U.S. Solar Photovoltaic System and Energy Storage Cost Benchmark: Q1 2020

U.S.
Solar Photovoltaic System and

Energy Storage Cost Benchmark:
Q1 2020
David Feldman, Vignesh Ramasamy, Ran Fu,
Ashwin Ramdas, Jal Desai, and Robert Margolis
National Renewable Energy Laboratory
NREL is a national laboratory of the U.S. Department of Energy Technical Report

Office of Energy Efficiency & Renewable Energy NREL/TP-6A20-77324
Operated by the Alliance for Sustainable Energy, LLC January 2021
This report is available at no cost from the National Renewable Energy
Laboratory (NREL) at www.nrel.gov/publications.
Contract No. DE-AC36-08GO28308

U.S. Solar Photovoltaic System and
Energy Storage Cost Benchmark:
Q1 2020
David Feldman, Vignesh Ramasamy, Ran Fu,
Ashwin Ramdas, Jal Desai, and Robert Margolis
National Renewable Energy Laboratory
Suggested Citation
Feldman, David, Vignesh Ramasamy, Ran Fu, Ashwin Ramdas, Jal Desai, and Robert
Margolis. 2021. U.S. Solar Photovoltaic System Cost Benchmark: Q1 2020. Golden, CO:
National Renewable Energy Laboratory. NREL/TP-6A20-77324.
https://www.nrel.gov/docs/fy21osti/77324.pdf.
NREL is a national laboratory of the U.S. Department of Energy Technical Report

Office of Energy Efficiency & Renewable Energy NREL/TP-6A20-77324
Operated by the Alliance for Sustainable Energy, LLC January 2021
This report is available at no cost from the National Renewable Energy National Renewable Energy Laboratory
Laboratory (NREL) at www.nrel.gov/publications. 15013 Denver West Parkway
Golden, CO 80401
Contract No. DE-AC36-08GO28308 303-275-3000 • www.nrel.gov
NOTICE
This work was authored by the National Renewable Energy Laboratory, operated by Alliance for Sustainable
Energy, LLC, for the U.S. Department of Energy (DOE) under Contract No. DE-AC36-08GO28308. Funding
provided by the U.S. Department of Energy Office of Energy Efficiency and Renewable Energy Solar Energy
Technologies Office. The views expressed herein do not necessarily represent the views of the DOE or the U.S.
Government.
This report is available at no cost from the National Renewable

Energy Laboratory (NREL) at www.nrel.gov/publications.
U.S. Department of Energy (DOE) reports produced after 1991

and a growing number of pre-1991 documents are available
free via www.OSTI.gov.
Cover Photos by Dennis Schroeder: (clockwise, left to right) NREL 51934, NREL 45897, NREL 42160, NREL 45891, NREL 48097,
NREL 46526.
NREL prints on paper that contains recycled content.

List of Acronyms
AC alternating current
ASP average selling price
BNEF Bloomberg New Energy Finance
BOS balance of system
CA NEM California Net Energy Metering
CdTe cadmium telluride
CF capacity factor
CPI Consumer Price Index
c-Si crystalline silicon
DC direct current
DOE U.S. Department of Energy
EPC engineering, procurement, and construction
FICA Federal Insurance Contributions Act
GPRA Government Performance and Reporting Act
HVAC heating, ventilating, and air conditioning
ITC investment tax credit
LBNL Lawrence Berkeley National Laboratory
LCOE levelized cost of energy
LCOS levelized cost of storage
LCOSS levelized cost of solar-plus-storage
Li-ion lithium-ion
MACRS Modified Accelerated Cost Recovery System
MLPE module-level power electronics
MM million
MWAC megawatts alternating current
MWDC megawatts direct current
NEC National Electrical Code
NEM net energy metering
NREL National Renewable Energy Laboratory
O&M operation and maintenance
PERC passivated emitter and rear cells
PII permitting, inspection, and interconnection
PPA power-purchase agreement
PV photovoltaic(s)
Q quarter
SETO Solar Energy Technologies Office (DOE)
SG&A selling, general, and administrative
TPO third-party ownership
USD U.S. dollars
VDC volts direct current
WAC watts alternating current
WDC watts direct current
Wp watts peak
iii
This report is available at no cost from the National Renewable Energy Laboratory (NREL) at www.nrel.gov/publications.
Executive Summary
This report benchmarks U.S. solar photovoltaic (PV) system installed costs as of the first
quarter of 2020 (Q1 2020). We use a bottom-up method, accounting for all system and project
development costs incurred during installation to model the costs for residential, commercial,
and utility-scale PV systems, with and without energy storage. We attempt to model typical
installation techniques and business operations from an installed-cost perspective. Costs are
represented from the perspective of the developer/installer; thus, all hardware costs represent the
price at which components are purchased by the developer/installer, not accounting for
preexisting supply agreements or other contracts. Importantly, the benchmark also represents the
sales price paid to the installer. Therefore, it includes profit in the cost of the hardware 1; the
profit the installer/developer receives is reported as a separate cost category on top of all other
costs to approximate the final retail price paid to the installer/developer. However, we do not
include any additional profit, such as a developer fee or price gross-up, which is common in the
marketplace. We adopt this approach owing to the wide variation in developer overhead and
profit in all three sectors (residential, commercial, and utility-scale), where project pricing
depends greatly on the region and project specifics such as local retail electricity rate structures,
local rebate and incentive structures, competitiveness of the environment, and overall project or
deal structures. Benchmarks also assume a business environment without any impact from novel
coronavirus pandemic. Finally, our benchmarks are national averages calculated using average
values across all states. Table ES-1 summarizes the first-order benchmark assumptions.
Table ES-1. Benchmark Assumptions
Unit Description
Values 2019 U.S. dollars (USD)a
System PV systems are quoted in direct current (DC) terms; inverter prices are converted by
Sizes DC-to-alternating current (AC) ratios; storage systems are quoted in terms of kilowatt-
hours or megawatt-hours (kWh or MWh) of storage or the number of hours of storage
at peak capacity.
PV Sector Description Size Range
Residential Residential rooftop systems, monocrystalline silicon modules 4kW–7 kW
Commercial Commercial rooftop with ballasted racking and fixed-tilt ground- 100 kW–2 MW
mounted systems, monocrystalline silicon modules
Utility-scale Ground-mounted systems, monocrystalline silicon modules, fixed- 5–100 MW

tilt and one-axis tracking
a The dollar-per-watt total cost values are benchmarked as three significant figures, because the model inputs,
such as module and inverter prices, use three significant figures.
1
Profit is one of the differentiators between “cost” (aggregated expenses incurred by a developer or installer to
build a system) and “price” (what an end user pays for a system).
iv
Based on our bottom-up modeling, the Q1 2020 PV cost benchmarks are:
• $2.71 per watt DC (WDC) (or $3.12/WAC) for residential PV systems

• $1.72/WDC (or $1.96/WAC) for commercial rooftop PV systems
• $1.72/WDC (or $1.91/WAC) for commercial ground-mount PV systems
• $0.94/WDC (or $1.28/WAC) for fixed-tilt utility-scale PV systems
• $1.01/WDC (or $1.35/WAC) for one-axis-tracking utility-scale PV systems
• $26,153–$28,371 for a 7-kW residential PV system with 3 kW/6 kWh of storage and
$35,591–$37,909 for a 7-kW residential PV system with 5 kW/20 kWh of storage
• $2.07 million–$2.13 million for a 1-MW commercial ground-mount PV system colocated
with 600 kW/2.4 MWh of storage
• $171 million–$173 million for a 100-MW PV system colocated with 60 MW/240 MWh
of storage.
Figure ES-1 puts our Q1 2020 PV-only benchmark results in context with the results of previous
National Renewable Energy Laboratory (NREL) benchmarking analyses. When comparing the
results across this period (2010–2020), it is important to note that:
1. Values are inflation-adjusted using the 2019 Consumer Price Index (CPI). Thus, historical
values from our models are adjusted and presented as real USD instead of nominal USD. In
previous year’s models, we inflation-adjusted values based on a partial year of CPI data.
For example, in the Q1 2018 benchmark report (Fu, Feldman, and Margolis 2018), all values
are quoted in 2018 USD; however, the inflation adjustment is based on the average CPI
Index of Q1 2018 (January through March 2018). Because the benchmark reports are
produced before the end of the calendar year, indexing them to the full-year average CPI in
that year is not possible. To better correct for inflation, in this year’s report, we quote values
in previous year’s dollars (2019 USD). In 2018, the CPI-All Urban Consumers Index is 248.8
for the first three months and 251.1 for the whole year (and 255.7 for 2019).
2. Cost categories are aggregated for comparison purposes. “Soft Costs—Others” represents
permitting, inspection, and interconnection (PII); transmission line (if any); sales tax; and
engineering, procurement, and construction (EPC)/developer overhead and profit. These
costs are broken out in the report for each subsector.
3. The current versions of our cost models make a few significant changes from the versions
used in our previous Q1 2018 benchmark report (Fu, Feldman, and Margolis 2018). To better
distinguish the historical cost trends over time from the changes to our cost models, we also
calculate Q1 2019 and Q1 2020 PV benchmarks using the Q1 2018 versions of the
residential, commercial, and utility-scale PV models. Appendix A provides a detailed
discussion of the changes made to the models between previous versions (Fu, Feldman, and
Margolis 2018) and this year’s versions.
4. Our Q1 2019 and Q1 2020 benchmarks use monocrystalline PV modules, whereas all
historical benchmarks used multicrystalline PV modules. This switch reflects the overall
trend occurring in the U.S. market.
5. For previous editions of this report, we assumed a land acquisition cost of $0.03/W. Based on
Wiser et al. (2020), which stated that most utility-scale PV projects do not own the land on
which the PV system is placed, we have reclassified land costs from an upfront capital
expenditure (land acquisition) to an operating expenditure (lease payments) for 2019 and 2020.
v
From 2010 to 2020, there was a 64%, 69%, and 82% reduction in the residential, commercial
rooftop, and utility-scale (one-axis) PV system cost benchmark, respectively. A significant
portion of that reduction can be attributed to total hardware costs (module, inverter, and
hardware balance of system [BOS]), with module prices dropping 85% over that period. Overall,
modeled PV installed costs across the three sectors have experienced different recent changes.
The inflation-adjusted system cost differences between Q1 2019 and Q1 2020 are a $0.06/WDC
reduction for residential PV, a $0.04/WDC reduction for commercial rooftop PV, and a
$0.01/WDC reduction for utility-scale PV. Table ES-2 shows the benchmarked values for all three
sectors and the drivers of cost decreases and increases.
vi
Figure ES-1. NREL PV system cost benchmark summary (inflation-adjusted), 2010–2020
* The current versions of our cost models make a few significant changes from the versions used in our Q1 2018 benchmark report (Fu, Feldman, and Margolis 2018) and
incorporate costs that had previously not been benchmarked in as much detail. To better distinguish the historical cost trends from the changes to our cost models, we also calculate
Q1 2019 and Q1 2020 PV benchmarks using the Q1 2018 versions of the residential, commercial, and utility-scale PV models. The “Additional Costs from Model Updates” category
represents the difference between modeled results. Using the previous costs models, the Q1 2019 and Q1 2020 benchmarks are calculated to be: Q1 2019 = $2.56/WDC and Q1
2020 = $2.47/WDC (residential PV); Q1 2019 = $1.71/WDC and Q1 2020 = $1.64/WDC (commercial PV); Q1 2019 = $0.94/WDC and Q1 2020 = $0.89/WDC (utility-scale PV, fixed-Tilt);
Q1 2019 = $1.01/WDC and Q1 2020 = $0.96/WDC (utility-scale PV, one-axis tracker). Appendix A provides a detailed discussion of the changes made to the models between last
year’s versions (Fu, Feldman, and Margolis 2018) and this year’s versions.
vii
Table ES-2. Comparison of Q1 2019 and Q1 2020 PV System Cost Benchmarks
Commercial Rooftop Utility-Scale PV,

Sector Residential PV
PV One-Axis Tracking
Q1 2019
benchmarks in $2.77 $1.76 $1.02
2019 USD/WDC
Q1 2020
Benchmarks in $2.71 $1.72 $1.01
2019 USD/WDC
Drivers of cost • Higher module • Higher module • Higher module

decrease efficiency (from efficiency efficiency
19.2% to 19.5%) • Lower material & • Lower material &
• Decrease in BOS equipment costs in equipment costs in
hardware and supply some categories some categories
chain costs • Movement of land
acquisition cost from
upfront capital
expenditures into
operation &
maintenance
Drivers of cost • Higher labor wages • Higher labor wages • Higher labor wages
increase • Higher module costs • Higher module costs • Higher steel prices
• Higher module and
inverter costs
Hardware costs remained relatively flat, year-on-year, in Q1 2020, as shown in Figure ES-1,
resulting in no change to the percentage of non-hardware, or “soft,” costs. 2 Figure ES-2 shows
the contribution of soft costs to total costs over time. 3 Also, soft costs and hardware costs
interact. For instance, module efficiency improvements have reduced the number of modules
required to construct a system of a given size, thus reducing hardware costs. This trend has also
reduced soft costs from direct labor and related installation overhead.
2
Soft cost = total cost – hardware (module, inverter, structural and electrical BOS) cost.
3
A stagnant or rising soft cost proportion in the last two years in Figure ES-2 indicates soft costs declined more
slowly than did hardware costs; it does not indicate soft costs increased on an absolute basis. Historical contributions
of soft costs to total utility-scale PV costs differ in this figure from previous versions, because values in previous
figures were representative of fixed-tilt systems, whereas these values are representative of a one-axis tracking
system.
viii
Figure ES-2. Modeled trend of soft cost as a proportion of total cost by sector, 2010–2020
Our bottom-up system cost models enable us to investigate regional variations, system
configurations (e.g., module-level power electronics [MLPE] versus non-MLPE, fixed-tilt versus
one-axis tracking, and small versus large system size), and business structures (e.g., small
installer versus national integrator, and EPC versus developer). Different scenarios result in
different costs, so consistent comparisons can only be made when cost scenarios are aligned.
The data in this annual benchmark report inform the formulation of and track progress toward
the U.S. Department of Energy Solar Energy Technologies Office’s (SETO’s) Government
Performance and Reporting Act cost targets.
The changes in installed cost—along with improvements in operation, system design, and
technology—have resulted in changes in the cost of electricity (Figure ES-3). Compared with
system prices when SETO’s levelized cost of energy (LCOE) targets were announced in 2010,
U.S. residential and commercial PV systems are 93% and 97% toward achieving the 2020
targets, respectively, and U.S. utility-scale PV systems achieved their 2020 SETO target three
years early. In recognition of both the transformative solar progress to date and the potential for
additional innovation, SETO extended its goals in 2016 to reduce the unsubsidized LCOE by
2030 to 3¢/kWh (utility-scale PV), 4¢/kWh (commercial PV), and 5¢/kWh (residential PV).
Continued research and development, public and private partnerships, and business innovations
are necessary to achieve SETO’s 2030 LCOE targets.
* * *
ix
Figure ES-3. NREL PV LCOE benchmark summary (inflation-adjusted), 2010–2020
We updated our methods and model structure in this year’s version; 2019 and 2020 LCOEs are higher than they
would have been using previous models. Appendix A provides a detailed discussion of the changes made to the
models between the previous versions (Fu, Feldman, and Margolis 2018) and this year’s versions. LCOE is
calculated for each scenario under a range of capacity factors, but all other values remain the same. 4 ITC = federal
investment tax credit.
We also conducted a cost analysis of PV-plus-storage systems. Figures ES-4 and ES-5 put our
Q1 2020 PV-plus-storage benchmark results in context with the results of previous NREL
benchmarking analyses. Figure ES-4 shows 9% and 8% reductions in utility-scale PV-plus-
storage benchmarks between 2018 and 2020 for DC-coupled and AC-coupled systems,
respectively. Approximately 28%–30% of total cost reductions can be attributed to lithium-ion
battery and bidirectional inverter cost reductions. Although there are some configuration
differences between AC-coupled and DC-coupled systems (e.g., the inverter, structural BOS, and
electrical BOS), the total cost difference between them is only 1%. For an actual project, cost
savings may not be the only factor in choosing between DC- or AC-coupling. Additional
factors—such as retrofit considerations, system performance (including energy loss due to
clipping), design flexibility, and operation and maintenance—should be considered.
4
Capacity factors were calculated using the locations: Phoenix, AZ (high solar resource); Kansas City, MO
(medium solar resource); and New York City, NY (low solar resource).
x
Figure ES-4. Utility-scale PV-plus-storage system cost benchmark summary (inflation-adjusted),
2018–2020, DC-coupled and AC-coupled
The Q1 2018 utility-scale PV-plus-storage benchmark (Fu, Remo, and Margolis 2018) was calculated at a different
time than the Q1 2018 utility-scale PV benchmark (Fu, Feldman, and Margolis 2018) and includes different
assumptions for PV system costs, including PV module costs. MM = million.
Figure ES-5 shows the 11% and 25% reductions in residential PV-plus-storage benchmarks
between 2016 and 2020 for AC-coupled less-resilient and more-resilient cases, respectively.
Most of these reductions can be attributed to reductions in the cost of PV modules and AC-
coupled batteries. The cost reductions occurred despite the rated capacity of the 22-module
system increasing from 5.6 kW to 7.0 kW between 2016 and 2020.
Figure ES-5. Residential PV-plus-storage system cost benchmark summary (inflation-adjusted),

