Download as pdf or txt
Download as pdf or txt
You are on page 1of 11

Resources, Conservation & Recycling 205 (2024) 107571

Contents lists available at ScienceDirect

Resources, Conservation & Recycling


journal homepage: www.elsevier.com/locate/resconrec

Full length article

Dynamics of large-scale solar PV adoption feedback effects: A technical,


economic, and environmental assessment
Mingcheng Ren a, Roozbeh Ghasemi a, Masoumeh Khalkhali b, Weiwei Mo a, c, *
a
Department of Civil and Environmental Engineering, University of New Hampshire, Durham, NH, United States
b
Hazen and Sawyer, San Diego, CA, United States
c
Carsey School of Public Policy, University of New Hampshire, NH, United States

A R T I C L E I N F O A B S T R A C T

Keywords: Large residential solar photovoltaic (PV) penetration has a compound effect on the grid load reductions, PV
Solar photovoltaic system hosts’ economic savings, and the achievable environmental benefits, which is not fully understood. This study
Residential adoption combines process-based energy balance modeling, life cycle assessment, regression analysis, and stochastic de­
Process-based energy balance modeling
mand simulations to assess the technical, economic, and environmental tradeoffs under increased residential
Life cycle assessment
solar PV adoption, using Boston, MA as a testbed. It was found that increased PV adoption may lead to a steeper
Stochastic energy demand modeling
Wholesale electricity rate ramp-up in the grid during winter months, but a flattened peak load curve during summer months, emphasizing
the need for seasonal time-of-use rates and energy storage. It also reduces electricity wholesale prices, lowering
PV hosts’ economic benefits by about $15 million under 100 % adoption. The largest buildings present the
highest load reduction (top 5.5 %) and environmental benefits (top 16.6 %), but they are the least cost efficient
(top 14.5 %), requiring tradeoff balance.

1. Introduction Residential PV studies in the past have focused on understanding PV


generation potential (Gagnon et al., 2016; Hofierka and Kaňuk, 2009;
The residential solar photovoltaic (PV) system adoption has Lazzeroni et al., 2015; Robinson et al., 2013; Villavicencio Gastelu et al.,
increased significantly in the US (EIA, 2021a), primarily due to the 2018), grid load fluctuations under PV adoption (Cheng et al., 2015;
reduced cost (Branker et al., 2011), environmental benefits (Sherwani Eftekharnejad et al., 2013; Thomson and Infield, 2007; Watson et al.,
et al., 2010), and strong policy incentives (e.g., Federal Energy Regu­ 2016; Westacott and Candelise, 2016), system optimization for peak
latory Commission Order 2222) (FERC, 2020; Li and Yi, 2014). The load reduction (Alam et al., 2016; Wang et al., 2020), and mitigation of
benefit of solar PV systems was further manifested during recent PV adoption’s load effect on the grid (Aleem et al., 2020; Mukwekwe
extreme climate events (e.g., 2021 U.S. Northwest heatwave and 2021 et al., 2017; Wajahat et al., 2019). These studies highlighted the
Texas winter storm), which resulted in high local/regional electricity importance of understanding and reducing the effect of increased PV
prices (EIA, 2021b; Zamuda et al., 2019). In both events, solar PV was adoption on the grid’s load fluctuations. Some studies have focused on
recognized as an effective energy supply for increased resiliency and for understanding the economic aspect of PV adoption, via investigating the
offsetting the potential effects of excessive high prices for energy users cost saving of individual PV adopters (Agnew et al., 2019; Lee et al.,
(Brown et al., 2016; Chesser et al., 2018). Nevertheless, increased PV 2018), potential utility revenue gains and losses (Satchwell et al., 2015),
penetration can potentially alter the centralized grid’s peak demand additional costs to improve distribution power quality due to load
pattern, causing a steeper ramp-up that is more difficult to manage ramp-up caused by increased PV adoption (McHenry et al., 2016), and
(Cheng et al., 2015). Furthermore, PV hosts’ cost saving may dissipate optimized PV array design under existing wholesale electricity price
with increasing PV penetration and decreasing grid sell prices (SEIA, profiles (Brown and O’Sullivan, 2019). These studies often utilize
2012). It is hence important to adopt a systematic approach in under­ averaged historical electricity production and rate data (Agnew et al.,
standing the influence of PV penetration on its technical, economic, and 2019; Satchwell et al., 2015), omitting the dynamic interactions be­
environmental performances to inform PV planning and management tween PV adoption/usage and the wholesale electricity price. Several
decisions. other studies were specifically focused on understanding the feedback

* Corresponding author at: 35 Colovos Road, 334 Gregg Hall, Durham, NH 03824.
E-mail address: weiwei.mo@unh.edu (W. Mo).

https://doi.org/10.1016/j.resconrec.2024.107571
Received 24 October 2023; Received in revised form 12 March 2024; Accepted 19 March 2024
Available online 22 March 2024
0921-3449/© 2024 Elsevier B.V. All rights reserved.
M. Ren et al. Resources, Conservation & Recycling 205 (2024) 107571

