Techno-Economic Analysis: Best Practices and Assessment Tools

SANDIA REPORT
SAND2020-13473
Printed December 2020
Techno-Economic Analysis: Best

Practices and Assessment Tools
Peter H. Kobos, Thomas E. Drennen, Alexander ‘Sasha’ Outkin, Erik K. Webb, Scott M.
Paap and Steven Wiryadinata.
Prepared by
Sandia National Laboratories
Albuquerque, New Mexico
87185 and Livermore,
California 94550
Issued by Sandia National Laboratories, operated for the United States Department of Energy by National
Technology & Engineering Solutions of Sandia, LLC.
NOTICE: This report was prepared as an account of work sponsored by an agency of the United States
Government. Neither the United States Government, nor any agency thereof, nor any of their employees, nor any of
their contractors, subcontractors, or their employees, make any warranty, express or implied, or assume any legal
liability or responsibility for the accuracy, completeness, or usefulness of any information, apparatus, product, or
process disclosed, or represent that its use would not infringe privately owned rights. Reference herein to any specific
commercial product, process, or service by trade name, trademark, manufacturer, or otherwise, does not necessarily
constitute or imply its endorsement, recommendation, or favoring by the United States Government, any agency
thereof, or any of their contractors or subcontractors. The views and opinions expressed herein do not necessarily
state or reflect those of the United States Government, any agency thereof, or any of their contractors.
Printed in the United States of America. This report has been reproduced directly from the best available copy.
Available to DOE and DOE contractors from

U.S. Department of Energy
Office of Scientific and Technical Information
P.O. Box 62
Oak Ridge, TN 37831
Telephone: (865) 576-8401

Facsimile: (865) 576-5728
E-Mail: reports@osti.gov
Online ordering: http://www.osti.gov/scitech
Available to the public from

U.S. Department of Commerce
National Technical Information Service
5301 Shawnee Rd
Alexandria, VA 22312
Telephone: (800) 553-6847

Facsimile: (703) 605-6900
E-Mail: orders@ntis.gov
Online order: https://classic.ntis.gov/help/order-methods/
2
ABSTRACT
A team at Sandia National Laboratories (SNL) recognized the growing need to maintain and
organize the internal community of Techno-Economic Assessment analysts at the lab. To meet this
need, an internal core team identified a working group of experienced, new, and future analysts to: 1)
document TEA best practices; 2) identify existing resources at Sandia and elsewhere; and 3) identify
gaps in our existing capabilities. Sandia has a long history of using techno-economic analyses to
evaluate various technologies, including consideration of system resilience. Expanding our TEA
capabilities will provide a rigorous basis for evaluating science, engineering and technology-oriented
projects, allowing Sandia programs to quantify the impact of targeted research and development
(R&D), and improving Sandia’s competitiveness for external funding options. Developing this
working group reaffirms the successful use of TEA and related techniques when evaluating the
impact of R&D investments, proposed work, and internal approaches to leverage deep technical and
robust, business-oriented insights. The main findings of this effort demonstrated the high-impact
TEA has on future cost, adoption for applications and impact metric forecasting insights via key
past exemplar applied techniques in a broad technology application space. Recommendations from
this effort include maintaining and growing the ‘best practices’ approaches when applying TEA,
appreciating the tools (and their limits) from other national laboratories and the academic
community, and finally a recognition that more proposals and R&D investment decisions locally at
Sandia, and more broadly in the research community from funding agencies, require TEA
approaches to justify and support well thought-out project planning.
3
ACKNOWLEDGEMENTS
The Techno-Economic Analysis (TEA) team thanks Director Carol Adkins and the Energy and
Homeland Security Portfolio office for supporting this Strategic Initiative. Thanks also go to David
Kistin, Liz Hillman, Michael Bernard, Greg Frye-Mason, Tito Bonano, Carmen Mendez, Amy
Halloran, the February 24, 2020 TEA Workshop participants for their ideas, contributions and
Geoff Klise for his review of this report.
4
Table of Contents
Abstract ................................................................................................................................ 3
Acknowledgements ............................................................................................................... 4
List of Figures ........................................................................................................................ 7
List of Tables ....................................................................................................................... 10
Executive Summary ............................................................................................................. 12
Acronyms and Definitions .................................................................................................... 15
1. An Introduction to Techno-Economic Analysis (TEA) ..................................................... 17
1.1. Evaluating the Economics of New Technologies ................................................................17
1.2. Levelized Cost of Energy Methodology .............................................................................21
1.3. Experience with LCOE at Sandia........................................................................................24
1.4. Estimating the LCOE for a new technology ........................................................................25
1.4.1. Mechanical System Costs.................................................................................................. 26
1.4.2. Scaling equations .............................................................................................................. 27
1.4.3. Other Additional Costs ...................................................................................................... 28
1.5. First- and Nth-of-a-kind Methodology ..............................................................................29
1.6. Additional Details on using LCOE Analysis for a New Technology: the SCO2 Brayton System
31
1.6.1. The sCO2 Brayton system ................................................................................................. 32
1.6.2. Example: The Estimated LCOE for a 100 MWe sCO2 system ........................................... 33
1.6.3. Sensitivity Analysis ............................................................................................................ 35
1.7. TEA, Resilience, and Planning ...........................................................................................38
1.7.1. Community Critical Service Availability and Burden Metrics ........................................... 39
1.7.2. Economic resilience metrics ............................................................................................. 40
1.7.3. Resilience in master planning frameworks ....................................................................... 42
2. Addditional TEA Examples: Objectives, Results, and Lessons Learned .......................... 44
2.1. Guiding Optimal Biofuels (TEA of Advanced Biofuels)........................................................44
2.2. Industrial process water treatment and reuse enabled by selective ion exchange materials
45
2.3. TEA of Solid-State Transformers .......................................................................................46
2.4. Microsystems-Enabled Photovoltaics (MEPV) ...................................................................46
2.5. Potential Strategies for Integrating Solar Hydrogen Production and Concentrating Solar Power
47
2.6. Pathways to Carbon Neutral Energy Systems at the University of California, Davis ............ 48
2.7. Industrial process water treatment and reuse enabled by selective ion exchange materials
49
2.8. Large Scale Geologic Storage of Hydrogen ........................................................................50
5
2.9. CO2 Capture, Management and Water Treatment .............................................................53
2.10. Economic Impacts of Nuclear Power Plants Accident Analysis ...........................................55
3. Illustrative TEA Strengths and Competencies at other National Labs ............................. 58
3.1. Argonne National Lab (ANL) .............................................................................................58
3.2. National Renewable Energy Lab (NREL) ............................................................................62
3.3. National Energy Technology Laboratory (NETL) .................................................................66
4. References ................................................................................................................... 71
Appendix A. Additional Details on TEA Studies Summarized in Main Report ...................... 76
A.1. Guiding Optimal Biofuels (TEA of Advanced Biofuels)........................................................76
A.2. Industrial process water treatment and reuse enabled by selective ion exchange materials
92
A.3. TEA of Solid-State Transformers ..................................................................................... 102
A.4. Microsystems-Enabled Photovoltaics (MEPV) ................................................................. 112
A.5. Potential Strategies for Integrating Solar Hydrogen Production and Concentrating Solar Power
127
A.6. Pathways to Carbon Neutral Energy Systems at the University of California, Davis .......... 141
A.7. Technical and economic feasibility of unitary, horizontal ground-loop geothermal heat pumps
for space conditioning in selected California climate zones ................................................. 156
A.8. DYMATICA ..................................................................................................................... 168
A.9. Economic Impacts of Nuclear Power Plants Accidents Analysis ........................................ 170
A.9.1. Cost-Based Model Overview........................................................................................... 171
A.9.2. GDP Impact Model (RDEIM) Overview ........................................................................... 172
A.9.3. NRC Economic Impacts Estimation and Cost-Benefit Analysis ....................................... 173
A.9.4. Simple Example ............................................................................................................... 175
A.9.5. Presentation and Use of Results ..................................................................................... 184
Appendix B. Collaborative Workshop and Participation ..................................................... 186
Distribution ....................................................................................................................... 188
6
LIST OF FIGURES
Figure 1. Lazard's estimated LCOE (Source: Lazard, 2019) ....................................................................18

