## Abstract

In this paper, a resiliency analysis is carried out to assess the energy, economic, and power outage survivability benefits of efficient and net-zero energy communities. The analysis addresses the appropriate steps to designing an energy-efficient and net-zero energy community using Phoenix, Arizona, as a primary location for weather and utility inputs. A baseline home is established using International Energy Conservation Code (IECC) 2018 code requirements. Three occupancy levels are evaluated in BEopt to provide diversity in the community’s building stock. The loads from the baseline, energy-efficient optimum, and net-zero energy optimum single-family homes are utilized to determine energy use profiles for various residential community types using occupancy statistics for Phoenix. Then, REopt is used to determine the photovoltaic (PV) and battery storage system sizes necessary for the community to survive a 72-hour power outage. The analysis results indicated that the baseline community requires a 544-kW PV system and 375-kW/1,564 kWh battery storage system to keep all electrical loads online during a 72-hour power outage. The energy-efficient community requires a 291-kW PV system and a 202-kW/820 kWh battery storage system while the net-zero energy community requires a 291-kW PV system and a 191-kW/880 kWh battery storage system. In this study, the economic analysis indicates that it is 31% more cost-effective to install a shared PV plus storage system than to install individual PV plus storage systems in an energy-efficient community. After analyzing the system sizes and costs required to survive various outage durations, it is found that only a 4% difference in net present cost exists between a system sized for a 24-hour outage and a 144-hour outage. In the event of a pandemic or an event that causes a community-wide lockdown, the energy-efficient community would only survive 6 h out of a 72-hour power outage during a time where plug loads are increased by 50% due to added laptops, monitors, and other office electronics. Finally, a climate sensitivity analysis is conducted for efficient communities in Naperville, Illinois, and Augusta, Maine. The analysis suggests that for a 72-hour power outage starting on the peak demand day and time of the year, the cost of resiliency is higher in climates with more heating and cooling needs as heating, ventilation, air conditioning, and cooling (HVAC) is consistently the largest load in a residential building.

## 1 Introduction

Net-zero energy (NZE) buildings and communities are gaining public interest as grid challenges and climate change become urgent issues. In 2019, the European Commission published a directive requiring all new buildings to be nearly zero-energy starting in 2021 to decarbonize the building stock by 2050 [1]. The state of California requires all new residential and commercial construction to be NZE by 2020 and 2030, respectively [2]. According to the Rocky Mountain Institute, the energy performance of NZE homes provides a breadth of benefits to builders, homeowners, utilities, and communities through reduced performance risks, affordability, and reduced energy insecurity. Despite the benefits, builder and consumer surveys indicate misperceptions about the cost and benefits of NZE homes, disincentivizing stakeholders from acting in their own long-term interests [3]. As of 2018, less than 20,000 residential buildings in the United States are achieving NZE with large multi-family buildings dominating the residential NZE stock [4]. While it is simple to consider NZE at an individual level, approaching NZE at the community scale can be advantageous for many reasons. Since not every building may be able to achieve the zero-energy goal, community-scale design allows for heat sharing and shared renewable resources [5].

A key step in designing an NZE community is minimizing demand through efficient construction materials, equipment, and controls. Climate plays a crucial role in NZE residential building design strategies; however, commonalities exist across various climate zones. Whether the community is in Ithaca, New York, or Honolulu, Hawaii, high-efficiency heat pumps and efficient appliances are considered in almost every case [6,7]. Climate affects envelope materials, which balance the interactions between the interior of the home and the outdoors. To meet the ambitious goal set by the European Commission, the European Technology Platform focuses research on advanced construction material solutions such as phase change material and aerogels, along with smart” glazing in hopes of minimizing the capacity and energy use of heating and cooling equipment [8]. D’Agostino and Parker modeled a residential building in 14 different cities across Europe, capturing the different climates present throughout the continent. The study found that in cold climates, low solar gain windows and low density, high thickness insulation are preferred design strategies. Since the performance of NZE buildings is sensitive to weather conditions, climate change was accounted for in the study by adjusting cooling setpoints in the model by 2 °C—the estimated global temperature increase by 2050. This temperature control adjustment was validated by comparing the energy usage of the home in a mild climate with the adjusted setpoint to the home in a warm climate with no setpoint adjustment [9]. While climate varies across the United States, NZE building design strategies are focused on the coastal climates of California and the Northeast. California boasts 44% of the total verified NZE buildings in the United States in 2020 [10]. For the 2015 Solar Decathlon, held in Irvine, California, the UC Davis team used efficiency strategies such as under-floor radiant heating and cooling and reduced window-to-wall ratios to eliminate duct losses and significant heat loss/gain during the winter/summer [11]. In contrast, an evaluation of NZE homes in Vermont and New Jersey indicated that passive solar heating, ground-source heat pumps, a tight building envelope, and solar thermal or instantaneous water heating are common design strategies in Northeastern United States [12]. While selecting the most efficient measures is an important design consideration, it is also valuable to understand how subsystems perform and interact. A virtual testbed was developed based on the Historic Green Village NZE Community (NZEC) in Anna Maria Island, Florida that can be used to design or operate an NZEC. The preliminary model for the study included a ground-coupled water-source heat pump system with solar thermal domestic water heaters coupled with the heat pumps for heat recovery [13].

Another key consideration for NZE design is power generation and energy storage. In the case of photovoltaic (PV) generation, residential peak loads tend to occur during morning and evening hours when the PV system is not producing. Therefore, some form of battery storage is usually required to achieve NZE results using onsite generation. The opposing schedule of residential buildings and PV generation makes storage and diverse building schedules important considerations in designing an NZE residential community. A study from Ajaei et. al. analyzed four different microgrid configurations and compared reliability, cost, expandability, and NZE performance for a mix-use community. They found that keeping critical loads and each building’s solar generation operating as an independent AC microgrid optimizes the reliability and expandability of the community [14]. Gong et. al. discusses controlling rooftop solar, batteries, and an electric water heater together as a virtual power plant. This system provided flexibility for the NZE home to supply power or behave like a controllable load with respect to the utility. Further, the system mitigates the problem of high load variation utilities experience from communities with high PV penetration [15].

