross 2016).Decarbonizing and retrofitting homes for energy efficiency can be very challenging (Less et al 2021) but can provide major health, safety, and operational cost benefits, particularly for the poor.Overall, this study emphasizes the importance of energy-efficient homes for achieving climate and justice goals, and highlights the need for policies and programs to make retrofits more affordable and accessible for all people in all homes.Strategies for efficiency are well known, well understood, and available, even in difficult climates and particularly for new buildings, though deployment continues to lag knowledge and retrofits can be especially challenging.

Limitations
Key limitations of this analysis include assumptions of static utility, building distribution, and demand curve structures: particularly assuming that much of the electricity grid and building stock would need to be newly built by 2050, it is unlikely that certain structural characteristics of the system (e.g.transmission and distribution loss rates, types of housing, how people use electricity) remain static.Peak loads are calculated based on estimated 15 min average consumption rather than instantaneous loads: as such, real peak loads could be somewhat higher.Load curves do not account for electric vehicle penetration, although for peak load estimates under extreme conditions, it is probably reasonable to assume that any noncritical loads would be switched off.Projected population estimates and grid estimates from home energy consumption data are uncertain.Cost estimates for envelope interventions are based on retrofits and accordingly might overestimate costs for new build residences.We assume residential load is effectively full load under extreme conditions.The analysis fundamentally assumes that electrification and building efficiency interventions would be successful, though the retrofitting of existing buildings in particular can be challenging.Peak load might not occur on the coldest day, depending on longer weather trends, whether people are mainly at home on that day, and other factors.Our use of a single case study enables high resolution analysis, but findings might not be generalizable, suggesting opportunities for more work across longer time periods and larger geographies.The scenarios investigated in this study (e.g. for population growth and residential load under various efficiency scenarios) rely on model data and are intended to illustrate potential outcomes under various conditions rather than predict the future with high accuracy.As such, the qualitative conclusion that deep residential energy efficiency could substantially and cost effectively reduce peak electricity loads under building electrification is more robust than specific quantitative results.

Conclusion
Residential building efficiency is a major and perhaps necessary component of deep decarbonization, particularly because of the increasingly clear role of electrification in reaching net-zero GHG emissions goals.Residences often drive load during emergency conditions, often related to widespread extreme temperatures that prompt difficult-to-shift coincident load for heating or cooling, while simultaneously being locations of resilience and adaptation to climate change.Our analysis, using a case study of an extreme cold day in Pierre, SD, suggests that careful attention to deep residential building efficiency could cut peak electricity load in half under decarbonized electrification-a critically important observation given that even in the most optimistic cases, peak loads might triple relative to what they are today.These interventions dramatically reduce the amount of supply side infrastructure that needs to be built, potentially at costs competitive with supplyside interventions, while also providing substantial human welfare benefits in the form of safety in extreme conditions and improved health and comfort with lower energy burdens in typical conditions.Building improvements are well understood, though challenging, with particular opportunities in places like Pierre that might see rapid growth in residential construction due to climate-induced mobility.Future work could evaluate implications of deep building efficiencies in other climate and growth settings, more closely investigate distributional impacts, and evaluate strategies for funding and deployment.As decarbonization accelerates, we have a unique opportunity to design homes efficiently, increasing resilience, reducing the pace of required electricity system buildout, and decarbonizing.
kilowatt-hour (kWh) demand over 15 min intervals.We convert these kWh/15 min data to end user load in megawatts (MW) for each 15 min interval, then to grid load in MW using the City of Pierre 2018 transmission and distribution loss rate (EIA 2019), to construct a single residential load curve for each building type, as follows: grid load (MW) utility transmission & distribution loss) .

Figure 2 .
Figure 2.Estimated 2018 and 2050 peak residential energy demand values for the Pierre, SD microarea by efficiency and/or electrification package for all residences, extreme winter day.

Figure 3 .
Figure 3. 2018 and 2050 electricity demand residential load shapes for the Pierre, SD microarea by efficiency and/or electrification package for all residences, extreme winter day.

Table 1 .
Estimated costs of building efficiency interventions by upgrade and package ($2023/average Pierre residence based on 2018 residence distribution or, where stated, $2023/kW peak reduction, extreme winter day).
a $2023/kW peak reduction, extreme winter day.