2016, 2019, and 2020
xi
Finally, for this year’s benchmark report, we derive a formula for the levelized cost of solar-plus-
storage (LCOSS) to better demonstrate the total cost of operating a PV-plus-storage plant, on a
per-MWh basis. Figure ES-6 shows the resulting LCOSS for a colocated AC-coupled PV-plus-
storage systems for each market segment, as well as the LCOE of standalone PV systems. For
residential PV-plus-storage, LCOSS is calculated to be $201/MWh without the federal ITC and
$124/MWh with the 30% ITC. For commercial PV-plus-storage, it is $113/MWh without the
ITC and $73/MWh with the 30% ITC. For utility-scale PV-plus-storage, it is $83/MWh without
the ITC and $57/MWh with the 30% ITC. 5
Figure ES-6. LCOSS for AC-coupled PV-plus-storage systems and LCOE for PV standalone
systems, by market segment, Q1 2020
LCOSS and LCOE are calculated for each scenario under a medium resource location. The LCOSS and LCOE
ranges are based on high and low capacity factor assumptions; all other values remain the same.
5
We use the same inputs and assumptions for the ITC and non-ITC cases, despite the fact that the inputs in
the LCOSS calculation assume the owner of the PV-plus-storage system operates the plant so they can claim the
ITC on the storage equipment. In reality, an owner would likely operate a PV-plus-storage system differently
without the ITC. Additionally, we assume projects can qualify as starting construction before 2020, allowing them to
claim a 30% ITC, instead of the 26% ITC for projects starting construction in 2020.
xii
Table of Contents
1 Introduction .................................................................................................................................... 1
2 Model Inputs and Sources ............................................................................................................ 3
2.1 Tracking the Sun Data Set ................................................................................................................ 3
2.2 Module Efficiency ............................................................................................................................ 3
2.3 PV System Size ................................................................................................................................ 5
2.4 Module-Level Power Electronics ..................................................................................................... 6
2.5 Small Installers versus National Integrators in the Residential PV Model....................................... 8
2.6 Inverter Prices and DC-to-AC Ratios ............................................................................................... 9
2.7 Module Prices ................................................................................................................................. 10
2.8 Battery Storage ............................................................................................................................... 12
2.9 PV LCOE Methods ........................................................................................................................ 13
3 Residential PV Model .................................................................................................................. 16
3.1 Residential Model Structure, Inputs, and Assumptions ................................................................. 16
3.2 Expanded “Other Soft Costs” Modeling ........................................................................................ 18
3.3 Residential Model Output .............................................................................................................. 20
3.4 Residential Model Output versus Reported Costs .......................................................................... 22
3.5 Retrofits versus New Construction ................................................................................................. 23
3.6 Additional Costs Typical of Residential PV Installation................................................................ 24
3.7 Residential PV Price Benchmark Historical Trends....................................................................... 25
3.8 Residential PV LCOE Historical Trends ........................................................................................ 27
4 Commercial PV Model ................................................................................................................. 30
4.1 Commercial Model Structure, Inputs, and Assumptions ................................................................ 30
4.2 Commercial Model Output............................................................................................................. 33
4.3 Commercial Rooftop PV Price Benchmark Historical Trends ....................................................... 37
4.4 Commercial PV LCOE Historical Trends ...................................................................................... 38
5 Utility-Scale PV Model ................................................................................................................. 42
5.1 Utility-Scale Model Structure, Inputs, and Assumptions ............................................................... 42
5.2 Utility-Scale Model Output ............................................................................................................ 45
5.3 Utility-Scale PV Price Benchmark Historical Trends .................................................................... 46
5.4 Utility-Scale PV LCOE Historical Trends ..................................................................................... 47
6 Residential Storage and PV-plus-Storage Model ..................................................................... 52
6.1 Residential Li-Ion Standalone Storage Cost Model ....................................................................... 53
6.2 Residential PV-plus-Storage System Cost Model .......................................................................... 55
6.3 Residential Model Output .............................................................................................................. 56
6.4 Residential PV-plus-Storage Price Benchmark Historical Trends ................................................. 57
6.5 Residential Levelized Cost of Solar-plus-Storage .......................................................................... 58
7 Commercial Storage and PV-plus-Storage Model ................................................................... 62
7.1 Commercial Li-Ion Standalone Storage Cost Model...................................................................... 62
7.2 Commercial PV-plus-Storage System Cost Model ........................................................................ 67
7.3 Commercial Model Output............................................................................................................. 67
7.4 Commercial Levelized Cost of Solar-plus-Storage ........................................................................ 69
8 Utility-Scale Storage and PV-plus-Storage Model ................................................................... 72
8.1 Utility-Scale Li-Ion Standalone Storage Cost Model ..................................................................... 72
8.2 Utility-Scale PV-plus-Storage System Cost Model........................................................................ 77
8.3 Utility-Scale Model Output ............................................................................................................ 80
8.4 Utility-Scale PV-plus-Storage Price Benchmark Historical Trends ............................................... 82
8.5 Utility-Scale Levelized Cost of Solar-plus-Storage ....................................................................... 83
9 Conclusions ................................................................................................................................. 86
References .......................................................................................................................................... 91
Appendix A. Changes in Methodology Between Q1 2018 and Q1 2020 Reports........................ 96
Appendix B. PV System LCOE Benchmarks in 2019 USD .......................................................... 102
xiii
List of Figures
Figure ES-1. NREL PV system cost benchmark summary (inflation-adjusted), 2010–2020 .............. vii
Figure ES-2. Modeled trend of soft cost as a proportion of total cost by sector, 2010–2020 ............... ix
Figure ES-3. NREL PV LCOE benchmark summary (inflation-adjusted), 2010–2020 ........................ x
Figure ES-4. Utility-scale PV-plus-storage system cost benchmark summary (inflation-adjusted),
2018–2020, DC-coupled and AC-coupled ............................................................................... xi
Figure ES-5. Residential PV-plus-storage system cost benchmark summary (inflation-adjusted), 2016,
2019, and 2020 ......................................................................................................................... xi
Figure ES-6. LCOSS for AC-coupled PV-plus-storage systems and LCOE for PV standalone systems,
by market segment, Q1 2020 ................................................................................................... xii
Figure 1. Capacity-weighted average module efficiency trends from the Tracking the Sun (LBNL)
and CA NEM data sets,a 2010–2020 .......................................................................................... 4
Figure 2. CA NEM and Tracking the Sun PV Installations by Technology, 2010–Q1 2020................. 5
Figure 3. Median PV system size trends from the Tracking the Sun data set,a 2010–2018 ................... 6
Figure 4. U.S. residential and commercial inverter market from the Tracking the Sun data set,a
2010–2018 ................................................................................................................................. 6
Figure 5. TPO market share in Tracking the Sun data set,a 2010–2018 ................................................. 8
Figure 6. Inverter prices from Wood Mackenzie, 2010–2020................................................................ 9
Figure 7. Ex-factory gate prices (spot prices) for U.S. and global multicrystalline and monocrystalline
modules, from Wood Mackenzie and SEIA (2020) data......................................................... 11
Figure 8. Total residential PV module market costs (2019 USD) ........................................................ 12
Figure 9. Ex-factory gate prices (spot prices) for Li-ion batteries by product, from BNEF (2018,
2019a, 2019b) data .................................................................................................................. 12
Figure 10. Q1 2020 residential, commercial, and utility-scale O&M costs by category...................... 14
Figure 11. Residential PV: Model structure ......................................................................................... 16
Figure 12. Q1 2020 U.S. benchmark: 7.0-kW residential PV system cost (2019 USD/WDC).............. 21
Figure 13. Sensitivity analysis for the Q1 2020 benchmark: Mixed 7.0-kW residential system cost
(2019 USD/WDC) ..................................................................................................................... 22
Figure 14. Q1 2020 NREL modeled cost benchmark (2019 USD/WDC) versus Q1 2020 company-
reported costs........................................................................................................................... 23
Figure 15. Q1 2020 NREL residential PV modeled cost benchmark (retrofit) versus
Q1 2020 NREL residential PV modeled cost benchmark (new construction) ........................ 24
Figure 16. Standard residential PV installation costs versus cost for systems with necessary additions
................................................................................................................................................. 25
Figure 17. NREL residential PV system cost benchmark summary (inflation adjusted),
2010–2020 ............................................................................................................................... 26
Figure 18. Q1 2019 cost for a residential multicrystalline PV system and Q1 2019 and Q1 2020
costs for a residential monocrystalline PV system .................................................................. 27
Figure 19. LCOE for residential PV systems, by region, with and without ITC, 2010–2020.............. 29
Figure 20. Commercial PV: Model structure ....................................................................................... 31
Figure 21. Q1 2020 U.S. benchmark: Commercial rooftop PV system cost (2019 USD/WDC) ........... 34
Figure 22. Q1 2020 U.S. benchmark: Commercial ground-mount PV system cost (2019 USD/WDC) 35
Figure 23. Sensitivity analysis for the Q1 2020 benchmark: 200-kW rooftop commercial PV
system cost (2019 USD/WDC) ................................................................................................. 36
Figure 24. Sensitivity analysis for the Q1 2020 benchmark: 500-kW commercial ground-mount
PV system cost (2019 USD/WDC) ........................................................................................... 36
Figure 25. NREL commercial rooftop PV system cost benchmark summary (inflation-adjusted),
2010–2020 ............................................................................................................................... 37
Figure 26. Q1 2019 cost for a commercial rooftop multicrystalline PV system and Q1 2019 and
Q1 2020 costs for a commercial rooftop monocrystalline PV system .................................... 38
xiv
Figure 27. LCOE for commercial rooftop PV systems, by region, with and without ITC,
2010–2020 ............................................................................................................................... 41
Figure 28. Utility-scale PV: Model structure ....................................................................................... 42
Figure 29. Percentage of U.S. utility-scale PV systems using tracking systems, 2010–2019 .............. 44
Figure 30. Q1 2020 U.S. benchmark: Utility-scale PV total cost (EPC + developer),
2019 USD/WDC........................................................................................................................ 45
Figure 31. Sensitivity analysis for the Q1 2020 benchmark: 100-MW one-axis utility-scale
PV system cost (2019 USD/WDC) ........................................................................................... 45
Figure 32. NREL utility-scale PV system cost benchmark summary (inflation-adjusted),
2010–2020 ............................................................................................................................... 46
Figure 33. Q1 2019 costs for utility-scale multicrystalline PV systems and Q1 2019 and Q1 2020
costs for utility-scale monocrystalline PV systems ................................................................. 47
Table 8. One-Axis Tracker and Fixed-Tilt Utility-Scale PV: LCOE Assumptions, 2010–2020
(2019 USD/WDC) ..................................................................................................................... 49
Figure 34. LCOE for utility-scale PV systems, by region, with and without ITC, 2010–2020
(fixed-tilt from 2010 to 2015, one-axis tracking from 2016 to 2020) ..................................... 51
Figure 35. Installed cost of residential storage only ............................................................................. 55
Figure 36. Modeled DC- and AC-coupled system configurations ....................................................... 56
Figure 37. Modeled total installed cost and price components for residential PV-plus-storage systems,
less-resilient versus more-resilient battery case (2019 USD) .................................................. 57
Figure 38. Residential PV-plus-storage system cost benchmark summary, 2016, 2019, and 2020 .... 58
Figure 39. U.S. residential LCOSS for an AC-coupled PV (7 kW) plus storage (3 kW/6 kWh,
2-hour duration) system and LCOE for a 7-kW standalone PV system, Q1 2020 .................. 61
Figure 40. Traditional commercial and utility-scale Li-ion battery energy storage components ......... 62
Figure 41. Battery system components ................................................................................................ 63
Figure 42. U.S. commercial Li-ion battery standalone storage costs for durations of 0.5–4.0 hours
(600 kWDC), Q1 2020 .............................................................................................................. 65
Figure 43. Modeled DC- and AC-coupled system configurations ....................................................... 67
Figure 44. Cost benchmarks for commercial PV-plus-storage systems (4-hour duration) in different
sites and the same site (DC-coupled and AC-coupled cases), Q1 2020 .................................. 68
Figure 45. U.S. commercial LCOSS for an AC-coupled PV (1 MW) plus storage (600 kW/2.4
MWh, 4-hour duration) system and LCOE for a 1-MW standalone PV system, Q1 2020 ..... 71
Figure 46. Structure of the bottom-up cost model for utility-scale standalone storage systems .......... 72
Figure 47. U.S. utility-scale Li-ion battery standalone storage costs for durations of 0.5–4.0 hours
(60 MWDC), Q1 2020............................................................................................................... 75
Figure 48. DC-coupled and AC-coupled PV-plus-storage system configurations ............................... 78
Figure 49. Cost benchmarks for PV-plus-storage systems (4-hour duration) in different sites and
the same site (DC-coupled and AC-coupled cases), Q1 2020 ................................................. 81
Figure 50. Utility-scale PV-plus-storage system cost benchmark summary 2018–2020, DC-coupled
and AC-coupled....................................................................................................................... 83
Figure 51. U.S. utility-scale LCOSS for an AC-coupled PV (100 MW) plus storage (60 MW/240
MWh, 4-hour duration) system and LCOE for a 100-MW PV standalone system, Q1 2020 . 85
Figure 52. NREL PV system cost benchmark summary (inflation-adjusted), 2010–2020 .................. 88
Figure 53. Modeled trend of soft cost as a proportion of total cost by sector, 2010–2020 .................. 89
Figure 54. NREL PV LCOE benchmark summary (inflation-adjusted), 2010–2020 .......................... 90
xv
List of Tables
Table ES-1. Benchmark Assumptions ................................................................................................... iv
Table ES-2. Comparison of Q1 2019 and Q1 2020 PV System Cost Benchmarks............................. viii
Table 1. Rapid-Shutdown Codes: Progress by State .............................................................................. 7
Table 2. Q1 2020 Inverter Price Conversion (2019 USD) ................................................................... 10
Table 3. Residential PV: Modeling Inputs and Assumptions ............................................................... 17
Table 4. Residential PV: LCOE Assumptions, 2010–2020 (2019 USD/WDC)..................................... 28
Table 5. Commercial PV: Modeling Inputs and Assumptions ............................................................. 31
Table 6. Commercial PV: LCOE Assumptions, 2010–2020 (2019 USD/WDC) ................................... 39
Table 7. Utility-Scale PV: Modeling Inputs and Assumptions ............................................................ 43
Table 9. Residential Storage-Only Modeling Inputs and Assumptions ............................................... 53
Table 10. Changes to Residential PV and Storage Models When PV and Storage Are Combined ..... 56
Table 11. Residential LCOSS Inputs and Assumptions ....................................................................... 59
Table 12. Commercial Li-ion Energy Storage System: Model Inputs and Assumptions ..................... 63
Table 13. Detailed Cost Breakdown for a 600-kW U.S. Commercial Li-ion Standalone Storage
System with Durations of 0.5–4 hours .................................................................................... 66
Table 14. Changes to Commercial PV and Storage Models When PV and Storage Are Combined ... 67
Table 15. Detailed Cost Breakdown for Commercial Li-ion PV-Plus-Storage Systems ..................... 69
Table 16. Commercial LCOSS Inputs and Assumptions ..................................................................... 70
Table 17. Utility-Scale Li-ion Energy Storage System: Model Inputs and Assumptions .................... 73
Table 18. Detailed Cost Breakdown for a 60-MW U.S. Utility-Scale Li-ion Standalone Storage
System with Durations of 0.5–4 hours .................................................................................... 76
Table 19. Cost Factors for Siting PV and Storage Together versus Separately ................................... 77
Table 20. Comparison of DC- and AC-Coupling for Utility-Scale PV-plus-Storage Systems ............ 79
Table 21. Detailed Cost Breakdown for Utility-Scale Li-ion PV-plus-Storage Systems ..................... 82
Table 22. Utility-Scale LCOSS Inputs and Assumptions..................................................................... 84
Table 23. Comparison of Q1 2019 and Q1 2020 PV System Cost Benchmarks.................................. 89
Table A-1. Comparison of Input Assumptions and Sources in the Q1 2018 Benchmark Report and
the Q1 2020 Benchmark Report .............................................................................................. 96
Table A-2. Comparison of Methods for Calculating Q1 2018 Residential PV Soft Costs in the Q1
2018 Benchmark Report and the Q1 2020 Benchmark Report ............................................... 98
Table A-3. Comparison of Q1 2020 Benchmark Costs, per Category, Calculated Using Previous
Report’s Model (Q1 2018) and the Current Model (Q1 2020) in 2019 USD........................ 101
Table B-1. NREL LCOE Summary (2019 cents/kWh) ...................................................................... 102
xvi
1 Introduction
This report continues previous tracking of photovoltaic (PV) cost reductions by benchmarking
the costs of U.S. residential, commercial, and utility-scale PV, energy storage, and PV-plus-
storage systems built in the first quarter (Q1) of 2020. 6 It was produced in conjunction with
several related research activities at the National Renewable Energy Laboratory (NREL) and
Lawrence Berkeley National Laboratory (LBNL), which are documented by Barbose and
Darghouth (2019), Bolinger, Seel, and Robson (2019), 7 Chung et al. (2015), Feldman et al.
(2015), and Fu et al. (2016).
Our benchmarking method includes bottom-up accounting for all necessary system and project-
development costs incurred when installing residential, commercial, and utility-scale systems,
and it models the Q1 2019 and Q1 2020 costs for such systems excluding any previous supply
agreements or contracts. In general, we attempt to model the typical installation techniques and
business operations from an installed-cost perspective, and our benchmarks are national
averages. The residential PV-only benchmark and the commercial rooftop PV-only benchmark
average costs by inverter type (string inverters, string inverters with direct current [DC]
optimizers, and microinverters), weighted by inverter market share. The residential PV-only
benchmark is further averaged across small installer and national integrator business models,
weighted by market share. All benchmarks include variations—accounting for the differences in
size, equipment, and operational use (particularly for storage)—that are currently available in the
marketplace. All benchmarks assume nonunionized construction labor; residential and
commercial PV systems predominantly use nonunionized labor, and the type of labor required
for utility-scale PV systems depends heavily on the development process. All benchmarks
assume the use of monofacial monocrystalline silicon PV modules. Benchmarks using cadmium
telluride (CdTe) or bifacial modules could result in significantly different results. 8 The data in
this annual benchmark report inform the formulation of and track progress toward the U.S.
Department of Energy (DOE) Solar Energy Technologies Office’s (SETO’s) Government
Performance and Reporting Act (GPRA) cost targets.
6
Previous cost benchmark reports include reports published for Q1 2018 PV (Fu, Feldman, and Margolis 2018),
2018 PV-plus-storage (Fu, Remo, and Margolis 2018), 2017 PV (Fu et al. 2017), 2016 PV (Fu et al. 2016), and 2015
utility-scale PV (Fu et al. 2015).
7
LBNL compares the bottom-up cost results of various entities, including our results.
8
In this report, we focus on the installation costs of crystalline-silicon modules, but a significant portion of U.S.
utility-scale PV systems use CdTe modules. From 2010–2019, CdTe accounted for approximately 30% of U.S.
utility-scale PV deployment (EIA 2020). This portion of the market is particularly noticeable given that CdTe
modules only represented 4% of global PV shipments over the same period. Similarly, a growing number of U.S.
systems are beginning to use bifacial modules, with transparent backs, which generate electricity from both sides of
the module—as opposed to traditional monofacial modules, which typically have opaque backsheets. Because of the
relative newness of bifacial modules, we do not have sufficient data on their current U.S. market share.
1
Our modeled costs can be interpreted as the sales price an engineering, procurement, and
construction (EPC) contractor or developer might charge for a system before any developer fee
or price gross-up (although our costs do include development costs). We use this approach
because of the wide variation in developer profits in all three sectors (residential, commercial,
and utility-scale), where project pricing depends highly on region and project specifics such as
local retail electricity rate structures, local rebate and incentive structures, competitive
environment, and overall project or deal structures.
The current versions of our cost models make a few significant changes from the versions used
in our previous Q1 2018 benchmark report (Fu, Feldman, and Margolis 2018). To better attribute
the historical cost trends over time from the changes to our cost models, we also calculate Q1
2019 and Q1 2020 PV benchmarks using the Q1 2018 versions of the residential, commercial,
and utility-scale PV models. Appendix A provides a detailed discussion of the changes made to
this year’s models from previous versions (Fu, Feldman, and Margolis 2018).
The remainder of the report is organized as follows. Section 2 describes our model inputs
and sources. Sections 3, 4, and 5 show specific model inputs and outputs for residential,
commercial, and utility-scale PV-only systems, including historical trends in system costs and
the levelized costs of energy. Sections 6, 7, and 8 show specific model inputs and outputs for
residential, commercial, and utility-scale PV-plus-storage systems, including a limited set of
historical trends in system costs and the levelized cost of PV-plus-storage. Finally, Section 9 puts
the results in context and offers conclusions.
2
2 Model Inputs and Sources
This section describes our model inputs and sources. Section 2.1 describes one of our main data
sources for the system characteristics and business models of residential and commercial PV, 9
the 2019 edition of Tracking the Sun (Barbose and Darghouth 2019). Sections 2.2 through 2.5
detail the inputs from the Tracking the Sun data set that affect PV system cost. Sections 2.6
through 2.8 detail our cost input assumptions for the highest-cost components (inverters,
modules, and battery packs), and Section 2.9 describes our levelized cost of energy (LCOE)
calculation methods.
2.1 Tracking the Sun Data Set

We use data from the 2019 edition of Tracking the Sun (Barbose and Darghouth 2019) to
benchmark generic system characteristics, such as module efficiency, system size, and direct- to
alternating-current (DC-to-AC) ratio for commercial and residential systems, as well as choice
of power electronics and installer type for residential PV standalone systems. Tracking the Sun is
based on a data set compiled primarily from state agencies, utilities, and other organizations that
administer PV incentive programs, solar renewable energy credit registration systems, or
interconnection processes. The full sample, from 30 states, includes most U.S. grid-connected
residential and nonresidential PV systems, but it excludes all ground-mounted PV systems larger
than 5 MWAC and therefore excludes U.S. utility-scale PV systems. In total, it consists of more
than 1.6 million individual PV systems installed through year-end 2018, including roughly
250,000 systems installed in 2018. These systems represent 81% of all U.S. residential and
nonresidential systems installed cumulatively through 2017 and 76% of installations in 2018.
The authors have also taken various steps to clean and standardize the raw data to ensure its
accuracy. Because the analysis—and publication—of the Q1 2019 and Q1 2020 modeled
benchmarks occur before 2019 and 2020 characteristic data are available, we use 2018 Tracking
the Sun values for the market share of inverters and national integrators, and we rely on the
California Net Energy Metering (CA NEM) Interconnection data set for module efficiency.
2.2 Module Efficiency

Figure 1 displays module efficiency data from the Tracking the Sun data set from 2010 to 2018,
along with data from the CA NEM Interconnection data set through Q1 2020. 10 Since 2010,
efficiencies for monocrystalline and multicrystalline modules have steadily improved, with the
capacity-weighted average multicrystalline module efficiency, for 60- and 72-cell modules,
increasing 0.3%–0.4% each year in absolute terms, on average. CA NEM values line up very
9
To represent commercial PV systems, we apply the characteristics of nonresidential PV systems in the Tracking
the Sun data set that are larger than 100 kW (and smaller than 5 MWAC for ground-mount systems). In addition to
rooftop commercial PV systems, the data set includes ground-mounted commercial PV systems as well as systems
for industrial applications, government sites, non-profits, and schools.
10
We use CA NEM data for the average module efficiency of utility-scale PV systems as well, even though the data
set does not include any utility-scale PV systems. We think this is justifiable, because PV modules can be used
for either application, and there is no comparable utility-scale PV data set to calculate average module efficiency in
that sector. That being said, there may be some discrepancies with real-world data, because significantly more CdTe
is deployed in the utility-scale PV sector than in the residential and commercial sectors. Module efficiency values
from CA NEM include only 60-cell and 72-cell modules from its data set to better correspond with the pricing data
we use. SunPower IBC panels, for example, are more efficient and have a different cell count, but also come at a
price premium.
3
closely with the national averages reported in Tracking the Sun. CA NEM reports a Q1 2020
capacity-weighted average monocrystalline module efficiency of 19.5%. Because module
selection may vary by region and sector, the capacity-weighted average module efficiencies (and
module prices) may be different in some regions and sectors. 11
Figure 1. Capacity-weighted average module efficiency trends from the Tracking the Sun (LBNL)
and CA NEM data sets,a 2010–2020
a Barbose and Darghouth (2019), CA NEM (2020)
In this year’s report, we model systems using monocrystalline PV modules rather than the
multicrystalline modules we modeled previously (Fu, Feldman, and Margolis 2018). When we
started benchmarking PV system prices in 2010, most U.S. PV systems used multicrystalline
modules. However, there has been an overall shift in the United States to using more
monocrystalline modules since 2016. For example, CA NEM reports that multicrystalline’s
percentage of installed PV systems in California peaked over the decade in 2013–2015 at
approximately 70%, but as of Q1 2020 had shrunk to 16% (Figure 2). Across the United States,
the percentage of distributed PV systems that installed multicrystalline modules dropped from
65% in 2015 to 11% in 2018 (Barbose and Darghouth 2019). Much of this shift can be attributed
to the rapid manufacturing expansion and associated reduction in price of passivated emitter and
rear cells (PERC) monocrystalline technology. Monocrystalline modules sell at a premium over
multicrystalline modules, but their higher efficiency can reduce the LCOE of PV systems.
We compare the 2019 total system cost between PV systems using monocrystalline modules and
those using multicrystalline modules in residential, commercial rooftop, and utility-scale PV
systems in Section 3.7.1, Section 4.3.1, and Section 5.3.1.
11
The residential sector has historically used a higher percentage of monocrystalline panels than other sectors.
4
Figure 2. CA NEM and Tracking the Sun PV Installations by Technology, 2010–Q1 2020
Sources: Barbose and Darghouth (2019), CA NEM (2020). c-Si = crystalline silicon.
2.3 PV System Size

Figure 3 displays median PV system sizes from the 2019 edition of Tracking the Sun. Residential
system sizes steadily increased from 2010 to 2018. As in previous years, we assume a 22-module
design for our residential PV system benchmark, which results in a system size of 6.3 kW, based
on the assumed 2018 average multicrystalline module efficiency. This is slightly smaller than the
2018 median size of 6.4 kW from Tracking the Sun; in 2019 and 2020, our residential PV system
benchmarks are larger (6.6 kW and 7.0 kW, respectively) owing to efficiency improvements over
time and the switch in our model to the use of monocrystalline modules. Commercial system
sizes have varied more, which likely reflects the wide range of users (e.g., office buildings,
malls, and retail stores). Limiting commercial systems to those larger than 100 kW (the
minimum size we model for commercial systems in this report and the vast majority of
commercial PV installed capacity each year), the median system size has ranged from 200 kW to
350 kW for rooftop systems and 400 kW to 1,200 kW for ground-mount applications. We use
200 kW and 500 kW as the baseline cases in our commercial rooftop and ground-mount PV
models, respectively.
5
Figure 3. Median PV system size trends from the Tracking the Sun data set,a 2010–2018
a Barbose and Darghouth (2019)
2.4 Module-Level Power Electronics

Microinverters and DC power optimizers are collectively referred to as module-level power
electronics (MLPE). By allowing designs with different roof configurations (e.g., orientations
and tilts), constantly tracking the maximum power point for each module, and providing rapid-
shutdown at the module level (required in some states), MLPE provide an optimized design
solution at the module level. In 2018, MLPE reached 86% of the total Tracking the Sun
residential data set (Figure 4).
Figure 4. U.S. residential and commercial inverter market from the Tracking the Sun data set,a
2010–2018
6
For residential system costs, we model the string inverter, power optimizer, and microinverter
options separately, and we use their market shares (14.6%, 49.8%, and 35.6%) in our Q1 2020
model for the weighted-average case. MLPE growth has been slower in the commercial rooftop
sector, although it has started to accelerate in the past few years, reaching a share of 55% in
2018. In past years, we only assumed string inverters for the commercial PV benchmark, rather
than weighting by MLPE share; this year, we also weight the commercial rooftop PV benchmark
by MLPE share (45% for three-phase string inverters, 39% for power optimizers, and 16% for
microinverters), because of changes to the National Electrical Code (NEC).
For safety reasons, rapid-shutdown codes were implemented at the array level in the 2014 NEC
and at the module level in the 2017 NEC, for rooftop PV systems. Although the 2014 NEC
required array-level rapid-shutdown, systems were also required to have the ability to reduce
system voltage quickly, so that no wires were energized more than five feet inside a building or
10 feet from a PV module array. Commercial rooftop PV systems accomplished this by having
the three-phase string inverters within 10 feet of the PV array to provide the disconnect, but
string-level inverters often cannot be located within 10 feet of a residential PV system.
Therefore, residential PV systems either required the installation of an additional combiner box
with a single array-level disconnect, or the more popular option: MLPE. Because the 2017 NEC
requires rapid shutdown at the module level for rooftop applications, the switch from the 2014
NEC to the 2017 NEC has a larger impact on commercial rooftop PV systems.
As of July 1, 2020, 34 states and Puerto Rico had adopted the 2017 or 2020 NEC rapid-shutdown
code (with California and South Carolina implementing it in 2020), and 11 states had adopted the
2014 NEC (Table 1).
Table 1. Rapid-Shutdown Codes: Progress by State
Code Rapid- State

Shutdown
R i t
Yes (module-
2020 NEC Massachusetts
level)
Alaska, Arkansas, California, Colorado, Connecticut, Georgia,
Hawaii, Idaho, Iowa, Kentucky, Maine, Michigan, Minnesota,
Yes (module- Nebraska, Nevada, New Hampshire, New Jersey, New Mexico,
2017 NEC North Carolina, North Dakota, Ohio (commercial), Oregon, Rhode
level)
Island, South Carolina, South Dakota, Tennessee, Texas, Utah,
Vermont, Washington, Wisconsin, Wyoming, West Virginia, and
Puerto Rico
Yes (array-level, Alabama, Delaware, Florida, Louisiana, Maryland, Montana, New
2014 NEC
within 10 feet) York, Ohio (residential), Oklahoma, Pennsylvania, Virginia
2011 NEC No Washington, D.C.
2008 NEC No Indiana
No statewide
No Arizona, Illinois, Kansas, Mississippi, and Missouri
NEC adoption
Source: NEMA (2020)
7
2.5 Small Installers versus National Integrators in the
Residential PV Model
Our residential PV benchmark is based on two different business structures: “small installer”
and “national integrator.” We define small installers as businesses that engage in lead generation,
sales, and installation but do not provide financing solutions in-house, although they may partner
with a larger company to offer customers loans, leases, or power-purchase agreements (PPAs).
National integrators perform all small installer functions, and they directly provide financing and
system monitoring for third-party-ownership (TPO) systems. 12 In our models, the difference
between small installers and national integrators is manifested in (1) differences in module and
inverter prices that are due to the buying power of national integrators and (2) differences in
customer acquisition, permitting, inspection, and interconnection (PII), and overhead cost
categories, where national integrators are modeled with higher expenses for customer acquisition
(relying less on referrals and spending more time on growing markets) and PII (due to higher
cancellation rates). Although national integrators provide financing solutions, we do not
incorporate financing costs into the benchmark.
As shown in Figure 5, residential TPO systems have lost market share to the direct business
model since 2015, led by small installers. We use 38% national integrator and 62% small
installer market shares to compute the national weighted-average case in our Q1 2020 residential
PV model.
Figure 5. TPO market share in Tracking the Sun data set,a 2010–2018
12
For modeling purposes, we separate the residential market into small installers and national integrators using
TPO market share. In reality, there are a wide range of business models and installer sizes. Most TPO providers are
national integrators and offer mostly TPO products. Even small installers that offer financing typically get that
financing from a larger, national company, so their costs would be borne by that model as well. In short, it is a
simplification, but we think a reasonable one.
8
2.6 Inverter Prices and DC-to-AC Ratios
As shown in Figure 6, we source inverter prices, including MLPE prices, from the Wood
Mackenzie (Wood Mackenzie 2020) database, which contains typical U.S. prices from Tier 1
suppliers to developers in the market. For Q1 2020 modeling, we convert the U.S. dollar (USD)
per WAC inverter prices from Wood Mackenzie (2020) to 2019 USD/WDC using the Consumer
Price Index (CPI) and the DC-to-AC ratios shown in Table 2. We model systems using an
average DC-to-AC ratio, but a wide variety of DC-to-AC ratios are reported for U.S. PV
systems.
Figure 6. Inverter prices from Wood Mackenzie, 2010–2020

Data are from Wood Mackenzie (2014a, 2014b, 2019a, 2020) and Wood Mackenzie and SEIA (2020). Data are also
supplemented, in 2010 and 2011, using revenue per-watt shipped data from Enphase (2019) for microinverters.
9
Table 2. Q1 2020 Inverter Price Conversion (2019 USD)
Inverter Type Sector USD/WAC DC-to-AC Ratioa USD/WDC
Single-phase
Residential PV (non-MLPE) 0.15 1.11 0.14
string inverter
Residential and commercial PV

Microinverter 0.34 1.16 0.29
(MLPE)
DC power optimizer,
single-phase string Residential PV (MLPE) 0.30 1.16 0.26
inverter
Three-phase
Commercial PV (non-MLPE) 0.08 1.11 0.07
string inverter
DC power optimizer,
three-phase string Commercial PV (MLPE) 0.14 1.16 0.12
inverter
Central inverter Utility-scale PV (fixed-tilt) 0.07 1.37 0.05
Central inverter Utility-scale PV (1-axis tracker) 0.07 1.34 0.05
All inverter prices include the cost of monitoring equipment.