While these reported studies address important design concepts, they do not quantify the resiliency potential of NZE communities. A study analyzing the result of using a Renewable Energy Hybrid System (REHS) in place of traditional diesel generators for a hospital building in New York sought to determine the resiliency of the building to natural disaster events [16]. The system is sized to maximize its economic benefits, rather than its energy production. The study results indicated that by supplementing the diesel generator with a direct current (DC) solar system and storage, the outage survivability increased from 0.9 days to 3.0 days. The added REHS is validated by comparing the mitigated cost of insurance and the value of lost load to the capital cost of the system. Another study focused on critical facilities that used REopt to determine the size and performance of an REHS plus battery storage to achieve 48-hour resiliency. It was found that for high usage buildings, implementing energy efficiency measures, a PV system, combined heat and power (CHP) plant, and thermal storage resulted in a net decrease between three and 14% in total energy costs over 20 years while meeting the goal of 48-hour resiliency [17]. A Department of Energy (DOE) report lists several case studies across the U.S. focused on sizing solar and storage to meet the 48-hour resiliency standard of critical facilities such as hospitals, schools, and police stations [18]. Of the six cases, one is a residential community in Alabama focused on high efficiency and resiliency. A recent study of the optimal control for renewable resource allocation and load scheduling of resilient communities identified various control strategies of a PV plus solar mixed building-type community. The study suggested a decentralized control strategy focused on allocating power generation and scheduling building loads to survive 48 h with the objective of either minimized unserved loads or maximized thermal comfort [19].

Studies about resiliency tend to lie in the realm of critical commercial infrastructure, and NZE studies tend to focus on design strategies and cost. However, as NZE homes and communities continue to be developed, it is important to understand the resilience potential of these communities as well as to optimize the resiliency levels of NZE communities located in areas exposed to natural disasters and unreliable grids. Community resiliency and energy conservation are currently being discussed as separate issues. This study bridges the gap between resiliency and NZE as it applies to residential communities in the U.S. through design, cost, and resiliency analyses. In this study, NZE is defined as a home or community producing as much energy with renewable energy as it consumes. The objectives of this study include designing a U.S. community made up of NZE homes to determine the baseline capital and energy costs, considering a 72-hour grid outage to size PV and battery storage equipment, and perform a series of sensitivity analyses to resiliency to determine design and cost considerations. Results from this study suggest the best pathway to design a cost-effective, resilient NZE community as well as address sensitivities such as outage duration, increased load, and climate.

## 2 Methodology

For this study, BEopt, a residential building design software, is chosen to simulate individual residential buildings, determine the appropriate energy efficiency measures (EEM), and perform optimization analyses [20]. BEopt is a tool developed by the National Renewable Energy Laboratory (NREL) and is widely used to design homes throughout the U.S. BEopt includes several prescriptive options ranging from efficiency, fuel type, and material. Each option has an associated cost that is used to determine the energy-related costs for the constructions and equipment for a home.

Phoenix, Arizona, is chosen as the primary location for the analysis. Phoenix is in International Energy Conservational Code (IECC) Climate Zone 2B with approximately 1083 heating degree days (HDD) and 4154 cooling degree days (CDD) based on 18 °C (65 °F) base temperature [21]. Phoenix has adopted the IECC 2018 building code as the residential building code. This code is used as a guide to develop the baseline building energy model for the simulation analysis. To understand the impact of various key EEMs and determine the optimal systems for an energy-efficient home in Phoenix, a series of parametric analyses are conducted using the BEopt tool. The parametric analyses are iterative, meaning, after an optimal solution is found from the parametric analysis, the optimal solution is applied to the next parametric analysis.

After the optimal building systems and materials are determined, three cost optimization scenarios are carried out in BEopt to consider different numbers of bedrooms and occupants. The optimization parameters are the same across all three scenarios, and the minimum cost optimization runs are set to stop at net-zero energy. The variables in the cost optimization runs include PV size and plug load multiplier. Therefore, four optimums are selected from each of the three runs; an energy-efficient (EE) home and an NZE home with average plug loads, and an EE home and NZE home with above average plug loads. Unmet hours are analyzed for each optimum point to ensure the model meets comfort standards.

The optimum solutions from BEopt are used to create diverse community load profiles. The load profiles are loaded into the REopt Lite web tool, another NREL product, that is used to estimate the PV and battery storage sizes required to sustain critical load during a 72-hour grid outage. REopt is used to compare the cost of resiliency for an EE community with no existing PV to an established NZE community with existing PV.

Finally, three sensitivity analyses are conducted using REopt: first, an analysis to determine the cost preparedness for various outage durations. Second, an analysis of the impact of increased plug loads on survivability and the cost of preparedness for crises that would increase residential loads. Last, a comparison of the cost of resiliency in varying climate zones and with varying utility offerings.

The methodology for this study is summarized in the flowchart:

## 3 Residential Building Simulation Analysis

### 3.1 Building Model Description.

In 2016, the city of Phoenix adopted ambitious 2050 Sustainability Goals and released construction plans for a “near NZE” single-family home that, “has the best potential for wide-spread adoption in the region (climate zone 2) [22].” As shown in Fig. 1, the 246 m2 (2650 ft2) floor plan from the single-family home construction plans is used to develop an energy model for the home used in the simulation analysis.

Southwest Gas Corporation provides natural gas to homes in Phoenix, and the current effective residential tariff rates of a fixed $10.70 per month and$0.92516 per therm are applied in the economic optimization analysis [23]. The electric service provider for residential buildings in Phoenix is Arizona Public Service (APS). The Residential Service Plan chosen for this study is the Saver Choice Plus plan. This time-of-use (TOU) plan is renewable energy compatible and does not charge for net metering. The customer pays a fixed $13 per month. From May to October, the cost of energy is$0.07798 per kWh during the off-peak hours from 8 p.m. to 3 p.m. and weekends and $0.1316 per kWh during the on-peak summer hours of 3 p.m. to 8 p.m. on weekdays. From November to April, the cost of energy is$0.11017 per kWh during the on-peak winter hours from 3 p.m. to 8 p.m. on weekdays [24]. The rate schedule is summarized in Table 1.