between PV adoption and electricity rates and reported retail electricity energy demand simulations using the bottom-up method was developed.
rate change under various residential PV adoption scenarios (e.g., Cai This modeling framework was applied to residential buildings in the
et al., 2013; Chesser et al., 2018; Thakur et al., 2018). Their simulations, metro area of Boston, Massachusetts of the United States as a testbed. PV
nevertheless, are limited in terms of reliance on a pre-defined set energy adoption scenarios were evaluated based on performance metrics,
demand and supply profiles without stochasticity, and the identified including off-, mid-, on-peak load reductions from the grid, life cycle
economic impacts were not contextualized with other types of impacts, cost (LCC), cumulative energy demand, carbon emissions, and water
such as grid and environmental effects. The environmental impact of PV consumption. This study intended to answer the following research
adoption has also been assessed by previous studies, focusing on quan­ questions: 1) How will increasing PV adoption influence the grid per­
tifying and predicting environmental impacts (e.g., carbon emission) of formance, PV hosts’ energy reliance on the grid, life cycle cost, and life
various PV adoption scenarios (Antonanzas and Quinn, 2021), investi­ cycle environmental impacts of the PV systems? 2) How does building
gating the key determinants of environmental impact (Blanc et al., size influence the efficiencies in achieving regional load reduction, cost
2008), and examining the influence system configurations and geo­ saving, and environmental benefits?
spatial locations (Lamnatou et al., 2016; Nikolakakis and Fthenakis,
2013; Tsoutsos et al., 2005). These environmental assessments typically 2. Methods
adopted a life cycle assessment (LCA) approach. Still, most of these LCA
studies remain static without considering the dynamics in the supply A process-based solar energy supply and demand simulation frame­
and demand balance. Unlike those that were focused on a single aspect work that runs on a 30 min time step was first developed on a building
of solar PV impacts, some studies included multiple types of impacts (e. scale. A community-scale model was then developed where solar energy
g., technical, economic, and environmental) (Bellocchi et al., 2019; supply was estimated based on the available solar radiation and rooftop
Deltenre et al., 2020; Imam et al., 2020; Jenniches and Worrell, 2019; area, and the household energy demand was estimated using a Markov
Thoy and Go, 2021). Tradeoffs were often found across the various types chain behavioral model on a 1 min time step using American Time Use
of impacts (Arcos-Vargas et al., 2018; Deltenre et al., 2020; Edalati et al., Survey (ATUS) data following the method presented in (Muratori et al.,
2016; Hosenuzzaman et al., 2015; Imam et al., 2020; Jenniches and 2013). We then estimated the load reduction, economic cost, and
Worrell, 2019; Jurasz et al., 2020; Korsavi et al., 2018; Li et al., 2018; environmental impacts associated with various adoption scenarios using
Thoy and Go, 2021). For instance, several studies have reported trade­ a life cycle framework, whenever possible. Particularly, the influence of
offs between the economic and environmental benefits that can be solar PV adoption on wholescale electricity price was estimated through
reaped through residential PV adoptions (Korsavi et al., 2018; Edalati an empirical approach and was factored into our cost estimations. The
et al., 2016; Li et al., 2018). Yet Arcos-Vargas et al. (2018) identified entire modeling framework was illustrated through its application in a
economic and environmental co-benefits of residential PV adoption. typical community in the City of Boston. The community-level results
Previous solar PV assessments typically relied on static, averaged were then extrapolated to the City of Boston for a simplified city-scale
demand data based on historical records. However, in order to capture analysis. Fig. 1 illustrates the integration of various models and
the interactions across the consumers, power grid, renewable energy methods in the overall methodological framework of this study.
generation, and storage, dynamic simulations of electricity demand is
required (McAvoy et al., 2021; Muratori et al., 2013; Shimoda et al., 2.1. Process-based simulation of electricity supply and demand
2020). There are two ways to simulate the temporal changes of energy
demand, a top-down method and a bottom-up method (Swan and The type of PV system considered in our assessment is grid-connected
Ugursal, 2009). The top-down method scales down historical regional polycrystalline silicon (poly-Si) PV system with no battery storage
energy demand data to a housing unit based upon macroeconomic or (Sharma et al., 2015). Panel size for each individual building was
climate indicators (e.g., family gross income, unemployment conditions, assumed to be 26 % of the available rooftop area (Gagnon et al., 2016;
energy price, and ambient temperature) (Dergiades and Tsoulfidis, Melius et al., 2013). Solar energy generation was simulated based on the
2008; Swan and Ugursal, 2009). The top-down method typically simu­ 30 min solar radiation, ambient temperature, and wind speed data ob­
lates energy demand on an annual basis. Its ability to provide finer tained from NREL for the year of 2015. The specific year was selected
resolution energy demand simulations (e.g., hourly or sub-hourly) for based on data completeness. The effect of urban microclimate was not
various household categories can be highly restricted, depending on the considered in this study. The PV system degradation rate was assumed to
availability and precision of observed energy demand data. For example, be 0.5 % per year (Köntges et al., 2016). The PV module efficiency was
Thomson and Infield (2007) simulated a regional electricity demand assumed to be 15 % (NREL, 2017), and the inverter efficiency was
profile by combining measured sample household load data with the assumed to be 95 % (HOMER, 2017). These assumptions are relatively
region’s house information (e.g., perimeter footprint, house type). Their conservative compared to the latest PV technologies. A more detailed
approach relies on a large amount of historical data and lacks trans­ description of the solar electricity generation model can be found in Ren
ferability and applicability to other areas. The bottom-up method, on the et al. (2020, 2021). The generated solar energy was prioritized for
other hand, utilizes physical or statistical models to simulate individual meeting the household’s electricity demand, while the surplus was sold
household’s energy consumption and then scale up to regional or na­ to the grid. Grid sell was assumed to be unconstrainted (Mass.gov,
tional levels (Arghira et al., 2012; Muratori et al., 2013; Swan and 2020). Time-of-use rates were not considered in this study as these are
Ugursal, 2009). For instance, Muratori et al. (2013, 2018) developed a currently not in place in the intended study area. Household electricity
Markov chain behavioral model to simulate occupants’ energy con­ demand was simulated as a sum of five types of uses following Muratori
sumption activities on a 10 min time step using American Time Use (2018) and Muratori et al. (2013), which include 1) heating, ventilation,
Survey (ATUS) data. Combined with other energy consumption esti­ and air conditioning (HVAC), 2) cold appliances, 3) occupants’ activities
mations such as heating, ventilation, and air conditioning (HVAC) and (e.g., cooking, dishwashing etc.), 4) lighting, and 5) other fixed uses.
cold appliances, fine resolution energy demand was stochastically Simulation for each type of demand was introduced in detail in the
simulated for 8 types of households and scaled up to a community with following sections.
200 residential households.
To gain a comprehensive understanding of the technical, economic, 2.1.1. HVAC
and environmental tradeoffs of solar PV adoption considering the dy­ HVAC electricity demand was estimated based upon HVAC config­
namic interactions of electricity supply and demand, an integrated uration, the targeted thermal comfort level, local climate, and housing
modeling framework that combines process-based energy balance characteristics (e.g., wall and window areas) using the thermal resis­
modeling, life cycle assessment, regression analysis, as well as stochastic tance theory (American Society of Heating, 2009). The HVAC system

2
M. Ren et al. Resources, Conservation & Recycling 205 (2024) 107571

Fig. 1. The overall methodological framework of the current study.

was assumed to be air-based given its wide application (U.S. Census sleeping, no-power activity, cleaning, laundry, cooking, automatic
Bureau, 2005). It can operate either in heating or cooling mode, which is dishwashing, leisure, away (working), and away (not working). The
determined by the difference between the ambient temperature and the model was first developed in a 1 min time step in alignment with the
targeted thermal comfort level in each time step. The targeted thermal ATUS time resolution, with sleeping as the initial status. The activity
comfort level was set at 21.1 ± 1 ◦ C. It has to be noted that residential transition from one time step to the next was estimated based on minute-
heating is often provided through fuels (e.g., natural gas, fuel oil) rather by-minute transition probability matrices following Eq. (1). These
than electricity and some households may not have cooling. The per­ transition probability matrices were calculated for 5 distinct occupant
centages of buildings using electricity for heating and cooling were types: working males, non-working males, working females, non-
obtained for the intended study area. We then randomly assign each working females, and children. The 1 min transition probability
building their heating and cooling fuel types (i.e., electricity, matrices were then multiplied together to generate 30 min transition
non-electricity) based on these percentages. Each selected building was probability matrices, which are consistent with the time step of other
assumed to use one centralized HVAC system no matter how many model components and reduce the computational burden.
households reside in the building. More details about the HVAC simu­
lation can be found in the SI. wk ⋅nd,m
(1)
i,j,k
pd,m
i,j = ∑ ∑ d,m
k j k ⋅ni,j,k
w
2.1.2. Cold appliances
Where,
The average annual electricity consumption of a U.S. residential
p represents the transition probability.
utility customer was 10,649 kWh in 2019 (EIA, 2020a). It was estimated
d indicates either weekdays (d = 1) or weekends (d = 0).
that about 6 % of the total residential electricity consumption goes to
m represents minutes of a day (m = 1,2,…,1440).
refrigeration (EIA, 2020b). Therefore, one household’s annual average
i and j are the states of the occupant (activity transition from i to j)
refrigeration electricity consumption was estimated to be about 639
and j = 1, 2, …, 9 (1. Sleep, 2. No-power activity, 3. Cleaning, 4.
kWh. Based on an average nominal power rating of 725 W for cold ap­
Laundry, 5. Cooking, 6. Automatic dishwashing, 7. Leisure, 8. Away,
pliances (Muratori et al., 2013) and assuming each household possessing
working, and 9. Away, not working);
one cold appliance, the averaged operation time of the cold appliance
k represents the identification of the respondent/occupant type.
was estimated to be 881 h. To develop the energy consumption pattern
wk is the weight placed on the respondent/occupant type from the
on a 30 min time step, we assume that each cold appliance runs for three
ATUS (the weight of data relative to the total population).
random 10 min intervals every 5 h following a Bernoulli distribution
(Weisstein, 2002). The impacts of changes in external temperature and nd,m
i,j,k is the number of transitions that respondent k makes from state i

occupants opening the doors of the cold appliances were neglected to state j during minute m of day d.
(Muratori et al., 2013). More details about the cold appliance simulation At each time step, the transition probability matrix gives the prob­
can be found in the SI. ability distribution of an occupant transitioning from a known activity i
to one of the 9 activities, j. A pseudorandom number between [0,1] was
2.1.3. Occupants’ activities generated to determine which specific activity will be assigned as j. As
Occupants’ activities were simulated using the Markov chain model such, each occupant’s sequence of activities for an entire year was
based on the 2019 American Time Use Survey (ATUS) data. An occu­ simulated. It has to be noted that the first three days of the simulated
pant’s activities were assumed to be a combination of 9 activities: activities were not used to minimize the influence of the selected initial