aWe updated the inverter DC-to-AC ratios using LBNL data (Bolinger, Seel, and Robson 2019; Barbose and
Darghouth 2019).
2.7 Module Prices

We assume an ex-factory gate (spot or first-buyer) price of $0.41/WDC for Tier 1 monocrystalline
modules in Q1 2020, based on Wood Mackenzie and SEIA (2020). As Figure 7 shows, U.S. spot
prices declined substantially between 2014 and 2016, and they approached global spot prices. In
2017, however, U.S. spot prices rose as global spot prices continued to decline. Several factors,
including U.S. policy on imported modules, may have contributed to the divergence between
U.S. and global spot prices. In early 2018, U.S. spot prices began to drop again; in Q1 2020,
U.S. module prices continued to fall, dropping close to their lowest recorded levels, but
monocrystalline modules were still trading at a significant premium over the global module
average selling price (ASP). In the past few years, the U.S. market has had such an increasing
demand for monocrystalline modules that by 2020 there was not enough demand for
multicrystalline modules to give an “apples-to-apples” comparison of U.S. spot pricing in Q1
2020; therefore, when comparing the two technologies, we model Q1 2019 costs. In Q1 2019, we
assume an ex-factory gate price of $0.40/WDC for Tier 1 monocrystalline modules and
$0.33/WDC for Tier 1 multicrystalline modules, based on Wood Mackenzie and SEIA (2020).
10
Although commercial and utility-scale PV developers typically can procure modules at or near
the spot price, residential national integrators and small installers incur additional supply chain
costs (Figure 8). Historical inventory can create a price lag (approximately six months) for the
market module price in the residential sector when the modules from previous procurements are
installed in today’s systems. In our Q1 2020 residential PV benchmark, this supply chain cost
equates to a $0.02/W (6%) premium. We assume small installers and national integrators are
both subject to a 15% ($0.06/W) premium on the spot price for module shipping and handling
(Fu, Feldman, and Margolis 2018). Small installers are subject to an additional 20% ($0.09/W)
premium owing to small-scale procurement (Bloomberg 2018), which is consistent with an
assumed 20% premium in the Q1 2017 residential PV benchmark (Fu et al. 2017). Both types
of companies are also subject to 5% sales tax (weighted national average), bringing the small
installer’s monocrystalline module cost to $0.61/W and the national integrator’s cost to $0.52/W.
Figure 7. Ex-factory gate prices (spot prices) for U.S. and global multicrystalline and
monocrystalline modules, from Wood Mackenzie and SEIA (2020) data
Global monocrystalline module prices before 2018 are from PVinsights (2019).
11
Figure 8. Total residential PV module market costs (2019 USD)
2.8 Battery Storage

As Figure 9 shows, lithium-ion (Li-ion) battery spot prices declined substantially (87%) between
2010 and 2019. From 2018 to 2019 alone, prices dropped 13%. The Li-ion battery pack price
from Bloomberg New Energy Finance (BNEF) refers to the volume-weighted average of
automotive and stationary storage. The stationary battery market has a slightly higher price.
There is also price variation for different battery durations. In previous years, we used the
volume-weighted average (i.e., the “Li-ion battery pack” price) because of a lack of data for
stationary storage with different durations. In this year’s report, we use BNEF (2019b) stationary
storage cost data, differentiated by market segment and hours of storage. Although not
referenced in this report, BNEF also provides commercial and utility battery rack data for 30-
minute and 2-hour storage products.
Figure 9. Ex-factory gate prices (spot prices) for Li-ion batteries by product, from BNEF (2018,
2019a, 2019b) data
12
2.9 PV LCOE Methods
Although LCOE is not a perfect metric to measure the competiveness of PV within the energy
marketplace, it does incorporate many PV metrics—beyond upfront installation costs—that are
important to energy costs. For a previous edition of this report (Fu et al. 2017), we performed
a literature review to determine inputs not already benchmarked in the report. When LCOE
assumptions were not found in the selected literature in a given year, straight-line changes
were assumed between any two values. This year, we inform the inputs using ongoing NREL
benchmarking work. We input these assumptions into NREL’s System Advisor Model, a
performance and financial model, 13 to calculate real LCOEs (considering inflation) for
various locations.
Annual Degradation
In January 2018, NREL and DOE interviewed nine independent engineers and PV project
financiers; they said they assume an annual PV module degradation of 0.7% per year. For certain
projects with specific project and system characteristics that have been well vetted, some
independent engineers assume a 0.5% annual degradation (Feldman, Jones-Albertus, and
Margolis 2018). Because this lower value only applies to specific projects, we benchmark the
higher degradation rate.
Operation and Maintenance

In fiscal year 2018, a PV operation and maintenance (O&M) working group that was convened
under the sponsorship of DOE’s SETO developed a model to calculate the cost associated with
PV system O&M (Walker et al. 2020). Measures of O&M in the cost model correlate to the PV
O&M services described by a best practices guide (NREL et al. 2018). Some of the O&M cost
drivers in the model are informed by actuarial failure and repair data from Sandia National
Laboratories (Klise et al. 2018), but current default values reflect the best judgement of the
working group for measures with unavailable data. In the current version of the model, labor
rates, inverter replacement cost, discount rate, inflation rate, and capital expenditures are
adjusted to fiscal year 2019. Apart from these updates, actuarial failure and repair data are
updated for a few measures (insulated-gate bipolar transistor matrix, broken modules, inverter
fan motors, inverter reboot, damaged racking, tracker controller, and tracker bearings) (Gunda
and Homan forthcoming). For this version, five additional line measures (land lease, property
taxes, insurance, asset management, and security) are added based on feedback collected by
LBNL from U.S. solar industry professionals (Wiser et al. 2020).
O&M costs in the Walker et al. (2020) O&M cost model include preventive maintenance,
scheduled at regular intervals with costs increasing at an inflationary rate, as well as corrective
maintenance to replace components. The model derives corrective maintenance by multiplying
the replacement cost, including labor, by the probability that a failure will occur each year based
on actuarial data. Component failure probabilities for each year are calculated using a Weibull,
log-normal, or other distribution based on actual data, when possible.
As shown in Figure 10, 133 measures in the cost model are sorted into nine O&M cost
categories: inverter replacement, operations administration, module replacement, components
13
See https://sam.nrel.gov/.
13
parts replacement, system inspection and monitoring, module cleaning and/or vegetation and
pest management, land lease, property tax, and insurance, asset management, and security.The
current benchmarks are $28.94/kWDC/yr (residential), $18.55/kWDC/yr (commercial; roof
mount), $18.71/kWDC/yr (commercial; ground mount), $16.32/kWDC/yr (utility-scale, fixed-tilt),
and $17.46/kWDC/yr (utility-scale, single-axis tracking).
Figure 10. Q1 2020 residential, commercial, and utility-scale O&M costs by category
System Losses
Energy losses occur between PV generation and output to the grid owing to AC and DC wiring
losses, soiling, inverter mismatches, and shading and snow loading for certain systems. We
aggregate the losses into two categories: 9.5% of electricity lost from preinverter derate (DC
losses) and 2.0% of energy lost from inverter efficiency (AC losses).
Based on data analyzed by NREL, previous system loss benchmarks are consistent with current
performance in the field, so these benchmarks have not been changed for 2020. We do assume
a higher-voltage inverter in this year’s utility-scale PV benchmark: 1,500 V rather than the 1,000
V used previously. However, increasing voltage typically has a negligible overall impact on
losses. On the DC side, increasing voltage reduces conductor losses per length of conductor, yet
system layouts typically move to longer string lengths, resulting in similar overall losses on the
DC side (although cost is reduced). On the AC side, AC loss factors have little to do with the DC
system voltage, so the AC losses will typically not change with higher DC voltages.
Financing
The 2019 and 2020 financing assumptions are based on Feldman, Bolinger, and Schwabe (2020);
financing costs in that report are lower than in previous benchmark work (Feldman, Lowder, and
Schwabe 2016; Feldman and Schwabe 2017, 2018). All data compiled for these reports are
14
derived from a combination of basic literature reviews, product research, and interviews with
industry professionals.
The financing values represent current financing structures, which depend on the investment tax
credit (ITC). Although the ITC represents a net positive for projects, financing costs (but not
LCOE) would be lower without the ITC (Feldman, Bolinger, and Schwabe 2020). In our
benchmark reports, we have historically reported LCOE with current financing costs (which are
based on owners using the ITC) without an ITC, which is incongruous. In this year’s report, we
calculate LCOE assuming long-term steady-state financing assumptions, with no ITC and with
interest rates higher than current historically low levels.
15
3 Residential PV Model
This section describes our residential PV model’s structure, inputs, and assumptions (Section
3.1); expanded “other soft costs” modeling” (3.2); model output (3.3); differences between
modeled output and reported costs (3.4); differences between retrofits and new construction
(3.5); additional costs typical of residential PV installation (3.6); and historical PV price (3.7)
and LCOE (3.8) trends.
3.1 Residential Model Structure, Inputs, and Assumptions

We model a 7.0-kW residential rooftop system using 60-cell, monocrystalline, 19.5%-efficient
modules from a Tier 1 supplier and a standard flush mount, pitched-roof racking system. Figure
11 presents the cost drivers and assumptions, cost categories, inputs, and outputs of the model.
Table 3 presents modeling inputs and assumptions in detail.
Figure 11. Residential PV: Model structure

BOS = balance of system
16
Table 3. Residential PV: Modeling Inputs and Assumptions
Category Modeled Value Description Sources

System size 7.0 kW Average installed size per Barbose and
system Darghouth 2019;
CA NEM 2020
Module efficiency 19.5% Average module efficiency CA NEM 2020
Module price $0.41/WDC Ex-factory gate (first buyer) Wood Mackenzie
price, Tier 1 monocrystalline and SEIA 2020
modules
Inverter price Single-phase string Ex-factory gate (first buyer) Wood Mackenzie
inverter: $0.14/WDC prices, Tier 1 inverters 2020; Wood
DC power optimizer Mackenzie and
single-phase string SEIA 2020
inverter: $0.26/WDC
Microinverter:
$0.29/WDC
Structural BOS $0.08/WDC Includes flashing for roof NREL 2020
(racking) penetrations and all rails and
clamps
Electrical BOS $0.18–$0.28/WDC Conductors, switches, Model assumptions,
Varies by inverter combiners and transition NREL 2020
option boxes, as well as conduit,
grounding equipment,
monitoring system or
production meters, fuses,
and breakers
Supply chain Varies by installer 15% costs and fees BLS 2019;
costs (percentage type and location associated with shipping and NREL 2020;
of equipment handling of equipment model assumptions
costs) Additional 6% cost for
historical inventory
Additional 20% small-scale
procurement for module-
related supply chain costs for
small installers
Additional 20% for inverter-
related supply chain costs for
small installers and 10% for
national integrators
Sales tax National average: Sales tax on the equipment RSMeans 2017
5.1%
Direct installation Electrician: $27.47 Modeled national average BLS 2019; NREL 2020
labor per hour labor rates
Laborer: $18.17
per hour
Hours vary by
inverter option
17
Burden rates Total nationwide Workers compensation, RSMeans 2017
(percentage of average: 18% federal and state
direct labor) unemployment insurance,
Federal Insurance
Contributions Act (FICA),
builder’s risk, and public
liability
PII $0.23/WDC for small Completed and submitted NREL 2020
installers applications, fees, design
$0.25/WDC for changes, and field inspection
national integrators
Varies by location
Sales and $0.38/WDC (small Initial and final drawing plans, NREL 2020
marketing installer) advertising, lead generation,
(customer $0.50/WDC (national sales pitch, contract
acquisition) integrator) negotiation, and customer
interfacing
Varies by location
Overhead $0.27/WDC (small Rent, building, equipment, NREL 2020
(general and installer) staff expenses not directly
administrative) $0.28/WDC (national tied to PII, customer
integrator) acquisition, or direct
installation labor
Varies by location
Profit (%) 17% Fixed percentage margin Fu et al. 2017
applied to all direct
costs including hardware,
installation labor, direct sales
and marketing, design,
installation, and permitting
fees
3.2 Expanded “Other Soft Costs” Modeling

In this year’s benchmark analysis, we expand our modeling of customer acquisition, engineering,
PII, and overhead. In addition to providing finer cost granularity, we include additional costs
borne by many U.S. installers that were not captured in previous editions; therefore, our
benchmarked soft costs in this report are higher than those in previous reports. The first four cost
categories estimate costs by looking at the necessary steps taken to sell, engineer, permit, and
interconnect a residential PV system. Each task requiring staff time is categorized by department
(e.g., sales and permitting). The last category, overhead, estimates costs by itemizing expenses of
all staff time and resources necessary to operate a residential PV installation company, but which
are not directly tied to a specific installation. Each method is described in detail below.
Customer acquisition costs are estimated by breaking down the process into the following
steps: advertisement, lead generation, qualifications/first sales pitch, and final sales pitch. Within
each step are various methods that may be used to acquire a customer. The cost contribution to
total PV system cost for each step is calculated by multiplying (1) the average cost per
occurrence (based on a fee or an hourly wage and the number of hours) by (2) the estimated
18
percentage of national sales that use this step divided by (3) the average conversion from this
step to an installed system. Multiplying the cost per occurrence by the estimated percentage of
national sales is done to provide a national average, whereas dividing by the average conversion
is needed to account for the costs incurred by the company from potential customers who do not
end up purchasing a PV system from the installer. The data in this section were compiled from
public securities filings (i.e., 10-Ks), analyst reports and presentations, as well as conversations
with those involved in the residential PV customer acquisition business.
Engineering costs summarize the costs associated with designing initial and final system plans.
The cost contribution to total PV system cost for each step is calculated by multiplying (1) the
average cost per occurrence (or hourly wage multiplied by the number of hours) by (2) the
estimated percentage of national sales that use this step divided by the (3) average conversion
from this step to an installed system. Multiplying the cost per occurrence by the estimated
percentage of national sales is done to provide a national average, whereas dividing by the
average conversion is needed to account for the costs incurred by the company from potential
customers who do not end up purchasing a PV system from the installer. The data in this section
were compiled by averaging costs reported from private conversations with residential PV
installers.
Permitting, inspection, and interconnection categories summarize the costs associated with
applying for and receiving a permit or interconnection agreement from an “authority having
jurisdiction.” The cost contribution to total PV system cost for each permitting or interconnection
step is calculated by multiplying (1) the average cost per occurrence (either a fee, cost [e.g.,
mileage], or hourly wage multiplied by the number of hours) by (2) the estimated percentage of
national sales that use this step divided by (3) the average conversion from this step to an
installed system. Multiplying the cost per occurrence by the estimated percentage of national
sales is done to provide a national average, whereas dividing by the average conversion is needed
to account for the costs incurred by the company from potential customers who do not end up
purchasing a PV system from the installer. The data in this section were compiled by averaging
costs reported from private conversations with residential PV installers.
Overhead summarizes the costs associated with providing the business platform and
infrastructure to sell, permit, install, and interconnect a residential PV system. It is divided into
two large categories: business expenses (e.g., rent, office equipment, and professional services)
and staff expenses. Staff expenses for national integrators include the “C-suite” executives,
treasurers, customer service staff, supply chain staff, and information technology staff; staff
expenses for small installers include principals, engineers, sales team members, and
administrators. The model also estimates the total staff time attributed directly to selling,
engineering, permitting, and interconnecting a PV system, as well as the percentage of time
for each position associated with direct project costs. The number of hours spent on projects is
calculated by multiplying the hours spent per PV system for an individual cost category by the
number of systems built in that year. For example, if 10 sales-hours were required per system
installed, on average, and 400 PV systems were installed in a year, total sales staff time spent on
projects in that year would be 4,000 hours. Furthermore, if sales staff spent two thirds of their
time directly on projects, total sales staff time would be 6,000 hours per year (or 4,000 divided
by two thirds). The 6,000 hours translates into three staff members (i.e., 6,000 divided by
2,000—the number of work-hours in a year [50 weeks, 40 hours per week—assuming two weeks
19
of vacation per year]). Annual salary is estimated by (1) multiplying the hourly wages for staff
associated directly with projects (based on private conversations with installers) by the number
of paid hours in a year (52 × 40) or (2) for those not directly associated with projects, by
estimating the average salary of that position in a company, using available published data. Base
salaries are also grossed to account for other corporate costs, such as benefits, FICA, and
bonuses. Total staff, per category, is based on the number of systems installed, the number of
megawatts installed, or the ratio of employee category per total staff size.
3.3 Residential Model Output

Figure 12 presents the U.S. national benchmark from our residential PV model. Market shares of
62% for small installers and 38% for national integrators are used to compute the national
weighted average. String inverter, power optimizer, and microinverter options are each modeled
individually, and the “mixed” case applies their market shares (14.6%, 49.8%, and 35.6%) as
weightings.
Figure 13 shows a sensitivity analysis for the mixed case, with cost categories that vary by
location and hardware specification. Inverter type has the largest impact on installed system cost,
with use of string inverters resulting in $2.47/WDC and use of microinverters resulting in
$2.83/WDC.
20
Figure 12. Q1 2020 U.S. benchmark: 7.0-kW residential PV system cost (2019 USD/WDC)
21
Figure 13. Sensitivity analysis for the Q1 2020 benchmark: Mixed 7.0-kW residential system cost
(2019 USD/WDC)
3.4 Residential Model Output versus Reported Costs

As shown in Figure 14, our bottom-up modeling approach yields a different cost structure than
those reported by public solar integrators in their corporate filings (e.g., Sunrun 2020, Vivint
Solar 2020). Because national integrators sell and lease PV systems, they practice a different
method of reporting costs than do businesses that only sell goods. Many of the costs for leased
systems are reported over the life of the lease rather than the period in which the system is sold;
therefore, determining the actual costs at the time of sale is difficult. Although Sunrun and Vivint
Solar report system costs in their corporate filings on a quarterly basis (but not “profit” per
system), the limited transparency in the public filings makes it difficult to determine the
underlying costs as well as the timing of those costs. Because of the lack of available reported
company costs, explaining these differences entirely is difficult, and this topic is worthy of future
research. Explanations of the difference in reported cost could include the following:
1. Reported companies may spend more on customer acquisition costs to grow market share.
2. Reported companies’ customer acquisition costs consist of leasing, loan, and cash purchase
options. Non-cash purchase options may have higher customer acquisition costs than the cash
purchase model in this report. National installers also have recently spent considerable effort
retraining sales teams as they have shifted focus toward offering customers a direct
ownership option rather than a lease or PPA. Retraining a sales staff can be a multi-month
process and add considerable expense (Wood Mackenzie and SEIA 2018). Moreover, fewer
systems may be sold during the transition process, which would increase customer
acquisition costs on a per-watt basis.
3. Part of the difference in installation costs could come from preexisting contracts or older
inventory that national integrators used in systems installed in Q1 2020.
22
Figure 14. Q1 2020 NREL modeled cost benchmark (2019 USD/WDC) versus Q1 2020 company-
reported costs
The PII cost category is included in sales and marketing.
3.5 Retrofits versus New Construction

As discussed by Ardani et al. (2018), the residential PV sector has a significant opportunity to
reduce costs by installing PV systems when new homes are constructed; this is unlike most of the
current market, in which existing homes are retrofitted with PV systems. For comparative
purposes, we build a “new construction” business structure, using the expanded modeling in this
year’s version of the residential PV model for customer acquisition, engineering, PII, and
overhead. The new construction case assumes residential PV systems are part of the standard
features of a new production home, which is akin to the legislation passed in California
mandating such a practice; some developers in other states also offer production homes in new
developments with residential PV systems as a default feature. 14 As indicated in Figure 15, new
builds are $0.65/W less expensive than retrofits; this is due to substantially lower customer
acquisition and PII costs, as well as reduced costs through efficiencies in labor and structural
BOS.
14
Many of the cost savings achieved by integrating solar into production homes may not translate to custom new
home PV projects, from third-party vendors, because there may still be a sales process and coordination between
firms for site work (e.g., framing, roofing, electrical, plumbing, and communications), the PV design may need to be
changed with design changes to home (e.g., installation of skylights), and there may be a longer time frame between
contract closing and the date the system is placed in service.
23
Figure 15. Q1 2020 NREL residential PV modeled cost benchmark (retrofit) versus
Q1 2020 NREL residential PV modeled cost benchmark (new construction)
Cost reduction for PV in new construction occurs for a variety of reasons. There are virtually
no customer acquisition costs, because the new home comes with a PV system as a standard
option. Some costs are borne by the installer (e.g., “holding the customer’s hand” through the PV
installation and interconnection process as well as PII costs). However, PII is significantly more
streamlined, because it is part of the larger permitting process for a home (or development), and
installers almost never incur costs due to customer cancellation. Finally, the installation process
takes less time because it is incorporated into building the roof and other parts of the new home,
while material cost savings are realized by building the roof and the electrical system seamlessly
with the PV system.
3.6 Additional Costs Typical of Residential PV Installation

Our benchmarking method includes bottom-up accounting for all necessary system and project-
development costs incurred when installing U.S. residential PV systems. This year, we calculate
additional hardware, installation labor, and roofing costs that are often incurred for many PV
systems. Because of requirements in some authorities having jurisdiction, or for a particular
building, additional hardware and installation labor costs must be incurred. These costs include
partial or full reroofing, adding another disconnect, upgrading a transformer, upgrading a main
panel, or being forced (for permitting or interconnection reasons) to install a smaller system than
originally designed. Not all U.S. projects must incur these costs, so the average additional
contribution to total PV system cost for each step is calculated by multiplying the average cost
per occurrence (either material costs or hourly wage multiplied by the number of hours) by the
estimated percentage of national sales that use this step, divided by the average conversion from
this step to an installed system. Figure 16 summarizes the results of this analysis. The extra cost
categories can add 10% to the benchmark system cost.
24
Figure 16. Standard residential PV installation costs versus cost for systems with necessary
additions
3.7 Residential PV Price Benchmark Historical Trends

NREL began benchmarking PV system costs in 2010 to track PV costs against SETO targets and
to examine cost-reduction opportunities for achieving these goals. 15 Since then, NREL has
produced eight additional benchmarks. The current version of our residential cost model makes a
few significant changes from the version used in our Q1 2018 benchmark report (Fu, Feldman,
and Margolis 2018). To better distinguish the historical cost trends over time from the changes to
our cost models, we also calculate Q1 2019 and Q1 2020 PV benchmarks using the Q1 2018
version. Appendix A provides a detailed discussion of the changes made to the models between
previous reports (Fu, Feldman, and Margolis 2018) and this year’s report. Figure 17 summarizes
the reduction in residential PV system cost benchmarks between 2010 and 2020. 16 The
“Additional Costs from Model Updates” category represents the difference between modeled
results calculated using the current model versus the previous model. Using the previous cost
model, the Q1 2019 and Q1 2020 benchmarks are $2.56/WDC and $2.47/WDC, respectively.
15
The original, overarching 2020 SETO goal for solar was to reach levelized cost parity with a new thermal plant,
which was estimated to be 6¢/kWh without subsidies, or a system installed cost of $1/W. SETO later separated
commercial and residential PV to have their own goals of costs below retail rates, which were estimated to be
7¢/kWh and 9¢/kWh, or system installed costs of $1.25/W and $1.50/W, respectively (all 2020 targets are quoted in
nominal USD). In recognition of both the transformative solar progress to date and the potential for additional
innovation, SETO extended its goals in 2016 to reduce the unsubsidized cost of energy by 2030 to 3¢/kWh, 4¢/kWh,
and 5¢/kWh for utility-scale PV, commercial PV, and residential PV (all 2030 targets are quoted in nominal USD).
16
Each year’s PV system cost benchmark corresponds to the NREL benchmark calculted in Q4 of the previous year
or Q1 of the current year (e.g., 2010 = Q4 2009, 2017 = Q1 2017).
25
Figure 17. NREL residential PV system cost benchmark summary (inflation adjusted),
2010–2020
* The current version of our cost model makes a few significant changes from the version used in our Q1 2018
benchmark report (Fu, Feldman, and Margolis 2018) and incorporates costs that had previously not been
benchmarked in as much detail. To better distinguish the historical cost trends from the changes to our cost models,
we calculate Q1 2019 and Q1 2020 PV benchmarks using the Q1 2018 version of the residential PV model. The
“Additional Costs from Model Updates” category represents the difference between modeled results. Using the
previous cost model, the Q1 2019 and Q1 2020 benchmarks are calculated to be $2.56/WDC and $2.47/WDC,
respectively.
As demonstrated in Figure 17, from 2010 to 2020, there was a 64% reduction in the residential
PV system cost benchmark. Approximately 57% of that reduction can be attributed to total
hardware costs (module, inverter, and hardware BOS), with module prices dropping 85% over
that period. An additional 20% can be attributed to labor costs, which dropped 84% over the
period. The final 22% is attributable to other soft costs, including PII, sales tax, overhead, and
net profit. 17 From 2019 to 2020, there was a 2% reduction in the residential PV system cost
benchmark.
Comparing Multicrystalline and Monocrystalline PV Systems