Under this plan, the customer is eligible to enroll in a Resource Comparison Proxy (RCP) export rate of $0.1045 per exported kWh paid out at the end of each calendar year [25]. Additionally, a 26% federal tax credit is applied to the capital cost of the PV system [26]. Table 2 lists the IECC 2018 requirements considered for the baseline energy model [27]. In addition to adhering to the most recent building codes, the baseline residential building utilizes a 15 seasonal energy efficiency ratio (SEER) central air-conditioning unit, 98% annual fuel utilization efficiency (AFUE) gas furnace, standard gas water heater, and standard appliances. ### 3.2 Parametric Analysis Results. The first parametric analysis varies the mechanical heating and cooling equipment. Several systems including central air conditioning, room air conditioning, electric baseboard, air source heat pump, and mini-split heat pump options are considered. The simulation analysis results are sorted by site energy consumption (MMBtu/year) and then by Life Cycle Cost ($). Accounting for cost-effectiveness, each option is analyzed to identify the most efficient option that meets comfort requirements. The optimal heating/cooling systems include 100% efficient electric baseboards and mini-split heat pumps.

The optimal heating and cooling systems are applied to the next parametric analysis targeting wall constructions. The variables for the second parametric analysis include studs, wall sheathing, exterior finish, and interzonal walls. The analysis process is repeated to determine the most efficient, cost-effective wall constructions. The optimal wall constructions include double wood studs with R-39 fiberglass batt, no wall sheathing, light wood exterior finish, and interzonal walls with R-19 fiberglass batt.

These wall construction results are carried over to the next parametric analysis, roof constructions. The variables of the roof constructions include unfinished attic, roof material, and radiant barrier. After analyzing the efficiency and cost-effectiveness of the parametric analysis results, optimal roof constructions are determined to be unvented roofs with R-38 fiberglass batt, white metal roof material, and a double-sided, foil radiant barrier.

The next parametric analysis determines the optimal floor construction and thermal mass. The variables of the analysis include slab, carpet, exterior wall mass, partition wall mass, and ceiling mass. The optimal floor construction consists of an R-30 whole slab, no carpet, 1/2” drywall exterior wall mass, no partition wall mass, and 5/8” drywall ceiling mass.

Finally, to determine the maximum photovoltaic (PV) system that would be required on an individual house, the rooftop area, 1080 square feet, is divided by 100 square feet per kilowatt to determine the maximum allowable rooftop PV system size is 10.8-kW. PV system sizes from 5 to 10 kW with 0.5-kW increments are considered for the parametric analysis. It is found that the smallest system required for the baseline home to reach NZE is 8.5 kW. This size is used as the maximum size option in the optimization analysis. Table 3 summarizes the incremental energy savings achieved by the five home design features based on the parametric analysis results. Based on these site energy savings, the estimated optimal efficient model for the home would save approximately 44% more energy compared to the baseline model. This is determined by simply adding the energy savings of each consecutive parametric analysis, excluding PV.

### 3.3 Optimization Analysis Results.

BEopt uses guidelines from the 2014 Building America House Simulation Protocols to determine the number of occupants in a single-family residential building [28]. For this study, three cases are established to account for the variance in occupancy level that typically exists within a neighborhood. Moreover, Eq. (1) is used to determine the number of bedrooms that will be selected to simulate three scenarios: single occupant, two occupants, and three or more occupants.
$Numberofoccupants=0.59×Nbr+0.87$
(1)
where Nbr is the number of bedrooms.

Three optimization scenarios are carried out to consider different numbers of bedrooms and occupants. The optimization parameters are the same across all three scenarios. The optimal heating/cooling systems and envelop constructions established in the parametric analyses are used as definite, set options except for PV. The parameters that are varied for the optimization analysis include water heater type, mechanical ventilation, and window areas. The design options that are set for the optimization analysis are detailed in Table 4. PV is varied from 0 to 8.5 kW.

Another analysis consideration addressed in the optimization options is a scenario where the occupants are working from home, such as the case of stay-at-home orders and lockdowns due to a pandemic, for example. Two options are selected to account for a stay-at-home schedule; average plug loads using a 1.0 load multiplier used as the reference case, and a plug load 1.5 times greater than the average to account for the increased use of equipment such as laptops, monitors, and other office electronics. A sensitivity analysis is conducted using these data in the Increased Plug Load section of this paper.

For each of the three optimization runs, two optimal points are selected; the energy-efficient optimal, meaning no onsite generation, and the NZE optimal which is the first optimal point beyond 100% energy savings. These optimal points are selected for the average plug load and 50% increased plug load scenarios for analysis. Figure 2 through Fig. 4 show the results from each case broken into four color-coded categories; average plug load with no PV, average plug load with PV, 1.5 times plug load with no PV, and 1.5 times plug load with PV.

The energy-efficient optimum is the point with the most savings and lowest energy-related costs. For each energy-efficient home, the optimal parameters include majority south-facing window areas, electric hot water heating, and the 2013 ASHRAE Standard 62.2 mechanical ventilation requirements with exhaust. For all the cases, the energy-efficient optimum averages around 41% site energy savings per year and is never more cost-effective than the baseline case of meeting minimum IECC 2018 code requirements. Contributing factors to the high cost of efficiency include the electric baseboards with an incremental present value of $1,357, the mini-split heat pumps with an incremental present value of$8,639, the 30-R slab with an incremental present value of $5,036, and the double wood studs with R-39 Fiberglass batt with an incremental cost of$8303. From the point where further improvement in the building envelope or equipment has a higher marginal cost than PV, the building design is held constant and PV capacity is increased to reach the point of zero net energy. Table 5 shows the rooftop PV system sizes added to the efficient solutions to reach NZE.