3
M. Ren et al. Resources, Conservation & Recycling 205 (2024) 107571

status. Each individual’s activity sequence was then converted into an of fuel sources used by the grid. Particularly, the fuel prices including
electricity demand pattern using the power consumption factors listed in natural gas (NG), coal, petroleum liquids, nuclear, hydro-electric, solar,
Table 1 For calculating household-level electricity demand, we first and wind for electric power sector were collected from EIA database
assigned each household one of the eight specific type as defined by the (EIA, 2020c). The fuel revenue was assumed to be the product of the fuel
US Census (i.e., single-male, single-female, husband-wife family with price and the amount of fuel consumption. The linear regression was
one child, husband-wife family without child, single-male family with conducted using JMP® Pro 15.0.0 using forward and backward
one child, singe-female family with one child, two-male, or two-female). stop-wise selection method with Akaike Information Criterion (AIC) as
The percent distribution of the eight types of households can be obtained the stopping rule. Table 2 presents the four tested models with different
from the US Census. Occupants were then assigned to each household independent variables as inputs. The model with the highest R2 value
according to the household type. The working and non-working condi­ (Model 3) was selected for subsequent analysis. The estimated wholesale
tions of the occupant were generated based upon the local labor force rate was then added with the flat distribution (7.04 cents/kWh) and
participation rate obtained from the U.S. Bureau of Labor Statistics. The transmission (3.52 cents/kWh) rates to estimate the new selling price. In
building-level electricity demand was calculated as a sum of all house­ the meantime, we assumed that the electricity buying rate remains flat
holds’ electricity consumption within the building. More information in this scenario as power utilities typically do not adjust their retail rate
about calculations related to occupants’ activity related to electricity on an hourly or daily basis (Eversource, 2023b). Since natural gas is the
demand can be found in the SI. dominant marginal fuel use for power generation at the wholesale
market based upon the ISO-NE grid mix database observation (ISO-NE,
2.1.4. Lighting and fixed demand 2020b,2020c, 2020d), this study assumed that the increase of solar PV
Lighting electricity consumption was modeled building upon the generation will replace the use of natural gas (ISO-NE, 2020d).
occupants’ activities estimated in the previous section. We assumed that
lighting was only on when there was at least one occupant in the house
and doing activities other than sleeping. Daytime and nighttime power 2.3. Impact assessment
factors for lighting were assumed to be 125 W and 330 W, respectively.
Daytime was assumed to be between sunrise and sunset, while nighttime 2.3.1. Load reduction
was assumed to be the rest of the day. The power factor of the fixed Load reduction was calculated as the sum of the solar energy used by
demand (i.e., constant electric consumption) was assumed to be 230 W the household and sold to the grid at each time step. It was then further
(Muratori et al., 2013). summed for 3 time periods for analysis: off-peak (0:00–8:00 and
19:00–24:00), mid-peak (8:00–14:00), and on-peak (14:00–19:00)
hours following a pilot study conducted by the Liberty Utilities (Teb­
2.2. The influence of solar energy generation on electricity cost betts, 2018). The percentage of solar electricity usage out of each
building’s total electricity demand was also calculated to reflect its
The influence of solar energy generation on electricity cost was reliance on or independency from the centralized grid.
investigated in two energy pricing scenarios. Scenario 1 (NM) represents
the net metering scenario in which the electricity retail rate and solar 2.3.2. Economic cost
energy selling rate are the same flat rate. In this scenario, the amount of We selected life cycle cost (LCC) as our target economic indicator
solar energy generation will not influence its selling price, which aligns using the net present value (NPV) method as detailed in Eq. (2). All
with the utility policy in place in Boston (Eversource, 2023a). An elec­ future costs were discounted to the year 2021 using an annual discount
tricity rate of 14.9 cents/kWh was adopted in our analysis based on rate of 6 % (Freyman, 2021). The capital cost was adapted from the PV
Boston’s electricity rate, which includes the power generation cost (4.34 installed prices provided in the Lawrence Berkeley National Labo­
cents/kWh), the distribution energy charge (7.04 cents/kWh), and the ratory’s Tracking the Sun 2023 Report (LBNL, 2023) based on the
transmission charge (3.52 cents/kWh) reported from Eversource in installed panel size of each building (details in Section 2.3 of the SI). The
Boston service area (Eversource, 2021). Scenario 2 (WS) is the wholesale O&M cost includes the cost for actual grid use and the savings from sales
pricing scenario in which the potential influence of solar energy gen­ to the grid from the PV system.
eration on electricity wholesale price was considered. Electricity ∫( ) ∫( ) ]
n=L [
wholesale price is typically updated on a 5 min timestep based on the ∑ ruse,t Euse,t dt − rsell,t Esell,t dt
LCC = Cc + (2)
supply and demand conditions in the wholesale electricity market n=0
(1 + i)n
(ISO-NE, 2020a), which further influences the selling price of the solar
Where,
PV owners. To capture this dynamic interaction between solar power
LCC represents the LCC of a PV system, $.
generation and solar PV owners’ selling price, an empirical model was
Cc is the capital cost of the PV system; $.
developed based on multi-linear regression using electricity wholesale
L is the life span of the PV system, 20 years.
rate and grid mix data obtained from the ISO New England between
ruse is the electricity rate for grid use, $/kWh.
October 2015 and September 2020 (ISO-NE, 2020b, 2020c, 2020d). The
Euse is the actual grid use, kW.
regression model predicted the New England region’s monthly whole­
rsell is the electricity rate for grid sell from the PV system, $/kWh
sale rate based on the prices, usages, and/or revenues of different types
Esell is the grid sell from the PV system, kW.
i is the discount rate, 6 %
Table 1
n is the year index.
Power consumption factors in the behavioral simulation.
Activity Power consumption (W) 2.3.3. Environmental impact
Sleeping 0 We applied life cycle assessment to estimate the cumulative energy
No-power activity 0 demand (CED), water and carbon footprints during the manufacturing,
Cleaning 1250
transportation, and operation stages of each solar PV system. The use
Laundry 3825
Cooking 1225 and selling of solar electricity during the operation stage was counted as
Automatic dishwashing 1800 an impact reduction. The maintenance and end-of-life stages were not
Leisure 300 included in our analysis due to their previously reported insignificance
Away, working 0 (Spanos et al., 2015). The cumulative energy demand V 1.09 method,
Away, not working 0
the Berger et al., 2014 Water Scarcity method, and the IPCC 2013 GWP

4
M. Ren et al. Resources, Conservation & Recycling 205 (2024) 107571

Table 2
Simulated models for regression analysis.
Parameters (Param.)1 and coefficients (Coef.) Model 1 Model 2 Model 3 (selected) Model 4

Input param. Coef. Input param. Coef. Input param. Coef. Input param. Coef.