Figure 2. Lazard's LCOE estimates 2009 - 2019 (Lazard, 2019) ..............................................................19
Figure 3. EIA's estimated regional LCOE to LACE values for new generating technologies
entering service in 2025 and 2040 (Source: EIA, 2020c) ...................................................................20
Figure 4. EIA projections of capacity additions and retirements in the Annual Energy Outlook 2020 .20
Figure 5. Regional variation in LACE for new generation entering service in 2025 (2019 $ per
MWh) (Source: EIA, 2020d) ..................................................................................................................21
Figure 6. Simplified Power Systems Lifecycle Analysis Tool interface (Drennen, 2016) .....................25
Figure 7. LCOE estimating tool within the supercritical Brayton Cycle Evaluation tool (Drennen,
2020)............................................................................................................................................................27
Figure 8. EIA (2016) cost estimates for a new NGCC ..............................................................................29
Figure 9. Theoretical Learning Curve (NETL, 2013). ................................................................................30
Figure 10. NOAK costs (typically $/kWe) as a function of number of units (n) and the learning
rate (R) ........................................................................................................................................................31
Figure 11. Estimated system configuration and efficiency from sBET ...................................................33
Figure 12. LCOE cost breakdown for 100 MWe Brayton system ...........................................................34
Figure 13. LCOE as a function of system size and turbine inlet temperature........................................35
Figure 14. LCOE and system efficiency as a function of turbine inlet temperatures and fuel costs ..36
Figure 15. Sensitivity analysis on recuperator effectiveness and fuel cost (100 MWe, 700°C) ............37
Figure 16. Sensitivity analysis on turbine efficiency (100 MWe, 700 °C) ................................................38
Figure 17. The burden to achieve all lifeline services is calculated based on a portfolio of 80
microgrids sited throughout Puerto Rico. (Jeffers et al. 2018) ...........................................................40
Figure 18. The avoided losses as a function of a microgrid location by county for a specific
disruption scenario ....................................................................................................................................41
Figure 19. Blue-sky Performance versus Resilience (Jeffers et al. 2020) .................................................42
Figure 20. The Geologic Hydrogen Storage model and assessment framework (Lord et al., 2014) ...51
Figure 21. Geologic Hydrogen Storage Cost Model. (levelized cost of Hydrogen, $/kg; Lord et
al., 2014)......................................................................................................................................................52
Figure 22. Total Capital costs, four city demand scenarios to meet 10, 25 or 100% of city
Hydrogen transportation demand (Lord et al., 2014) ..........................................................................52
Figure 23. Water, Energy and Carbon Sequestration Simulation Model (WECSsim©) Assessment
Framework (Klise et al., 2013; Kobos et al., 2016) ..............................................................................53
Figure 24. Base Case CO2 capture, storage and water treatment costs for a large, coal-fired power
station (Kobos et al., 2011) ......................................................................................................................54
Figure 25. United States National Power Plant water demand and extracted saline formation
water for all coal and natural gas fired power stations (Kobos et al., 2016).....................................54
Figure 26. United States national cost curve for CO2 capture and storage for all coal and natural
gas fired power station system (Kobos et al., 2016).............................................................................55
Figure 27. Sample output from the web-based version of GREET (Greenhouse Gases) ...................59
7
Figure 28. Sample output from the web-based version of GREET (Energy Usage) ............................60
Figure 29. Representative output from Argonne's AFLEET model .......................................................61
Figure 30. NREL's simplified levelized cost of energy calculator ............................................................63
Figure 31. Sample output from JEDI model for a 100 MWe CSP project in New Mexico .................66
Figure 32. Greenhouse gas emissions associated with the current grid mix in the southwest US ......68
Figure 33. Eutrophication potential associated with the current grid mix in the southwest US .........69
Figure 34. Mercury emissions associated with the current grid mix in the southwest US ....................69
Figure 35. Process pathways evaluated in this study for the conversion of cellulosic feedstocks to
ethanol and biodiesel (FAEE) .................................................................................................................79
Figure 36. Monte Carlo analysis to explore the impact of uncertainty in conversion process model
parameters ..................................................................................................................................................82
Figure 37. (a) Minimum fuel selling price ($ per ethanol-equivalent liter), (b) Fuel yield (ethanol-
equivalent liters per metric ton of dry biomass), (c) Net electricity generation (MW) for a
2,000 metric ton/day conversion facility, and (d) process water consumption (liters of water
per liter of ethanol-equivalent fuel produced) for all pathways to ethanol and FAEE. .................84
Figure 38. Net GHG emissions (g CO2 equivalents per MJ of fuel produced) for all pathways to
ethanol and FAEE ....................................................................................................................................85
Figure 39. Contributions to the variance in minimum fuel selling price for process to produce (a)
ethanol and (b) FAEE from switchgrass using dilute acid pretreatment..........................................86
Figure 40. Comparison of costs for unit operations and major inputs for the ‘most likely’ input
parameters for ethanol and FAEE production from switchgrass using dilute acid
pretreatment ...............................................................................................................................................87
Figure 41. Minimum fuel selling price ($/ethanol-equiv liter) for the EtOH and FAEE scenarios ...88
Figure 42. Comparison of costs for the ‘most likely’ parameter values of the ethanol baseline
scenario and FAEE best case scenario by unit operation and major inputs ....................................89
Figure 43. Novel silica getter materials potentially enable the reuse of produced water for cooling
applications in thermoelectric power generation..................................................................................93
Figure 44. Process flow diagram for HTC calcination, silica removal, and post-treatment
conditioning ...............................................................................................................................................96
Figure 45. Single parameter sensitivity analysis on minimum water treatment price (MWTP)............98
Figure 46. Stochastic minimum water treatment price distributions for four scenarios given
potential improvements on HTC materials........................................................................................ 100
Figure 47: Transformer voltage levels........................................................................................................ 103
Figure 48: Example calculation of dry-type transformer capacity rating based on the device
efficiency curve ....................................................................................................................................... 106
Figure 49: Components of total purchase price for the representative SST architectures ................ 108
Figure 50. Price vs Rdson for Cree SiC MOSFETS ................................................................................... 109
Figure 51: Comparison of SST and conventional transformer retail price versus capacity.
Conventional transformer price gathered from manufacturer’s website in July, 2019 ................ 110
Figure 52. Schematic representation of a microsystems-enabled photovoltaics (MEPV) concept. . 113
Figure 53. Mathematical equation for calculating the Levelized Cost of Electricity (LCOE). NPV
refers to the net present value of the quantities in brackets. ........................................................... 114
8
Figure 54. MEPV cost modeling framework and components ............................................................. 115
Figure 55. Energy flows and losses in an MEPV system (FOV: Field of View) ................................. 117
Figure 56. Identification of the optimal region of concentration ratio and cell size for an early
MEPV module concept......................................................................................................................... 118
Figure 57. Optical system costs as a function of cell size and concentration ratio. ............................ 118
Figure 58. Schematic representations of (a) an early MEPV module design and (b) the final
module design (Prototype 4). ............................................................................................................... 119
Figure 59. Output of a break-even analysis comparing EVA and PDMS as lens material ................ 120
Figure 60. Cost of MEPV cell placement with "pick-and-place" tools. ............................................... 120
Figure 61. Contour plot of cell placement costs, based on 250 µm cells, 200X concentration ratio,
5-year tool life, and 500 cells per placement cycle ............................................................................ 121
Figure 62. Components of estimated (2013) and projected (2020) cell fabrication cost for the final
MEPV design.......................................................................................................................................... 122
Figure 63. LCOE estimates for MEPV and "hybrid" modules in two representative locations. ..... 123
Figure 64. 2013 (Prototype 3) and 2020 (Prototype 4) MEPV module cost estimates ...................... 124
Figure 65. The contributions of non-module system costs as a function of module efficiency,
with estimates of MEPV module cost for multiple prototypes ...................................................... 125
Figure 66. Flow schematic of a molten-salt central receiver system. EPG: Electric power
generation system. Source: SAND2011-9320. ................................................................................... 129
Figure 67. Energy flow diagram of solar energy capture and conversion to heat in a representative
power tower with molten salt heat transfer fluid and nine hours of thermal storage .................. 131
Figure 68. Estimated future hydrogen production costs for four representative hydrogen
production technologies, from H2A models. .................................................................................... 132
Figure 69. Estimated current and future hydrogen production cost for PEM electrolysis as a
function of electricity cost .................................................................................................................... 132
Figure 70. Schematic representation of Case 1 of Scenario 2 ................................................................ 133
Figure 71. Schematic representation of Case 2 of Scenario 2. .............................................................. 134
Figure 73. Comparison of estimated hydrogen costs for Scenario 2 Cases ......................................... 135
Figure 74. Schematic representation of a high-temperature thermochemical metal oxide hydrogen
production cycle. .................................................................................................................................... 136
Figure 76. Schematic representation of Case 2 of Scenario 3. ............................................................... 137
Figure 77. Schematic representation of Case 3 of Scenario 3. ............................................................... 137
Figure 78. Hydrogen production costs versus electricity price for Scenario 3 Cases ......................... 138
Figure 79. Contributions to hydrogen production costs versus electricity price for Scenario 3
Cases reflecting an “optimal” future hydrogen production process and the 2015 state of the
art yielding more waste heat ................................................................................................................. 139
Figure 80. Hydrogen production costs versus electricity price for Scenario 3 Cases reflecting an
optimized process (solid lines) and the 2015 state of the art (dashed)........................................... 139
Figure 81. Projected annual cash flow over the analysis period............................................................. 151
Figure 82. Sensitivity of Baseline to changes in carbon cost and escalation rate ................................ 152
9
Figure 83. Base case cumulative present cost (PC) .................................................................................. 153
Figure 84. Alt_1 cumulative present cost (PC)......................................................................................... 154
Figure 87. Schematic of the modeled lodging facility consisting of four 4-story buildings ............... 159
Figure 88. Schematic of the modeled 2-story multifamily apartment ................................................... 159
Figure 89. LCOS for modeled lodging facility.......................................................................................... 164
Figure 90. LCOS for modeled multifamily facility ................................................................................... 165
Figure 91. Inland lodging sensitivity .......................................................................................................... 166
Figure 92. Coastal lodging sensitivity ......................................................................................................... 166
Figure 93. Illustrative DYMATICA assessment tailored to fit the need of the customer. ................ 169
Figure 94. Annual direct loss in GDP ($M) to regional economy. All values are discounted to
2011 (year 1) using the social discount rate. ....................................................................................... 178
Figure 95. Direct annual losses as percent of regional GDP. ................................................................ 179
Figure 96. Regional annual GDP with and without the disruption. All values are discounted to
2011 (year 1) using the social discount rate. ....................................................................................... 180
Figure 97. National GDP with and without disruption. All values are discounted to 2011 (year 1
in the plot) using the social discount rate. .......................................................................................... 182
Figure 98. GDP losses as a % of national GDP ...................................................................................... 183
Figure 99. Direct, total, indirect, and induced annual losses at the national level. All values are
discounted to 2011 (year 1 in the plot) using the social discount rate. .......................................... 184
LIST OF TABLES
Table 1. Operating assumptions for 100 MWe reference case. ................................................................33
Table 2. Cost comparison of silica removal technologies. .........................................................................97
Table 3. Comparison of the components of LCOE for non-concentrating PV, CPV, and MEPV
technologies. (O&M: Operations and Maintenance.) ....................................................................... 114
Table 4. Energy System for Each Scenario ............................................................................................... 145
Table 5. Phasing and Economic Summary of Alternatives .................................................................... 149
Table 6. Summary of loads and facility area.............................................................................................. 161
Table 7. Summary of System Configuration ............................................................................................. 163
Table 8. Summary of Energy Use and Emissions Reductions ............................................................... 164
Table 9. Employment by industry in affected counties........................................................................... 175
Table 10. Fraction of each county affected............................................................................................... 176
Table 11. Lost employment by grid element and industry ..................................................................... 176
Table 12. Grid element recovery schedule ................................................................................................ 176
Table 13. Value added per worker for each industry ............................................................................... 176
Table 14. Value added multipliers of Type I and Type II....................................................................... 177
Table 15. Regional Type I and Type II multipliers adjusted for double counting .............................. 177
Table 16. Summary of direct losses for the region. All values are discounted to 2011 (year 1 in the
table) using the social discount rate. .................................................................................................... 177
10
Table 17. GDP losses at the national level................................................................................................ 181
Table 18. Annual losses summary at national level. All values are discounted to 2011 (year 1)
using the social discount rate................................................................................................................ 183
Table 19. GDP losses in the first accident year........................................................................................ 185
Table 20. GDP losses in year 4, assuming TN = 3. .................................................................................. 185
This page left blank
11
EXECUTIVE SUMMARY
A Techno-Economic Analysis (TEA) approach assesses the value of a given technology and guides
investment and resource allocation decisions.
Sandia has a long history of using techno-economic analyses to evaluate various technologies.
Expanding our TEA capabilities will provide a rigorous basis for evaluating science, engineering and
technology-oriented projects, allowing Sandia programs to quantify the impact of targeted research
and development (R&D), and improving Sandia’s competitiveness for external funding options.
The goals of this project include:

• Documenting the use of TEA across Sandia and lessons learned about the applicability of
this analysis to furthering market development or securing additional funding
• Identifying a common set of ‘core equations’ and existing tools at Sandia, and elsewhere, that
have broad applicability for Sandia teams
• Identifying metrics for consideration of resilience goals within a TEA framework
• Providing recommendations to expand development of Sandia’s TEA capabilities to leverage
current subject matter expert (SME) capabilities to serve as laboratory-wide capability for
technology assessments; and to serve as an illustration-in-practice on how to appropriately
apply TEA in proposal development.
As part of this project, we convened a workshop on February 24, 2020 that included Sandia
technical staff who had either used TEA approaches or who wanted to learn how TEA could be
applied to their projects/programs. While the core team was aware of a variety of TEA capabilities
and efforts prior to the workshop, this well-attended workshop uncovered additional resources,
ideas, and an overwhelming desire to coordinate on future TEA. Participants were invited to provide
written input and examples for this project, many of which are included in this report.
While the majority of the TEA examples are in the energy sector, other examples included water
resource projects and carbon sequestration technologies. We provide summaries of several
representative projects in the main body of this report; additional details about each of these projects
are included in the appendix.
Our review of the projects includes several key lessons learned:
• Leverage previous analyses and tools whenever possible. This includes previous
published work and tools. Many of these TEA tools are documented in this report.
• TEA should be incorporated at an early stage to guide technology development. TEA

should not be conducted at the end of the project for the purpose of generating a cost
12
estimate for a point design. TEA can reveal obstacles and opportunities that significantly
impact the direction of research and development efforts.
• Assess the projected performance of the technology at commercial scale (vs. bench
or pilot scale). Apply the Nth plant assumption (assume an “Nth of a kind” (NOAK)
facility versus a “first-of a kind” (FOAK) facility) when proposing a production process for
analysis, and consider the impact on raw materials costs (either increasing or decreasing).
• Point estimates of the metrics of interest are not sufficient. An early-stage technology
will rarely be competitive with incumbent technologies at its initial inception. An
understanding of the uncertainties associated with the initial cost estimate, the important
drivers of performance, and potential pathways to improve performance (measured by cost
or other metrics) is necessary to more fully assess the technology and guide future research
efforts.
• TEA should assist the customer in recognizing/constructing/modifying the value

proposition for a given technology. Identifying cost drivers and trends can point to
promising applications of the technology, especially when combined with knowledge of
constraints and opportunities in potential markets.
• Consider including life cycle assessment (LCA) as part of a comprehensive TEA.

Several life-cycle assessment (LCA) tools, documented in this report, make it possible, for
example, to quantify water consumption and any associated greenhouse gas or other
emissions of interest.
• Be aware of fundamental constraints on the system of interest imposed by physics,

chemistry, etc. This is particularly important when exploring pathways for improving the
performance of the technology of interest.
From these lessons, several application areas identified include:
• Apply these lessons for internal business operations;

• Leverage current SME capabilities to serve as a laboratory-wide capability for
technology assessments.
• Serve as an illustration-in-practice on how to appropriately apply TEA in proposal
development.
13
The key recommendations and suggested future activities identified from both the lessons learned
and application areas include:
• Developing a tutorial on LCOE and life cycle cost assessment (LCA) best practices.
We consider these tools essential to help SNL researchers develop proposals and evaluate
technology commercialization potential.
• Utilizing a more targeted quantitative approach across programs to highlight higher-

impact programs and efforts. Internal investments and Business Development (BD)
efforts would benefit from this more targeted approach, particularly in energy technologies
and their future integration for a lower-carbon and resilient energy system.
• Providing resources, such as subject matter experts, and tools to help Sandia
technical staff teams conduct initial TEA for use in ARPA-E and DOE funding
proposals. Sponsors are increasingly requiring technical proposals to provide a TEA as part
of their initial proposals and subsequent funding justification. Providing support and tools
will improve the competitive position of our teams.
This report is organized as follows. The first section summarizes cost metrics commonly used for
evaluating energy systems, including documenting key equations and examples from past and
ongoing projects at Sandia National Laboratories. The second section summarizes several examples
of TEAs performed by salient Sandia SMEs. While these examples focus primarily on energy
technologies, they offer key lessons learned and illustrate the approaches taken to meet the goals of
their respective TEAs. The third section summarizes relevant TEA resources at other national
laboratories that are particularly useful for Sandia researchers. Additional details about the various
TEA projects summarized in the report are included in the Appendix A. The full list of participants
in the TEA workshop is included in Appendix B.
14
ACRONYMS AND DEFINITIONS
Abbreviation Definition
COTS commercial off-the- shelf technologies
CF capacity factor
CNG compressed natural gas
CRF capital recovery factor
DOE U.S. Department of Energy
EIA Energy Information Administration
EPC engineering, procurement, and construction
EV electric vehicle
FCR fixed charge rate
FAEE fatty acid ethyl esters
FOAK first-of-a-kind
kWe kilowatt electric
kWh kilowatt hour
GREET greenhouse gases, regulated emissions, and energy use in transportation