The hourly indoor air temperature and relative humidity are analyzed to understand the potential unmet hours for each optimal point as shown in Fig. 5. Unmet hours are defined in this study as hours where the indoor air temperature is greater than or less than the heating or cooling setpoint by 5%. It is found that there are no unmet hours for any of the optimal solutions. The indoor relative humidity is also compared to the dry-bulb temperature using ASHRAE Standard 55 Graphic Comfort Zone Method [29]. It is found that all optimal cases are within reason of the comfort threshold considering the dry Arizona climate. It is important to note that the average indoor conditions across all cases are around 23 °C (73 °F) with a relative humidity of 35%, which is within the required comfort boundaries shown in Fig. 6. There is no attempt to control the indoor humidity resulting in approximately 3400 h of the year where the relative humidity is outside the desired range. Figure 7 illustrates the correlation between outdoor and indoor relative humidity. The indoor relative humidity tends to follow the outdoor relative humidity curve more closely during the winter months when temperature and outdoor relative humidity are low.

The annual energy consumption and energy-related costs increase as occupancy increases. Figure 8 depicts the annual site energy use broken into end uses. As occupancy increases, increasing the metabolic heat dissipated, cooling and hot water loads increase while heating loads decrease. A three or more-occupant baseline home has a 14% higher cooling load, 37% higher hot water load, and 18% lower heating load than a single occupant baseline home.

Table 6 summarizes the annual site energy use and annual energy-related costs for each home type. Similar to energy use, the energy-related costs increase as occupancy increases. For each occupancy level, the energy-related costs decrease for each home type with the lowest energy-related costs occurring at the NZE optimums. This is due to the cost-effectiveness of PV resulting from the net metering and federal tax credit incentives as discussed in more detail below.

To better understand the effects of the federal tax credit and net metering on the cost of PV, additional scenarios are investigated and applied to the single occupant house optimization case. First, the tax credit and net metering are not considered. This scenario analyzes the outcome and cost-effectiveness of PV when no incentives exist for producing excess energy through onsite generation. Next, the tax credit is applied without net metering to understand the cost of PV in an area with no net metering options. Finally, the net metering rate is applied, but no tax credit. The federal solar tax credit is set to expire in 2022 unless it is renewed. This scenario accounts for when the tax credit is eliminated for PV systems installed post-2022. The scenarios are listed in Table 7 detailing the user inputs for each scenario.

Figure 9 summarizes the optimization analysis results from all four scenarios showing only the optimal Pareto paths. Note that net metering only affects the energy-related costs after the energy savings have exceeded 100%. The tax credit plays a more significant role in decreasing the overall cost of PV, making the optimum solution with a PV system of 4.5 kW or greater more cost-effective than the baseline case. This sensitivity analysis shows that without the federal solar tax credit, the cost to build NZE home would increase, likely making it less cost-effective than a baseline home design.

## 4 Residential Community Prototypes

Phoenix household size statistics from 2017 are used to determine the number of households to assign to each occupancy tier; single resident, two residents, and three or more residents [30]. The modeled community consists of 50 homes, the size of an average block or neighborhood in Phoenix. Table 8 shows the percentage breakdown and a consequential number of homes assigned to each occupancy level that make up the community building stock.

The hourly consumption data for each home type obtained from the BEopt simulation analysis are multiplied by the associated number of homes in the community. The hourly consumption data are then added together to create a diversified community load profile for each home design including baseline, energy-efficient, and NZE. Table 9 lists the rooftop PV size and associated annual onsite generation for each occupancy level. The NZE community in total generates 424,659 kWh per year.

Table 10 summarizes the annual consumption, peak demand, and load factor for each community type. The NZE community consumption is the net total of hourly gross consumption less hourly PV generation. The NZE community load factor does not account for negative demand or hours at which the PV systems are supplying energy to the grid.

Figure 10 compares the duration curves for the baseline, energy-efficient, and NZE residential community cases. The peak demand for each case varies insignificantly; however, the duration of peak load decreases dramatically when comparing the baseline case to the energy-efficient and NZE cases. The NZE curve shows that for approximately 5800 h of the year, the load is provided by the grid and for the remaining 2960 h, the demand is met by onsite PV generation. Figure 11 illustrates the natural gas duration load curve for the baseline case, the only case that uses natural gas as a fuel source.

To better understand the impact of energy efficiency and PV on the hourly load profile, the week of August 1 through August 7 is chosen for further analysis. During this summer week, the community’s heating load is minimal providing a direct comparison of total energy use since the baseline case utilizes natural gas for heating rather than electricity. Figure 12 shows the hourly load profiles for each community type. The peak daily electricity demand for the baseline and energy-efficient cases occurs at 5:00 p.m. each day in the summertime. The peak demands for the NZE case occur at either 7:00 p.m. or 9:00 p.m. each day with peak PV production occurring around 1:00 p.m. each day.

A detailed comparison of a typical summer day (August 1) and a typical winter day (December 1) for each community is shown in Fig. 13. A secondary axis is added to the graphs to measure the baseline heating load for each sample day. The baseline electricity consumption drops below the energy-efficient energy consumption during winter months while cooling loads diminish. Meanwhile, heating loads met by electrical equipment for the efficient and NZE communities and natural gas equipment for the baseline community increase.

## 5 Resiliency Analysis

In this study, REopt Lite, a free web tool created by the National Renewable Energy Laboratory, is used to simulate the resiliency of the baseline, energy-efficient, and NZE community during a 72-hour grid outage [31]. The resilient solution for each case is set to cover 100% of the load over the 72-hour grid outage period. A three-day duration is a minimum period that the U.S. Department of Defense advises citizens to prepare for during an emergency with a long-term grid outage [32]. For the analysis, the outage is selected to start on July 15 at 5:00 p.m., the peak hour of an average load day, of each year and last for 72 h. Table 11 is a list of generation and storage cost inputs used in the REopt resiliency analysis. In a 2018 report, the cost of a 6.2-kW residential PV system was $2.70 per watt, and the cost of a commercial 200-kW system was$1.83 per watt [33]. The BEopt optimization analysis for individual homes considered $3.00 per watt for the economic analysis of PV systems. To keep consistency between BEopt and REopt economic analyses,$3000 per kW is used as a conservative PV system capital cost for the REopt resiliency analysis. Battery costs are derived from a 2019 report that projects the cost of lithium-ion battery storage systems over the next 30 years [34]. The Saver Choice Plus plan is selected as the utility rate for the community in Phoenix, Arizona. Another cost that is not taken into account in REopt is the cost of microgrid compatibility.