AIC 440.79 451.61 395.1 406.75


BIC 451.77 459.26 407.61 416.11
R-square Adj 0.7 0.62 0.86 0.82
Intercept – -6.34 – 91.6 – -30.7 – -9.34
Coal use, MWh x 0 x 0
Coal price, $/MWh x 0 x 0
Coal revenue, $ x 0
NG use, MWh x 5.59E-06 x 3.15E-06
NG price, $/MWh x 5.64 x 2.79
NG revenue, $ x 5.92E-07
Hydro use, MWh x 0 x 0
Hydro price, $/MWh x 0 x 0
Hydro revenue, $ x 0
Nuclear use, MWh x 0 x 0
Nuclear price, $/MWh x -4.42 x 0
Nuclear revenue, $ x 0
Oil use, MWh x 7.70E-05 x 6.14E-05
Oil price, $/MWh x 0 x 0.36
Oil revenue, $ x 1.30E-06
Refuse, MWh x 0 x 0
Refuse revenue, $ x 0
Solar use, MWh x -4.59E-05 x 0
Wind use, MWh x 1.06E-04 x 5.10E-05
Solar/Wind price, $/MWh x 0 x 0
Solar revenue, $ x 0
Wind revenue, $ x 3.03E-06

20a method embedded in SimaPro were used for estimating CED, water number, living area, number of floors, and number of household units
footprint, and carbon footprint, respectively. The SimaPro entries, unit within the building. Each building’s rooftop area was calculated by
cost and environment impact of the PV system components are provided dividing the living area by the number of floors (COB, 2019). We
in the SI. For the use phase, time-varying carbon emission factors of the assumed that 15.3 % (Mass.gov, 2018) and 79 % (EIA, 2009) of the
regional grid supply were used (Fig. 2) following (Ren et al., 2021). buildings use electricity for heating and for cooling, respectively. The
These carbon emission factors were calculated based upon the 2019 New communities’ labor force participation rates for males and females were
England utility fuel mix profile obtained from the Independent System assumed to be 69.2 % and 57.4 % respectively (BLS, 2020). Daily sunrise
Operator-New England (ISO-NE) database (ISO-NE, 2020d). The CED and sunset times were obtained from the National Oceanic and Atmo­
and water factors were estimated using the U.S. electricity grid supply spheric Administration (NOAA, 2021).
entry from SimaPro. The technical, economic, and environmental results simulated for
the 145 buildings in the selected community representing the average
3. Case study condition were extrapolated to a total of 68,692 residential buildings in
the city of Boston simply based on the 145:68,692 building number
The City of Boston was selected as our testbed due to its solar po­ ratio. It has to be noted that the Boston building dataset contains 68,698
tential and strong policy incentives (Burns and Kang, 2012; Eid et al., buildings and 6 buildings were removed from the analysis given missing
2014; Heeter et al., 2014). We tested our modeling framework on two building story information from these data entries. We sorted the
scales. First, a community of 209 households (145 buildings) located in buildings based upon their available rooftop area for PV installation. A
the City of Boston, MA was modeled, which has similar household type larger rooftop area for PV installation usually represents a higher po­
distribution as compared to the City of Boston (Table S2 of the SI). The tential for solar generation, therefore incentivizing these PV hosts may
detailed building level information was obtained from Boston’s GIS data lead to higher economic and environmental benefits (Gagnon et al.,
portal (COB, 2019), which include attributes such as street name and 2016; Ren et al., 2020). Based on an order of buildings with larger to
smaller rooftop areas, we investigated the influence of varied solar PV
adoption rate on the load, economic, and environmental performances.
We also compared our simulated demand pattern for the City of Boston
with the ISO-NE residential demand dataset and found our simulation
could effectively capture the features of the actual demand pattern in
terms of the monthly average consumption and seasonality (Figure S19
of the SI).

4. Results and discussion

Below we discuss the extrapolated results for the City of Boston. The
community-level results present a similar trend, which are detailed in
Section 3 of the SI.

4.1. Load reductions

Fig. 2. Carbon emission factors of ISO-NE grid supply over a year. Fig. 3 presents the load reductions and household energy

5
M. Ren et al. Resources, Conservation & Recycling 205 (2024) 107571

4
Off-peak Direct sol
solar use
Demand, 100 GWh
Mid-p
peak
Mid-peak
3.5

Load reduction, 100 GWh


peak
off-peak 3.80 On-peak
( )Solar energy use onsite
onsit %
3 mid-peak
peak 2.28
2.52
2.5 on-peak
peak 3.51
2.07 (63.5%))
annual
al total 9.59
2
1.48 (68.2%))
1.5
(69.5%))
0.9
0.94
1 0 79
0.79
0.79 0 77
0.77
0.77 (19.8%)
(67.4%)
4%))
0.55 (21.9%)
(21.9%))

0.5 0 29
0.29
0.29 ((22.8%)
22.8%))
0 23
0.23
0.23 0 28
0.28
0.28
09
0.09 ((22.2%)
22.2%))0.16
(5.1%))
(5.0%) (5.5%)) ((5.5%)
5.5%))
0
25% 50% 75% 100%
Adoption percentage
Fig. 3. Load reductions of four PV adoption percentages during off-, mid-, and on-peak periods. Percentages in parentheses were calculated as the amount of solar
energy used onsite divided by the PV hosts’ total electricity demand.

independence results during off-peak, mid-peak, and on-peak periods


when the top 25 %, 50 %, 75 %, and 100 % buildings with the largest
rooftop areas in the City of Boston adopt solar PV. Overall, the house­
hold, distributed rooftop solar PV generation can provide a maximum
load reduction of 374 GWh under 100 % adoption, which can meet a
maximum of about 39 % of the residential electricity demand. This
number would be only about 12 % if the commercial and industrial
electricity demands are considered (EIA, 2022). The top 25 % of the
buildings represent 31 % of the load reduction potential, while the
bottom 25 % of the buildings represent only about 18 % of the load
reduction potential. Direct solar use, or in other words, the
self-consumption of the generated solar energy, is about 1.3 times of the
selling of the excess solar energy to the grid when the top 25 % buildings
are considered. This number is about 1.7 times when all buildings are
considered. This aligns with the fact that larger buildings tend to pro­
duce more excess solar energy for sale. Out of the four studied adoption
rates, the solar hosts’ average electricity independence is the highest
under the 50 % adoption rate, as indicated by the percentages of solar
energy used onsite (calculated as the ratio of direct solar use and Fig. 4. Load reduction change rates under different adoption percentages. The
load reduction change per adoption percent change was calculated as (change
household’s total energy demand) shown in parentheses in Fig. 3. This is
of direct solar energy consumption onsite + change of grid sell from the PV
because larger buildings tend to have a larger electricity demand that is
systems)/increase of adoption percentage.
more difficult to be solely met by solar energy given the potential
mismatch in the timing of solar energy generation and demand, while
load reduction benefits, given their large solar generation potential. The
smaller buildings tend to have a smaller solar energy generation ca­
change rate remains relatively stable with a slight downward trend
pacity to meet its demand. Mid-sized buildings, on the other hand, are
between the 5.5 % and 88.3 % adoption rates. After 88.3 % adoption
shown to have the largest potential to utilize their generated solar en­
rate, the rate of load reduction change drops quickly again, indicating
ergy onsite. From a temporal perspective, mid-peak hours present the
the significant ineffectiveness of the bottom 11.7 % buildings in
highest load reduction and percent solar energy use onsite regardless of
providing load reduction benefits. When looking at the off-peak, mid-
the adoption rate. This is because the mid-peak hours typically coincide
peak, and on-peak hours separately, it is observed that the mid-peak
with the time with the highest solar radiation, and hence the largest
hours have the largest influence on the load reduction changes. Over­
solar generation potential. On-peak and off-peak hours, on the other
all, although the largest load reduction is achieved at 100 % PV adop­
hand, are usually in the late afternoon and early morning/night
tion, the top 5.5 % buildings present the highest efficiency in load
respectively when the solar radiation is more limited. For the ISOs, this
reduction.
indicates the importance of implementing energy storage and
Fig. 5 further presents the simulated monthly and hourly patterns of
time-varying energy system control to match the solar use and solar grid
residential grid use under various adoption rates. Fig. 5(a) shows that PV
feed-in time with the on-peak window to achieve the optimal on-peak
adoption presents the largest load reduction from the centralized grid
grid load reduction. Uncontrolled PVs without energy storage may not
during July and August, and the smallest load reduction during
be the optimal solution for the energy users who intend to increase
November to January. This is largely influenced by the seasonal varia­
energy security through eliminating grid reliance by PV adoption.
tions in solar generation potential. The largest benefit is seen with the
Looking at the load reduction change rates under varied adoption
adoption of the top 50 % of buildings. On a typical winter day,
percentages (Fig. 4), the load reduction change per adoption percent
increasing PV adoption could potentially lead to a steeper ramp-up
change rapidly decreases during the initial 5.5 % of PV adoption. This
curve (duck curve) and adding undesired stress on the centralized
indicates the high effectiveness of the top 5.5 % buildings in providing