HTC hydrotalcite
LACE levelized avoided cost of energy
LCA Life-cycle analysis
LCOE levelized cost of energy
LCOS Levelized cost of service
MACCS MELCOR Accident Consequence Code System
MACRS modified accelerated cost recovery system

MEPV microsystems-enabled photovoltaics
MMBtu million British thermal units
MWe megawatt electric
MWh megawatt hour
NGCC natural gas combined cycle
15
Abbreviation Definition
NOAK nth-of-a-kind
NPV net present value
NRC Nuclear Regulatory Council
NREL National Renewable Energy Laboratory
O&M operating and maintenance
PHEV plug-in hybrid electric vehicle
PPA purchase power agreement
PV photovoltaic
RDEIM Regional Disruption Economic Impact Model
SAMDA severe accident mitigation design alternatives
sBET supercritical Brayton economic tool
SME subject matter expert
SST solid-state power transformer
TEA techno-economic analysis
WACC weighted average cost of capital
WTW well-to-wheel
16
1. AN INTRODUCTION TO TECHNO-ECONOMIC ANALYSIS (TEA)
The goal of this section is to summarize cost metrics commonly used for evaluating energy systems,
including documenting key equations and examples from past and ongoing projects at Sandia
National Laboratories.
Sandia has a long history of using techno-economic analyses for evaluating alternative energy
pathways. These analyses have also shown beneficial in helping researchers prioritize research
objectives. For example, the Brayton cycle technology team has effectively used TEA to focus on
component-level research that could speed the commercialization of that technology. This Brayton
example will be the key exemplar to illustrate ‘Best Practices’ and ‘Core Equations’.
1.1. Evaluating the Economics of New Technologies

The most common economic metric for comparing energy technologies is the levelized cost of
energy (LCOE). LCOE calculations estimate the per unit ($/kWh) cost of production over the
economic lifetime of the technology. Specifically, this calculation takes the capital cost, associated
financing costs, taxes, O&M, and fuel costs and calculates per unit production costs.
The Energy Information Administration (EIA) projects by 2025 the technologies with the lowest
LCOE will be solar PV (32.8 $/MWh), wind (34.1 $/MWh), and natural gas combined cycle
(NGCC) (36.6$/MWh). 1 Note that these estimates for solar PV and wind are significantly lower
than those in the 2019 report for capacity anticipated coming online in 2023: NGCC (36.2$/MWh),
solar PV (48.8 $/MWh), and wind (42.8 $/MWh). As the EIA notes, there are huge regional
differences in estimated LCOEs; for example, the LCOE for new onshore wind capacity ranges
from 28.7 $/MWh in the region with the best available wind resource to 62.7 $/MWh for the region
with either low-quality wind or higher estimated capital costs.
The EIA estimates are largely consistent with estimates from other industry sources. For example,
Lazard – a widely quoted consulting firm – estimates unsubsidized onshore wind costs ranging from
28 to 54 $/MWh, utility scale solar PV ranging from 32 to 44 $/MWh, and NGCC ranging from 44
to 68 $/MWh), Figure 1 . Lazard also documents how quickly the LCOE for solar and wind have
changed over the past decade. In 2009, the lowest cost option was NGCC ($83/MWh). Solar PV
was more than 4X higher - $359/MWh. By 2019, the estimated LCOE for utility-scale solar PV had
dropped to 41 $/MWh, below NGCC (56 $/MWh), and comparable to onshore wind ($40/MWh),
Figure 2. Based on these estimates, costs for new coal-fired plants have remained fairly constant
($109/MWh in 2019). Estimates for new nuclear plants have increased ($155/MWh in 2019), likely
due to the actual experience in constructing the Vogtle plants in Georgia.
1These estimates are capacity-weighted averages defined as the “average levelized cost per technology, weighted by the
new capacity coming online in each region (EIA, 2020d, Table 2).” They do not include tax credits, which in 2025 are
only available for solar and reduce the capacity weighted average to 30.39 $/MWh.
17
Figure 1. Lazard's estimated LCOE (Source: Lazard, 2019)
18
Figure 2. Lazard's LCOE estimates 2009 - 2019 (Lazard, 2019)
Based solely on the LCOE metric, the lowest cost options for new generating plants, on average, are
utility scale PV and onshore wind. However, as both options are non-dispatchable resources due to
the intermittent nature of the resources (without including energy storage), decisions cannot be
made based solely on this one cost metric.
The levelized avoided cost of energy (LACE) is increasingly used in conjunction with LCOE as it
better quantifies a power plant’s value to the grid. For example, power delivered to the grid during
peak demand times is more valuable than power delivered during non-peak times. As solar PV is
often available during peak demand times (sunny afternoons when air conditioning loads are high),
solar PV will often have a higher LACE than LCOE. Conversely, wind in certain regions may have a
lower LACE than LCOE if the power is delivered to grid in non-peak times. If comparing two
technologies, the project with the higher LACE-to-LCOE ratio will be more attractive.
Specifically, if the LACE-to-LCOE ratio is greater than one, then that technology is attractive to
build. Figure 3 shows the estimated LACE to LCOE ratio for NGCC, onshore wind, and solar PV
coming online in 2025 (lighter circles) and 2040 (darker circles). The dashed line separates the
economically attractive projects from the economically unattractive options. Each circle represents a
specific region. These estimates suggest that most NGCC plants planned for 2025 are economically
viable and that the estimated LACE does not change significantly by 2040 as NGCC is a mature
technology. Many of the onshore wind plants are not economically attractive in 2025 but are
expected to become more economically attractive by 2040. For solar PV, most installations are
economically attractive in 2025 and the LACE continues to fall by 2040 as the technology matures.
The main takeaway from this figure is that solar PV has a higher LACE-to-LCOE ratio than
onshore wind because of the time the power is delivered to the grid.
19
The EIA uses this LACE to LCOE ratio in their projections of capacity additions going forward,
although they note that policies and other non-economic drivers also are considered, Figure 4.
Figure 3. EIA's estimated regional LCOE to LACE values for new generating technologies entering
service in 2025 and 2040 (Source: EIA, 2020c)
Figure 4. EIA projections of capacity additions and retirements in the Annual Energy Outlook 2020
20
Figure 5 shows the regional variation in estimated LACE for new generating technologies coming
online in 2025 (EIA, 2020d). For example, estimated LACE for solar PV range from a low of 28.36
$/MWh to 34.1 $/MWh.
Figure 5. Regional variation in LACE for new generation entering service in 2025 (2019 $ per MWh)
(Source: EIA, 2020d)
The next section documents the levelized cost of energy methodology.
1.2. Levelized Cost of Energy Methodology
As discussed above, the LCOE approach is one metric commonly used to compare alternatives 2.
LCOE calculations estimate the per unit ($/kWh) cost of production over the economic lifetime of
the technology. Specifically, this calculation takes the capital costs, associated financing costs, O&M,
and fuel costs and calculates a per unit production cost. The LCOE is often used as an economic
measure of energy costs as it allows for comparison of technologies with different capital and
operating costs, construction times, and plant load factors.
The levelized COE calculation is given by:
𝐼𝐼 ∗ 𝐹𝐹𝐹𝐹𝐹𝐹 𝑂𝑂&𝑀𝑀 𝐹𝐹
𝐿𝐿𝐿𝐿𝐿𝐿𝐿𝐿 = + + (1)
𝐸𝐸 𝐸𝐸 𝐸𝐸
where: 𝐼𝐼 = total financed capital costs

𝐹𝐹𝐹𝐹𝐹𝐹 = fixed charge rate
2Rhodes (2016) provides a good overview of the LCOE methodology.