In this year’s report, we model systems using monocrystalline PV modules, unlike previous
editions of this report (Fu et al. 2018), for which we modeled multicrystalline PV modules. In
the past few years, the U.S. market has had an increasing demand for monocrystalline modules;
by 2020, there is not enough demand for multicrystalline modules to give an apples-to-apples
comparison of U.S. spot pricing. Figure 18 compares Q1 2019 residential PV system pricing
when using monocrystalline versus multicrystalline modules, and it shows the change in price
of a residential PV system using monocrystalline modules between Q1 2019 and Q1 2020.
17
Although the residential PV system model always assumes a 22-panel design for all years, the rated size of the
system increases owing to improvements in efficiencies. Therefore, some of the cost reduction can be attributed to
an increase in system size.
26
Figure 18. Q1 2019 cost for a residential multicrystalline PV system and Q1 2019 and Q1 2020
costs for a residential monocrystalline PV system
As shown in Figure 18, in Q1 2019 there was a $0.06/W system price premium from using
multicrystalline modules over monocrystalline modules for residential PV systems. The total
system cost reductions achieved by increasing efficiency with monocrystalline modules
outweighed the premium in monocrystalline module price. Residential PV systems using
monocrystalline modules achieved a $0.06/W (2%) reduction in price from Q1 2019 to Q1 2020.
3.8 Residential PV LCOE Historical Trends

Assumptions for the residential PV LCOE benchmarks from 2010 to 2020 are summarized in
Table 4. In addition to a 64% reduction in installed cost from 2010 to 2020, O&M costs declined
49%, annual degradation declined 30%, equity discount rate declined 32%, debt interest rate
declined 27%, and debt fraction increased 57%.
27
Table 4. Residential PV: LCOE Assumptions, 2010–2020 (2019 USD/WDC)
2010 2011 2012 2013 2014 2015 2016 2017 2018 2019 2020
Benchmark Report
Installed cost ($/W) 7.53 6.62 4.67 4.09 3.60 3.36 3.16 2.94 2.78 2.77 2.71
Inverter loading ratio 1.10 1.11 1.12 1.13 1.13 1.14 1.15 1.15 1.15 1.15 1.15
Ongoing NREL Benchmarking
Annual degradation (%) 1.00 0.95 0.90 0.85 0.80 0.75 0.75 0.75 0.70 0.70 0.70
O&M expenses ($/kW-yr) 56 49 42 36 31 26 25 25 22 27 29
Preinverter derate (%) 90.0 90.1 90.2 90.3 90.4 90.5 90.5 90.5 90.5 90.5 90.5
Inverter efficiency (%) 94.0 94.8 95.6 96.4 97.2 98.0 98.0 98.0 98.0 98.0 98.0
Inflation rate (%) 2.5 2.5 2.5 2.5 2.5 2.5 2.5 2.5 2.5 2.5 2.5
Market Case
Equity discount rate (real) (%) 9.0 8.6 8.3 7.9 7.6 7.3 6.9 6.9 6.9 6.1 6.1
Debt interest rate (%) 5.5 5.4 5.3 5.2 5.0 4.9 4.8 4.8 4.8 4.0 4.0
Debt fraction (%) 34.2 35.2 36.1 37.1 38.1 39.0 40.0 40.0 40.0 53.7 53.7
Steady-State Financing (No ITC)
Equity discount rate (real) (%) — — — — — — — — — — 6.1
Debt interest rate (%) — — — — — — — —— — — 5.0
Debt fraction (%) — — — — — — — — — — 71.8
All 2010–2018 data are from Fu, Feldman, and Margolis (2018), and they are adjusted for inflation. Residential PV system LCOE assumes:
(1) System lifetime of 30 years (8) Module tilt angle of 25 degrees, and an azimuth of 180 degrees
(2) Federal tax rate of 21% (9) Debt with a term of 18 years
(3) State tax rate of 6% (10) $1.1 million of upfront financial transaction costs for a $100 million TPO
transaction of a pool of residential projects
(4) Modified Accelerated Cost Recovery System (MACRS)
depreciation schedule (11) 2019 and 2020 financial assumptions from Feldman, Bolinger, and
Schwabe (2020).
(5) No state or local subsidies
(6) A working capital and debt service reserve account for six months of
operating costs and debt payments (earning an interest rate of 1.75%)
7) Three-month construction loan, with an interest rate of 4% and a fee of 1%
of the cost of the system
28
Using these assumptions, we calculate the residential PV LCOE—with and without the 30%
federal ITC—for a high solar resource (capacity factor [CF]: 21.6%), medium solar resource
(CF: 17.6%), and low solar resource (CF: 16.4%) (Figure 19). 18 From 2010 to 2020, residential
PV LCOE declined 74% (1% between 2019 and 2020), resulting in an unsubsidized LCOE of
$0.11–$0.14/kWh ($0.07–$0.09/kWh when including the federal ITC). This reduction is 93%
toward achieving SETO’s 2020 residential PV LCOE goal from the residential PV system price
when the goal was announced in 2010. 19 We also calculate PV LCOE without the ITC using
steady-state financing assumptions. Under these assumptions, unsubsidized residential PV LCOE
ranges from $0.10–$0.14/kWh in Q1 2020.
Figure 19. LCOE for residential PV systems, by region, with and without ITC, 2010–2020
We updated our methods and model this year; 2019 and 2020 LCOEs are higher than they would have been using
previous models. Appendix A provides a detailed discussion of the changes made to the models between the
previous version (Fu, Feldman, and Margolis 2018) and this year’s version. LCOE is calculated for each scenario
under a range of CFs, but all other values remain the same.
18
CFs are calculated based on Phoenix, AZ (high solar resource), Kansas City, MO (medium solar resource), and
New York, NY (low solar resource).
19
In 2019 USD, the 2020 SETO target is $0.106/kWh, and the residential LCOE in a medium resource area (without
the ITC) is $0.509/kWh in 2010 and $0.135/kWh in 2020; see Appendix B. Progress toward the SETO target is
calculated as follows: (0.509 – 0.135)/(0.509 – 0.106) = 93%.
29
4 Commercial PV Model
This section describes our commercial PV model’s structure, inputs, and assumptions (Section
4.1) and output (4.2) as well as trends in historical PV price (4.3) and LCOE (4.4).
4.1 Commercial Model Structure, Inputs, and Assumptions

We model both a 200-kW, 1,000-volt DC (VDC), commercial-scale flat-roof system using a
ballasted racking solution on a membrane roof, 20 and a 500-kW, 1,000-VDC commercial-scale
fixed-tilt ground-mount system using driven-pile foundations; the ground-mount system is larger
because U.S. ground-mount systems are larger than rooftop systems on average. Owing to the
adoption of the 2017 and 2020 NEC in many states, three-phase string inverter, power optimizer,
and microinverter options are each modeled individually for the commercial rooftop model, and
the “mixed” case applies their market shares (45%, 39%, and 16%, respectively) as weightings.
Because the 2017 NEC only requires rapid shutdown at the module level for rooftop
applications, the commercial ground-mount system only models three-phase string inverters.
Both models use monocrystalline 19.5%-efficient modules from a Tier 1 supplier.
We also model a range of system sizes, from 100 kW to 2 MW. Figure 20 presents a schematic
of our commercial-scale system cost model. Table 5 presents the detailed modeling inputs and
assumptions. We separate our cost estimate into EPC and project-development functions.
Although some firms engage in both activities in an integrated manner, and potentially achieve
lower cost and pricing by reducing the total margin across functions, we believe the distinction
can help separate and highlight the specific cost trends and drivers associated with each function.
20
A penetrating PV mounting system can have higher energy yield (in kilowatt-hours per kilowatt) than a ballasted
racking solution owing to wider tilt-angle range allowance. However, we do not model this system type, because its
market share has declined owing to the additional flashing and sealing work required, roof warranty issues, and the
difficulty of replacing such systems.
30
Figure 20. Commercial PV: Model structure
SG&A = selling, general, and administrative
Table 5. Commercial PV: Modeling Inputs and Assumptions

System size 200 kW (rooftop) and Average installed size Barbose and
500 kW (ground- per system Darghouth 2019
mount); range (100
kW–2 MW)
Module 19.5% Average monocrystalline CA NEM 2020
efficiency module efficiency
Module price $0.41/WDC Ex-factory gate (first Wood Mackenzie
buyer) ASP, Tier 1 and SEIA 2020
monocrystalline modules
Inverter price Three-phase string Ex-factory gate prices Wood Mackenzie 2020;
inverter: $0.07/WDC (first buyer) ASP, Tier 1 Wood Mackenzie and
DC power optimizer inverters SEIA 2020
three-phase string
inverter: $0.12/WDC
(rooftop only)
Microinverter:
$0.29/WDC (rooftop
only)
31
Structural $0.11–$0.17/WDC; Ex-factory gate prices; MEPS 2019;
components assumes national flat-roof ballasted model assumptions;
(racking) average wind and snow racking system or fixed- NREL 2019
loadinga; varies by tilt ground-mount racking
racking type (ground- system
mount versus rooftop
ballasted)
Electrical $0.13–$0.24/WDC Conductors, conduit and Model assumptions;
components fittings, transition boxes, NREL 2020; RSMeans 2017
switchgear, panel
boards, and other parts
EPC 13% Costs and fees NREL 2020
overhead associated with EPC
(percentage overhead, inventory,
of equipment shipping, and handling
costs)
Sales tax National average: 5% Sales tax on equipment RSMeans 2017
costs
Direct Electrician: $27.47 per Modeled labor rate BLS 2019; NREL 2020
installation hour assumes national
labor Laborer: $18.17 average nonunionized
per hour labor rates
Burden rates Total nationwide Workers compensation, RSMeans 2017

(percentage average: 18% federal and state
of direct unemployment
labor) insurance, FICA,
builders’ risk, public
liability
PII $0.11/WDC For construction permits NREL 2020
fee, interconnection
study fees for existing
substation, testing, and
commissioning
Developer $0.30–$0.36/W Includes overhead Model assumptions;
overhead Varies by system size expenses such as NREL 2020
(30% developer payroll, facilities, travel,
overhead) legal fees,
administrative, business
development, finance,
and other corporate
functions
Contingency 4% Estimated as markup on NREL 2020
EPC cost; value
represents actual cost
overruns above
estimated cost
32
Profit 7% Applies a fixed NREL 2020
percentage margin to all
costs, including
hardware, installation
labor, EPC overhead,
and developer overhead
a Racking companies currently meet the national standard, so there is not as much differentiation by state
in the market within rooftop systems. The ground-mount racking system requires more material, equipment,
and labor compared than the ballasted racking system. However, installation of ground-mount PV systems
at utility scale helps reduce the BOS cost of these systems owing to economies of scale.
4.2 Commercial Model Output

Figure 21 presents the U.S. national benchmarks from our commercial PV models. We model
different system sizes because of the wide scope of the commercial sector, which comprises a
diverse customer base occupying a variety of building and property sizes. Economies of scale—
driven by hardware, labor, and related markups—are evident here. As system sizes increase, the
per-watt cost to build systems decreases. As shown in Figure 21 and Figure 22, commercial
rooftop applications have lower costs than commercial ground-mount systems for several smaller
system sizes. However, the difference in price decreases as system size increases, and ground-
mount systems have lower costs for system sizes of 1 and 2 MW. Compared with rooftop
systems, ground-mount applications have higher material, equipment, and labor costs associated
with pile-driven mounting. As PV system size increases, the per-watt cost of pile-driven
mounting is significantly reduced through economies of scale. Ground-mount commercial PV
systems also benefit from lower inverter costs owing to the rapid shutdown requirements for
commercial rooftop systems.
Figure 23 and Figure 24 show sensitivity analyses for the 200-kW rooftop system and 500-kW
ground-mount system, with cost categories that vary by location and hardware specification. For
the rooftop system, inverter type has the largest impact on installed system cost. For the ground-
mount system, material location factor and equipment location factor have the largest impacts.
33
Figure 21. Q1 2020 U.S. benchmark: Commercial rooftop PV system cost (2019 USD/WDC)
34
Figure 22. Q1 2020 U.S. benchmark: Commercial ground-mount PV system cost (2019 USD/WDC)
35
Figure 23. Sensitivity analysis for the Q1 2020 benchmark: 200-kW rooftop commercial PV system
cost (2019 USD/WDC)
Figure 24. Sensitivity analysis for the Q1 2020 benchmark: 500-kW commercial ground-mount PV
system cost (2019 USD/WDC)
36
4.3 Commercial Rooftop PV Price Benchmark Historical Trends
The current version of our commercial cost model makes a few significant changes from the
version used in our Q1 2018 benchmark report (Fu, Feldman, and Margolis 2018). To better
distinguish the historical cost trends from the changes to our cost models, we also calculate Q1
2019 and Q1 2020 PV benchmarks using the Q1 2018 version. Appendix A provides a detailed
discussion of the changes made to the models between the previous report (Fu, Feldman, and
Margolis 2018) and this year’s report. Figure 25 summarizes the reduction in commercial PV
system cost benchmarks between 2010 and 2020. The “Additional Costs from Model Updates”
category represents the difference between modeled results calculated using the current model
versus the previous model. Using the previous cost model, the Q1 2019 and Q1 2020
benchmarks are calculated to be $1.71/WDC and $1.64/WDC, respectively. Figure 25 shows a
69% reduction in commercial PV system cost benchmarks between 2010 and 2020. 21
Approximately 78% of that reduction can be attributed to total hardware costs (module, inverter,
and hardware BOS), and module prices dropped 85% over that period. The final 22% is
attributable to labor and soft costs, including PII, sales tax, overhead, and net profit. From 2019
to 2020, there was a 2.4% reduction in the commercial rooftop PV system cost benchmark,
largely driven by reductions in inverter and BOS hardware costs.
Figure 25. NREL commercial rooftop PV system cost benchmark summary (inflation-adjusted),
2010–2020
we calculate Q1 2019 and Q1 2020 PV benchmarks using the Q1 2018 version of the commercial rooftop PV model.
The “Additional Costs from Model Updates” category represents the difference between modeled results. Using the
previous costs model, the Q1 2019 and Q1 2020 benchmarks are calculated to be $1.71/WDC and $1.64/WDC,
respectively.
21
Each year’s PV system cost benchmark corresponds to the NREL benchmark calculated in Q4 of the previous
year or Q1 of the current year (e.g., 2010 = Q4 2009; 2017 = Q1 2017).
37
For the same reasons described in Section 3.7.1, we compare commercial rooftop system pricing
using monocrystalline and multicrystalline PV modules. Figure 26 compares Q1 2019 system
pricing between commercial rooftop systems using the different module types, and it shows the
change in price of a commercial rooftop PV system using monocrystalline PV modules between
Q1 2019 and Q1 2020.
Figure 26. Q1 2019 cost for a commercial rooftop multicrystalline PV system and Q1 2019 and
Q1 2020 costs for a commercial rooftop monocrystalline PV system
As shown in Figure 26, in Q1 2019 there was a $0.06/W system price premium for using
monocrystalline PV modules over multicrystalline PV modules in commercial rooftop PV
systems. The system cost reductions achieved by increased monocrystalline module efficiency
were counterbalanced by the higher module price. Commercial rooftop PV systems using
monocrystalline modules achieved a $0.04/W (2.4%) reduction in price from Q1 2019 to Q1
2020.
4.4 Commercial PV LCOE Historical Trends

Assumptions for the commercial PV LCOE benchmarks from 2010 to 2020 are summarized in
Table 6. In addition to the 69% reduction in installed cost for commercial rooftop PV from 2010
to 2020, O&M costs declined 46%, annual degradation declined 30%, equity discount rate
declined 32%, debt interest rate declined 27%, and debt fraction increased 57%.
38
Table 6. Commercial PV: LCOE Assumptions, 2010–2020 (2019 USD/WDC)
2010 2011 2012 2013 2014 2015 2016 2017 2018 2019 2020
Rooftop (200 kW)
Installed cost ($/W) 5.57 5.18 3.57 2.90 2.89 2.40 2.29 1.94 1.77 1.76 1.72
O&M expenses ($/kW-yr) 35 32 29 26 23 20 19 19 18 19 19
Ground-Mount (500 kW)
Installed cost ($/W) — — — — — — — — — — 1.72
Inverter loading ratio — — — — — — — — — — 1.11
Annual degradation (%) — — — — — — — — — — 0.70
O&M expenses ($/kw-yr) — — — — — — — — — — 18.71
Preinverter derate (%) — — — — — — — — — — 90.5
Inverter efficiency (%) — — — — — — — — — — 98.0
Financing Assumptions
Inflation rate (%) 2.5 2.5 2.5 2.5 2.5 2.5 2.5 2.5 2.5 2.5 2.5
Market Case
Debt fraction (%) 34.2 35.2 36.1 37.1 38.1 39.0 40.0 40.0 40.0 53.8 53.8
Steady-State financing
Debt interest rate (%) — — — — — — — — — — 5.0
Debt fraction (%) — — — — — — — — — — 71.8
All 2010–2018 data are from Fu, Feldman, and Margolis (2018), and they are adjusted for inflation. Commercial PV system LCOE assumes:
(1) System lifetime of 30 years
39
(2) Federal tax rate of 21%
(3) Sate tax rate of 6%
(4) MACRS depreciation schedule
(6) A working capital and debt service reserve account for six months of operating costs and debt payments (earning an interest rate of 1.75%)
(7) Six-month construction loan, with an interest rate of 4% and a fee of 1% of the cost of the system
(8) Module tilt angle of 10 degrees and an azimuth of 180 degrees
(9) Debt with a term of 18 years
(10) $1.1 million of upfront financial transaction costs for a $100 million TPO transaction of a pool of commercial projects
(11) 2019 and 2020 financial assumptions from Feldman, Bolinger, and Schwabe (2020).
40
Using these assumptions, we calculate the commercial PV LCOE—with and without the 30%
federal ITC—for a high solar resource (Phoenix, CF: 20.4%), medium solar resource (Kansas
City, CF: 16.4%), and low solar resource (New York City, CF: 15.3%) (Figure 27). From 2010
to 2020, commercial rooftop PV LCOE declined 77% (3% between 2019 and 2020), resulting in
an unsubsidized LCOE of $0.08–$0.10/kWh ($0.05–$0.07/kWh when including the federal
ITC). This reduction is 97% toward achieving SETO’s 2020 commercial PV LCOE goal from
the commercial system price when the goal was announced in 2010. 22 Commercial ground-
mount PV systems, which we began benchmarking this year, are calculated to have a 2020
unsubsidized LCOE of $0.07–$0.09/kWh ($0.05–$0.06/kWh when including the federal ITC).
We also calculate PV LCOE without the ITC using steady-state financing assumptions. Under
these assumptions, the commercial rooftop PV LCOE ranges from $0.07–$0.10/kWh, and the
commercial ground-mount PV LCOE ranges from $0.07–$0.10/kWh in Q1 2020.
Figure 27. LCOE for commercial rooftop PV systems, by region, with and without ITC, 2010–2020
We updated our methods and model this year; 2019 and 2020 LCOEs are higher than they would have been using
previous models. Appendix A provides a detailed discussion of the changes made to the models between the
previous version (Fu, Feldman, and Margolis 2018) and this year’s version. LCOE is calculated for each scenario
under a range of CFs, but all other values remain the same.
22
In 2019 USD, the 2020 SETO target is $0.082/kWh, and the commercial LCOE in Kansas City (without the ITC)
is $0.397/kWh in 2010 and $0.093/kWh in 2020; see Appendix B. Progress toward the SETO target is calculated as
follows: (0.397 – 0.093)/(0.397 – 0.082) = 97%.
41
5 Utility-Scale PV Model
This section describes our utility-scale PV model’s structure, inputs, and assumptions (Section
5.1) and output (5.2) as well as trends in historical PV price (5.3) and LCOE (5.4).
5.1 Utility-Scale Model Structure, Inputs, and Assumptions

We model a baseline 100-MW, 1,500-VDC utility-scale system using 72-cell, monocrystalline
19.5%-efficient modules from a Tier 1 supplier and three-phase central inverters. We model
both fixed-tilt and one-axis tracking on ground-mounted racking systems using driven-pile
foundations. In addition, we separate our cost estimates into EPC and project-development
functions. Although some firms engage in both activities in an integrated manner, we believe the
distinction can help separate and highlight the specific cost trends and drivers associated with
each function. We also model a range of system sizes, from 5 MW to 100 MW. Figure 28
presents a schematic of our utility-scale system cost model, and Table 7 details its assumptions
and inputs.
Figure 28. Utility-scale PV: Model structure
42
Table 7. Utility-Scale PV: Modeling Inputs and Assumptions

System size 100 MW; range: A large utility-scale system capacity Model assumption
5 MW–100 MW
Module 19.5% Average monocrystalline CA NEM 2020
efficiency module efficiency
Module price $0.41/WDC Ex-factory gate (first buyer) price, Wood Mackenzie
Tier 1 monocrystalline modules and SEIA 2020;
NREL 2020
Inverter price $0.05/WDC (fixed- Ex-factory gate (first buyer) price, Wood Mackenzie and
tilt) Tier 1 inverters SEIA 2020; Bolinger,
$0.05/WDC (one- DC-to-AC ratio = 1.37 for fixed-tilt Seel, and Robson 2019
axis tracker) and 1.34 for one-axis tracker
Structural $0.12/WDC for a Fixed-tilt racking or one-axis MEPS 2019;
components 100-MW system tracking system model assumptions;
(racking) NREL 2020
Electrical $0.07–$0.13/WDC Model was upgraded to a 1,500-VDC Model assumptions;
components Varies by system system that includes conductors, NREL 2020;
size conduit and fittings, transition boxes, RSMeans 2017
switchgear, panel boards, onsite
transmission, and other electrical
connections
EPC 8.67%–13% for Costs associated with EPC SG&A, NREL 2020
overhead equipment and warehousing, shipping, and logistics
(percentage material (except
of equipment for transmission
costs) line costs); 23%–
69% for labor
costs; varies by
system size and
labor activity
Sales tax National Sales tax on equipment costs RSMeans 2017
average: 5%
Direct Electrician: Modeled labor rate assumes BLS 2019; NREL 2020
installation $27.47 per hour national average nonunionized labor
labor Laborer: $18.17
per hour
Burden rates Total nationwide Workers compensation, federal and RSMeans 2017
(percentage average: 18% state unemployment insurance,
of direct FICA, builders’ risk, public liability
labor)
PII $0.03–$0.07/WDC For construction permits fee, NREL 2020
Varies by system interconnection, testing, and
size commissioning
43
Transmission $0.00–$0.02/WDC System size < 10 MW uses 0 miles Model assumptions;
line Varies by system for gen-tie line NREL 2020
(gen-tie line) size System size > 200 MW uses five
miles for gen-tie line
System size = 10–200 MW uses
linear interpolation
Developer 2%–12% Includes overhead expenses such Model assumptions;
overhead Varies by system as payroll, facilities, travel, legal NREL 2020
size (100 MW fees, administrative, business
uses 2%; 5 MW development, finance, and other
uses 12%) corporate functions
Contingency 3% Estimated as markup on EPC cost NREL 2020

Profit 5%–8% Applies a percentage margin to all NREL 2020
Varies by system costs including hardware,
size (100 MW installation labor, EPC overhead,
uses 5%; 5 MW and developer overhead
uses 8%)
Figure 29 shows the percentage of U.S. utility-scale PV systems using tracking systems for
2010–2019. Although the data include one-axis and dual-axis tracking systems in the same
“tracking” category, there are many more one-axis trackers than dual-axis trackers (EIA 2020).
Cumulative tracking system installation reached 65% in 2019, with 82% of new installations in
2019 having tracking. Based on these trends, we use fixed-tilt systems to calculate LCOE
benchmarks from 2010 to 2015, and we use one-axis tracking systems for 2016 to 2020 (see
Section 5.4).
Figure 29. Percentage of U.S. utility-scale PV systems using tracking systems, 2010–2019
Source: EIA (2020)
44
5.2 Utility-Scale Model Output
Figure 30 shows the U.S. national benchmark (EPC + developer) for fixed-tilt and one-axis
tracker systems, using nonunionized labor. Figure 31 shows a sensitivity analysis for the one-
axis system benchmark, with cost categories that vary by location and hardware specification.
Equipment location factor has the largest impact on installed system cost.
Figure 30. Q1 2020 U.S. benchmark: Utility-scale PV total cost (EPC + developer), 2019 USD/WDC
Figure 31. Sensitivity analysis for the Q1 2020 benchmark: 100-MW one-axis utility-scale PV
system cost (2019 USD/WDC)
45
5.3 Utility-Scale PV Price Benchmark Historical Trends
Figure 32 shows the 80% (fixed-tilt) and 82% (one-axis tracking) reductions in utility-scale PV
system cost benchmarks between 2010 and 2020. 23 Approximately 70% (fixed-tilt) and 64%
(one-axis tracking) of those reductions can be attributed to total hardware costs, with module
prices dropping 85% over that period. An additional 11% (fixed-tilt) to 12% (one-axis tracking)
reduction can be attributed to labor, which dropped over that period. For previous editions of this
report, we assumed a land acquisition cost of $0.03/W. Based on Wiser et al. (2020), which
stated that most utility-scale PV projects do not own the land on which the PV system is placed,
we have reclassified land costs from an upfront capital expenditure (land acquisition) to an
operating expenditure (lease payments) for 2019 and 2020. Therefore, approximately 1% of the
reduction in cost is attributed to the reclassification of land costs. The final 20% (fixed-tilt) and
25% (one-axis tracker) is attributable to other soft costs, including PII, sales tax, overhead, and
net profit.
Figure 32. NREL utility-scale PV system cost benchmark summary (inflation-adjusted), 2010–2020
we calculate Q1 2019 and Q1 2020 PV benchmarks using the Q1 2018 versions of the utility-scale PV model. The
“Additional Costs from Model Updates” category represents the difference between modeled results. Using the
previous costs model, the Q1 2019 and Q1 2020 benchmarks are calculated to be $0.94/WDC and $0.89/WDC (fixed-
tilt) as well as $1.01/WDC and $0.96/WDC (one-axis), respectively.
From 2019 to 2020, overall there was a 1% reduction in the cost benchmarks for both utility-
scale PV systems (fixed-tilt and one-axis tracking).
23
Each year’s PV system cost benchmark corresponds to the NREL benchmark calculted in Q4 of the previous year
or Q1 of the current year (e.g., 2010 = Q4 2009; 2017 = Q1 2017).
46
For the same reasons described in Section 3.7.1, we compare utility-scale PV system pricing
using monocrystalline and multicrystalline PV modules. Figure 33 compares Q1 2019 system
pricing between fixed-tilt and one-axis tracking utility-scale PV systems using the different
module types, and it shows the change in price of fixed-tilt and one-axis tracking utility-scale PV
systems using monocrystalline PV modules between Q1 2019 and Q1 2020.
Figure 33. Q1 2019 costs for utility-scale multicrystalline PV systems and Q1 2019 and Q1 2020
costs for utility-scale monocrystalline PV systems
As shown in Figure 33, in Q1 2019 there was a $0.05/WDC system price premium for using
monocrystalline PV modules over multicrystalline PV modules in utility-scale PV systems.
The system cost reductions achieved by increased monocrystalline module efficiency were
counterbalanced by the higher module price. The price of utility-scale PV systems using
monocrystalline modules decreased by $0.01/WDC from Q1 2019 to Q1 2020.
5.4 Utility-Scale PV LCOE Historical Trends