According to a 2018 microgrid cost study conducted by NREL, the average total cost of a community microgrid is $2.1 million per megawatt [35]. This cost includes the cost of the generation equipment, microgrid controller, additional infrastructure, and soft costs such as construction and commissioning. Since the cost of the PV system and battery storage are already accounted for in the REopt analysis, this value is considered less the cost of generation and storage equipment. Since PV and battery storage can only supply electrical loads, natural gas loads that apply to the baseline community are not addressed in the resiliency analysis. The assumption is that when there is a power outage, the natural gas supply will remain functional for the baseline community. REopt automatically identifies sizes for the PV system and battery storage based on user input parameters and desired goals. However, an iterative process is used in this study to determine the optimal battery storage size. Without constraints, REopt Lite provides battery storage solutions with durations of eight or more hours, which is unrealistic for a community-scale storage system. Instead, the recommended storage energy capacity is divided by four to estimate the appropriate battery power capacity to meet a maximum four-hour duration, the maximum duration seen in utility-scale battery storage [36]. REopt recommends a 544-kW PV system and 375-kW/1564 kWh battery storage system to sustain 100% of the load for the baseline community during an annual outage from July 15 at 5:00 p.m. to July 18 at 5:00 p.m. The entire system has a net present cost of$604,635 and could sustain the community electrical load for an average of 3251 h up to 7481 h without the grid.

For the energy-efficient community, REopt suggests a 291-kW PV system and 202-kW/820 kWh battery storage system to survive an annual 72-hour outage event. The system has a net present cost of $244,665 and an average survivability rating of 297 h, with a maximum survivability of 1750 h. The baseline and energy-efficient communities, consisting of community shared PV and battery storage systems, would require a microgrid capable feeder for grid isolation. When the NZE community is simulated in REopt using the community hourly load profile and total rooftop PV, the PV is treated as a shared resource rather than individual home resources. This skews the results to show the PV as more effective than it would be in a community where each house is served by its own rooftop PV system. To maintain the rooftop PV output modeled in BEopt, the NZE community resilience is determined by analyzing each occupancy load profile individually and then extrapolating respectively to the number of each different occupancy home in the community. The existing rooftop PV systems are applied to the gross electricity consumption for the resiliency analysis. The NZE community, consisting of homes equipped with rooftop PV and battery storage, would require each home to be microgrid compatible. Table 12 lists key results from the three occupancy levels considered for the NZE resilient community. In total, the PV system of the NZE community will increase from 256 kW to 291 kW with an added battery storage system equivalent of 191 kW/880 kWh. The total net present cost for the added systems is$196,812. The average survivability of the community is estimated to be the average of the individual average survivability ratings for each occupancy level which is 245 h. The maximum survivability of the NZE community is conservatively estimated to be the lowest maximum survivability of the three occupancy levels which is 714 h. The hourly power profiles for each occupancy level of the NZE community are combined to create a resilient, NZE community. The community hourly power profile during the simulated annual outage event is shown in Fig. 14. While the graph depicts a community-level response to the outage, it is important to note that the NZE community is designed for each home to serve itself with a rooftop PV system and home battery storage system.

The net present energy-related costs of the home construction, the cost of the PV and battery storage systems used to achieve 72-hour resiliency for each community type, and the estimated cost of microgrid compatibility are calculated and shown in Table 13. The results suggest that the most cost-effective option to design a resilient, NZE community is to build energy-efficient homes with shared solar generation as opposed to individual NZE homes with rooftop solar. In this study, the cost of installing a shared PV plus storage system to achieve 72-hour resilience level costs 31% less than purchasing individual PV and storage systems for a resilient, NZE community. The NZE community with individual PV plus storage has a net present cost of 42% less than the cost of adding PV plus storage to a community built to minimum code requirements.

Figure 15 represents the hourly power profile for the resilient, energy-efficient community during a simulated 72-hour outage event. The hourly power profiles are almost identical between the NZE community, which is modeled using rooftop solar and individual home battery storage, and the energy-efficient community which is modeled using a shared ground-mounted PV system and battery storage system. The NZE community requires an overall larger amount of battery storage to meet the 72-hour resiliency requirement than the energy-efficient community since each storage system serves individual homes rather than one storage system shared by the community.

## 6 Sensitivity Analyses

### 6.1 Outage Duration.

In this section, additional analyses are considered using the community type to better understand the main factors affecting the design of resilient communities. First, the impact of the outage duration is evaluated. Specifically, three outage durations in addition to the 72-hour outage are simulated: 24 h, 144 h (6 days), and 336 h (14 days). Table 14 lists the PV system and battery storage system sizes required to meet the respective outage durations.

An anomaly that is less intuitive is that the 24-hour outage scenario requires a larger battery storage system than the 72-hour and 144-hour outage scenarios. The community is served by the battery storage system for most of the outages as shown in Fig. 16. The storage system is nearly depleted by 5:00 a.m. when the PV system begins serving the community’s load. Since the PV system is rated at 165 kW, the majority of the energy produced serves the electric load while slowly charging the battery storage system to approximately 40% state of charge before the storage system must again serve the community during the final peak demand hours. For this reason, a higher capacity for the battery storage system is necessary to supplement the smaller PV system that does not have the capacity to charge the batteries to a full state of charge during an outage. In the case of the 72-hour and 144-hour outages, the PV system is large enough to charge a smaller battery storage system to 100% state of charge before the battery beings serving the load each day of the outage.

Figure 17 shows four net present costs versus outage duration points and the quadratic polynomial trend line, also shown as Eq. 2. The cost of a resilient community in Phoenix, Arizona, or a similar climate can easily be estimated using this empirical equation for outage durations between 24 and 336 h.
$Netpresentcost()=0.5416R2−37.129R+182,853$
(2)
where R is the resiliency goal or outage duration (hours).