21
𝐸𝐸 = annual plant output (i.e. kWh)
𝑂𝑂&𝑀𝑀 = fixed and variable operating and maintenance costs
𝐹𝐹 = feedstock costs (i.e. natural gas, biomass)
We assume that capital expenditures are uniformly distributed over the construction period. 3 The
financed capital cost (I) is multiplied by a fixed charge rate (FCR), which includes assumptions about
state and federal taxes, the depreciation period (as defined by the Modified Accelerated Cost
Recovery System (MACRS) methodology), and other exogenous costs.
The FCR is calculated using:
𝐶𝐶𝐶𝐶𝐶𝐶[1 − 𝑏𝑏𝑏𝑏 ∑𝑀𝑀 𝑛𝑛

𝑛𝑛=1 𝑉𝑉𝑛𝑛 / (1 + 𝑟𝑟𝑤𝑤𝑤𝑤𝑤𝑤𝑤𝑤 ) − 𝑡𝑡𝑐𝑐 ] (2)
𝐹𝐹𝐹𝐹𝐹𝐹 = + 𝑝𝑝1 + 𝑝𝑝2
(1 − 𝑇𝑇)
where: 𝐶𝐶𝐶𝐶𝐶𝐶 = capital recovery factor

𝑏𝑏 = fraction of investment that can be depreciated (initially is 100%)
𝑇𝑇 = effective tax rate (default 37.6% (federal, 34%; state, 6%)) 4
M = depreciation period (3 to 20 years; default depends on technology)
𝑉𝑉𝑛𝑛 = fraction of depreciable base in year n (initially 100%)
𝑟𝑟𝑤𝑤𝑤𝑤𝑤𝑤𝑤𝑤 = real weighted average cost of capital
𝑡𝑡𝑐𝑐 = tax credit (initially zero)
𝑝𝑝1 = annual insurance cost (initially zero)
𝑝𝑝2 = other taxes (initially zero)
MACRS is an accelerated depreciation method utilized in the U.S. and allows for faster depreciation
of capital investments than allowed by straight-line methodologies. Accelerated depreciation
methods allow firms to take tax-deductible depreciation expenses earlier in the life of a capital
expenditure, giving them an upfront tax advantage for new investments. In the U.S., most utility
type investments use either a 15 or 20-year depreciation schedule. Certain investments, such as
renewables, can use a five-year depreciation schedule. Quicker depreciation schedules effectively
lower the annual capital requirements for these investments (the CRF (equation 4) is lowered as
number of years allowed for depreciation decreases).
The FCR typically ranges from 0.11 and 0.17 and represents the percentage of capital costs that
must be recovered each year to cover all investment costs, including return on debt and equity. For
3Capital expenditures are often not uniform over the construction period. For example, a more detailed LCOE
calculator developed at Sandia assumes that for a five-year construction period, the percent breakdown of financed
capital is 10%, 30%, 25%, 20% and 15% respectively over the five-year period (Drennen and Andruski, 2012). Future
versions of the Brayton Tool may want to use this modified methodology.
4 The 2017 revisions to the tax code changed the corporate tax rate to 21%.
22
example, for a $1 million capital investment and an FCR of 0.15, the annual capital requirement for
that investment is $150,000.
An important part of LCOE calculations is the percentage of the capital investment that is either
debt or equity financed. The real weighted average cost of capital (𝑟𝑟𝑤𝑤𝑤𝑤𝑤𝑤𝑤𝑤 ) takes into account the
debt-to-equity ratio and their specific financing rates. Debt financing refers to the part of the
investment that is financed through traditional financing options, such as those from banks or
bonds, and equity financing can include owner or investor financing.
The 𝑟𝑟𝑤𝑤𝑤𝑤𝑤𝑤𝑤𝑤 is calculated by:
𝐸𝐸 𝐷𝐷 (3)
𝑟𝑟𝑤𝑤𝑤𝑤𝑤𝑤𝑤𝑤 = ∗ 𝑟𝑟𝑒𝑒 + ∗ 𝑟𝑟𝑑𝑑 ∗ (1 − 𝑇𝑇)
𝑉𝑉 𝑉𝑉
where: 𝐸𝐸/𝑉𝑉 = percent of total project equity financed

𝑟𝑟𝑒𝑒 = equity financing rate
𝐷𝐷/𝑉𝑉 = percent of total project debt financed
𝑟𝑟𝑑𝑑 = debt financing rate (pre-tax)
𝑉𝑉 = capital cost
𝑇𝑇 = effective tax rate
Assumptions about the debt/equity financing split are technology specific. For example, for a plant
that assumes a 50%/50% debt/equity financing, with a debt financing rate of 4.5% and equity
financing rate of 12.0%, the 𝑟𝑟𝑤𝑤𝑤𝑤𝑤𝑤𝑤𝑤 is 7.4%.
The CRF is calculated using:
(1 + 𝑟𝑟𝑊𝑊𝑊𝑊𝑊𝑊𝑊𝑊 )𝑛𝑛 (4)

𝐶𝐶𝐶𝐶𝐶𝐶 = 𝑟𝑟𝑊𝑊𝑊𝑊𝑊𝑊𝑊𝑊 ∗
(1 + 𝑟𝑟𝑊𝑊𝑊𝑊𝑊𝑊𝑊𝑊 )𝑛𝑛 − 1
where:
𝑟𝑟𝑤𝑤𝑤𝑤𝑤𝑤𝑤𝑤 = real weighted average cost of capital
n = economic plant life (initially 20 years).
23
1.3. Experience with LCOE at Sandia
There have been several tools developed at Sandia that incorporated this LCOE methodology,
including, but not limited to:
• U.S. Energy and Greenhouse Gas Model (USEGM) 5
• Hydrogen Futures Model (H2Sim) 6
• Electricity Cost Generation Simulation Model (GenSim) 7
• Alternative Liquid Fuels Model (AltSim) 8
• Power Systems Lifecycle Analysis Tool (Power L-CAT) 9
To make this methodology more widely available to Sandia colleagues and others, Drennen (2016)
developed a simplified LCOE calculator for new energy technologies. This tool is available from the
author. Figure 6 shows the summary screen from the model. The numbers shown here have not
been updated to include most recent capital and operating costs for new plants. Note that the tool
includes the option to include up to three custom options as well as consideration of the
transmission and distribution costs, which are not normally included in the LCOE calculation.
5 Drennen et al., 2002.

24
Figure 6. Simplified Power Systems Lifecycle Analysis Tool interface (Drennen, 2016)
1.4. Estimating the LCOE for a new technology
The LCOE framework provides a powerful tool for understand the potential economic
competitiveness of new or significantly improved technologies. For example, knowing that the
LCOE for new NGCC plants ranges from 33.35 $/MWh to 45.31 $/MWh (3.34 to 4.53
cents/kWh) means that any base-load energy technology needs to be able to compete at that level in
the absence of mandates or other considerations.
Of course, the challenge with estimating the LCOE for new technologies is the uncertainty
associated with new technologies. In addition, we know that estimated costs will decline as
experience increases. There are also additional costs that are often overlooked when first estimating
LCOE, ranging from contingency to engineering and owner’s costs. This section provides advice for
25
developing initial estimates, drawing widely on work done by Sandia’s Brayton team (Drennen and
Lance, 2019; Drennen, 2020).
1.4.1. Mechanical System Costs
Typically for a new energy technology, one needs estimates at the component level (i.e., heat
exchangers, steam generator, turbomachinery), etc. The estimates for commercial off-the-shelf-
technologies (COTS) will be known with much greater certainty than for newly developed
technologies. Uncertainty ranges should be specified for each component.
Work on supercritical CO2 Brayton systems at Sandia serves as a useful example of using the LCOE
methodology to determine the market viability of a new energy technology. Specifically, Sandia
researchers developed a techno-economic modeling tool to evaluate and optimize Brayton system
configurations. The supercritical Brayton Economic Tool (sBET) calculates key system performance
and LCOE based on user-defined input on key variables such as system size, recuperator
effectiveness, and turbine inlet temperatures. The goal for this integrated tool is to allow system
designers to understand the tradeoffs associated with various key design decisions. For example,
increasing turbine inlet temperatures results in higher system efficiencies, but also requires
components made from higher-cost alloys that raise the overall system cost. sBET allows one to
analyze whether this increase in system efficiency is economically justified.
The basic structure of sBET is illustrated in
Figure 7. The major components of the mechanical system include the heat source, the various heat
exchangers, and the turbomachinery components. For each listed component, the team solicited
estimates from a wide range of vendors. For the major components, the team used these estimates
to derive cost formulas (power law form) and cost scaling factors based on a total of 129 vendor
estimates. The authors’ approach included weighting of vendor estimates to take account of such
factors as commercial availability and estimate details.
Additional details about sBET and lessons learned from using this methodology are discussed in
section 4 of this report.
26
Levelized Cost of Energy Calculator
Subsystem Cost ($1000) $/kWe Net (FOAK) % Total Cost R-Value $/kWe Net (NOAK)
Heat Source 17,382 173.82 11% 0.06 133
Heat Exchangers
High temp recuperator 13,587 135.87 9% 0.06 104
Low temp recuperator 14,539 145.39 10% 0.06 111
Primary Heat Exchanger - 0.00 0% 0.04 0.00
Heat rejection (Air Coolers/Condensers) 6,518 65.18 4% 0.04 55
Turbomachinery
Turbine 3,035 30.35 2% 0.06 23
Compressors 9,313 93.13 6% 0.06 71
Turbine stop valve 4,600 46.00 3% 0.06 35
Turbine governor valve 4,700 47 3% 0.06 36
Gear box 606 6 0% 0.06 5
Turbomachinery control 791 8 1% 0.06 6
Other instrumentation 396 4 0% 0.06 3
Inventory control 375 4 0% 0.06 3
Generator 1,723 17 1% 0.06 13
Total Mechanical 77,565 776 598
Electrical, Instrumentation, Control 4,776 47.76 3% 0.02 44
Facilities (includes major infrastructure) 10,787 108 7% 0.01 103
Project indirects 23,714 237 16% 0.01 227
Total EPS 116,842 1,168 973
Contingency 11,684 117 8% 0.01 112
Owner's Costs 23,368 233.68 15% 0.01 224
Total Project Costs 151,895 1,519 100% 1,308
Levelized Cost of Energy ($/kWhe) FOAK $ 0.055 NOAK $ 0.052