Assumptions for the utility-scale PV LCOE benchmarks from 2010 to 2020 are summarized in
Table 8. In addition to the 82% reduction in the installed cost of utility-scale (one-axis) systems
from 2010 to 2020, O&M costs declined 40%, annual degradation declined 30%, equity discount
rate declined 31%, debt interest rate declined 27%, and debt fraction increased 52%.
47
Using these assumptions, we calculate the utility-scale PV LCOE—with and without the 30%
federal ITC—for a high solar resource (Phoenix, CF: 21.6% for fixed-tilt and 25.2% for one-
axis), medium solar resource (Kansas City, CF: 17.3% for fixed-tilt and 19.6% for one-axis), and
low solar resource (New York City, CF: 16.2% for fixed-tilt and 18.1% for one-axis) (Figure
34). We use fixed-tilt systems for LCOE benchmarks from 2010 to 2015 and then switch to one-
axis tracking systems from 2016 to 2020 to reflect the market share change in Figure 29.
From 2010 to 2020, utility-scale PV LCOE declined 83% (0% between 2019 and 2020),
resulting in an unsubsidized LCOE of $0.04–$0.05/kWh ($0.025–$0.035/kWh when including
the federal ITC). This reduction signifies the achievement of SETO’s 2020 utility-scale PV
goal. 24 We also calculate PV LCOE without the ITC using steady-state financing assumptions.
Under these assumptions, utility-scale (one-axis and fixed-tilt) PV LCOE ranges from $0.04–
$0.05/kWh in Q1 2020.
24
The 2020 utility-scale goal is not adjusted for inflation, because wholesale electricity prices were relatively flat,
and in some cases declined, from 2010 to 2020. The goal is shown in Appendix B along with the detailed utility-
scale LCOE values over time.
48
Table 8. One-Axis Tracker and Fixed-Tilt Utility-Scale PV: LCOE Assumptions, 2010–2020 (2019 USD/WDC)
2010 2011 2012 2013 2014 2015 2016 2017 2018 2019 2020
One-Axis Tracker
Installed cost ($/W) 5.66 4.79 3.29 2.50 2.25 2.08 1.63 1.16 1.16 1.02 1.01
O&M expenses ($/kW-yr) 29 28 26 25 24 22 22 21 15 17 17
Fixed-Tilt
Installed cost ($/W) 4.75 4.08 2.77 2.13 1.97 1.93 1.53 1.08 1.08 0.95 0.94
O&M expenses ($/kW-yr) 29 27 25 23 21 19 19 18 13 16 16
Financing Assumptions
Inflation rate (%) 2.5 2.5 2.5 2.5 2.5 2.5 2.5 2.5 2.5 2.5 2.5
Market Case
Debt fraction (%) 34.2 35.2 36.1 37.1 38.1 39.0 40.0 40.0 40.0 51.9 51.9
Steady-State Financing
Debt interest rate (%) — — — — — — — — — — 5.0
Debt fraction (%) — — — — — — — — — — 71.8
All 2010–2018 data are from Fu, Feldman, and Margolis (2018), and they are adjusted for inflation. Utility-scale PV system LCOEs assume:
(1) System lifetime of 30 years
49
(2) Federal tax rate of 21%
(3) State tax rate of 6%
(4) MACRS depreciation schedule
(6) A working capital and debt service reserve account for six months of operating costs and debt payments (earning interest of 1.75%)
(7) Six-month construction loan with an interest rate of 4% and a fee of 1% of the cost of the system
(8) System size of 100 MW
(9) Debt with a term of 18 years
(10) $1.1 million of upfront financial transaction costs
(11) 2019 and 2020 financial assumptions from Feldman, Bolinger, and Schwabe (2020).
50
*
Figure 34. LCOE for utility-scale PV systems, by region, with and without ITC, 2010–2020 (fixed-tilt
from 2010 to 2015, one-axis tracking from 2016 to 2020)
We updated our methods and model this year; 2019 and 2020 LCOEs are higher than they would have been
using previous models. Appendix A provides a detailed discussion of the changes made to the models
between the previous version (Fu, Feldman, and Margolis 2018) and this year’s version. LCOE is calculated
for each scenario under a range of CFs, but all other values remain the same.
51
6 Residential Storage and PV-plus-Storage Model
To analyze component costs and system prices for PV-plus-storage installed in Q1 2020, we
adapt NREL’s component- and system-level modeling approach for standalone PV. For this
report, system configuration refers to four characteristics that determine a PV-plus-storage
system’s functionality:
• PV system capacity (kW)

• Battery energy capacity (kWh)
• Battery power capacity (kW)
• Whether the battery is DC- or AC-coupled. 25
Customer preference for specific characteristics is based on several factors, including cost, load
profile, and planned use of the system for load shifting (storing energy in one period for use in a
later period). In general, customers who have loads with high peaks of short duration may desire
a high-power (high-kW) battery capable of meeting the high peak. Customers who have flatter
loads with lower peaks of longer duration may prefer a high-energy (high-kWh) battery capable
of longer-duration energy discharge.
A PV array, a battery, and a battery-based inverter are the fundamental components of every PV-
plus-storage system. Additional component requirements are determined by whether the system
is DC- or AC-coupled 26: a DC-coupled system often requires a charge controller to step down
the PV output voltage to a level that is safe for the battery, whereas an AC-coupled system
requires a grid-tied inverter to feed PV output directly to the customer’s load or the grid. 27 For a
detailed discussion of the differences and considerations related to DC- versus AC-coupled
system configurations, see Ardani et al. (2017).
Based on our industry interviews, increasing numbers of end users are willing to pay a premium
for larger, more-resilient PV-plus-storage systems with enhanced back-up power capabilities,
owing to the increased occurrence of superstorms and natural disasters. This decision may not
always be driven by economics, given the higher costs of PV-plus-storage systems today;
however, consumer-adoption motivations extend beyond economics to concerns about security,
safety, and resiliency (EuPD Research and Greentech Media 2016).
When considering PV-plus-storage for enhanced back-up power, optimal system configurations
and technology choices are determined by the system application. We model a larger, more-
resilient PV-plus-storage system (7-kW PV plus 5-kW/20-kWh storage) designed for daily PV
self-consumption and enhanced back-up capabilities. The average U.S. home uses about 30 kWh
of electricity each day, with large variations based on location and season. Assuming an average
household could cut its electricity use by two thirds in an emergency, it would need to meet
25
NREL’s modeled DC-coupled system includes a single dual-function inverter that is tied to both the PV array and
the battery. In our AC-coupled system, to charge a battery, PV power is first converted (DC to AC) through a grid-
tied inverter and then converted (AC to DC) through a battery-based inverter.
26
Our discussion is simplified to explain the basic technical differences between AC- and DC-coupled systems.
However, the decision to use AC- or DC-coupling might also be driven by non-technical factors such as policy,
contractual obligations, and economics.
27
Some Li-ion battery packs have built-in safety controls, such as those integrated in a battery management system,
but some do not. For consistency, our model assumes there is a dedicated charge controller.
52
10 kWh of demand each day. At this rate, our more-resilient system could provide back-up
electricity for an average of 35 hours without PV recharging. In contrast, our less-resilient
battery system (3-kW/6-kWh storage) could only provide back-up electricity for an average of
10 hours without PV recharging. 28 If 30% of the PV system’s average output were available to
charge the battery each day, the more-resilient battery system could provide back-up electricity
for about four days, compared with about one day for the less-resilient battery system. 29 The
higher power of the more-resilient battery system (5 kW)—compared with the less-resilient
battery system (3 kW)—would also enable the more-resilient battery system to meet higher peak
electricity demands during a grid outage (e.g., to run a refrigerator).
Sections 6.1 and 6.2 present the residential storage and PV-plus-storage cost models, Section 6.3
shows the model outputs, Section 6.4 compares PV-plus-storage benchmark trends over time,
and Section 6.5 benchmarks the levelized cost for a residential PV-plus-storage system.
6.1 Residential Li-Ion Standalone Storage Cost Model

The residential storage market is predominantly composed of fully integrated storage kits, which
include Li-ion battery packs, inverters, field wiring, disconnect, and casing. Although this
equipment is sold as one product, we model these components separately to compare costs across
storage kit sizes and configurations. Table 9 presents the detailed modeling inputs and
assumptions for the residential standalone storage costs.
Table 9. Residential Storage-Only Modeling Inputs and Assumptions
Category Modeled Value Description

System size 3-kW/6-kWh storage Less-resilient system
5-kW/20-kWh storage More-resilient system
Battery pack $253/kWh Battery pack only

cost
Battery-based $174/kWh 6-kW, 48-V bidirectional inverter (less

inverter cost resilient)
8-kW, 48-V bidirectional inverter (more
resilient)
Electrical • $1,830 (DC-coupled) Revenue-grade meter, communications

BOS cost • $1,520 (AC-coupled) device, AC main panel, DC disconnect,
maximum power point tracking, charge
Assumes higher electrical BOS costs
controller, subpanel (breaker box) for
for DC-coupled systems that are due
critical load, conduit, wiring, DC cable
to the need for a charge controller
28
These calculations assume 80% depth of discharge for the batteries and 90% inverter efficiency. Even in these
simplified scenarios, the actual amount of time that the system could provide back-up electricity would depend
on the battery’s charge level and the time of day at the time of the outage as well as the home’s load profile.
29
This is based on 2016 results using NREL’s PVWatts for a 5.6-kW PV system located in Denver (5.6 kW was
the calculated system size for a 22-module PV system in 2016, based on average module efficiency). This modeled
system generates 8,179 kWh per year (average, 22.4 kWh per day). Thus, we assume this same 5.6-kW PV array
will generate an average of 6.7 kWh per day when only 30% of the total PV resource is available owing to severe
weather conditions. Since 2016, our residential PV benchmark has increased in capacity, but battery storage capacity
has remained flat. The storage capacities benchmarked here conform with what we observed in the marketplace.
53
Supply-chain 5% of cost of equipment Includes costs of inventory, shipping,
costs and handling of equipment
Sales tax 5.1% (national average) Sales tax on the equipment
Installation labor Electrician: $27.47 per hour Assumes national average pricing
cost Laborer: $18.17 per hour
AC systems require more hours of
work to integrate with an existing
inverter and monitoring system
Engineering fee $99 Engineering design and professional
engineer-stamped calculations and
drawings
PII $297 permit fee 20–32 hours (DC-coupled/AC-coupled)
$594–$951 in labor of commissioning and interconnection
labor, and permit fee
Sales and $0.61/WDC 20 hours more time for DC system, and
marketing 32 hours more for AC system, per
(customer closed sale, associated with selling a
acquisition) storage systems versus selling a PV
system
Overhead $0.28/WDC Rent, building, equipment, staff
(general and expenses not directly tied to PII,
administrative) customer acquisition, or direct
installation labor
Profit (%) 17% Fixed percentage margin applied to all
direct costs including hardware,
installation labor, direct sales and
marketing, design, installation, and
permitting fees
As demonstrated in Figure 35, the kit for a 3-kW/6-kWh storage system costs approximately
$4,200–$4,600, with a total installed cost of $11,823 (DC-coupled) to $12,287 (AC-coupled).
The kit for a 5-kW/20-kWh storage system costs approximately $10,400–$10,800, with a total
installed cost of $21,471 (DC-coupled) to $22,041 (AC-coupled). 30
30
We assume all batteries are installed inside the home. Installation of batteries outside would require additional
BOS hardware, such as a concrete pad and associated container. Such additional BOS hardware would add to the
benchmarked price of our modeled systems.
54
Figure 35. Installed cost of residential storage only
6.2 Residential PV-plus-Storage System Cost Model

We model a 7-kW PV system coupled with a 3-kW/6-kWh or 5-kW/20-kWh storage system,
using the same PV assumptions we used with our standalone PV system. Figure 36 provides a
schematic of typical DC- and AC-coupled PV systems with battery back-up. Table 3, Table 9,
and Table 10 present modeling inputs and assumptions.
55
Figure 36. Modeled DC- and AC-coupled system configurations
Figure is simplified for illustrative purposes.
Table 10. Changes to Residential PV and Storage Models When PV and Storage Are Combined

Electrical BOS 90% of the combined BOS costs for PV Duplicative parts are removed.
and battery standalone systems
Installation 90% of the combined BOS costs for PV Duplicative work is removed.
labor and battery standalone systems
Sales and 20 hours more time for DC system, and Additional explanation, calculations, and
marketing 32 hours more for AC system, per a lower close rate, and the AC system
closed sale, associated with selling a requires more customer site assessment.
PV system with storage
6.3 Residential Model Output

Figure 37 compares cost and price components for a standalone PV system as well as PV-plus-
storage systems with less-resilient (3-kW/6-kWh) and more-resilient (5-kW/20-kWh) battery
systems. With DC-coupling, the price of the more-resilient system is $35,591, which is $9,438
(36%) more than the price of the DC-coupled less-resilient system. With AC-coupling, the price
of the more-resilient battery system is $37,909, which is $9,538 (34%) more than the price of the
DC-coupled less-resilient battery system. The premium is due to the more-resilient systems’
higher battery, inverter, BOS, and labor costs plus indirect costs (profit, sales tax, and supply-
chain costs).
56
Figure 37. Modeled total installed cost and price components for residential PV-plus-storage
systems, less-resilient versus more-resilient battery case (2019 USD)
6.4 Residential PV-plus-Storage Price Benchmark Historical Trends

Figure 38 shows the 11% and 25% reductions in residential PV-plus-storage benchmarks
between 2016 (Ardani et al. 2017) and 2020, for the AC-coupled less-resilient and more-resilient
cases, respectively. 31 The reduction is due to a 26% reduction in PV module costs, 38% and 44%
reduction in costs associated with the storage system kit (including a bidirectional inverter), a
16% reduction in hardware BOS, and a 34% and 65% reduction in labor costs. These cost
reductions are partially offset by 18% and 10% increases in other soft costs (including PII, sales
tax, overhead, and net profit). Other soft costs increased between 2016 and 2020 because of a
change in methodology and because the rated capacity of the 22-module system increased from
5.6 kW to 7.0 kW between 2016 and 2020. From 2019 to 2020, the residential PV-plus-storage
system cost benchmarks decreased by 5%, mostly owing to lower storage system kit prices.
31
Each year’s PV system cost benchmark corresponds to the NREL benchmark calculated in Q1 of the current year
(e.g., 2016 = Q1 2016). Figure 38 only shows AC-coupled system costs to more easily demonstrate the historical
trends; the cost of DC-coupled systems follows the same historical trends.
57
Figure 38. Residential PV-plus-storage system cost benchmark summary,
2016, 2019, and 2020
The 2016 benchmarks differ from those originally published, because values are adjusted for inflation.
6.5 Residential Levelized Cost of Solar-plus-Storage

For this year’s benchmark report, we derive a formula for the levelized cost of solar-plus-storage
(LCOSS) to contextualize our upfront PV-plus-storage system benchmarks and better represent
the total cost of operating a PV-plus-storage system, on a per-kWh basis. BNEF (2019c) and
Lazard (2018) performed similar LCOSS calculations. None of these LCOSS calculations,
including the one in this report, attempts to value the electricity generated by these systems or
the different ways they may operate. Storage value calculations for residential applications
require integrating storage dispatch into building load, retail rates, and requirements. In addition,
residential storage systems perform various functions—such as shifting load, reducing peak
demand, and providing emergency power—depending on location, regulations, and customer
preferences; our analysis represents the load-shifting use case. For a detailed discussion of
residential storage value, see Fitzgerald et al. (2015), DiOrio et al. (2015), and Darghouth et al.
(2019). Similar to LCOE, LCOSS does not focus on value but rather can help track
improvements to all costs associated with residential PV-plus-storage systems over time (as
opposed to just upfront costs), and the metric can provide limited comparisons with other
dispatchable electricity generation technologies (e.g., PV-plus-generator systems). Table 11 lists
our model inputs and assumptions for calculating residential LCOSS.
58
Table 11. Residential LCOSS Inputs and Assumptions
Model Component Model Input Description

7-kW PV plus 3-kW/6-kWh
System size
storage system
2020 residential PV-plus-storage
Initial investment $28,371
benchmark, AC-coupled
20% of the batteries are replaced after
First follow-on 10 years due to battery capacity dropping
investments (inverter, $240 in year 10 20%. We assume costs for battery and
battery replacements) bidirectional inverters drop 20% in the
next 10 years.
Second follow-on 20 years due to battery capacity dropping
investments (inverter, $180 in year 20 20%. We assume costs for battery and
next 20 years.
Real discount rate 3.1% Consistent with LCOE formula
Tax rate 25.7% 21% federal, 6% state
Residual value $0
High resource: 1,892 MWh/MW
Initial annual PV Medium resource: 1,546
system production MWh/MW
Low resource: 1,440 MWh/MW
Percentage of High resource: 25%
generated solar Assumes a 75% discharge per day for a 2-
Medium resource: 31%
electricity fed to hour, 3-kW battery
battery Low resource: 33%
Roundtrip energy
losses from 10%
PV/battery/grid
Roundtrip energy
losses from 8%
grid/battery/grid
Battery charged solely by PV due to ITC
Charging cost $0
considerations
Assumes storage O&M adds $10/kW-yr to
O&M ($/kW/yr) $39
PV costs
Annual PV
0.70%
degradation
Annual electricity
0
purchased from grid
System lifetime 30 years
Inflation 2.5%
59
We use these inputs to calculate LCOSS as follows:
Equation 1. LCOSS formula

𝐿𝐿𝐿𝐿𝐿𝐿𝐿𝐿𝐿𝐿
𝐹𝐹 𝑛𝑛 𝑁𝑁 (𝐷𝐷 + 𝐷𝐷𝐷𝐷)
𝑛𝑛
𝑁𝑁 (𝑂𝑂 + 𝐶𝐶 + 𝐼𝐼)
𝑛𝑛
𝑅𝑅𝑅𝑅 𝑛𝑛 𝑁𝑁 (𝑃𝑃)𝑛𝑛
𝐸𝐸 + − ∑ 𝑛𝑛−1 × (𝑇𝑇) + ∑ 𝑛𝑛−1 × (1 − 𝑇𝑇) − × (1 − 𝑇𝑇) + ∑ 𝑛𝑛−1 × (1 − 𝑇𝑇)
(1 + 𝑅𝑅)𝑛𝑛 (1 + 𝑅𝑅𝑅𝑅)𝑛𝑛 (1 + 𝑅𝑅𝑅𝑅)𝑛𝑛 (1 + 𝑅𝑅)𝑛𝑛 (1 + 𝑅𝑅𝑅𝑅)𝑛𝑛
=
𝑃𝑃 × (1 − 𝐷𝐷𝐷𝐷)𝑛𝑛 𝑃𝑃 × (1 − 𝐷𝐷𝐷𝐷)𝑛𝑛 𝐺𝐺
� ∑𝑁𝑁 × (1 − 𝐵𝐵) + ∑𝑁𝑁 × (𝐵𝐵) × (1 − 𝐿𝐿𝐿𝐿) + ∑𝑁𝑁 𝑛𝑛−1 (1 + 𝑅𝑅)𝑛𝑛 × (1 − 𝐿𝐿𝐿𝐿)� × (1 − T)
𝑛𝑛−1 (1 + 𝑅𝑅)𝑛𝑛 𝑛𝑛−1 (1 + 𝑅𝑅)𝑛𝑛
E = Initial equity investment of solar and storage

I = Debt interest payments
P = Debt principal payments
C = Charging cost
F = Follow-on investments (inverter, battery replacements)
D = Depreciation of solar and storage (which may include depreciation from follow-on investments)
R = Real discount rate
Rn = Nominal discount rate
T = Tax rate
O = O&M
Dr = Degradation of PV
Rv = Residual value
P = Initial annual system production
B = Percentage of generated solar electricity fed to battery
Lp = Roundtrip energy losses from PV-storage-grid
Lg = Roundtrip energy losses from grid-storage-grid
G = Annual electricity purchased from grid. 32
32
If the ITC is claimed, we assume the initial investment is reduced by 30% and depreciation is reduced by 15%. We assume projects can qualify as starting
construction before 2020, allowing them to claim a 30% ITC, instead of the 26% ITC for projects starting construction in 2020.
60
Figure 39 shows the LCOSS for a residential AC-coupled PV (7 kW) plus storage (3 kW/6 kWh,
2-hour duration) system, as well as the LCOE of a 7-kW standalone PV system. LCOSS is
calculated to be $201/MWh without the federal ITC and $124/MWh with the 30% ITC for the
PV-plus-storage system, with a medium resource for PV electricity production. 33 The PV-plus-
storage LCOSS is $74/MWh higher than the standalone-PV LCOE without the ITC, and
$47/MWh higher with a 30% ITC.
Figure 39. U.S. residential LCOSS for an AC-coupled PV (7 kW) plus storage (3 kW/6 kWh, 2-hour
duration) system and LCOE for a 7-kW standalone PV system, Q1 2020
LCOSS is calculated for each scenario with a medium CF (representing Kansas City); LCOSS and LCOE ranges
based on high and low CF assumptions; all other values remain the same.
33
We do not change the inputs and assumptions between the ITC and non-ITC cases, despite the fact that the inputs
in the LCOSS calculation assume the owner of the PV-plus-storage system is operating the plant in such a way that
they can claim the ITC on the storage equipment. In reality, an owner would likely operate a PV-plus-storage system
differently without the ITC.
61
7 Commercial Storage and PV-plus-Storage Model
To analyze component costs and system prices for commercial PV-plus-storage installed in Q1
2020, we adapt NREL’s component- and system-level modeling approach for standalone PV in
the same manner as we did for the residential PV-plus-storage system. This is the first year in
which we analyzed commercial PV-plus-storage, and therefore we have no historical analysis
from which to compare.
Customer preference for specific characteristics is based on several factors, including cost, load
profile, and planned use of the system for load shifting (storing energy in one period for use in a
later period). In general, customers who have loads with high peaks of short duration may desire
a high-power (high-kW) battery capable of meeting the high peak. Customers who have flatter
loads with lower peaks of longer duration may prefer a high-energy (high-kWh) battery capable
of longer-duration energy discharge.
Sections 7.1 and 7.2 present the commercial storage and PV-plus-storage cost models, Section
7.3 shows the model outputs, and Section 7.4 benchmarks the LCOSS for a commercial PV-plus-
storage system
7.1 Commercial Li-Ion Standalone Storage Cost Model

To reduce installation costs, some battery manufacturers may combine Li-ion battery cells,
a battery management system, and the battery inverter in one compact unit (Sonnen Batterie
2018) as an AC battery. However, in this report, we focus on traditional DC batteries typically
configured with the components shown in Figure 40 and Figure 41.
Battery cells → modules → packs → racking

system (DC)
Storage container
(HVAC system, thermal management,
monitors and controls, fire suppression,
switchgear, and energy management system)
Power conversion system

(bidirectional inverter to convert AC to DC for
battery charging and DC to AC for discharging)
Transformer (to step up 480-V inverter output

to 12–66 kV)
Figure 40. Traditional commercial and utility-scale Li-ion battery energy storage components
HVAC = heating, ventilating, and air conditioning
62
Figure 41. Battery system components
Source: 2018 North American Generator Forum/Energy Systems Integration Group Workshop
Table 12 lists our model inputs and assumptions for a commercial energy storage system. We
determine the battery size (600 kWDC) 34 using an inverter loading ratio of 1.3 and a PV/storage
size ratio of 1.67, based on Denholm, Eichman, and Margolis (2017).
Table 12. Commercial Li-ion Energy Storage System: Model Inputs and Assumptions
Model Modeled Value Description Sources