It is important to note that the 24-hour resiliency solution was not an NZE solution, and there is only a 4% difference in cost between the 24-hour resiliency solution and the 144-hour resiliency solution.

REopt reports that for a case in which plug loads increase 50%, the 72-hour resiliency requirement cannot be met with the existing PV and battery storage systems. Instead, the resiliency is reduced to 6 h. A simulation is run with no constraints to determine the suggested PV system and battery storage system sizes for the work-from-home energy-efficient community case. Figure 20 compares cost and survivability indicators between the average and work-from-home scenarios. The suggested PV system is rated at 349 kW with a 244-kW/978 kWh battery storage system resulting in a 20% increase in solar and storage capacities relative to the baseline case.

### 6.3 Alternate Climates.

Phoenix, Arizona, is an ideal location for a resilient, NZE community. Therefore, it is valuable to understand the impact of climate on the cost of an NZE, resilient community. A sensitivity analysis is conducted to compare an efficient, resilient community in Phoenix to efficient, resilient communities in two other climate zones. The selected 72-hour outage for each location begins at the annual peak load of the respective community. The chosen cities have a residential building code that is less than or equal to IECC 2018 requirements [37,38]. Therefore, the IECC 2018 baseline home is used as the baseline for these analyses. The baseline home is updated with EEMs to simulate an efficient, resilient community for each city.

The first alternative community is in a suburb of Chicago, Illinois, called Naperville. Naperville is in IECC climate zone 5A with an average of 6007 HDD and 977 CDD [39]. The city of Naperville’s energy efficiency recommendations are used to adjust the EEMs to accommodate the climate in the Chicago area [40]. The efficient home is modeled in BEopt for each occupancy level with the EEMs listed in Table 15, Chicago weather data, and Naperville utility rates. The major adjustments to the efficient home include darker exterior materials with higher absorptivity, dehumidification parameters, and high R-value attic insulation as recommended by the city of Naperville.

Figure 21 details the site energy use for the efficient community in Naperville and Phoenix. The hot water load in Naperville makes up most of the community’s consumption, with cooling as the second-highest consumption. The total community site energy use in Naperville is 26% higher than the community in Phoenix. The increased site energy use is attributed solely to heating, cooling, and hot water loads.

Naperville owns the city’s electric utility, and residents pay electric bills directly to the city with a monthly fee of $15.60 and an electricity rate of$0.1090 per kWh [41]. The city of Naperville does not offer a TOU rate comparable to Phoenix. The city allows residential PV generation and battery storage and credits the customer for a monthly excess generation until the end of the year. If by the end of the year the system generated more than the home consumed, there is no financial incentive offered for a surplus of energy generation.

The 72-hour power outage is selected to start on the peak demand day and hour for the year in Naperville which is February 6th at 10:00 a.m. The Phoenix outage is selected to start on January 21 at 1:00 p.m., the peak demand occurrence for the year in Phoenix. These outage times represent worst-case scenarios as the outage starts when the community’s demand is highest for the year. Table 16 compares the system requirements and costs to survive a 72-hour power outage starting on the peak day of the year in Phoenix, Arizona, and Naperville, Illinois.

Figure 22 is a visual comparison of the hourly power profiles of Phoenix and Naperville during the respective 72-hour outages. The y-axes are identically scaled to illustrate the difference in peak demand between the two communities. While the demand is higher in Chicago, another factor that dramatically increases the net present cost is that Naperville does not pay for excess energy that is generated by a system sized to survive a 72-hour outage starting at the highest demand of the year.

The second alternative community is in Augusta, Maine. Augusta is in IECC climate zone 6A with an average of 6954 HDD and 518 CDD [42]. Residential energy efficiency recommendations from Efficiency Maine, mainly consisting of insulation guidance, are used to adjust the EEMs used for the Phoenix home to be appropriate for a home in Augusta, Maine [43]. The efficient home model is simulated in BEopt for each occupancy level using the EEMs listed in Table 17, weather data, and utility rates for Augusta, Maine. The insulating materials are the major difference between a cooling-dominated climate such as Phoenix and a heating-dominated climate like Augusta.

Figure 23 details the site energy use for the efficient community in Augusta and Phoenix. The heating load in Augusta makes up most of the community’s consumption, and the total community site energy use is 43% higher than the community in Phoenix. The community in Augusta uses only 60 kWh per year for cooling; however, the heating demand makes up almost 40% of the total community load.

The electric utility that serves Augusta is Central Maine Power. A residential TOU rate is applied to the economic analysis in BEopt. The rate schedule is shown in Table 18. Central Maine Power applies kWh credits to residential customer bills for rooftop and community shared PV systems; however, the utility does not provide dollar credits for an excess generation [44].

The 72-hour power outage is selected to start on the peak demand day and hour for the year in Augusta, which is January 19 at 10:00 a.m. Table 19 compares the system requirements and costs to survive a 72-hour power outage starting on the peak day of the year in Phoenix, Arizona, and Augusta, Maine. For an efficient community, the cost of surviving a 72-hour power outage during the annual peak demand period in Augusta, Maine, is about 14 times more than the cost in Phoenix, Arizona.

Figure 24 is a visual comparison of the hourly power profiles of Phoenix and Augusta during the respective 72-hour outages. The y-axes are identically scaled to illustrate the difference in peak demand that drives the cost of resiliency up for the community located in Augusta, Maine, compared to the community in Phoenix, Arizona.

These findings suggest that in climates that have high heating and cooling degree days, the cost of an NZE, resilient community increases. The net present costs are also affected by net metering policies, where communities connected to utilities that pay for excess generation will have a lower net present cost than communities without such programs.