6
M. Ren et al. Resources, Conservation & Recycling 205 (2024) 107571

Fig. 5. Simulated residential grid use by month (a), in a typical winter day (b), and in a typical summer day (c) under different PV adoption rates.

grid. This is a result of a combination of a large load reduction during the constantly higher solar grid sell price compared to the time varying WS
mid-peak hours and very limited load reductions during on- and off-peak rates. It also indicates that with the increasing solar energy in the grid
hours. On a typical summer day, however, increasing PV adoption can mix, electricity wholesale price is likely to drop, which reduces the
effectively flatten the peak load curve. This indicates the significance of economic saving that can be reaped by the PV hosts, although our
implementing seasonal time-of-use rate designs to achieve an optimal analysis suggests that the reduction is likely to be small. Under a 100 %
peak shave outcome for the utilities, especially during the winter adoption rate, the difference between the NM and WS scenarios amounts
months. Energy storage, peak-oriented renewable resources, and de­ to $15 million dollars. The rate of LCC change is the fastest when the PV
mand response programs may play a significant role in peak demand adoption rate is relatively small (0–14.5 %) under both rate designs. This
control during the winter months (Lazar, 2016). is mainly because of the large infrastructure cost associated with the
larger rooftop area acquired by the top buildings, as well as the large
generation to demand ratio. The lowest LCC per building is achieved at
4.2. Economic costs around 69.7 % adoption rate.

Fig. 6 presents the total operational cost savings as well as the


average LCCs of the simulated PV systems per building under different 4.3. Environmental impacts
adoption rates. The NM rate design consistently presents a slightly lower
LCC regardless of the adoption rate as compared to the WS rate design. Fig. 7 presents the average life cycle CED, water consumption, and
This is because the PV hosts using the NM design can take advantage of a carbon emission effects of the simulated PV systems per building under

Fig. 6. (a) Operational cost savings from direct onsite solar use and grid sell of the net metering (NM) and wholesale (WS) price designs under different PV adoption
percentages; (b) Life cycle cost change per adoption percent change under NM and WS price design scenarios.

7
M. Ren et al. Resources, Conservation & Recycling 205 (2024) 107571

Fig. 7. (a) Life cycle environmental costs per PV-adopted building under different adoption percentages; (b) life cycle environmental cost change per adoption
percent change.

different adoption rates. Results show all simulated PV systems present carbon emission curve, which aligns with the load reduction results. On
life cycle environmental benefits, as indicated by the negative values in a typical summer day, increasing PV adoption is likely to shift the carbon
Fig. 7(a). The life cycle environmental benefit per building decreases emission peak time as well as to shorten the peak duration. While carbon
quickly within the first 16.6 % of adoption (Fig. 7(b)). After that, the management is typically less sensitive to daily or monthly patterns, our
environmental benefit stabilizes with a slight decreasing trend. This results confirm the varying seasonal and hourly benefits that might be
indicates that the top 16.6 % of buildings can provide the highest provided by solar PV installations, which informs more precise future
environmental benefit compared with the remaining buildings. carbon management strategies.
Fig. 8 further presents the simulated daily, monthly, and hourly
patterns of the carbon footprint of residential buildings under various
4.4. Load reduction, economic, and environmental tradeoffs
adoption rates. Fig. 8(a) and (b) show that the highest daily carbon
footprint is observed around July, while the monthly trend shows two
Our analysis shows strategies for effective load and environmental
prominent peaks, one during July and August, and the other during
impact reductions may be at odds with strategies for effective cost
January and February. This is resulted from a combination of high
reduction. While the load reduction and the environmental benefits per
electricity demands during both peaks, as well as a “dirtier” grid mix
building are the highest in the buildings with the largest rooftop areas,
during summer months, as reflected in the unit carbon emission factors
their life cycle economic costs are also the highest. This aligns with
shown in Fig. 1. A high solar PV adoption can significantly decrease the
findings from previous studies (Korsavi et al., 2018; Li et al., 2018).
carbon footprint, especially for the summer months; however, the
When examining the changes in these three indicators concerning
decrease is the most effectively seen in the top 50 % buildings. On a
adoption, the adoption of the top 5.5 % buildings presents the fastest
typical winter day, increasing PV adoption is likely to steepen the hourly
increase in load reduction while the adoption of the top 16.6 % buildings

Fig. 8. (a) Daily carbon emission and (b) monthly carbon emission of residential grid use over a year under different PV adoption percentages; carbon emission of
grid use (c) in a typical winter day and (d) a typical summer day under different PV adoptions.

8
M. Ren et al. Resources, Conservation & Recycling 205 (2024) 107571

presents the fastest increase in environmental impact reduction. least life cycle cost efficiency based on per adoption percent change. It is
Nevertheless, the adoptions in the top 14.5 % of buildings also represent hence crucial to explore potential incentives and financial models that
the fastest decrease in life cycle cost. These buildings are primarily alleviate the potential economic barriers for buildings with large rooftop
multi-unit apartment buildings. It has been reported that solar PV areas or to develop strategies that enhance the load reduction and
adoption in such buildings could potentially be hindered by the financial environmental efficiencies of PV systems implemented in buildings with
hurdle imposed by the large capital cost and the complexity in the more limited rooftop space. It is also important to identify strategies for
governance and/or the competition of the rooftop area (Roberts et al., effectively managing the steeper ramp-up as a result of an increased
2019). To balance this tradeoff, it is important to investigate potential solar PV adoption, especially during the winter months. Future studies
incentives or financial models to reduce the economic barrier for may further include energy storage systems and related management
buildings with large rooftop areas to adopt solar PV systems, or to strategies to allow for a more comprehensive understanding of different
develop strategies that increase the load reduction and environmental types of PV systems and the influence of their large penetrations on load
efficiencies of PV systems installed in buildings with smaller rooftop reduction, economic, and environmental aspects.
areas. Another tradeoff observed was that while the 100 % PV adoption
provides the largest overall load reduction and environmental benefits, CRediT authorship contribution statement
it presents sub-optimal energy independence from the centralized grid.
Mid-sized buildings were found to be the most effective in utilizing their Mingcheng Ren: Writing – review & editing, Writing – original
generated solar energy for meeting their own energy demand. As energy draft, Visualization, Validation, Methodology, Investigation, Formal
independence could be associated with system resilience, strategies to analysis, Data curation. Roozbeh Ghasemi: Writing – review & editing,
increase energy independence, such as demand response management, Investigation, Formal analysis. Masoumeh Khalkhali: Formal analysis.
innovative prosumer pricing policies, as well as energy trading and Weiwei Mo: Writing – review & editing, Visualization, Supervision,
sharing strategies are desirable for achieving enhanced energy resilience Resources, Project administration, Methodology, Investigation, Funding
(Boiarkin et al., 2023). PV adoption in Boston is also likely to cause a acquisition, Conceptualization.
steeper ramp-up in the centralized grid system during winter months;
however, they are likely to flatten the curve during summer months. Declaration of competing interest
Given this, it is important to identify strategies to counter the effect of
steeper ramp up as a result of an increased solar PV adoption during the The authors declare that they have no known competing financial
winter months. Such strategies may include utilizing peak-oriented interests or personal relationships that could have appeared to influence
renewable energy resources (e.g., impoundment hydropower), the work reported in this paper.
enhancing energy storage capacities (e.g., battery systems, pumped
storage, water and wastewater pumping loads), implementing demand Data availability
response programs, among others (Lazar, 2016).
Data will be made available on request.
5. Conclusions