Calculated Fuel Cost ($/kWh) $ 0.026 $ 0.026
Figure 7. LCOE estimating tool within the supercritical Brayton Cycle Evaluation tool (Drennen,
2020)
1.4.2. Scaling equations
While we developed specific scaling terms for Brayton systems of various sizes, the National
Renewable Energy Laboratory (NREL) provides a recommended method for scaling various system
components (NREL, 2014). Specifically, they recommend that cost estimates for systems of
different sizes can be calculated using a simple exponential scaling relationship:
𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆 𝑛𝑛
𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶 𝑛𝑛𝑛𝑛𝑛𝑛 = 𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝑏𝑏𝑏𝑏𝑏𝑏𝑏𝑏 ∗ �𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆new � (5)
base
where n is a scaling factor. NREL notes that scaling factors for power systems typically range from
0.5 to 0.8.
27
1.4.3. Other Additional Costs
In addition to the mechanical and electrical systems, there are several other categories of costs that
must be included in the total cost estimates prior to using the LCOE methodology. The EIA (2013,
2016, 2020b) provides a methodology for estimating total cost estimates for a wide range of utility
scale electricity generating plants.
This methodology includes the following categories:

• Mechanical equipment supply (major equipment)
• Electrical and instrumentation and control (transformers, switch gear, etc.)
• Civil and structural costs (site preparation, underground utilities, structural steel supply,
and on-site building construction)
• Project indirect costs (engineering, labor, construction management)
• Fees and contingency
• Owners costs (development costs, feasibility and engineering studies, legal fees,
insurance, electrical interconnection
The sum of the first four categories is referred to as the Engineering, Procurement, and
Construction (EPC) cost. The EIA uses this methodology to estimate total project costs, excluding
financing, for a wide range of technologies. Several cost components are simply some percentage of
either the estimated mechanical and electrical costs or an assumed percentage of the EPC. For
example, the fees and contingency fees are usually approximately 10% of the EPC and owner’s costs
are 20%.
As an example, consider the following example for a new NGCC generating facility from the EIA
2016 document, Figure 8. This new NGCC has a rated capacity of 429 MWe and a heat rate of 6300
Btu/kWh. The mechanical costs (major equipment and balance of plant) and installation account for
just 45% of the total cost. The additional costs include:
• Civil and construction costs: 10.5% of mechanical + electrical

• Project indirect: 35.4% of mechanical + electrical
• Fees and contingency: 10.5% of EPC costs
• Owner’s costs: 20.0% of EPC costs
Including all costs results in an estimated total cost (excluding finance) of $473,605,000 (2106 $).
Based on a rated nominal capacity of 429,000 kWe, the total project cost is estimated at 1,104
$/kWe. Based on this estimate, the capital cost component of the LCOE (first term in equation 1) is
about 1.8 cents/kWh. 10 The O&M and fuel costs increase the total LCOE to around 4 cents/kWh.
10 Estimated using the default assumptions from the Simplified Power Systems Lifecycle Analysis Tool (Drennen, 2016).
28
Figure 8. EIA (2016) cost estimates for a new NGCC
1.5. First- and Nth-of-a-kind Methodology
Costs for technologies typically decrease as more plants are built. Figure 9 illustrates the basic
theory. Prior to actual construction of a new type of facility, costs usually increase as cost estimates
become more inclusive. Costs can increase further as the second or third plant is built as alternative
configurations and/or materials are tested. As additional plants are built, and economies of scale are
recognized, cost begins to fall. Accurately incorporating this methodology into a costing model
29
requires consideration at the subcomponents level; experimental or newer technologies will
experience higher rates of technology learning than will be the case for off the shelf technologies.
Costs are estimated for both the first-of-a-kind (FOAK) and Nth-of-a-kind (NOAK) plant using a
methodology developed by NETL (2013).
Figure 9. Theoretical Learning Curve (NETL, 2013).

NETL (2013) defines NOAK costs as:
𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝑁𝑁𝑁𝑁𝑁𝑁𝑁𝑁 = 𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐹𝐹𝐹𝐹𝐹𝐹𝐹𝐹 ∗ 𝑋𝑋 −𝑏𝑏 (6)
Where 𝑋𝑋 is the cumulative number of units and 𝑏𝑏 is the learning rate exponent, which is further
defined as:
log (1−𝑅𝑅)
𝑏𝑏 = log (2)
(7)
where 𝑅𝑅 is a technology-specific learning rate. The NETL methodology suggests 𝑅𝑅 values as high as
0.06 for experimental technologies (e.g., fuel cells) and in the range of 0.01 for mature technologies
(e.g., buildings, steam turbines, instrumentation). Figure 10 illustrates the relationship between costs
and the number of units constructed (𝑛𝑛) and the learning rate (𝑅𝑅). Components with an 𝑅𝑅 value of
0.6 show an approximate 25% decrease in costs for 𝑛𝑛 = 20; more mature components (𝑅𝑅 = 0.1),
see much smaller cost reductions (a couple percent).
30
Figure 10. NOAK costs (typically $/kWe) as a function of number of units (𝒏𝒏) and the learning rate
(𝑹𝑹)
The sBET example, Figure 7, shows the use of FOAK and NOAK in the LCOE framework. In this
example, the estimated LCOE for a first-of-a-kind facility is 0.055 $/kWh; this estimate drops to
0.052 $/kWh for the nth-of-a-kind plant. The component-level R-values shown in this figure feed
directly into equation 7.
EIA uses a modified version of this methodology. Their methodology is fully documented in EIA
(2020c). Additional details on learning curves and in particular how they relate to renewable energy
costs are discussed in Kobos et al. 2006.
1.6. Additional Details on using LCOE Analysis for a New Technology: the
SCO2 Brayton System
In this section, we describe how LCOE analysis was used to help the sCO2 Brayton team estimate
the LCOE for this emerging technology to better understand the potential market opportunities and
to help prioritize research needs to further reduce overall system cost. This example was briefly
discussed in section 1.4 of this report. Here we provide additional details, results, and sensitivity
analyses that demonstrate the value of the methodology for evaluating emerging technologies. For
31
example, if an emerging technology has estimated LCOE costs significantly greater than existing
options, it will be difficult for that new technology to compete in the market, unless the new
technology is able to offer additional benefits that could be monetarized to offset the higher plant-
level LCOE.
1.6.1. The sCO2 Brayton system
Researchers at Sandia National Laboratories and elsewhere are developing supercritical CO2 Brayton
cycle systems for use in a wide range of power systems. The physical properties of the working fluid
significantly reduce volumetric flow, which means key components, such as turbomachinery, are
significantly smaller, and hence less expensive to build.
The supercritical Brayton Economic Tool (sBET) calculates key system performance and levelized
cost of energy (LCOE) based on user-defined input on key variables such as system size, recuperator
effectiveness, and turbine inlet temperatures. This integrated tool allows system designers to
understand the tradeoffs associated with various design decisions. For example, increasing turbine
inlet temperatures results in higher system efficiencies, but also requires components made from
higher-cost alloys that raise the overall system cost. sBET allows one to analyze whether this
increase in system efficiency is economically justified.
The sBET tool integrates the basic LCOE methodology with an existing Brayton cycle evaluation
tool developed at Sandia – the RCBC Evaluation and Trade Studies Tool (RETS) (Pasch, 2016).
RETS is a sCO2 recompression closed Brayton cycle (RCBC) modeling tool that calculates key
system performance characteristics based on user-defined inputs.
To develop the model, we had to define and procure estimates for each of the main components of
the plant. Since many of the components for this plant are not commercially available, there was
considerable uncertainty associated with the initial cost estimates. Despite these uncertainties, the
initial LCOE estimates helped the team focus their research priorities. For example, we were able to
demonstrate it was unlikely that the efficiency gains associated with operating at higher turbine inlet
temperatures would offset the higher-cost alloys required for the turbomachinery and high
temperature recuperators.
A previous report (Drennen and Lance, 2019) documented sBET’s structure, assumptions, and
preliminary results. That report noted the importance of further refining the costing methodology in
order to increase the confidence in the estimates. An interlaboratory effort in 2018 and 2019 led to
updated component estimations (Weiland et al., 2019) that were used to update and refine sBET.
Whereas the initial estimates relied on simple component cost models developed by Carlson et al.
(2017) for 1 – 100 MWe Brayton systems integrated with concentrated solar power (CSP) systems,
Weiland et al. (2019) expanded on this work “by leveraging the collective resources of the U.S.
Department of Energy (DOE) national laboratories with sCO2 component vendor costs spanning
multiple applications (nuclear, fossil, solar) and size ranges (5 – 750 MWe).” The updated study
provided new cost formulas (power law form) and cost scaling factors based on a total of 129
32
vendor estimates. The authors’ approach included weighting of vendor estimates to take account of
such factors as commercial availability and estimate details.
1.6.2. Example: The Estimated LCOE for a 100 MWe sCO2 system
As an example, consider the 100 MWe sCO2 Brayton system with operating characteristics
summarized in Table 1. Using sBET, the estimated system efficiency is 40.0%, Figure 11, and the
estimated LCOE is 5.5 cents/kWh and 5.2 cents/kWh for a first-of-a-kind (FOAK) and nth-of-a-
kind (NOAK) plant, respectively, Figure 7.
Table 1. Operating assumptions for 100 MWe reference case.