Component
Battery total 600 kWDC Baseline case to match a 1-MW PV system NREL 2020
size
Battery size 2.4 MWh per 40-ft 1 container NREL 2020
per container container
Li-ion battery 0.5 hours: $242/kWh Ex-factory gate (first buyer) prices BNEF 2019b
price 1 hour: $223/kWh
2 hours: $198/kWh
4 hours: $194/kWh
Duration 0.5–4.0 hours Duration determines energy (MWh) NREL 2020
Battery central $0.06/W Ex-factory gate (first buyer) prices Wood Mackenzie
inverter price 2019
Electrical BOS $0.19/W Includes conduit, wiring, DC cable, energy NREL 2020
management system, switchgear,
transformer, and monitor and controls for
each container. Costs impacted by the
number of containers, transformers, and row
spacing
Structural BOS $0.10/W Includes foundation, battery containers, and NREL 2020
inverter house. Costs impactedby the number
of containers, inverters, transformers, and the
spacing between containers
34
For a 1 MW PV system with an inverter loading ratio of 1.3 and PV/storage size ratio of 1.67, maximum
deliverable power at point of interconnection is 1.37 MWAC (1-MW/1.3 + 1 MW/1.67) for AC coupled systems and
770 kWAC (1 MW/1.3) for DC coupled systems.
63
Model Modeled Value Description Sources
Component
Installation Electrician: $27.47 National average modeled labor rate BLS 2019
labor per hour assumes nonunionized labor
Laborer: $18.17
per hour
Sales tax 5% (national Sales tax on the equipment RSMeans 2017
average)
EPC overhead 8.67% for equipment Costs associated with EPC SG&A, NREL 2020
and profit and material; 23%– warehousing, shipping, and logistics
69% for labor costs;
varies by system
size, labor activity,
and location
Developer 6% of total Includes overhead expenses such as payroll, NREL 2020
cost: developer installation cost facilities, travel, legal fees, administrative,
overhead business development, finance, and other
corporate functions
Developer $0.06/W Construction permits fee, interconnection NREL 2020
cost: PII study, interconnection inspection, and
interconnection fee
Developer 4% Estimated as markup on the total EPC cost NREL 2020
cost:
contingency
Developer 5% Applies a percentage margin to all costs NREL 2020
cost: including hardware, installation labor, EPC
EPC/developer overhead, and developer overhead
net profit
We use these inputs to calculate energy storage cost via the following equation 35:
$
𝐸𝐸𝐸𝐸𝐸𝐸𝐸𝐸𝐸𝐸𝐸𝐸 𝑠𝑠𝑠𝑠𝑠𝑠𝑠𝑠𝑠𝑠𝑠𝑠𝑠𝑠 𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑖 𝑐𝑐𝑐𝑐𝑐𝑐𝑐𝑐 � �=
𝑘𝑘𝑘𝑘ℎ
$ 𝑂𝑂𝑂𝑂ℎ𝑒𝑒𝑒𝑒 𝑐𝑐𝑐𝑐𝑐𝑐𝑐𝑐 𝑐𝑐𝑐𝑐𝑐𝑐𝑐𝑐𝑐𝑐𝑐𝑐𝑐𝑐𝑐𝑐𝑐𝑐𝑐𝑐 ($) 𝑠𝑠𝑠𝑠𝑠𝑠ℎ 𝑎𝑎𝑎𝑎 𝑏𝑏𝑏𝑏𝑏𝑏𝑏𝑏𝑏𝑏𝑏𝑏𝑏𝑏 𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑖 𝑎𝑎𝑎𝑎𝑎𝑎 𝑙𝑙𝑙𝑙𝑙𝑙𝑙𝑙𝑙𝑙
𝐵𝐵𝐵𝐵𝐵𝐵𝐵𝐵𝐵𝐵𝐵𝐵𝐵𝐵 𝑐𝑐𝑐𝑐𝑐𝑐𝑐𝑐 � �+
𝑘𝑘𝑘𝑘ℎ 𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆 𝑠𝑠𝑠𝑠𝑠𝑠𝑠𝑠𝑠𝑠𝑠𝑠 𝑠𝑠𝑠𝑠𝑠𝑠𝑠𝑠 (𝑘𝑘𝑘𝑘) × 𝐷𝐷𝐷𝐷𝐷𝐷𝐷𝐷𝐷𝐷𝐷𝐷𝐷𝐷𝐷𝐷 (ℎ𝑜𝑜𝑜𝑜𝑜𝑜𝑜𝑜)
Figure 42 and Table 13 show the resulting $/kWh costs for 600-kW Li-ion energy storage
systems, which vary from $469/kWh (4-hour duration) to $2,167/kWh (0.5-hour duration). The
battery cost accounts for 41% of total system cost in the 4-hour system, but only 11% in the 0.5-
hour system. At the same time, non-battery cost categories account for an increasing proportion
of the system cost as duration declines.
35
This equation is only for the energy storage installation cost calculation. For levelized cost of storage (LCOS), the
equation would be different. LCOS is not covered in this report.
64
Figure 42. U.S. commercial Li-ion battery standalone storage costs for durations of 0.5–4.0 hours
(600 kWDC), Q1 2020
65
Table 13. Detailed Cost Breakdown for a 600-kW U.S. Commercial Li-ion Standalone Storage System with Durations of 0.5–4 hours
600-kW, 4-hour Duration, 600-kW, 2-hour Duration, 600-kW, 1-hour Duration, 600- 600-kW, 0.5-hour
2,400-kWh 1,200-kWh kWh Duration, 300-kWh
Total Total Total Total
Model Component $/kWh $/W $/kWh $/W $/kWh $/W $/kWh $/W
Cost ($) Cost ($) Cost ($) Cost ($)
Li-ion battery 465,600 192 0.78 237,600 196 0.40 133,800 221 0.22 72,600 240 0.12
Battery central inverter 36,000 15 0.06 36,000 30 0.06 36,000 59 0.06 36,000 119 0.06
Structural BOS 62,012 26 0.10 62,012 51 0.10 62,012 102 0.10 62,012 205 0.10
Electrical BOS 115,618 48 0.19 115,618 95 0.19 115,618 191 0.19 115,618 382 0.19
Installation labor & equipment 151,596 63 0.25 151,596 125 0.25 151,596 250 0.25 151,596 500 0.25
EPC overhead 79,475 33 0.13 79,475 66 0.13 79,475 131 0.13 79,475 262 0.13
Sales tax 42,432 18 0.07 29,208 24 0.05 23,188 38 0.04 19,638 65 0.03
∑ EPC cost 952,734 393 1.59 711,510 587 1.19 601,689 993 1.00 536,940 1,772 0.89
Permitting fee 7,507 3 0.01 7,507 6 0.01 7,507 12 0.01 7,507 25 0.01
Interconnection fee 27,846 11 0.05 27,846 23 0.05 27,846 46 0.05 27,846 92 0.05
Contingency 38,455 16 0.06 28,806 24 0.05 24,414 40 0.04 21,824 72 0.04
Developer overhead 57,683 24 0.10 43,209 36 0.07 36,620 60 0.06 32,735 108 0.05
EPC/developer profit 52,836 22 0.09 39,569 33 0.07 33,529 55 0.06 29,967 99 0.05
∑ Developer cost 184,327 76 0.31 146,937 121 0.24 129,915 214 0.22 119,879 396 0.20
∑ Total energy storage
1,137,060 469 1.90 858,447 708 1.43 731,604 1,207 1.22 656,818 2,167 1.09
system cost
66
7.2 Commercial PV-plus-Storage System Cost Model
We model a 1-MW commercial fixed-tilt ground-mount PV plus 600-kW storage system, with
0.5 hours (300 kWh), 1 hour (600 kWh), 2 hours (1.2 MWh), and 4 hours (2.4 MWh) of storage,
using the same PV assumptions we used with our standalone PV system. Figure 43 provides a
schematic of typical DC- and AC-coupled PV systems with battery back-up. Table 5, Table 12,
and Table 14 present modeling inputs and assumptions.
Figure 43. Modeled DC- and AC-coupled system configurations

Figure is simplified for illustrative purposes.
Table 14. Changes to Commercial PV and Storage Models When PV and Storage Are Combined

Electrical BOS 90% of the combined BOS costs for PV Duplicative parts are removed
and battery standalone systems
Installation 90% of the combined BOS costs for PV Duplicative work is removed
labor and battery standalone systems
Sales and 20 hours more time for DC system, and Additional explanation, calculations, and
marketing 32 hours more for AC system, per a lower close rate; also, the AC system
closed sale, associated with selling a requires more customer site assessment
PV system with storage
7.3 Commercial Model Output

Figure 44 summarizes our model results for several system types and configurations:
• Standalone 1-MW commercial fixed-tilt ground-mount PV system ($1.59 million)
67
• Standalone 600-kW/2.4-MWh, 4-hour-duration energy storage system ($1.13 million)
• Colocated DC-coupled PV (1-MW) plus storage (600 kW/2.4 MWh, 4-hour duration) system
($2.13 million)
• Colocated AC-coupled PV (1-MW) plus storage (600 kW/2.4 MWh, 4-hour duration) system
($2.07 million)
• PV (1-MW) plus storage (600 kW/2.4 MWh, 4-hour duration) system with PV and storage
components sited in different locations ($2.72 million).
Table 15 shows detailed costs for the three PV-plus-storage configurations. Colocating the PV
and storage subsystems produces cost savings by reducing costs related to site preparation,
permitting, interconnection, installation labor, hardware (via sharing of hardware such as
switchgears, transformers, and controls), overhead, and profit. The cost of the colocated AC-
coupled system is 24% lower than the cost of the system with PV and storage sited separately.
Using DC-coupling rather than AC-coupling results in a 2.8% higher total cost, which is the net
result of cost differences between DC-coupling and AC-coupling in the categories of solar
inverter, structural BOS, electrical BOS, labor, EPC and developer overhead, sales tax,
contingency, and profit. For an actual project, however, cost savings may not be the only factor
in choosing DC- or AC-coupling. Additional factors—such as retrofit considerations, system
performance (including energy loss due to clipping), design flexibility, and O&M—should be
considered.
Figure 44. Cost benchmarks for commercial PV-plus-storage systems (4-hour duration) in
different sites and the same site (DC-coupled and AC-coupled cases), Q1 2020
68
Table 15. Detailed Cost Breakdown for Commercial Li-ion PV-Plus-Storage Systems
Total Cost
1-MW PV Plus 1-MW PV Plus 1-MW PV Plus
600-kW/2.4-MWh 600-kW/2.4-MWh 600-kW/2.4-MWh
Battery, DC- Battery, AC- Battery, in
Model Component Coupled, Colocated Coupled, Colocated Different Sites
PV module $405,877 $405,877 $405,877
Li-ion battery $460,917 $460,917 $460,917
Solar inverter — $71,347 $71,347
Bidirectional inverter $35,638 $35,638 $35,638
Structural BOS $179,759 $173,284 $173,285
Electrical BOS $225,088 $190,036 $298,378
Installation labor & equipment $271,097 $217,553 $294,560
EPC overhead $161,386 $129,511 $175,354
Sales tax $82,924 $84,816 $91,688
∑ EPC cost $1,822,686 $1,768,978 $2,007,044
Land acquisition 0 0 0
Permitting fee $7,507 $7,657 $15,014
Interconnection fee $27,846 $28,403 $55,691
Transmission line 0 0 0
Contingency $55,741 $54,151 $82,038
Developer overhead $55,741 $54,151 $386,876
EPC/developer profit $157,562 $153,067 $170,144
∑ Developer cost $304,396 $297,429 $709,407
∑ Total energy storage
$2,127,082 $2,066,408 $2,716,451
system cost
7.4 Commercial Levelized Cost of Solar-plus-Storage

For this year’s benchmark report, we calculate the LCOSS for our commercial PV-plus-storage
system, with the same formula and caveats as we use for our residential PV-plus-storage system
(see Section 6.5). Table 16 lists our model inputs and assumptions for the commercial PV-plus-
storage system.
69
Table 16. Commercial LCOSS Inputs and Assumptions

1-MW fixed-tilt ground-mount PV
System size plus 600-kW/2.4-MWh storage
system
2020 commercial PV-plus-storage
Initial investment $2,066,408
First follow-on 10 years due to battery capacity dropping
investments (inverter, $73,747 in year 10 20%. We assume costs for battery and
next 10 years.
Second follow-on 20 years due to battery capacity dropping
investments (inverter, $55,310 in year 20 20%. We assume costs for battery and
next 20 years.
Residual value $0
High resource area: 1,894
MWh/MW
Initial annual Medium resource area: 1,541
Low resource area: 1,438
MWh/MW
Percentage of generated High resource area: 35%

Assumes a 75% discharge per day for a 4-
solar electricity fed to Medium resource area: 43%
hour, 600-kW battery
battery Low resource area: 46%
Roundtrip energy losses
10%
from PV/battery/grid
Roundtrip energy losses
8%
from grid/battery/grid
Battery is charged solely by PV due to
Charging cost $0
ITC considerations
Assumes storage O&M adds $10/kW-yr
O&M ($/kW/yr) $29
to PV costs
Annual PV degradation 0.70%
Annual electricity
0
purchased from grid
Inflation 2.5%
70
We use these inputs to calculate LCOSS via Equation 1. Figure 45 shows the resulting LCOSS
for a commercial AC-coupled fixed-tilt ground-mount PV (1 MW) plus storage (600 kW/2.4
MWh, 4-hour duration) system, as well as the LCOE of a 1-MW fixed-tilt ground-mount
standalone PV system. LCOSS is calculated to be $113/MWh without the federal ITC and
$73/MWh with the 30% ITC for commercial PV-plus-storage, with a medium resource for PV
electricity production. The PV-plus-storage LCOSS is $37/MWh higher than the standalone-PV
LCOE without the ITC, and $27/MWh higher with a 30% ITC.
Figure 45. U.S. commercial LCOSS for an AC-coupled PV (1 MW) plus storage (600 kW/2.4 MWh, 4-
hour duration) system and LCOE for a 1-MW standalone PV system, Q1 2020
LCOSS is calculated for each scenario with a medium CF (representing Kansas City); LCOSS and LCOE ranges
based on high and low CF assumptions; all other values remain the same.
71
8 Utility-Scale Storage and PV-plus-Storage Model
Figure 46 shows the detailed bottom-up cost structure of our standalone utility-scale storage
model, which uses a structure similar to our previously developed PV cost model (Fu et al. 2015,
2016, 2017; Fu, Feldman, and Margolis 2018; Fu, Remo, and Margolis 2018). Total system
upfront capital costs are broken into EPC costs and developer costs. EPC non-hardware, or
“soft,” costs are driven by labor rates and labor productivities. We adapt engineering-design and
cost-estimating models from RSMeans (2017) to determine the EPC hardware costs (including
module/battery racking, mounting, wiring, containerization, and foundation) and related EPC
soft costs (including related labor and equipment hours required in any given U.S. location).
Sections 8.1 and 8.2 present the utility-scale storage and PV-plus-storage cost models, Section
8.3 shows the model outputs, Section 8.4 compares PV-plus-storage benchmark trends over time,
and Section 8.5 benchmarks the LCOSS for a utility-scale PV-plus-storage system.
Figure 46. Structure of the bottom-up cost model for utility-scale standalone storage systems
8.1 Utility-Scale Li-Ion Standalone Storage Cost Model

The major storage components we model for utility-scale standalone storage systems are the
same as those summarized in Figure 40 and Figure 41 for the commercial standalone storage
model. Table 17 lists our model inputs and assumptions for such a utility-scale energy storage
72
system. We determine the battery size (60 MWDC) 36 using an inverter loading ratio of 1.3 and a
PV/storage size ratio of 1.67, based on Denholm, Eichman, and Margolis (2017).
Table 17. Utility-Scale Li-ion Energy Storage System: Model Inputs and Assumptions
Model Modeled Value Description Source

Component
Battery total 60 MWDC Baseline case to match a 100-MW NREL 2020
size PV system
Battery size 2.5 MWh per 40-ft Assumption to compute the number of NREL 2020
per container container containers
Li-ion battery 0.5 hours: $242/kWh Ex-factory gate (first buyer) prices BNEF 2019b
price 1 hour: $223/kWh
2 hours: $198/kWh
4 hours: $194/kWh
Duration 0.5–4.0 hours Duration determines energy (MWh) NREL 2020
Battery central $0.06/W Ex-factory gate (first buyer) prices Wood Mackenzie
inverter price 2019
Inverter size 2.5 MW per inverter Used to determine the number of battery NREL 2020
inverters
Electrical BOS $0.07–$0.14/W Includes conduit, wiring, DC cable, energy NREL 2020
management system, switchgear,
transformer, and monitor and controls for
each container. Determined by the number
of containers, transformers, and row
spacing.
Structural BOS $0.01–$0.05/W Includes foundation, battery containers, and NREL 2020
inverter house. Determined by the number
of containers, inverters, transformers, and
the spacing between containers.
Installation Electrician: $27.47 National average modeled labor rate BLS 2019
labor per hour assumes nonunionized labor
Laborer: $18.17
per hour
Sales tax 5% (national Sales tax on the equipment RSMeans 2017
average)
EPC overhead 8.67% for equipment Costs associated with EPC SG&A, NREL 2020
and profit and material; 23%– warehousing, shipping, and logistics
69% for labor costs;
varies by system
size, and labor
activity
36
For a 100-MW PV system with an inverter loading ratio of 1.3 and PV/storage size ratio of 1.67, maximum
deliverable power at point of interconnection is 137 MWAC (100 MW/1.3 + 100 MW/1.67) for AC coupled systems
and 77 MWAC (100 MW/1.3) for DC coupled systems.
73
Model Modeled Value Description Source
Component
Developer 3% of total Includes overhead expenses such as NREL 2020
cost: developer installation cost payroll, facilities, travel, legal fees,
overhead administrative, business development,
finance, and other corporate functions
Developer $0.03/W Construction permits fee, interconnection NREL (2020
cost: PII study, interconnection inspection, and
interconnection fee
Developer 3% Estimated as markup on the total EPC cost NREL 2020)
cost:
contingency
Developer 5% Applies a percentage margin to all costs NREL 2020
cost: including hardware, installation labor, EPC
EPC/developer overhead, and developer overhead
net profit
We use these inputs to calculate energy storage cost via the following equation 37:
$
𝐸𝐸𝐸𝐸𝐸𝐸𝐸𝐸𝐸𝐸𝐸𝐸 𝑠𝑠𝑠𝑠𝑠𝑠𝑠𝑠𝑠𝑠𝑠𝑠𝑠𝑠 𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑖 𝑐𝑐𝑐𝑐𝑐𝑐𝑐𝑐 � �=
𝑘𝑘𝑘𝑘ℎ
$ 𝑂𝑂𝑂𝑂ℎ𝑒𝑒𝑒𝑒 𝑐𝑐𝑐𝑐𝑐𝑐𝑐𝑐 𝑐𝑐𝑐𝑐𝑐𝑐𝑐𝑐𝑐𝑐𝑐𝑐𝑐𝑐𝑐𝑐𝑐𝑐𝑐𝑐 ($) 𝑠𝑠𝑠𝑠𝑠𝑠ℎ 𝑎𝑎𝑎𝑎 𝑏𝑏𝑏𝑏𝑏𝑏𝑏𝑏𝑏𝑏𝑏𝑏𝑏𝑏 𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑖 𝑎𝑎𝑎𝑎𝑎𝑎 𝑙𝑙𝑙𝑙𝑙𝑙𝑙𝑙𝑙𝑙
𝐵𝐵𝐵𝐵𝐵𝐵𝐵𝐵𝐵𝐵𝐵𝐵𝐵𝐵 𝑐𝑐𝑐𝑐𝑐𝑐𝑐𝑐 � �+
𝑘𝑘𝑘𝑘ℎ 𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆 𝑠𝑠𝑠𝑠𝑠𝑠𝑠𝑠𝑠𝑠𝑠𝑠 𝑠𝑠𝑠𝑠𝑠𝑠𝑠𝑠 (𝑘𝑘𝑘𝑘) × 𝐷𝐷𝐷𝐷𝐷𝐷𝐷𝐷𝐷𝐷𝐷𝐷𝐷𝐷𝐷𝐷 (ℎ𝑜𝑜𝑜𝑜𝑜𝑜𝑜𝑜)
Figure 47 and Table 18 show the resulting costs for 60-MW Li-ion energy storage systems,
which vary from $341/kWh (4-hour duration) to $845/kWh (0.5-hour duration). While the per-
energy-unit battery cost increases as system duration decreases, the total battery cost—and the
proportion of the cost attributed to the battery—decrease as system duration decreases. For
example, the battery cost accounts for 56% of total system cost in the 4-hour system, but only
28% in the 0.5-hour system. At the same time, non-battery cost categories account for an
increasing proportion of the system cost as duration declines.
37
This equation is only for the energy storage installation cost calculation. For LCOS, the equation would be
different. LCOS is not covered in this report.
74
Figure 47. U.S. utility-scale Li-ion battery standalone storage costs for durations of 0.5–4.0 hours
(60 MWDC), Q1 2020
75
Table 18. Detailed Cost Breakdown for a 60-MW U.S. Utility-Scale Li-ion Standalone Storage System with Durations of 0.5–4 hours
60-MW, 4-hour Duration, 60-MW, 2-hour Duration, 60-MW, 1-hour Duration, 60-MW, 0.5-hour Duration,
240-MWh 120-MWh 60-MWh 30-MWh
Total Cost Total Cost Total Cost Total Cost
Model Component $/kWh $/W $/kWh $/W $/kWh $/W $/kWh $/W
($) ($) ($) ($)
Li-ion battery 46,560,000 192 0.78 23,760,000 196 0.40 13,380,000 221 0.22 7,260,000 240 0.12
Battery central inverter 3,600,000 15 0.06 3,600,000 30 0.06 3,600,000 59 0.06 3,600,000 119 0.06
Structural BOS 3,173,302 13 0.05 1,853,216 15 0.03 1,193,174 20 0.02 863,152 28 0.01
Electrical BOS 8,599,517 35 0.14 6,087,485 50 0.10 4,831,469 80 0.08 4,203,461 139 0.07
Installation labor
4,694,348 19 0.08 3,706,099 31 0.06 3,211,975 53 0.05 2,964,913 98 0.05
& equipment
EPC overhead 2,354,557 10 0.04 1,623,195 13 0.03 1,257,513 21 0.02 1,074,673 35 0.02
Sales tax 3,807,403 16 0.06 2,236,341 18 0.04 1,509,970 25 0.03 1,092,844 36 0.02
∑ EPC cost 72,789,126 300 1.21 42,866,336 354 0.71 28,984,101 478 0.48 21,059,043 695 0.35
Land acquisition 0 0 0.00 0 0 0.00 0 0 0.00 0 0 0.00
Permitting fee 295,289 1 0.00 295,289 2 0.00 295,289 5 0.00 295,289 10 0.00
Interconnection fee 1,849,475 8 0.03 1,849,475 15 0.03 1,849,475 31 0.03 1,849,475 61 0.03
Contingency 2,265,878 9 0.04 1,359,264 11 0.02 938,331 15 0.02 698,347 23 0.01
Developer overhead 1,603,157 7 0.03 961,708 8 0.02 663,889 11 0.01 494,095 16 0.01
EPC/developer profit 3,940,146 16 0.07 2,366,604 20 0.04 1,636,554 27 0.03 1,219,812 40 0.02
∑ Developer cost 9,953,946 41 0.17 6,832,340 56 0.11 5,383,539 89 0.09 4,557,019 150 0.08
∑ Total energy
82,743,072 341 1.38 49,698,676 410 0.83 34,367,640 567 0.57 25,616,062 845 0.43
storage system cost
76
8.2 Utility-Scale PV-plus-Storage System Cost Model
Here we combine our energy storage cost model with our PV system cost model in various
configurations, including (1) colocated PV-plus-storage systems versus PV and storage systems
located in different places and (2) DC-coupled versus AC-coupled battery configurations for the
colocated PV-plus-storage systems. As shown in Table 19, colocation enables sharing of several
hardware components by the PV and energy storage systems, which can reduce costs. Colocation
can also reduce soft costs related to site preparation, land acquisition, installation labor,
permitting, interconnection, and EPC/developer overhead and profit.
Table 19. Cost Factors for Siting PV and Storage Together versus Separately
PV-plus-Storage
Model Component Colocated PV-plus-Storage
at Different Sites
Site preparation 38 Once Twice
Land acquisition cost Lower Higher
Hardware sharing between PV Yes (step-up transformer, switchgear, No
and energy storage monitor, and controls)
Installation labor cost Lower (due to hardware sharing and Higher
single labor mobilization)
EPC/developer overhead and profit Lower (due to lower labor cost, BOS, Higher
and total system cost)
Interconnection and permitting Once Twice
When PV and battery storage are colocated, the subsystems can be connected by either a DC-
coupled or an AC-coupled configuration (Figure 48). A DC-coupled system needs only one
bidirectional inverter, connects battery storage directly to the PV array, and enables the battery to
charge and discharge from the grid. On the other hand, an AC-coupled system needs both a PV
inverter and a bidirectional inverter, and there are multiple conversion steps between DC and AC
to charge or discharge the battery. Also, the transmission line could be used for both PV and
battery storage systems.
The advantages of the DC-coupled system include the following:
1. A DC-coupled system uses only a single bidirectional inverter (Table 20), thus reducing costs
for the inverter, inverter wiring, and inverter housing.
2. Because of the extra conversion between DC and AC, an AC-coupled system may have
lower roundtrip efficiency for battery charging than a DC-coupled system, which charges the
battery directly. However, as power electronics are becoming more efficient, the actual
efficiency difference is becoming smaller (Enphase 2019).
3. Because the battery is connected directly to the PV array, excess PV generation that would
otherwise be clipped by an AC-coupled system at the inverter level can be sent directly to the
battery, which could improve system economics (DiOrio and Hobbs 2018).
38
Site preparation is a subcategory of labor cost, so it is not shown in the cost breakdown chart.
77
DC-Coupled System
Solar PV System
DC DC to DC DC Battery Pack
DC Converter (Charge and Discharge)
DC
Bidirectional
Inverter AC
Grid
(DC → AC or
AC → DC)
Solar PV System
AC-Coupled System
DC
PV Inverter
AC
Grid
(DC → AC) AC AC
Bidirectional Inverter AC
(DC → AC or
AC → DC)
DC DC
Battery Pack
DC
(Charge and Discharge)
Figure 48. DC-coupled and AC-coupled PV-plus-storage system configurations
78
Table 20. Comparison of DC- and AC-Coupling for Utility-Scale PV-plus-Storage Systems
Model Component DC-Coupled Configuration AC-Coupled Configuration