## 7 Conclusion

While it is intuitive that an NZE community is resilient, the actual survivability and system optimization are rarely discussed. This study addresses the appropriate steps to designing resilient energy-efficient and NZE communities using Phoenix, Arizona, as a primary location. A baseline home is established using IECC 2018 residential code requirements. Three occupancy levels are evaluated in BEopt to provide diversity in the community’s building stock. The plug load is adjusted to simulate a work-from-home scenario. The optimization analyses for each occupancy level show that the maximum site energy savings due to EEMs and no onsite generation are approximately 41% for each home type. With the 26% federal solar tax credit and a competitive net metering rate, an NZE home has a lower net present cost than the baseline home. Without the federal tax credit, which is due for renewal in 2022, the cost of solar becomes less competitive. Each community’s 72-hour resiliency analysis results in a PV system that produces more energy annually than the energy served to the community by the grid. Thus, the 72-hour resiliency solution for each community results in an NZE community. Comparing results from the energy-efficient and NZE communities modeled in REopt, it is found that for the same resiliency, a community-scale shared PV and battery storage system has a lower net present cost than equipping each home in the community with rooftop PV and small battery storage units. Sensitivity analyses suggest that a resilient community design can be diminished by factors such as an increased load from a work-from-home order or applying the design to a different climate zone. It should also be noted that design decisions can impact system sizes and costs. For instance, the efficient community in Phoenix designed to survive a 72-hour period starting on the peak demand day of the year costs 47% more than the community in Phoenix designed to survive 72-hours over three average load days.

Notable limitations identified in this study include the inability to adjust occupancy schedules in BEopt. The only way to adjust building occupancy is to adjust the number of bedrooms in the home. This limitation affected the work-from-home sensitivity analysis as the best way to simulate increased occupancy in the home during the workweek is to increase the plug load multiplier. A more desirable approach would be to adjust the occupancy schedule if that option were available. Another limitation that affected the accuracy of the study is the inability to define a battery storage duration. REopt allows the user to adjust minimum and maximum power and energy capacity; however, this requires the user to know the optimal battery storage size prior to running the simulation. The REopt results without constraints produce unrealistic battery storage sizes with durations ranging from 7 h to 11 h. While the approach used in this study is consistent for each community resiliency analysis, the system sizes and costs have an unknown margin of error due to manually manipulating battery storage sizes to be of a reasonable duration. A final notable limitation of both BEopt and REopt is the fact that neither software provides a means for modeling electric vehicle (EV) charging.

It is recommended that future analyses compare resiliency simulations in REopt to equivalent simulations in a program like HOMER. An interesting future analysis would include finding the minimum resiliency at which an energy-efficient community reaches NZE in various climates. Additionally, EV charging should be taken into consideration for community load sizing and demand response.

## Acknowledgment

I would like to acknowledge Dr. Wangda Zuo, and Dr. Kyri Baker for their advice, recommendations, and review of this work.

## Conflict of Interest

There are no conflicts of interest.

## Data Availability Statement

The data sets generated and supporting the findings of this article are obtainable from the corresponding author upon reasonable request.