This study presents a modeling framework that combines process- Acknowledgements


based energy balance modeling, LCA, regression analysis, as well as
stochastic energy demand simulations to investigate various technical, We acknowledge the National Science Foundation’s support via a
economic, and environmental tradeoffs under increased residential solar CBET award (#1706143). The views, findings, and conclusions
PV adoption. This modeling framework was applied to residential expressed in this study are those of the authors and do not necessarily
buildings in the metro area of Boston, Massachusetts. We found that the reflect the views of the National Science Foundation.
household, distributed rooftop solar PV generation can provide a
maximum load reduction of 374 GWh under 100 % adoption rate, which
Supplementary materials
is about 39 % of the total residential electricity demand. While the top
5.5 % of buildings with the largest rooftop areas provide the highest load
Supplementary material associated with this article can be found, in
reduction potential, the solar hosts’ average electricity independence is
the online version, at doi:10.1016/j.resconrec.2024.107571.
the highest under the 50 % adoption rate. From a temporal perspective,
mid-peak hours present the highest load reduction and percent solar
References
energy use onsite regardless of the adoption rate. On a typical winter
day, increasing PV adoption could potentially lead to a steeper ramp-up Agnew, S., Smith, C., Dargusch, P., 2019. Understanding transformational complexity in
curve in the centralized grid, while on a typical summer day, increasing centralized electricity supply systems: modelling residential solar and battery
PV adoption may effectively flatten the peak load curve. This indicates adoption dynamics. Renew. Sustain. Energy Rev. 116, 109437 https://doi.org/
10.1016/j.rser.2019.109437.
the significance of implementing seasonal time-of-use rate designs as Alam, M.J.E., Muttaqi, K.M., Sutanto, D., 2016. Effective utilization of available PEV
well as energy storage systems. Our study also confirmed that an battery capacity for mitigation of solar PV impact and grid support with integrated
increased solar adoption is likely to lead to a decreased electricity V2G functionality. IEEE Trans. Smart Grid 7, 1562–1571. https://doi.org/10.1109/
TSG.2015.2487514.
wholesale price, which reduces the economic saving that can be reaped Aleem, S.A., Suhail Hussain, S.M., Ustun, T.S., 2020. A review of strategies to increase PV
by the PV hosts, although our analysis suggests that the reduction is penetration level in smart grids. Energies 13. https://doi.org/10.3390/en13030636.
likely to be small (about $15 million overall under 100 % adoption). American Society of Heating, 2009. American Society of Heating, Refrigerating and Air-
Conditioning Engineers, Inc., 2009 ASHRAE handbook-fundamentals, I-P edition
From an environmental perspective, we found that the top 16.6 % of [WWW document].
buildings can provide the highest environmental benefit compared with Antonanzas, J., Quinn, J.C., 2021. Net environmental impact of the PV industry from
the remaining buildings. Furthermore, solar PV adoption is likely to 2000-2025. J. Clean. Prod. 311, 127791 https://doi.org/10.1016/j.
jclepro.2021.127791.
impact the daily carbon emission curves differently in summer and
Arcos-Vargas, A., Cansino, J.M., Román-Collado, R., 2018. Economic and environmental
winter months. This finding may benefit more precise carbon manage­ analysis of a residential PV system: a profitable contribution to the Paris agreement.
ment. Overall, it was found that strategies aimed at achieving substan­ Renew. Sustain. Energy Rev. 94, 1024–1035. https://doi.org/10.1016/j.
tial load and environmental impact reductions may conflict with those rser.2018.06.023.
Arghira, N., Hawarah, L., Ploix, S., Jacomino, M., 2012. Prediction of appliances energy
focused on cost-effectiveness. The largest buildings that present the most use in smart homes. Energy 48, 128–134. https://doi.org/10.1016/j.
significant load reduction and environmental benefit also present the energy.2012.04.010.