Number of inverters 1 (bidirectional inverter for battery) 2 (bidirectional inverter for battery
plus grid-tied inverter for PV),
resulting in higher costs for the
inverter, inverter wiring, and
inverter housing
Battery rack size Smaller (because battery is directly Larger
connected to PV),a resulting in more
HVAC and fire-suppression systems
required
Structural BOS More (due to smaller battery Less
rack size)
Electrical BOS Less (but needs additional DC-to-DC More (due to additional wiring
converters) for inverters)
Installation labor cost More (due to smaller battery rack size Less
and more skilled labor and labor
hours required for DC work)
EPC overhead More (due to higher installation Less
labor cost)
Sales tax Less More (due to higher total
hardware costs)
EPC/developer profit Less More (due to higher total EPC and
developer costs)
a Because a PV system is not directly connected to a battery in an AC-coupled configuration, the battery
racks are fewer and larger; this configuration is less costly than a DC-coupled system in which multiple
distributed battery racks are deployed and managed. For example, using five smaller battery racks rather
than one large rack requires five fire-suppression systems and five air conditioning systems.
The advantages of the AC-coupled system include the following:
1. Because the battery racks are not directly connected to the PV system in AC-coupled
systems, these systems can use larger battery racks and thus reduce the number of HVAC and
fire-suppression systems in the containers. This feature also reduces installation labor costs
compared with DC-coupled systems.
2. For a retrofit (i.e., adding battery storage to an existing PV array), an AC-coupled battery
may be more practical than a DC-coupled battery, because DC-coupled systems require
installers to replace the existing PV inverter with a bidirectional inverter. Thus, the additional
costs that are due to replacing the inverter and rewiring the system could make retrofit costs
higher for a DC-coupled system than for an AC-coupled system (Ardani et al. 2017). In
addition, AC-coupled systems enable the option of upgrading the PV and battery separately,
because these systems are independent of one another.
79
3. Because AC-coupled systems have separate PV and battery systems, installers have more
flexibility to adjust the battery location. For instance, DC-coupled systems require batteries
to be installed next to the bidirectional inverter, and the resulting need for maintenance crews
to enter the PV field can make maintenance more time consuming. Because AC-coupled
systems can have batteries located outside the PV field, maintenance work can be quicker
and easier.
8.3 Utility-Scale Model Output

Figure 49 (page 81) summarizes our model results for several system types and configurations:
• Standalone 100-MW PV system with one-axis tracking ($101 million)

• Standalone 60-MW/240-MWh, 4-hour-duration energy storage system ($83 million)
• Colocated DC-coupled PV (100-MW) plus storage (60-MW/240-MWh, 4-hour-duration)
system ($173 million)
• Colocated AC-coupled PV (100-MW) plus storage (60-MW/240-MWh, 4-hour-duration)
system ($171 million)
• PV (100-MW) plus storage (60-MW/240-MWh, 4-hour-duration) system with PV and
storage components sited in different locations ($183 million).
Table 21 shows detailed costs for the three PV-plus-storage configurations. Colocating the PV
and storage subsystems produces cost savings by reducing costs related to site preparation, land
acquisition, permitting, interconnection, installation labor, hardware (via sharing of hardware
such as switchgears, transformers, and controls), overhead, and profit. The cost of the colocated
AC-coupled system is 7% lower than the cost of the system with PV and storage sited separately.
Using DC-coupling rather than AC-coupling results in a 1% higher total cost, which is the net
result of cost differences between DC-coupling and AC-coupling in the categories of solar
inverter, structural BOS, electrical BOS, labor, EPC and developer overhead, sales tax,
contingency, and profit. For an actual project, however, cost savings may not be the only factor
in choosing DC- or AC-coupling. Additional factors—such as retrofit considerations, system
performance (including energy loss due to clipping), design flexibility, and O&M—should be
considered.
80
Figure 49. Cost benchmarks for PV-plus-storage systems (4-hour duration) in different sites and
the same site (DC-coupled and AC-coupled cases), Q1 2020
81
Table 21. Detailed Cost Breakdown for Utility-Scale Li-ion PV-plus-Storage Systems
Total Cost
100-MW PV Plus 100-MW PV Plus 100-MW PV Plus
60-MW/240-MWh 60-MW/240-MWh 60-MW/240-MWh
Battery, DC- Battery, AC- Battery, in
Model Component Coupled, Colocated Coupled, Colocated Different Sites
PV module $40,587,666 $40,587,666 $40,587,666
Li-ion battery $46,091,749 $46,091,749 $46,091,749
Solar inverter — $5,171,344 $5,171,344
Bidirectional inverter $3,563,795 $3,563,795 $3,563,795
Structural BOS $15,908,348 $15,289,203 $15,342,164
Electrical BOS $13,384,607 $10,336,576 $15,855,408
Installation labor & equipment $15,537,385.79 $13,417,123.80 $15,757,821
EPC overhead $7,905,594 $6,826,782 $8,017,754
Sales tax $7,645,117 $7,741,319 $8,097,671
∑ EPC cost $150,624,262 $149,025,558 $158,485,372
Permitting fee $198,395 $198,395 $396,790
Interconnection fee $2,784,560 $2,784,560 $5,569,120
Transmission line $1,669,331 $1,669,331 $1,669,331
Contingency $4,658,296 $4,610,335 $4,980,515
Developer overhead $4,658,296 $4,610,335 $3,523,821
EPC/developer profit $8,229,657 $8,144,926 $8,688,259
∑ Developer cost $22,198,536 $22,017,883 $24,827,837
∑ Total system cost
$172,822,798 $171,043,440 $183,313,209
8.4 Utility-Scale PV-plus-Storage Price Benchmark Historical Trends

Figure 50 shows 9% and 8% reductions in utility-scale PV-plus-storage benchmarks between
2018 (Fu, Remo, and Margolis 2018) and 2020 (this report), for DC-coupled and AC-coupled
systems. For the DC-coupled system, approximately 28% of that reduction can be attributed to
the Li-ion battery plus bidirectional inverter, while electrical and structural BOS decreased
system cost by 13%; an additional 17% can be attributed to lower labor costs, and the final 42%
is attributable to other soft costs, including PII, sales tax, overhead, and net profit. For the AC-
coupled system, approximately 30% of the reduction can be attributed to the Li-ion battery plus
bidirectional inverter, and 4% to electrical and structural BOS; an additional 16% can be
attributed to lower labor costs, and the final 49% is attributable to other soft costs, including PII,
sales tax, overhead, and net profit.
82
Figure 50. Utility-scale PV-plus-storage system cost benchmark summary
2018–2020, DC-coupled and AC-coupled
MM = million
8.5 Utility-Scale Levelized Cost of Solar-plus-Storage

For this year’s benchmark report, we calculate the LCOSS for utility-scale PV-plus-storage, with
the same formula and caveats as we use for our residential and commercial PV-plus-storage
systems (see Section 6.5). BNEF (2019c) and Lazard (2018) have performed similar LCOSS
calculations. None of these LCOSS calculations, including the ones in this report, attempts to
value the electricity generated by these systems or the different ways they may operate. Storage
value calculations require integrating storage dispatch into regional capacity expansion, load, or
reliability models. For a detailed discussion of storage value, see Balducci et al. (2018), Denholm
et al. (2019), Frew et al. (2018), and Schmidt et al. (2019). Similar to LCOE, LCOSS does not
focus on value but rather can help track improvements to all costs of a utility-scale PV-plus-
storage system over time (as opposed to just upfront costs), and the metric can provide limited
comparisons with other dispatchable electricity generation technologies (e.g., natural gas).
Table 22 lists our model inputs and assumptions for calculating utility-scale LCOSS.
83
Table 22. Utility-Scale LCOSS Inputs and Assumptions
100-MW PV plus 60-MW / 240-

System size MWh battery storage, AC-
coupled
2019 utility-scale PV-plus-storage

Initial investment $171 million
20% of batteries replaced after 10 years
First follow-on
due to battery capacity dropping 20%. We
investments (inverter, $7.4 million in year 10
assume costs for battery and bidirectional
battery replacements)
inverters drop 20% in the next 10 years.
20% of batteries replaced after 20 years
Second follow-on
due to battery capacity dropping 20%. We
investments (inverter, $5.5 million in year 20
assume costs for battery and bidirectional
battery replacements)
inverters drop 40% in the next 20 years.
Residual value $0
High resource area: 2,185
MWh/MW
Initial annual Medium resource area: 1,707
Low resource area: 1,572
MWh/MW
Percentage of High resource area: 30%
generated solar Assumes a 75% discharge per day for a 4-
Medium resource area: 39%
electricity fed to hour, 60-MW battery
battery Low resource area: 42%
Roundtrip energy
losses from 10%
PV/battery/grid
Roundtrip energy
losses from 8%
grid/battery/grid
Battery is charged solely by PV due to
Charging cost $0
ITC considerations
Assumes storage O&M adds $10/kW-yr
O&M ($/kW/yr) $27
to PV costs
PV Degradation 0.70%
Annual electricity
0
purchased from grid
Inflation 2.5%
84
We use these inputs to calculate LCOSS via Equation 1. Figure 51 shows the resulting LCOSS
for a colocated AC-coupled PV (100 MW) plus storage (60 MW/240 MWh, 4-hour duration)
system, as well as the LCOE of a 100-MW PV-standalone system, with one-axis tracking.
LCOSS is calculated to be $83/MWh without the federal ITC and $57/MWh with the 30% ITC,
with a medium resource for PV electricity production. 39 Based on these calculations, PV-plus-
storage LCOSS is $40/MWh higher than standalone-PV LCOE without the ITC, and $28/MWh
higher with a 30% ITC. Bolinger, Seel, and Robson (2019) reported a storage premium of $10–
$15/MWh for PPAs with a 30% ITC, for systems that have a 4-hour battery sized to 50%–75%
of the PV capacity.
Figure 51. U.S. utility-scale LCOSS for an AC-coupled PV (100 MW) plus storage (60 MW/240 MWh,
4-hour duration) system and LCOE for a 100-MW PV standalone system, Q1 2020
LCOSS is calculated for each scenario with a medium CF (representing Kansas City);
LCOSS and LCOE ranges based on high and low CF assumptions; all other values remain the same.
39
We do not change the inputs and assumptions between the ITC and non-ITC cases, despite the fact that the inputs
in the LCOSS calculation assume the owner of the PV-plus-storage system is operating the plant such that they can
claim the ITC on the storage equipment. In reality, an owner would likely operate a PV-plus-storage system
differently without the ITC. Additionally, we assume projects can qualify as starting construction before 2020,
allowing them to claim a 30% ITC, instead of the 26% ITC for projects starting construction in 2020.
85
9 Conclusions
NREL’s bottom-up cost models can be used to assess the costs of PV and storage systems using
various configurations. They can also estimate future potential cost-reduction opportunities for
PV and PV-plus-storage systems, thus helping guide research and development aimed at
advancing cost-effective system configurations. The data in this annual benchmark report inform
the formulation of and track progress toward SETO’s GPRA cost targets.
Based on our bottom-up modeling, the Q1 2020 cost benchmarks are:
• $2.71/WDC (or $3.12/WAC) for residential PV systems

• $1.72/WDC (or $1.96/WAC) for commercial rooftop PV systems
• $1.72/WDC (or $1.91/WAC) for commercial ground-mount PV systems
• $0.94/WDC (or $1.28/WAC) for fixed-tilt utility-scale PV systems
• $1.01/WDC (or $1.35/WAC) for one-axis-tracking utility-scale PV systems 40
• $26,153–$28,371 for a 7-kW residential PV system with 3 kW/6 kWh of storage and
$35,591–$37,909 for a 7-kW residential PV system with 5 kW/20 kWh of storage
• $2.07 million–$2.13 million for a 1-MW commercial ground-mount PV system colocated
with 600 kW/2.4 MWh of storage
• $171 million–$173 million for a 100-MW PV system colocated with 60 MW/240 MWh
of storage.
Overall, modeled installed costs of PV and storage systems continued to decline between Q1
2019 and Q1 2020. Figure 52 puts our Q1 2020 benchmark results in context with the results
of previous NREL benchmarking analyses. When comparing the results across this period,
note that:
1. Values are inflation-adjusted using the CPI (2019). Thus, historical values from our models
are adjusted and presented as real USD instead of nominal USD.
2. Cost categories are aggregated for comparison purposes. “Soft Costs – Others” represents:
A. PII
B. Transmission line (if any)
C. Sales tax
D. EPC/developer overhead and profit.
3. The current versions of our cost models make a few significant changes from the versions
used in our Q1 2018 benchmark report (Fu, Feldman, and Margolis 2018). To better
distinguish the historical cost trends over time from the changes to our cost models, we also
calculate Q1 2019 and Q1 2020 PV benchmarks using the Q1 2018 versions of the
residential, commercial, and utility-scale PV models. Appendix A provides a detailed
discussion of the changes made to the models between previous versions (Fu, Feldman, and
Margolis 2018) and this year’s versions.
The dollar-per-watt total cost value is benchmarked as three significant figures, because the model inputs, such as
40
module and inverter prices, use three significant figures.
86
4. Our Q1 2019 and Q1 2020 benchmarks use monocrystalline PV modules, whereas all
historical benchmarks used multicrystalline PV modules. This switch reflects the overall
trend occurring in the U.S. market.
5. For previous editions of this report, we assumed a land acquisition cost of $0.03/W. Based on
which the PV system is placed, we have reclassified land costs from an upfront capital
expenditure (land acquisition) to an operating expenditure (lease payments) for 2019 and
2020.
From 2010 to 2020, there was a 64%, 69%, and 82% reduction in the residential, commercial
rooftop, and utility-scale (one-axis) PV system cost benchmark, respectively (Figure 52). The
inflation-adjusted system cost differences between Q1 2019 and Q1 2020 are a $0.06/WDC
reduction for residential PV, a $0.04/WDC reduction for commercial rooftop PV, and a
$0.01/WDC reduction for utility-scale PV. Table 23 (page 89) shows the benchmarked values for
all three sectors and drivers of cost decreases and increases.
BOS hardware cost reductions in Q1 2020 were counterbalanced by higher module costs, and
soft costs remained relatively unchanged, year over year (Figure 18, Figure 26, Figure 33); this
resulted in a steady percentage of soft costs as a percentage of total costs (Figure 53). 41 The
historical increase in soft cost proportion for residential and commercial PV systems in Figure 53
indicates soft costs declined more slowly than did hardware costs over time; it does not indicate
soft costs increased on an absolute basis.
Soft costs and hardware costs interact with each other. For instance, module efficiency
improvements have reduced the number of modules required to construct a system of a given
size, thus reducing hardware costs. This trend has also reduced soft costs from direct labor and
related installation overhead.
Also, our bottom-up system cost models enable us to investigate regional variations, system
configurations (e.g., MLPE versus non-MLPE, fixed-tilt versus one-axis tracker, and small
versus large system size). In addition, we consider business structures (e.g., small installer versus
national integrator, and EPC versus developer). Different scenarios result in different costs, so
consistent comparisons can only be made when cost scenarios are aligned.
41
Soft cost = total cost – hardware (module, inverter, structural and electrical BOS) cost.
87
Figure 52. NREL PV system cost benchmark summary (inflation-adjusted), 2010–2020
* The current versions of our cost models make a few significant changes from the versions used in our Q1 2018 benchmark report (Fu, Feldman, and Margolis 2018) and
incorporate costs that had previously not been benchmarked in as much detail. To better distinguish the historical cost trends from the changes to our cost models, we also
calculate Q1 2019 and Q1 2020 PV benchmarks using the Q1 2018 versions of the residential, commercial, and utility-scale PV models. The “Additional Costs from Model
Updates” category represents the difference between modeled results. Using the previous costs models, the Q1 2019 and Q1 2020 benchmarks are calculated to be: Q1
2019 = $2.56/WDC and Q1 2020 = $2.47/WDC (residential PV); Q1 2019 = $1.71/WDC and Q1 2020 = $1.64/WDC (commercial PV); Q1 2019 = $0.94/WDC and Q1 2020 =
$0.89/WDC (utility-scale PV, fixed-tilt); Q1 2019 = $1.01/WDC and Q1 2020 = $0.96/WDC (utility-scale PV, one-axis tracker). Appendix A provides a detailed discussion of the
changes made to the models between last year’s versions (Fu, Feldman, and Margolis 2018) and this year’s versions.
88
Table 23. Comparison of Q1 2019 and Q1 2020 PV System Cost Benchmarks
Commercial Rooftop Utility-Scale PV,

Sector Residential PV
PV One-Axis Tracking
Q1 2019
benchmarks in $2.77 $1.76 $1.02
2019 USD/WDC
Q1 2020
Benchmarks in $2.71 $1.72 $1.01
2019 USD/WDC
Drivers of cost • Higher module • Higher module • Higher module

decrease efficiency (from efficiency efficiency
19.2% to 19.5%) • Lower material & • Lower material &
• Decrease in BOS equipment costs in equipment costs in
hardware and supply some categories some categories
chain costs • Movement of land
acquisition cost from
upfront capital
expenditures into
O&M
Drivers of cost • Higher labor wages • Higher labor wages • Higher labor wages
increase • Higher module costs • Higher module costs • Higher steel prices
• Higher module and
inverter costs
Figure 53. Modeled trend of soft cost as a proportion of total cost by sector, 2010–2020
89
Finally, the reduction in installed cost—along with improvements in operation, system design,
and technology—have resulted in significant LCOE reductions (Figure 54). Compared with
system prices when SETO’s LCOE targets were announced in 2010, U.S. residential and
commercial rooftop PV systems are 93% and 97% toward achieving the 2020 targets,
respectively, and U.S. utility-scale PV systems achieved their 2020 SETO target three years
early. In recognition of both the transformative PV progress to date and the potential for
additional innovation, SETO extended its goals in 2016 to reduce the unsubsidized LCOE by
2030 to 3¢/kWh (utility-scale PV), 4¢/kWh (commercial PV), and 5¢/kWh (residential PV).
Continued research and development, public and private partnerships, and business innovations
are necessary to achieve SETO’s 2030 LCOE targets.
* * *
Figure 54. NREL PV LCOE benchmark summary (inflation-adjusted), 2010–2020