## Abbreviations

• ASHRAE =

American Society of Heating, Refrigerating and Air-Conditioning Engineers

•
• LED =

light emitting diode

•
• SHGC =

solar heat gain coefficient

## References

1.
European Commission
, “
NZEB
,”
2019
. https://ec.europa.eu/energy/.
2.
California Public Utilities Commission
, “
Zero Net Energy
,” State of California,
2020
. https://www.cpuc.ca.gov/ZNE/
3.
Peterson
,
A.
,
Gartman
,
M.
, and
Corvidae
,
J.
,
2019
,
The Economics of Zero-Energy Homes: Single-Family Insights
,
Rocky Mountain Institute
,
Basalt, CO
.
4.
Team Zero
,
2018
,
2018 Inventory of Residential Projects on the Path to Zero in the U.S. and Canada
,
Team Zero
,
Richmond, CA
.
5.
Kallushi
,
A.
,
Harris
,
J.
,
Miller
,
J.
,
Johnston
,
M.
, and
Ream
,
A.
,
2012
, “
Think Bigger: Net-Zero Communities
,”
2012 ACEEE Summer Study on Energy Efficiency in Buildings
,
Pacific Grove, CA
.
6.
U.S. Department of Energy
,
2015
,
EcoVillage: A Net Zero Energy Ready Community
,
DOE Building America
,
Washington, DC
.
7.
Ching
,
T.
,
2012
, “
Kalaeloa NZE Community
.”
8.
Magrini
,
A.
,
Lentini
,
G.
,
Cuman
,
S.
,
Bodrato
,
A.
, and
Marenco
,
L.
,
2020
, “
From Nearly Zero Energy Buildings (NZEB) to Positive Energy Buildings (PEB): The Next Challenge—The Most Recent European Trends With Some Notes on the Energy Analysis of a Forerunner PEB Example
,”
Dev. Built Environ.
,
3
, p.
100019
.
9.
D'Agostino
,
D.
, and
Parker
,
D.
,
2018
, “
A Framework for the Cost-Optimal Design of Nearly Zero Energy Buildings (NZEBs) in Representative Climates Across Europe
,”
Energy
,
149
, pp.
814
829
.
10.
Hobart
,
S.
,
2019
,
2019 Zero Energy Buildings Count Nears 600
,
New Buildings Institute
,
Portland, OR
.
11.
Alemi
,
P.
, and
Loge
,
F.
,
2017
, “
Energy Efficiency Measures in Affordable Zero net Energy Housing: A Case Study of the UC Davis 2015 Solar Decathlon Home
,”
Renewable Energy
,
101
, pp.
1242
1255
.
12.
Hoque
,
S.
,
2010
, “
Net Zero Energy Homes: An Evaluation of Two Homes in the Northeastern United States
,”
J. Green Build.
,
5
(
2
), pp.
79
90
.
13.
He
,
D.
,
Huang
,
S.
,
Zuo
,
W.
, and
Kaiser
,
R.
,
2016
, “
Towards to the Development of Virtual Testbed for Net Zero Energy Communities
,”
SimBuild 2016: Building Performance Modeling Conference
,
Salt Lake City, UT
,
Aug. 10–12
, pp.
125
132
.
14.
Ajaei
,
F. B.
, and
Jafer
,
M.
,
2017
, “
Hybrid AC/DC Microgrid Configurations for[Q13] a Net-Zero Energy Community
,”
2019 IEEE/IAS 55th Industrial and Commercial Power Systems Technical Conference (I&CPS)
,
,
May 5–8
.
15.
Gong
,
H.
,
Rallabandi
,
I.
,
Colliver
,
D.
,
Duerr
,
S.
, and
Ababei
,
C.
,
2018
, “
Net Zero Energy Houses With Dispatchable Solar PV Power Supported by Electric Hater Heater and Batter Energy Storage
,”
2018 IEEE Energy Conversion Congress and Exposition (ECCE)
,
Portland, OR
,
Sept. 23–27
.
16.
Anderson
,
K.
,
Laws
,
N. D.
,
Marr
,
S.
,
Lisell
,
L.
,
Jimenez
,
T.
,
Case
,
T.
,
Li
,
X.
,
Lohmann
,
D.
, and
Cutler
,
D.
,
2018
, “
Quantifying and Monetizing Renewable Energy Resilience
,”
Sustainability
,
10
(
4
), pp.
933
946
.
17.
Better Buildings
,
2019
,
How Distributed Energy Resources Can Improve Resilience in Public Buildings: Three Case Studies and a Step-by-Step Guide
,
U.S. Department of Energy
,
Washington, DC
.
18.
Office of Energy Efficiency & Renewable Energy
,
2019
,
Energy Efficiency and Distributed Generation for Resilience: Withstanding Grid Outages for Less
,
U.S. Department of Energy
,
Washington, DC
.
19.
Wang
,
J.
,
Garifi
,
K.
,
Baker
,
K.
,
Zuo
,
W.
,
Zhang
,
Y.
,
Huang
,
S.
, and
Vrabie
,
D.
,
2020
, “
Optimal Renewable Resource Allocation and Load Scheduling of Resilient Communities
,”
Energies
,
13
(
21
), p.
5683
.
20.
National Renewable Energy Laboratory
,
2017
,
Building Energy Optimization Tool (BEopt)
,
National Renewable Energy Laboratory
,
Golden, CO
.
21.
Western Regional Climate Center
. “
Phoenix City, Arizona: Period of Record General Climate Summary
,” 28 July 2006. https://wrcc.dri.edu
22.
Marlene Imizrain & Associates Architects
,
2016
,
Homenz: Sustainable Single Family Home
,
City of Phoenix
,
Phoenix, AZ
.
23.
Zub
,
E.
,
2020
,
Tariff Schedules Applicable to Gas Service of Southwest Gas Corporation
,
Southwest Gas Corporation
,
Las Vegas, NV
.
24.
Arizona Public Service
.
2020
, “
Residential Service Plans
,”
25.
Hobbick
,
J.
,
2020
,
Rate Rider RCP
,
Arizona Public Service
,
Phoenix, AZ
.
26.
Office of Energy Efficiency & Renewable Energy
,
2020
,
Homeowner's Guide to the Federal Tax Credit for Solar Photovoltaics
,
U.S. Department of Energy
,
Washington, DC
.
27.
International Code Council
,
2017
, “
2018 International Energy Conservation Code
.”
28.
Wilson
,
E.
,
Metzger
,
C.
,
Horowitz
,
S.
, and
Hendron
,
R.
,
2014
,
2014 Building America House Simulation Protocols
,
National Renewable Energy Laboratory
,
Golden, CO
.
29.
American Society of Heating, Refrigerating and Air-Conditioning Engineers, Inc.
, “
ASHRAE Standard 55: Thermal Environmental Conditions for Human Occupancy
,”
2010
.
30.
City-Data.com
. “
Phoenix, AZ (Arizona) Houses and Residents
,”
2017
. https://www.city-data.com
31.
National Renewable Energy Laboratory
,
2020
,
REopt Lite
,
U.S. Department of Energy
,
Washington, DC
, https://reopt.nrel.gov/tool/
32.
The President's National Infrastructure Advisory Council
,
2018
,
Surviving a Catastrophic Power Outage
,
NIAC
,
Washington, DC
.
33.
Fu
,
R.
,
Feldman
,
D.
, and
Margolis
,
R.
,
2018
,
U.S. Solar Photovoltaic System Cost Benchmark: Q1 2018
,
National Renewable Energy Laboratory
,
Golden, CO
.
34.
Cole
,
W.
, and
Frazier
,
A.
,
2019
,
Cost Projections for Utility-Scale Battery Storage
,
National Renewable Energy Laboratory
,
Golden, CO
.
35.
Giraldez
,
J.
,
Flores-Espino
,
F.
,
MacAlpine
,
S.
, and
Asmus
,
P.
,
2018
,
Phase I Microgrid Cost Study: Data Collection and Analysis of Microgrid Costs in the United States
,
NREL
,
Golden, CO
.
36.
Office of Energy Efficiency & Renewable Energy
,
2019
,
Solar-Plus-Storage 101
,
U.S. Department of Energy
,
Washington, DC
, https://www.energy.gov
37.
City of Naperville, Illinois
, “
Building Codes
,”
2020
. https://www.naperville.il.us
38.
State of Maine
,
2020
,
Building Codes
,
Department of Public Safety, Office of State Fire Marshal
,
Augusta, ME
, https://www.maine.gov
39.
Western Regional Climate Center
, “
Period of Record General Climate Summary for Chicago Univ, Illinois
,”
2012
. https://wrcc.dri.edu/cgi-bin/cliMAIN.pl?il1572
40.
City of Naperville, Illinois
, “
Residential Energy Efficiency Rebates
,”
2020
. https://www.naperville.il.us/
41.
City of Naperville, Illinois
, “
Electric Rates
,”
2020
. https://www.naperville.il.us
42.
South Turner Maine Weather
, “
Degree Days Summary
,”
2020
. http://www.southturnermaineweather.com
43.
Efficiency Maine
,
2020
,
Weatherization
,
Efficiency Maine
,
Augusta, ME
, https://www.efficiencymaine.com
44.
Maine Public Utilities Commission
,
2019
,
Net Energy Billing (NEB)
,
State of Maine
,
Hallowell, ME
, https:///www.maine.gov