Turbine inlet temperature (°C) 550
Cooling outlet temperature (°C) 33
Compressor input pressure (MPa) 8.45
Compressor discharge pressure (MPa) 25.0
Recuperator approach temperatures (°C) 10
Turbine efficiency (%) 85
Main compressor efficiency (%) 82
Secondary compressor efficiency (%) 78
Power output (MWe) 100
Figure 11. Estimated system configuration and efficiency from sBET
33
Figure 7 shows the detailed component cost breakdown for the 100 MWe Brayton system. The
estimated system cost for a nth-of-a-kind facility is 1,308 $/kWe. Of that total, the recuperators
account for 19% of the component costs. In terms of total LCOE, fuel costs account for 49% of
the total, Figure 12. The actual system costs account for just 28% of the final LCOE costs. In
addition to the fuel costs, other major cost components include the operating and maintenance
(O&M) costs (7%), contingency and owner’s fees (11%), and project indirects (8%). 11
Figure 12. LCOE cost breakdown for 100 MWe Brayton system
11
sBET utilizes the EIA methodology discussed in section 2.4.3. sBET assumes indirect costs are a
fixed 28.8% of total mechanical and electrical costs. Owner’s costs are 20% of the EPC.
34
1.6.3. Sensitivity Analysis
These initial LCOE estimates suggest that the new Brayton system can compete with the current
low cost production technologies –as discussed in section 1.1. This section shows how detailed
sensitivity analysis can help researchers prioritize research needs and to better understand the
tradeoffs associated with achieving higher system efficiencies by increasing operating temperatures
or by increasing specific component efficiencies.
For example, we used sBET to test the relationship between system size, turbine inlet temperatures,
and LCOE, Figure 13. Estimated LCOE falls significantly as system size increases, with the sharpest
declines coming as system size increases from 10 to 50 MWe. For this range of plants and these
operating assumptions, increasing system temperatures does not lead to lowered LCOE. For 10
MWe units, the higher operating temperature adds about 1.5 cents/kWh to the estimated cost. This
cost difference narrows as system size increases, but this analysis suggests the higher operating
temperatures do not result in lower LCOE and that the team should focus on commercialization of
lower-temperature units, at least in the near future.
Figure 13. LCOE as a function of system size and turbine inlet temperature
Figure 14 provides additional insight about the tradeoffs associated with higher operating
temperatures. Specifically, this figure demonstrates the relationship between turbine inlet
temperature, fuel cost, and system efficiency. Increasing the turbine inlet temperature increases
35
system efficiency; going from 550 to 700 °C improves overall system efficiency by 6.8 %. However,
for natural gas at 3.00 $/MMBtu, estimated LCOE is minimized at a turbine inlet temperature of
600 °C. For natural gas prices of 7.00 $/MMBtu, the LCOE is minimized for a turbine inlet
temperature of 650 °C. These results suggest that despite higher efficiencies achieved at higher
operating temperatures, the added system component costs do not offset the increased efficiency.
Figure 14. LCOE and system efficiency as a function of turbine inlet temperatures and fuel costs
As shown in Figure 7, the recuperators account for 19% of the total system cost. sBET allowed the
team to test the relationship between recuperator effectiveness and estimated LCOE, Figure 15. The
results show that the optimal design for the recuperators is dependent on natural gas prices. At
lower natural gas prices (3.00 – 5.00 $/MMBtu), the LCOE is minimized for recuperator
effectiveness around 91%. Beyond that the LCOE begins increasing. For more expensive natural gas
(7.00 $/MMBtu), LCOE is minimized for a recuperator effectiveness of 93%.
36
Figure 15. Sensitivity analysis on recuperator effectiveness and fuel cost (100 MWe,
700°C)
The final example shows the relationship between the turbine efficiency, system efficiency, and
LCOE for a 100 MWe Brayton system operating at 700 °C and with natural gas at $3.00/MMBtu,
Figure 16. Each one percent improvement in turbine efficiency translates into a 0.4% increase in
overall system efficiency and a 1% decrease in estimated LCOE 12, suggesting that it indeed makes
sense to focus on designing and commercializing the most efficient turbine possible.
This analysis assumes the cost estimation for turbomachinery is valid over this efficiency range;
12
whether this is a valid assumption requires additional vendor discussions and estimates.
37
Figure 16. Sensitivity analysis on turbine efficiency (100 MWe, 700 °C)
This type of sensitivity analysis was extremely useful for helping the Brayton team understand the
various tradeoffs and to focus their research efforts. Further, demonstrating this type of analysis to
our DOE sponsors showed that the Sandia team understood the importance of focusing on
designing a commercially viable system – effectively making the case for additional funding for
Brayton system research.
1.7. TEA, Resilience, and Planning
To adequately integrate resilience goals within a TEA framework, an understanding of how

technologies perform during major disruptions is necessary. Three categories of performance-based,
consequence-focused resilience metrics have been proposed that quantify impacts to national
security, the economy, and community wellbeing respectively. These metrics can be used to evaluate
performance subject to past disruptions, or in a probabilistic manner in expectation of future
disruptions. For an example of the community metrics and methodology, see Jeffers et al. (2017)
and Jeffers et al. (2018).
38
Sandia treats resilience as a multifaceted characteristic of socio-technical systems that can
characterize the economic performance and losses, incurred or avoided, at the level of individual
firms, but can also reflect the societal performance at the level of the affected communities. The
economic performance metrics include avoided business losses and the community-level metrics
include various burden metrics to reflect access and availability of food and other essential
infrastructure services.
The resilience-enhancing investments can be applied to improve specific aspects of economic (e.g.
business access to electric power and an improved ability to operate in the event of a hurricane) or
community performance (e.g. access to clean water in the event of a hurricane). This can be done as
a one-off improvement to address a specific aspect of resilience (e.g. adding microgrids after a
hurricane destroyed a part of the electric power network) or it can be done on the community
planning stage (see Wachtel, 2013; Jeffers et al., 2018; Shandiz et al., 2020; and Jeffers et al., 2020).
Resilience can also apply at the level of the entire nation and contribute to national security overall.
This type of TEA becomes necessary because there are multiple technological solutions with
different performance characteristics that can be used to improve resilience. These performance
metrics can be both technological, such as the electric power availability expectation in case of
different disruptions, as well as sociological, such as the overall burden of getting food in the event
of the disruption. The TEA allows quantifying these metrics, evaluating tradeoffs for example
between the economic and societal performance, and generates metrics that can be used in selecting
different technological options and decision-making in general.
These different aspects of resilience and their connections to the TEA are outlined in the following
sections.
1.7.1. Community Critical Service Availability and Burden Metrics