LCOE is calculated for each scenario under a low CF (New York City), medium CF (Kansas City), and high CF
(Phoenix), but all other values remain the same. Appendix A provides a detailed discussion of the changes made to
the models between last year’s versions (Fu, Feldman, and Margolis 2018) and this year’s versions.
90
References
Ardani, Kristen, Jeffrey J. Cook, Ran Fu, and Robert Margolis. 2018. Cost-Reduction Roadmap
for Residential Solar Photovoltaics (PV), 2017–2030. Golden, CO: National Renewable Energy
Laboratory. NREL/TP-6A20-70748. https://www.nrel.gov/docs/fy18osti/70748.pdf.
Ardani, Kristen, Eric O’Shaughnessy, Ran Fu, Chris McClurg, Joshua Huneycutt, and Robert
Margolis. 2017. Installed Cost Benchmarks and Deployment Barriers for Residential Solar
Photovoltaics with Energy Storage: Q1 2016. Golden, CO: National Renewable Energy
Laboratory. NREL/TP-7A20- 67474. https://www.nrel.gov/docs/fy17osti/67474.pdf.
Balducci, Patrick J., M. Jan E. Alam, Trevor D. Hardy, and Di Wu. 2018. “Assigning Value to
Energy Storage Systems at Multiple Points in an Electrical Grid.” Energy Environ. Sci. 11: 1926.
Barbose, Galen, and Naïm Darghouth. 2019. Tracking the Sun: Pricing and Design Trends for
Distributed Photovoltaic Systems in the United States: 2019 Edition. Berkeley, CA: Lawrence
Berkeley National Laboratory. https://emp.lbl.gov/tracking-the-sun.
Bloomberg. 2018. Bloomberg Professional Services. Accessed May 20, 2018:
https://www.bloomberg.com/professional/.
BLS (U.S. Bureau of Labor Statistics). 2019. “National Occupational Employment and Wage
Estimates United States.” Accessed October 18, 2019:
https://www.bls.gov/oes/current/oes_nat.htm.
BNEF (Bloomberg New Energy Finance). 2019a. “2019 Lithium-Ion Battery Price Survey.”
December 3, 2019.
———. 2019b. “Energy Storage System Costs Survey 2019.” October 14, 2019.
———. 2019c. “LCOE Data: Current LCOE Range ($/MWh, Nominal): United States,
2019 H2.”
———. 2018. “Energy Storage System Costs Survey 2019.” September 4, 2018.
Bolinger, Mark, Joachim Seel, and Dana Robson. 2019. Utility-Scale Solar: Empirical Trends in
Project Technology, Cost, Performance, and PPA Pricing in the United States: 2019 Edition.
Berkeley, CA: Lawrence Berkeley National Laboratory. https://emp.lbl.gov/publications/utility-
scale-solar-empirical-trends.
CA NEM (California net energy metering). 2020. “NEM Currently Interconnected Data Set.”
Accessed August 11, 2020: http://www.californiadgstats.ca.gov/downloads.
Case, Tyler. 2012. “US Cost of Doing Business: Costs Fall in 2010.” Moody’s Analytics
Regional Financial Review, September 2012.
Chung, Donald, Carolyn Davidson, Ran Fu, Kristen Ardani, and Robert Margolis. 2015. U.S.
Photovoltaic Prices and Cost Breakdowns: Q1 2015 Benchmarks for Residential, Commercial,
and Utility-Scale Systems. NREL/TP-6A20-64746. Golden, CO: National Renewable Energy
Laboratory. https://www.nrel.gov/docs/fy15osti/64746.pdf.
91
Darghouth, Naïm, Galen Barbose, and Andrew Mills. 2019. Implications of Rate Design for the
Customer-Economics of Behind-the-Meter Storage. Berkeley, CA: Lawrence Berkeley National
Laboratory. https://emp.lbl.gov/publications/implications-rate-design-customer.
Denholm, Paul, Josh Eichman, and Robert Margolis. 2017. Evaluating the Technical and
Economic Performance of PV Plus Storage Power Plants. NREL/TP-6A20-68737. Golden, CO:
National Renewable Energy Laboratory. https://www.nrel.gov/docs/fy17osti/68737.pdf.
Denholm, Paul, Jacob Nunemaker, Pieter Gagnon, and Wesley Cole. 2019. The Potential
for Battery Energy Storage to Provide Peaking Capacity in the United States. NREL/TP-6A20-
74184. Golden, CO: National Renewable Energy Laboratory.
DiOrio, Nicholas, Aron Dobos, and Steven Janzou. 2015. Economic Analysis Case Studies of
Battery Energy Storage with SAM. NREL/TP-6A20-64987. Golden, CO: National Renewable
Energy Laboratory. https://www.nrel.gov/docs/fy16osti/64987.pdf.
DiOrio, Nicholas, and Will Hobbs. 2018. Economic Dispatch for DC-Connected Battery Systems
on Large PV Plants. 10th PVPMC, Albuquerque, NM. https://pvpmc.sandia.gov/download/6559/.
EIA (United States Energy Information Administration). 2020. “2019ER Form EIA-860 detailed
data with previous form data.” Released June 2, 2020. Washington, DC: EIA.
Enphase. 2019. Enphase quarterly presentations. Accessed October 14, 2019:
https://investor.enphase.com/events-and-presentations
EuPD Research and Green Tech Media. 2016. “PV Installer Survey USA 2015/16.” Accessed
October 2016.
Feldman, David, Mark Bolinger, and Paul Schwabe. 2020. Current and Future Costs of
Renewable Energy Project Finance Across Technologies. NREL/TP-6A20-76881. Golden, CO:
Feldman, David, Rebecca Jones-Albertus, and Robert Margolis. 2018. Impact of Research
and Development, Analysis, and Standardization on PV Project Financing Costs. NREL/TP-
6A20-70939. Golden, CO: National Renewable Energy Laboratory.
Feldman, David, Galen Barbose, Robert Margolis, Mark Bolinger, Donald Chung, Ran Fu,
Joachim Seel, Carolyn Davidson, Naïm Darghouth, and Ryan Wiser. 2015. Photovoltaic System
Pricing Trends: Historical, Recent, and Near-Term Projections: 2015 Edition. NREL/PR-6A20-
Feldman, David, Barry Friedman, and Robert Margolis. 2013. Financing, Overhead, and Profit:
An In-Depth Discussion of Costs Associated with Third-Party Financing of Residential and
Commercial Photovoltaic Systems. NREL/TP-6A20-60401. Golden, CO: National Renewable
Energy Laboratory. https://www.nrel.gov/docs/fy14osti/60401.pdf.
Feldman, David, and Paul Schwabe. 2018. Terms, Trends, and Insights on PV Project Finance in
the United States, 2018. NREL/TP-6A20-72037. Golden, CO: National Renewable Energy
92
Feldman, David, and Paul Schwabe. 2017. Terms, Trends, and Insights: PV Project Finance in
the United States, 2017. Golden, CO: National Renewable Energy Laboratory. NREL/TP-6A20-
70157. https://www.nrel.gov/docs/fy18osti/70157.pdf.
Feldman, David, Travis Lowder, and Paul Schwabe. 2016. Terms, Trends, and Insights: PV
Project Finance in the United States, 2016. NREL/BR-6A20-66991. Golden, CO: National
Renewable Energy Laboratory. http://www.nrel.gov/docs/fy16osti/66991.pdf.
Fitzgerald, Garrett, James Mandel, Jesse Morris, and Herve Touati. 2015. The Economics of
Battery Energy Storage: How Multi-Use, Customer-Sited Batteries Deliver the Most Services
and Value to Customers and the Grid. Rocky Mountain Institute. https://rmi.org/wp-
content/uploads/2017/03/RMI-TheEconomicsOfBatteryEnergyStorage-FullReport-FINAL.pdf.
Frew, Bethany, Wesley Cole, Nina Vincent, Andrew Reimers, and Robert Margolis. 2018.
Impact of Dynamic Storage Capacity Valuation in Capacity Expansion Models: Preprint.
NREL/CP-6A20-71462. Golden, CO: National Renewable Energy Laboratory.
Fu, Ran, David Feldman, and Robert Margolis. 2018. U.S. Solar Photovoltaic System Cost
Benchmark: Q1 2018. NREL/TP-6A20-72399. Golden, CO: National Renewable Energy
Fu, Ran, Timothy Remo, and Robert Margolis. 2018. 2018 U.S. Utility-Scale Photovoltaics-Plus-
Energy Storage System Costs Benchmark. NREL/TP-6A20-71714. Golden, CO: National
Renewable Energy Laboratory. https://www.nrel.gov/docs/fy19osti/71714.pdf.
Fu, Ran, David Feldman, Robert Margolis, Mike Woodhouse, and Kristen Ardani. 2017. U.S.
Solar Photovoltaic System Cost Benchmark: Q1 2017. NREL/TP-6A20-68925. Golden, CO:
Fu, Ran, Donald Chung, Travis Lowder, David Feldman, Kristen Ardani, and Robert
Margolis. 2016. U.S. Solar Photovoltaic System Cost Benchmark: Q1 2016. NREL/TP-6A20-
Fu, Ran, Ted L. James, Donald Chung, Douglas Gagne, Anthony Lopez, and Aron Dobos. 2015.
“Economic Competitiveness of U.S. Utility-Scale Photovoltaic Systems in 2015: Regional Cost
Modeling of Installed Cost ($/W) and LCOE ($/kWh).” Presented at the IEEE 42nd Photovoltaic
Specialist Conference, New Orleans, LA. https://ieeexplore.ieee.org/document/7356261.
Gunda, T., and R. Homan. (Forthcoming). Evaluation of Component Reliability in Photovoltaic
Systems using Field Failure Statistics. Albuquerque, NM: Sandia National Laboratories.
Klise, G., O. Lavrova, and R. Gooding. 2018. PV System Component Fault and Failure
Compilation and Analysis. SAND Report No. 2018-1743. Albuquerque, NM: Sandia National
Laboratories. https://prod-ng.sandia.gov/techlib-noauth/access-control.cgi/2018/181743.pdf.
Lazard. 2018. Lazard’s Levelized Cost of Storage Analysis: Version 4.0. November 2018.
MEPS International Ltd. 2019. “MEPS: North American Carbon Steel Purchasing Price Index
(SPPI).” Accessed October 18, 2019. http://www.meps.co.uk/N.Amer%20Index.htm.
93
NEMA (National Electrical Manufacturers Association). 2020. Implementation of the National
Electrical Code, NEMA Field Representatives. April 1, 2020. Accessed June 23, 2020.
https://www.nema.org/Technical/FieldReps/Documents/Adoption%20of%20the%20National%2
0Electrical%20Code%20by%20State%20or%20local%20jurisdiction.pdf.
NREL (National Renewable Energy Laboratory). 2020. NREL dialogues and interviews with
solar industry collaborators. Golden, CO: National Renewable Energy Laboratory.
NREL (National Renewable Energy Laboratory), Sandia National Laboratories, SunSpec
Alliance, and the SunShot National Laboratory Multiyear Partnership (SuNLaMP) PV O&M
Best Practices Working Group. 2018. Best Practices for Operation and Maintenance of
Photovoltaic and Energy Storage Systems, 3rd Edition. NREL/TP-7A40-73822. Golden, CO:
PVinsights. 2019. PVinsights database. Accessed October 14, 2019: http://pvinsights.com.
———. 2020. PVinsights database. Accessed August 1, 2020.
RSMeans. 2017. RSMeans Building Construction Cost Data 2017. RS Means Catalog No.
60012.
Schmidt, Oliver, Sylvain Melchior, Adam Hawkes, and Iain Staffell. 2019. “Projecting the
Future Levelized Cost of Electricity Storage Technologies.” Joule 3(1): 81–100.
SolarEdge. 2019. SolarEdge quarterly presentations. Accessed October 14, 2019:
https://investors.solaredge.com/financial-information/presentations
Sonnen Batterie. 2018. “Sonnen Batterie Smart Energy Storage for Homeowners.” Accessed
April 10, 2018: https://sonnen-batterie.com/en-us/sonnenbatterie.
Sunrun. 2020. Sunrun quarterly presentations. Accessed August 1, 2020:
Tesla. 2020. Tesla quarterly presentations. Accessed August 1, 2020: https://ir.tesla.com/events
Vivint Solar. 2020. Vivint Solar quarterly presentations. Accessed August 1, 2020:
http://investors.vivintsolar.com/company/investors/events-and-
presentations/presentations/default.aspx.
Walker, Andy, Eric Lockhart, Jal Desai, Kristen Ardani, Geoff Klise, Olga Lavrova, Tom Tansy,
Jessie Deot, Bob Fox, and Anil Pochiraju. 2020. Model of Operation and Maintenance Costs for
Photovoltaic Systems. NREL/TP-5C00-74840. Golden, CO: National Renewable Energy
Wiser, Ryan H, Mark Bolinger, and Joachim Seel. 2020. “Benchmarking Utility-Scale PV
Operational Expenses and Project Lifetimes: Results from a Survey of U.S. Solar Industry
Professionals.” Berkeley, CA: Lawrence Berkeley National Laboratory.
https://emp.lbl.gov/publications/benchmarking-utility-scale-pv.
Wood Mackenzie and SEIA (Solar Energy Industries Association). 2020. U.S. Solar Market
Insight Report, Q2 2020. Washington, D.C.: Solar Energy Industries Association.
https://www.woodmac.com/research/products/power-and-renewables/us-solar-market-insight/.
94
———. 2018. U.S. Solar Market Insight Report, Q2 2018. Washington, D.C.: Solar Energy
Industries Association. https://www.woodmac.com/research/products/power-and-renewables/us-
solar-market-insight/.
Wood Mackenzie. 2020. The Global PV Inverter and MLPE Landscape, H1 2020.
———. 2019. Solar-Plus-Storage System Architectures: Design, Pricing, and Economics in
the U.S.
———. 2014a. The Microinverter and DC Optimizer Landscape 2014. January 2014.
———. 2014b. The Microinverter and DC Optimizer Landscape 2014, Updated March 2014.
March 2014.
95
Appendix A. Changes in Methodology Between
Q1 2018 and Q1 2020 Reports
Since 2010, NREL has performed PV system benchmark calculations. Each year we endeavor
to improve the modeling to better characterize the U.S. market and the costs associated with
installing (and operating, in the case of LCOE) residential, commercial, and utility-scale PV
systems. This year, to better distinguish the historical cost trends from the changes to our cost
models, we also calculate Q1 2019 and Q1 2020 PV benchmarks using the Q1 2018 versions of
the residential, commercial, and utility-scale standalone PV models. This appendix summarizes
the major changes we made in the models between the publication of the Q1 2018 and Q1 2020
reports.
Different Data Sources

We changed our data sources for several inputs to (1) create more consistency and transparency
across years, and (2) incorporate sources that include data from a larger part of the U.S. market.
Table A-1 summarizes the differences in data sources and the associated inputs in benchmarking
PV system costs between this year’s report and the 2018 report.
Table A-1. Comparison of Input Assumptions and Sources in the Q1 2018 Benchmark Report and
the Q1 2020 Benchmark Report
Model Input Q1 2018 Model: Sources Q1 2020 Model: Sources
Inverter cost PVinsights (2019, 2020) for string Wood Mackenzie and SEIA (2020)
and central inverters; public for all inverter types; the switch to
corporate filings from Enphase one data source provides more
(2019) and SolarEdge (2019) were consistency across years and
used to calculate costs for between sectors. Q1 2019: $0.14/W
microinverters and DC optimizers, residential string inverters; $0.30/W
respectively, using revenue per watt residential power optimizers plus
shipped. Q1 2019: $0.12/W string inverters; $0.34/W
residential string inverters; $0.24/W microinverters; $0.09/W three-
power optimizers plus string phase commercial; $0.15/W three-
inverters; $0.36/W microinverters; phase commercial with power
$0.06/W commercial; $0.04/W optimizers; $0.06/W utility-scale. Q1
utility-scale. Q1 2020: $0.10/W 2020: $0.15/W residential string
residential string inverters; $0.23/W inverters; $0.30/W residential power
power optimizers plus string optimizers plus string inverters;
inverters; $0.35/W microinverters; $0.34/W microinverters; $0.078/W
$0.05/W commercial; $0.03/W three-phase commercial; $0.14/W
utility-scale. three-phase commercial with power
optimizers; $0.069/W utility-scale.
Module efficiency California’s NEM Interconnected California’s NEM Interconnected

Data Set, using average module Data Set, using capacity-weighted
power for the previous year’s PV average module efficiency of 60-cell
system in the residential sector for and 72-cell monocrystalline or
the residential PV model—311 multicrystalline modules for PV
watts-peak (Wp, 2018) and 319 Wp systems installed in that year.
(2019)—divided by the average Monocrystalline: 19.2% (2019) and
96
Model Input Q1 2018 Model: Sources Q1 2020 Model: Sources
module size for the entire data set 19.5% (Q1 2020). Multicrystalline:
(1.64 m2 (2018) and 1.65 m2 17.3% (2019) and 17.4%
(2019)), giving an estimated module (Q1 2020).
efficiency in residential PV systems
of 19.0% (2018) and 19.4% (2019).
Using average module power for the
previous year’s PV system in the
commercial sector for the
commercial PV model (330 Wp
(2018) and 343 Wp (2019)), divided
by the average module size for the
entire data set (1.64 m2 (2018) and
1.65 m2 (2019)), giving an estimated
module efficiency in commercial PV
systems of 20.1% (2018) and 20.8%
(2019).
Module price Market-share-weighted-average U.S. monocrystalline silicon spot

monocrystalline and multicrystalline price from Wood Mackenzie and
spot price from Wood Mackenzie SEIA (2020) in the first quarter of
and SEIA (2020) in the first quarter the year. Q1 2019: $0.40/W.
of the year. Q1 2019: $0.39/W. Q1 Q1 2020: $0.41/W.
2020: $0.38/W.
Residential inverter California’s NEM Interconnection Tracking the Sun data set;
market share Data Set, using the percentage of these data include a broader
market penetration of the previous representation of the United States
year (by installed capacity), than just California. Residential
assuming all microinverters were market share of string inverters,
represented by Enphase inverters, power optimizers, and
DC optimizers were represented by microinverters for latest year
SolarEdge inverters, and the available: 2018 (used for both
remainder were string inverters. Q1 2019 and Q1 2020 benchmark):
Residential market share of string 14.6%, 49.8%, and 35.6%.
inverters, power optimizers, and
microinverters: 2018 (61.8%, 16.8%,
21.4%); 2019 (59.2%, 18.8%,
22.0%).
Residential business Corporate filings from Sunrun TPO market share from Tracking
structure market share (2020), Tesla (2020), and Vivint the Sun data set to estimate
Solar (2020) to estimate national national integrator market share,
integrator Q1 2019 (33%) and Q1 with the remainder classified as
2020 (30%) market share; Wood small installers; the TPO data
Mackenzie and SEIA (2020) to include other national integrators
estimate the remainder, classified besides Sunrun, Tesla, and Vivint.
small installers. The 2018 (latest year available)
TPO market share of 38% is used
for both Q1 2019 and Q1 2020.
97
Different Methodology for Calculating Residential Overhead, Customer Acquisition,
and PII
In this year’s version of our benchmark analysis, we expand our modeling of customer
acquisition, engineering, PII, and overhead. In addition to providing finer granularity to costs, we
include costs borne by many installers throughout the United States. Table A-2 summarizes the
current and previous methods.
Table A-2. Comparison of Methods for Calculating Q1 2018 Residential PV Soft Costs in the Q1
2018 Benchmark Report and the Q1 2020 Benchmark Report
Residential Q1 2018 Model: Summary of Method Q1 2020 Model: Summary of Method

Soft Cost (Value) (Value)
PII A permit fee of $200 and six hours of An itemized list of steps and
staff time ($0.05/W). associated labor and other costs
needed to design the initial and final
system plans, apply for and receive a
permit and interconnection agreement,
multiplied by the estimated percentage
of national sales that use this step,
divided by the average conversion
from this step to an installed system
(to account for the cost of lost sales)
($0.24/W). Many of these costs were
not captured in previous editions.
Sales and Data from the 2013 report (Feldman et An estimated breakdown of the
marketing al. 2013), which calculated sales and necessary steps and associated labor
(customer marketing costs by estimating the and other costs of customer
acquisition) headcount, salaries, benefits, and acquisition, including advertisement,
taxes of the sales, engineering, and lead generation, qualifications/first
marketing departments, as well as sales pitch, and final sales pitch. Each
vehicle costs ($0.37/W). possible customer acquisition pathway
is multiplied by the estimated
percentage of national sales that use
this step, divided by the average
conversion from this step to an
installed system (to account for the
cost of lost sales) ($0.43/W).
Overhead Data from the 2013 report (Feldman et al. An update of previous cost categories,
(general and 2013), which calculated overhead costs including the overhead of the larger
administrative) by estimating the headcount, salaries, staff associated with customer
benefits, and taxes of the management, acquisition and PII; the model also
human resources, project management, assumes 10% is added to the base
administration, supply chain, information salary to account for training expenses
technology, and customer service ($0.27/W).
departments, as well as rent and other
office expenses, professional services,
insurance, taxes, dues, and memberships
($0.34/W).
98
Incorporation of MLPE into the Commercial Rooftop PV Model
MLPE are growing to be a substantial part of the commercial PV rooftop market in the United
States owing in part to the adoption by many states of the 2017 NEC, which requires rooftop PV
systems to have module-level rapid shutdown. In past years, we only assumed string inverters for
the commercial PV benchmark, only weighting the residential PV system benchmark by MLPE
market share. This year, we also weight the commercial rooftop PV benchmark by MLPE market
share (45% for three-phase string inverters, 39% for power optimizers, and 16% for
microinverters).
Annual Updates of Installation Labor Rates

In previous year’s models, we adjusted the 2012 labor rates by inflation, using the CPI “All
Urban Consumers” series. For this year’s model, we pull each year’s labor rate directly from the
Bureau of Labor Statistics.
Elimination of State Variations from the Cost of Doing Business

In previous year’s models, we adjusted installation labor rates and supply chain costs for each
state by the state’s cost of doing business, as reported by Case (2012). For this year’s report, we
do not use a cost of doing business adjustment because all costs are use national averages.
Changes to Calculating Number of Modules in Commercial and Utility-Scale PV Models

Previous year’s models calculated the number of modules in a system by dividing system size
by 310.4 W, rounded to the nearest whole number. We update the model to calculate the number
of modules by dividing system size by module efficiency and module area, and rounding any
fraction up to the closest whole number.
Changes to the Cost Classification of Land Acquisition for Utility-Scale PV Models

For previous editions of this report, we assumed a land acquisition cost of $0.03/W. Based on
which the PV system is placed, we reclassify land costs from an upfront capital expenditure
(land acquisition) to an operating expenditure (lease payments) for 2019 and 2020.
Switching from Weighting Costs by State PV Installation Levels

In previous year’s models, our national average benchmarks were calculated by weighting the
state averages of sales tax, labor rates, wind load, snow load, and material and equipment
location factor by the amount of PV capacity installed in each state in the previous year for that
sector (utility-scale, commercial, residential). We update the model to use national average labor
rates and the average values of state sales tax, wind load, snow load, and material and equipment
location factor.
Changes to Reported Dollar Year Calculation

In previous year’s models, we adjusted values for inflation based on a partial year of CPI data.
For example, in the Q1 2018 benchmark report (Fu, Feldman, and Margolis 2018), all values
are quoted in $2018; however, the inflation adjustment is based on the average CPI Index of Q1
2018 (January through March 2018). Because the benchmark reports are produced before the end
of the calendar year, indexing them in that year is not possible. To better correct for inflation, in
this year’s report, we quote values in previous year’s dollars ($2019). In 2018, the CPI-All Urban
99
Consumers Index is 248.8 for the first three months and 251.1 for the whole year (and 255.7
for 2019).
The changes summarized in this appendix result in Q1 2019 and Q1 2020 benchmarks with
different results than would have been calculated using the previous edition’s models and
assumptions. For example, the 2020 total residential PV installed-cost benchmark calculated
using the Q1 2018 model is $2.47/WDC, whereas the same benchmark calculated using the
Q1 2020 model is $2.71/WDC (7% higher).
Table A-3 summarizes the impacts these changes have on each cost category in the residential,
commercial, and utility-scale PV benchmarks for Q1 2020.
100
Table A-3. Comparison of Q1 2020 Benchmark Costs, per Category, Calculated Using Previous
Report’s Model (Q1 2018) and the Current Model (Q1 2020) in 2019 USD
Residential PV Commercial Rooftop PV Utility-Scale, Fixed-Tilt Utility-Scale, One-Axis

(2020) (2020) (2020) (2020)
Q1
Q1 2018 Q1 2020 % Q1 2020 % Q1 2018 Q1 2020 % Q1 2018 Q1 2020 %
2018
Model Model Change Model Change Model Model Change Model Model Change
Model
Module $0.376 $0.406 5% $0.376 $0.406 5% $0.376 $0.406 5% $0.376 $0.406 5%
Inverter $0.217 $0.250 12% $0.045 $0.123 169% $0.022 $0.051 127% $0.022 $0.052 127%
Structural BOS $0.084 $0.084 -3% $0.112 $0.110 -4% $0.087 $0.080 -10% $0.130 $0.122 -9%
Electrical BOS $0.190 $0.228 17% $0.133 $0.133 -3% $0.088 $0.073 -19% $0.088 $0.073 -19%
Supply chain costs $0.254 $0.261 0% — — — — — — — — —
Installation labor $0.242 $0.187 -25% $0.159 $0.148 -9% $0.094 $0.102 5% $0.102 $0.111 6%
PII $0.050 $0.238 366% $0.100 $0.106 3% $0.033 $0.030 -13% $0.033 $0.030 -13%
Transmission Line
— — — — — — $0.019 $0.017 -13% $0.019 $0.017 -13%
(if any)
Sales and marketing
$0.360 $0.428 15% — — — — — — — — —
(customer acquisition)
Overhead $0.327 $0.274 -18% $0.526 $0.492 -8% $0.068 $0.068 -3% $0.077 $0.076 -3%
Contingency — — — $0.040 $0.044 7% $0.024 $0.026 3% $0.026 $0.027 2%
Profit $0.296 $0.292 -4% $0.107 $0.113 2% $0.042 $0.045 2% $0.046 $0.048 2%
Sales tax $0.077 $0.063 -21% $0.045 $0.046 1% $0.038 $0.041 4% $0.041 $0.043 4%
Total price $2.474 $2.710 7% $1.642 $1.720 2% $0.891 $0.937 2% $0.959 $1.005 2%
101
Appendix B. PV System LCOE Benchmarks in 2019 USD
Table B-1. NREL LCOE Summary (2019 cents/kWh)
Steady-
Market Financing Rates
State Financing
2010
2011
2012
2013
2014
2015
2016
2017
2018
2019
2020
2020
Reporting Year
Goal 2030
2020 Goal
Q4 2009
Q4 2010
Q4 2011
Q4 2012
Q4 2013
Q1 2015
Q1 2016
Q1 2017
Q1 2019
Q1 2020
Q1 2020
Q12018
Benchmark Date
Residential (6.9 kW)

High resource (CF 21.6%), no ITC 41.6 35.0 24.1 20.2 16.9 15.0 13.8 12.9 12.0 11.2 11.0 10.5 — —
Medium resource (CF 17.6%), no ITC 50.9 42.8 29.5 24.7 20.8 18.4 16.9 15.8 14.8 13.7 13.5 12.8 10.6 5.3
Low resource (CF 16.4%), no ITC 54.7 46.0 31.7 26.6 22.3 19.7 18.1 17.0 15.9 14.7 14.5 13.8 — —
High resource (CF 21.6%), ITC 27.6 23.2 16.1 13.5 11.3 9.9 9.1 8.6 8.0 7.1 7.1 — — —
Medium resource (CF 17.6%), ITC 33.9 28.4 19.8 16.5 13.8 12.1 11.2 10.5 9.8 8.7 8.7 — — —
Low resource (CF 16.4%), ITC 36.4 30.5 21.2 17.7 14.8 13.0 12.0 11.3 10.5 9.4 9.3 — — —
Commercial Rooftop (200 kW)

102
Commercial Ground-Mount (500 kW)
High resource (CF 21.6%), no ITC — — — — — — — — — — 7.1 6.7 — —
Medium resource (CF 17.6%), no ITC — — — — — — — — — — 8.7 8.2 — —
Low resource (CF 16.4%), no ITC — — — — — — — — — — 9.3 8.8 — —
High resource (CF 21.6%), ITC — — — — — — — — — — 4.5 — — —
Medium resource (CF 17.6%), ITC — — — — — — — — — — 5.6 — — —
Low resource (CF 16.4%), ITC — — — — — — — — — — 6.0 — — —
Utility-Scale (100 MW One-Axis Tracking)
Utility-Scale (100 MW Fixed-Tilt)

Medium resource (CF 17.3%), no ITC 27.7 23.2 15.7 12.0 10.8 10.1 8.1 6.1 5.8 4.9 4.9 4.6 — —
a 2020 residential and commercial SETO goals are adjusted for inflation using the CPI; the 2020 utility-scale goal was left unchanged, because wholesale prices were
relatively flat, and in some cases declined, from 2010 to 2020.
103

U.S. Solar Photovoltaic System and Energy Storage Cost Benchmark: Q1 2020

Uploaded by

Copyright:

Available Formats

U.S. Solar Photovoltaic System and Energy Storage Cost Benchmark: Q1 2020

Uploaded by

Document Information

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

U.S. Solar Photovoltaic System and Energy Storage Cost Benchmark: Q1 2020

Uploaded by

Copyright:

Available Formats

U.S.

Solar Photovoltaic System and

NREL is a national laboratory of the U.S. Department of Energy Technical Report

Contract No. DE-AC36-08GO28308

NREL is a national laboratory of the U.S. Department of Energy Technical Report

This report is available at no cost from the National Renewable

U.S. Department of Energy (DOE) reports produced after 1991

NREL prints on paper that contains recycled content.

Values 2019 U.S. dollars (USD)a

PV Sector Description Size Range

Residential Residential rooftop systems, monocrystalline silicon modules 4kW–7 kW

Utility-scale Ground-mounted systems, monocrystalline silicon modules, fixed- 5–100 MW

• $2.71 per watt DC (WDC) (or $3.12/WAC) for residential PV systems

Commercial Rooftop Utility-Scale PV,

Drivers of cost • Higher module • Higher module • Higher module

Figure ES-5. Residential PV-plus-storage system cost benchmark summary (inflation-adjusted),

2.1 Tracking the Sun Data Set

2.2 Module Efficiency

2.3 PV System Size

2.4 Module-Level Power Electronics

Table 1. Rapid-Shutdown Codes: Progress by State

Code Rapid- State

Figure 6. Inverter prices from Wood Mackenzie, 2010–2020

Inverter Type Sector USD/WAC DC-to-AC Ratioa USD/WDC

Residential and commercial PV

Central inverter Utility-scale PV (fixed-tilt) 0.07 1.37 0.05

Central inverter Utility-scale PV (1-axis tracker) 0.07 1.34 0.05

All inverter prices include the cost of monitoring equipment.

2.7 Module Prices

2.8 Battery Storage

Operation and Maintenance

3.1 Residential Model Structure, Inputs, and Assumptions

Figure 11. Residential PV: Model structure

Category Modeled Value Description Sources

3.2 Expanded “Other Soft Costs” Modeling

3.3 Residential Model Output

3.4 Residential Model Output versus Reported Costs

3.5 Retrofits versus New Construction

3.6 Additional Costs Typical of Residential PV Installation

3.7 Residential PV Price Benchmark Historical Trends

Comparing Multicrystalline and Monocrystalline PV Systems

3.8 Residential PV LCOE Historical Trends

4.1 Commercial Model Structure, Inputs, and Assumptions

Table 5. Commercial PV: Modeling Inputs and Assumptions

Category Modeled Value Description Sources

Burden rates Total nationwide Workers compensation, RSMeans 2017

4.2 Commercial Model Output

4.4 Commercial PV LCOE Historical Trends

5.1 Utility-Scale Model Structure, Inputs, and Assumptions

Figure 28. Utility-scale PV: Model structure

Category Modeled Value Description Sources

Contingency 3% Estimated as markup on EPC cost NREL 2020

5.4 Utility-Scale PV LCOE Historical Trends

• PV system capacity (kW)

6.1 Residential Li-Ion Standalone Storage Cost Model

Table 9. Residential Storage-Only Modeling Inputs and Assumptions

Category Modeled Value Description

Battery pack $253/kWh Battery pack only

Battery-based $174/kWh 6-kW, 48-V bidirectional inverter (less