9
M. Ren et al. Resources, Conservation & Recycling 205 (2024) 107571

Bellocchi, S., Manno, M., Noussan, M., Vellini, M., 2019. Impact of grid-scale electricity Imam, A.A., Al-Turki, Y.A., Sreerama Kumar, R., 2020. Techno-economic feasibility
storage and electric vehicles on renewable energy penetration: a case study for Italy. assessment of grid-connected PV systems for residential buildings in Saudi Arabia-A
Energies 12. https://doi.org/10.3390/en12071303. case study. Sustainability 12. https://doi.org/10.3390/su12010262.
Blanc, I., Belloin-Saint-Pierre, D., Payet, J., Jacquin, P., Adra, N., Mayer, D., 2008. ISO-NE, 2020a. ISO-New England, Wholesale load cost estimator [WWW document].
Espace-PV: key sensitive parameters for environmental impacts of grid-connected PV ISO-NE, 2020b. ISO-New England, Markets and operations, iso express, energy, load, and
systems with LCA. In: 23rd Eur. Photovolt. Sol. Energy Conf. Exhib., p. 3779. demand reports [WWW document].
Boiarkin, V., Rajarajan, M., Al-Zaili, J., Asif, W., 2023. A novel dynamic pricing model ISO-NE, 2020c. ISO-New England, Operations reports, daily generation by fuel type
for a microgrid of prosumers with photovoltaic systems. Appl. Energy 342, 121148. [WWW document].
https://doi.org/10.1016/J.APENERGY.2023.121148. ISO-NE, 2020d ISO-New England, Operations reports, dispatch fuel mix [WWW
BLS, 2020. U.S. Bureau of Labor Statistics, Employment Projections, Civilian labor force document].
participation rate by age, sex, race, and ethnicity. https://www.bls.gov/emp/tables Jenniches, S., Worrell, E., 2019. Regional economic and environmental impacts of
/civilian-labor-force-participation-rate.htm (accessed 3/1/2024). renewable energy developments: solar PV in the Aachen Region. Energy Sustain.
Branker, K., Pathak, M.J.M.M., Pearce, J.M., 2011. A review of solar photovoltaic Dev. 48, 11–24. https://doi.org/10.1016/j.esd.2018.10.004.
levelized cost of electricity. Renew. Sustain. Energy Rev. 15, 4470–4482. https:// Jurasz, J., Ceran, B., Orłowska, A., 2020. Component degradation in small-scale off-grid
doi.org/10.1016/j.rser.2011.07.104. PV-battery systems operation in terms of reliability, environmental impact and
Brown, M.A., Johnson, E., Matisoff, D., Staver, B., Beppler, R., 2016. Impacts of solar economic performance. Sustain. Energy Technol. Assess. 38, 100647 https://doi.
power on electricity rates and bills. ACEEE Summer Study Energy Effic. Build. 6, org/10.1016/j.seta.2020.100647.
1–13. Korsavi, S.S., Zomorodian, Z.S., Tahsildoost, M., 2018. Energy and economic
Brown, P.R., O’Sullivan, F.M., 2019. Shaping photovoltaic array output to align with performance of rooftop PV panels in the hot and dry climate of Iran. J. Clean. Prod.
changing wholesale electricity price profiles. Appl. Energy 256, 113734. https://doi. 174, 1204–1214. https://doi.org/10.1016/j.jclepro.2017.11.026.
org/10.1016/j.apenergy.2019.113734. Köntges, M., Altmann, S., Heimberg, T., Jahn, U., Berger, K.A., 2016. Mean degradation
Burns, J.E., Kang, J.-S.S., 2012. Comparative economic analysis of supporting policies for rates in PV systems for various kinds of PV module failures. Proceedings of the 32nd
residential solar PV in the United States: solar renewable energy credit (SREC) European Photovoltaic Solar Energy Conference and Exhibition. Munich,, Germany.
potential. Energy Policy 44, 217–225. https://doi.org/10.1016/j. Lamnatou, C., Baig, H., Chemisana, D., Mallick, T.K., 2016. Environmental assessment of
enpol.2012.01.045. a building-integrated linear dielectric-based concentrating photovoltaic according to
Cai, D.W.H., Adlakha, S., Low, S.H., De Martini, P., Mani Chandy, K., 2013. Impact of multiple life-cycle indicators. J. Clean. Prod. 131, 773–784. https://doi.org/
residential PV adoption on retail electricity rates. Energy Policy 62, 830–843. 10.1016/j.jclepro.2016.04.094.
https://doi.org/10.1016/j.enpol.2013.07.009. Lazar, J., 2016. Teaching the “Duck” to fly - second edition - regulatory assistance project
Cheng, D., Mather, B., Seguin, R., Hambrick, J., Broadwater, R.P., 2015. PV impact [WWW document]. URL https://www.raponline.org/knowledge-center/teaching-th
assessment for very high penetration levels. In: 2015 IEEE 42nd Photovolt. Spec. e-duck-to-fly-second-edition/(accessed 2.12.24).
Conf. PVSC 2015, 6, pp. 295–300. https://doi.org/10.1109/PVSC.2015.7356170. Lazzeroni, P., Olivero, S., Stirano, F., Repetto, M., 2015. Impact of PV penetration in a
Chesser, M., Hanly, J., Cassells, D., Apergis, N., 2018. The positive feedback cycle in the distribution grid: a Middle-East study case. In: 2015 IEEE 1st Int. Forum Res.
electricity market: residential solar PV adoption, electricity demand and prices. Technol. Soc. Ind. RTSI 2015 - Proc, pp. 353–358. https://doi.org/10.1109/
Energy Policy 122, 36–44. https://doi.org/10.1016/j.enpol.2018.07.032. RTSI.2015.7325123.
COB, 2019. Boston maps open data geospatial datasets [WWW document]. LBNL, L.B.N.L., 2023. Tracking the sun | Energy markets & policy [WWW document].
Deltenre, Q., De Troyer, T., Runacres, M.C., 2020. Performance assessment of hybrid PV- URL https://emp.lbl.gov/tracking-the-sun (accessed 2.12.24).
wind systems on high-rise rooftops in the Brussels-Capital Region. Energy Build. 224, Lee, M., Hong, T., Jeong, K., Kim, J., 2018. A bottom-up approach for estimating the
110137 https://doi.org/10.1016/j.enbuild.2020.110137. economic potential of the rooftop solar photovoltaic system considering the spatial
Dergiades, T., Tsoulfidis, L., 2008. Estimating residential demand for electricity in the and temporal diversity. Appl. Energy 232, 640–656. https://doi.org/10.1016/j.
United States, 1965-2006. Energy Econ. 30, 2722–2730. https://doi.org/10.1016/j. apenergy.2018.09.176.
eneco.2008.05.005. Li, C., Zhou, D., Zheng, Y., 2018. Techno-economic comparative study of grid-connected
Edalati, S., Ameri, M., Iranmanesh, M., Tarmahi, H., Gholampour, M., 2016. Technical PV power systems in five climate zones, China. Energy 165, 1352–1369. https://doi.
and economic assessments of grid-connected photovoltaic power plants: iran case org/10.1016/j.energy.2018.10.062.
study. Energy 114, 923–934. https://doi.org/10.1016/j.energy.2016.08.041. Li, H., Yi, H., 2014. Multilevel governance and deployment of solar PV panels in U.S.
Eftekharnejad, S., Vittal, V., Heydt, G.T., Keel, B., Loehr, J., 2013. Impact of increased cities. Energy Policy 69, 19–27. https://doi.org/10.1016/j.enpol.2014.03.006.
penetration of photovoltaic generation on power systems. IEEE Trans. Power Syst. Mass.gov, 2020. Commonwealth of Massachusetts, Net metering guide [WWW
28, 893–901. https://doi.org/10.1109/TPWRS.2012.2216294. document].
EIA, 2022. EIA-Massachusetts state energy profile [WWW document]. Mass.gov, 2018. Mass.gov, How Massachusetts households heat their homes [WWW
EIA, 2021a. U.S. Energy Information Administration, Today in Energy, U.S. solar document].
photovoltaic module shipments up 33% in 2020 [WWW document]. McAvoy, S., Grant, T., Smith, C., Bontinck, P., 2021. Combining life cycle assessment and
EIA, 2021b. U.S. Energy Information Administration, Electricity explained, electricity system dynamics to improve impact assessment: a systematic review. J. Clean. Prod.
generation, capacity, and sales in the United States [WWW document]. 315, 128060 https://doi.org/10.1016/j.jclepro.2021.128060.
EIA, 2020a. EIA, U.S. Energy Information Administration, How much electricity does an McHenry, M.P., Johnson, J., Hightower, M., 2016. Why do electricity policy and
American home use? [WWW document]. competitive markets fail to use advanced PV systems to improve distribution power
EIA, 2020b. EIA, U.S. Energy Information Administration, How is electricity used in U.S. quality? J. Sol. Energy 2016, 1–10. https://doi.org/10.1155/2016/5187317.
homes? [WWW document]. Melius, J., Margolis, R., Ong, S., 2013. Estimating rooftop suitability for PV : a review of
EIA, 2020c. U.S. Energy Information Administration, Natural gas, natural gas monthly methods, patents, and validation techniques. NREL Tech. Rep. 35.
[WWW document]. Mukwekwe, L., Venugopal, C., Davidson, I.E., 2017. A review of the impacts and
EIA, 2009. Household energy use in Massachusetts [WWW document]. mitigation strategies of high PV penetration in low voltage networks. In: Proc. - 2017
Eid, C., Guillén, J.R., Marín, P.F., Hakvoort, R., 2014. The economic effect of electricity IEEE PES-IAS PowerAfrica Conf. Harnessing Energy, Inf. Commun. Technol. Afford.
net-metering with solar PV: consequences for network cost recovery, cross subsidies Electrif. Africa, PowerAfrica 2017, pp. 274–279. https://doi.org/10.1109/
and policy objectives. Energy Policy 75, 244–254. PowerAfrica.2017.7991236.
Eversource, 2023a. EVERSOURCE - Massachusetts net metering [WWW document]. Muratori, M., 2018. Impact of uncoordinated plug-in electric vehicle charging on
Eversource, 2023b. Eversource - electric supply rates [WWW document]. residential power demand. Nat. Energy 3, 193–201.
Eversource, 2021. 2021 Summary of Eastern Massachusetts electric rates for Cambridge Muratori, M., Roberts, M.C., Sioshansi, R., Marano, V., Rizzoni, G., 2013. A highly
service area [WWW document]. resolved modeling technique to simulate residential power demand. Appl. Energy
FERC, 2020. Federal Energy Regulatory Commission, News media contact : Craig Cano, 107, 465–473. https://doi.org/10.1016/j.apenergy.2013.02.057.
mediadl@ferc.gov, Docket No . RM18-9-000 FERC Order No . 2222 : a new day for NREL, 2017. National Renewable Energy Laboratory, Sustainable Energy, PVWatts®
distributed energy resources. Calculator. https://pvwatts.nrel.gov/index.php (accessed 3/1/2024).
Freyman, T., 2021. How do you value a renewable energy project? [WWW document]. Nikolakakis, T., Fthenakis, V., 2013. Modeling the environmental impact of PV and wind
Gagnon, P., Margolis, R., Melius, J., Phillips, C., Elmore, R., 2016. Rooftop Solar large scale penetration in regional grids. In: Conf. Rec. IEEE Photovolt. Spec. Conf.,
Photovoltaic Technical Potential in the United States. A Detailed Assessment. pp. 2326–2330. https://doi.org/10.1109/PVSC.2013.6744942.
National Renewable Energy Lab.(NREL), Golden, CO (United States). NOAA, 2021. National Oceanic and Atmospheric Administration, Global Monitoring
HOMER, 2017. HOMER Energy, HOMER Grid 1.1. https://www.homerenergy.com/prod Laboratory, Sunrise table for 2021, Sunset table for 2021 [WWW document].
ucts/grid/docs/1.1/index.html (accessed 3/1/2024). Ren, M., Mitchell, C.R., Mo, W., 2021. Managing residential solar photovoltaic-battery
Heeter, J., Gelman, R., Bird, L., 2014. Status of Net Metering: Assessing the Potential to systems for grid and life cycle economic and environmental co-benefits under time-
Reach Program Caps. National Renewable Energy Lab.(NREL), Golden, CO (United of-use rate design. Resour. Conserv. Recycl. 169, 105527 https://doi.org/10.1016/j.
States). resconrec.2021.105527.
Hofierka, J., Kaňuk, J., 2009. Assessment of photovoltaic potential in urban areas using Ren, M., Mitchell, C.R., Mo, W., 2020. Dynamic life cycle economic and environmental
open-source solar radiation tools. Renew. Energy 34, 2206–2214. https://doi.org/ assessment of residential solar photovoltaic systems. Sci. Total Environ. 722 https://
10.1016/j.renene.2009.02.021. doi.org/10.1016/j.scitotenv.2020.137932.
Hosenuzzaman, M., Rahim, N.A., Selvaraj, J., Hasanuzzaman, M., Malek, A.B.M.A., Roberts, M.B., Bruce, A., MacGill, I., 2019. Opportunities and barriers for photovoltaics
Nahar, A., 2015. Global prospects, progress, policies, and environmental impact of on multi-unit residential buildings: reviewing the Australian experience. Renew.
solar photovoltaic power generation. Renew. Sustain. Energy Rev. 41, 284–297. Sustain. Energy Rev. 104, 95–110. https://doi.org/10.1016/J.RSER.2018.12.013.
https://doi.org/10.1016/j.rser.2014.08.046.

10
M. Ren et al. Resources, Conservation & Recycling 205 (2024) 107571

Robinson, S.A., Stringer, M., Rai, V., Tondon, A., 2013. GIS-integrated agent-based model Thoy, E.J.W., Go, Y.I., 2021. Enhancement and validation of building integrated PV
of residential solar PV diffusion. In: 32nd USAEE/IAEE North Am. Conf.. July 28-31, system: 3D modelling, techno-economics and environmental assessment. Energy
2013 1–19. Built Environ. https://doi.org/10.1016/j.enbenv.2021.05.001.
Satchwell, A., Mills, A., Barbose, G., 2015. Quantifying the financial impacts of net- Tsoutsos, T., Frantzeskaki, N., Gekas, V., 2005. Environmental impacts from the solar
metered PV on utilities and ratepayers. Energy Policy 80, 133–144. https://doi.org/ energy technologies. Energy Policy 33, 289–296. https://doi.org/10.1016/S0301-
10.1016/j.enpol.2015.01.043. 4215(03)00241-6.
SEIA, 2012. Solar Energy Industries Association, Potential impact of solar PV on U.S. Census Bureau, 2005. U.S. Census Bureau, American housing survey for the United
electricity markets in Texas [WWW document]. States: 2005, H 150/05 edition [WWW document].
Sharma, S., Jain, K.K., Sharma, A., 2015. Solar cells: in research and applications—a Villavicencio Gastelu, J., Melo Trujillo, J.D., Padilha-Feltrin, A., 2018. Hierarchical
review. Mater. Sci. Appl. 06, 1145–1155. https://doi.org/10.4236/ Bayesian model for estimating spatial-temporal photovoltaic potential in residential
msa.2015.612113. areas. IEEE Trans. Sustain. Energy 9, 971–979. https://doi.org/10.1109/
Sherwani, A.F., Usmani, J.A., Varun, 2010. Life cycle assessment of solar PV based TSTE.2017.2768824.
electricity generation systems: a review. Renew. Sustain. Energy Rev. 14, 540–544. Wajahat, M., Khalid, H.A., Bhutto, G.M., Leth Bak, C., 2019. A comparative study into
https://doi.org/10.1016/j.rser.2009.08.003. enhancing the pv penetration limit of a lv cigre residential network with distributed
Shimoda, Y., Yamaguchi, Y., Iwafune, Y., Hidaka, K., Meier, A., Yagita, Y., grid-tied single-phase pv systems. Energies 12, 2964.
Kawamoto, H., Nishikiori, S., 2020. Energy demand science for a decarbonized Wang, H., Wang, J., Piao, Z., Meng, X., Sun, C., Yuan, G., Zhu, S., 2020. The optimal
society in the context of the residential sector. Renew. Sustain. Energy Rev. 132, allocation and operation of an energy storage system with high penetration grid-
110051 https://doi.org/10.1016/j.rser.2020.110051. connected photovoltaic systems. Sustainability 12. https://doi.org/10.3390/
Spanos, C., Turney, D.E., Fthenakis, V., 2015. Life-cycle analysis of flow-assisted nickel su12156154.
zinc-, manganese dioxide-, and valve-regulated lead-acid batteries designed for Watson, J.D., Watson, N.R., Santos-Martin, D., Wood, A.R., Lemon, S., Miller, A.J.V.,
demand-charge reduction. Renew. Sustain. Energy Rev. 43, 478–494. 2016. Impact of solar photovoltaics on the low-voltage distribution network in New
Swan, L.G., Ugursal, V.I., 2009. Modeling of end-use energy consumption in the Zealand. IET Gener. Transm. Distrib. 10, 1–9. https://doi.org/10.1049/iet-
residential sector: a review of modeling techniques. Renew. Sustain. Energy Rev. 13, gtd.2014.1076.
1819–1835. https://doi.org/10.1016/j.rser.2008.09.033. Weisstein, E.W., 2002. Bernoulli distribution.
Tebbetts, H., 2018. Liberty utilities (Granite State Electric) Corp. TOU rate calculation. Westacott, P., Candelise, C., 2016. Assessing the impacts of photovoltaic penetration
Thakur, J., Rauner, S., Darghouth, N.R., Chakraborty, B., 2018. Exploring the impact of across an entire low-voltage distribution network containing 1.5 million customers.
increased solar deployment levels on residential electricity bills in India. Renew. IET Renew. Power Gener. 10, 460–466. https://doi.org/10.1049/iet-rpg.2015.0535.
Energy 120, 512–523. https://doi.org/10.1016/J.RENENE.2017.12.101. Zamuda, C.D., Larsen, P.H., Collins, M.T., Bieler, S., Schellenberg, J., Hees, S., 2019.
Thomson, M., Infield, D.G., 2007. Impact of widespread photovoltaics generation on Monetization methods for evaluating investments in electricity system resilience to
distribution systems. IET Renew. Power Gener. 1, 33–40. https://doi.org/10.1049/ extreme weather and climate change. Electr. J. 32, 106641 https://doi.org/
iet-rpg:20060009. 10.1016/j.tej.2019.106641.

11

You might also like