Rather than focus on whether an individual asset or building remains available during an outage
event, Sandia’s energy resilience work has shifted to focus on enabling critical lifeline services,
critical economic producers, or critical missions to stay online. In Jeffers et al., 2020, the team
outlines how to map critical infrastructure sectors to the lifeline services they provide. The goal is
then to ensure an adequate amount of each lifeline service is available during an emergency event to
serve the area of interest. Multiple types of buildings can provide the same type of service at
different levels; for example, a grocery may provide a high level of food while a gas station provides
a low level of food. Similarly, each building or asset may provide more than one service, such as a
shelter that provides food, water, shelter, and some medical supplies. Because the mapping is not
one-to-one between buildings/assets and services, creating a mapping matrix and then focusing on
enabling services gives both analysts and communities flexibility in deciding where to increase
resilience so that needs may be met, while keeping costs associated with microgrids, backup
generation, etc. as low as possible.
39
Similarly, for defense energy applications, Sandia’s resilience work focuses on enabling critical
missions rather than buildings/assets. The theory for doing so is based in the mission assurance
field, in which missions are decomposed into mission essential functions, and interdependencies
between these functions are described mathematically. These mission essential functions can also be
dependent on the threat itself – for instance, an intelligence mission changes its goals during an
intentionally-driven hazard. Next, the dependence of each mission essential function on energy loads
throughout the planning region is quantified. Finally, models are run that calculate mission
performance based on energy system performance. These can be run stochastically to incorporate a
wide distribution of system conditions as well as uncertainty.
For community resilience, based on where each individual is located within the community, there is
a cost associated with obtaining a service, both in the amount of effort required to access that
service and in their economic ability to afford the service. Sandia has developed the social burden
metric to calculate the effort and ability needed to obtain services. The social burden metric is
discussed and evaluated in Jeffers et al., 2018. The burden metric includes access to lifeline services
such as food, communications, emergency services, shelter, and others. Jeffers et al. (2018)
demonstrates that appropriately chosen microgrid locations can reduce burden in events of
hurricanes and other disruptions, and that the metric can be compared to cost for large portfolios of
multiple microgrids. Figure 17 shows that the burden can be calculated spatially based on
demographics and variable population needs.
Figure 17. The burden to achieve all lifeline services is calculated based on a portfolio of 80
microgrids sited throughout Puerto Rico. (Jeffers et al. 2018)
1.7.2. Economic resilience metrics

Jeffers et al. (2018) also consider the economic aspects of disruption and evaluate microgrid
locations based on the avoided economic losses as a result of introducing the microgrids. This study
also investigates the placement of microgrids as an option in hurricane disruption preparation and
40
takes into account that the microgrids will also provide benefit during normal operations. This
evaluation is based on enumerating the disruption scenarios and in estimating the avoided GDP
losses associated with different microgrid placements. The avoided losses are estimated using a
comparison of normal operations and disrupted operations based on a set of pre-defined disruption
scenarios using the RDEIM model, described in Bixler et al., 2020. Two example studies, discussed
in Jeffers et al. (2018) show that the returns on microgids in the form of expected avoided GDP
losses vary significantly between different possible microgrid locations, primarily because different
locations will have different sets of industry in their proximity and therefore different abilities to
support business operations in the event of disruption. The difference in returns on a single
micgrogrid for different counties in Puerto Rico is shown in Figure 18 for a notional disruption
scenario.
Figure 18. The avoided losses as a function of a microgrid location by county for a specific
disruption scenario
The analysis in Jeffers et al. (2018) shows that the decision makers have the ability to explore
possible microgrid locations and other resilience-enhancing measures, and to evaluate and take into
account the tradeoffs between the economic and societal burden measures.
41
1.7.3. Resilience in master planning frameworks
Resilience can also be an aspect of community planning and development across multiple inter-
dependent infrastructure sectors, as addressed in Wachtel (2013), Jeffers et al. (2018), Shandiz et al.
(2020), and Jeffers et al. (2020). Shandiz et al. (2020) develops a multi-layered energy resilience
framework that evaluates a set of metrics for energy master planning of communities, including
evaluating the effects of local-scale energy generation systems. These metrics are discussed in the
context of possible disruptions and disturbances. The different layers of energy resilience include
engineering-designed resilience, operational resilience, and community-societal resilience.
Meanwhile, Jeffers et al. (2020) specifies a framework for implementing resilience within an energy
master planning framework that considers both blue-sky and black-sky metrics to enable analysts
and stakeholders to evaluate options that co-optimize performance for both every day and
emergency operations. Co-optimizing for both situations allows stakeholders to maximize their
investments and potentially help offset costs typically associated with black-sky only strategies. A
sample output of solutions for the integrated framework is shown in Figure 19 where resilience
performance describes how far the design is beyond resilience requirements and blue-sky
performance consists of a life cycle cost analysis. Pareto-efficient solutions are shown in green.
Figure 19. Blue-sky Performance versus Resilience (Jeffers et al. 2020)
This planning and wide-scale aspect of resilience is also reflected in Sandia’s integrated methodology
for energy and infrastructure resilience analysis being developed for the Resilient Energy Systems
(RES) Strategic Initiative. Taking a high-level approach to integrating the many resilience
42
frameworks Sandia has developed for multiple domains such as energy, water, and cyber, the RES
framework outlines six key steps for resilience analysis, including both technical and economic
considerations. These six steps include defining the scope and goals, defining metrics for the
analysis, conducting the baseline analysis, defining and analyzing resilience mitigations, and doing an
improvement analysis to evaluate the real-world effectiveness of implemented mitigations.
Regardless of the approach to integrating resilience into planning frameworks, the metrics used to
evaluate potential solutions and mitigation approaches need to capture both system performance
and cost.
Sample performance metrics might include energy availability, PV penetration, water pressure levels,
pipeline damage, communication service availability, etc.
Sample cost metrics might include the purchase cost of equipment, cost of physical mitigation
strategies, maintenance and repair costs, economic costs associated with avoided business losses and
with other measures of socio-economic losses.
43
2. ADDDITIONAL TEA EXAMPLES: OBJECTIVES, RESULTS, AND
LESSONS LEARNED
The following examples offer an abbreviated narrative for several TEAs performed by salient Sandia
subject matter experts (SMEs). The examples include primarily examples in energy technology, but
offer key lessons learned and illustrate the approaches taken to meet the goals of their respective
TEAs.
2.1. Guiding Optimal Biofuels (TEA of Advanced Biofuels)

TEA Staff: Scott Paap (PI)
Project Dates: February 2010-February 2012
Customer/Funding Source
This work was funded under the Early Career LDRD program to support new staff members.
Goals and Objectives

The goal of this stand-alone TEA project was to understand the near-term state of advanced
biofuels technology relative to biomass-based ethanol, by pursuing the following objectives:
1. Conduct an assessment of the relative economic and environmental performance of
representative processes to produce ethanol and a water-immiscible biofuel from
lignocellulosic material under the current state of technology development.
2. Investigate the effect of uncertainty and variability in process parameters and the relative
importance of these parameters in determining the process performance.
3. Explore the long-term prospects for improving the performance of the fermentation and
product recovery sections of the water-immiscible biofuel production processes, and the
implications for their relative competitiveness versus the pathways for producing cellulosic
ethanol.
Conclusions
The results of this analysis indicated that biochemical processes to produce fatty acid ethyl esters
(FAEE) from switchgrass were uncompetitive with similar processes to produce ethanol under the
existing state of technology development, in terms of both economic and environmental metrics.
Beyond this initial assessment, the study identified the important drivers of performance for both
process types and explored options for improving the performance of the FAEE processes through
research and development. The study found that significant challenges confront potential efforts to
close the gap in performance with cellulosic ethanol processes.
44
2.2. Industrial process water treatment and reuse enabled by selective ion
exchange materials
TEA Staff: Brandon Heimer and Scott Paap

Project Dates: October 2015- September 2017
This work was a task within the LDRD project “Waste Water for Power Generation via Energy
Efficient Selective Silica Separations”, under the Energy and Climate (EC) Investment Area (IA).

The goal of this TEA was to understand and maximize the near-term potential of a novel silica
separation technology to compete with conventional technologies for water treatment, by pursuing
the following objectives:
1. Design a representative commercial-scale process utilizing the novel technology for silica
removal.
2. Determine the economic performance of this process to remove silica from a representative
water stream given the current state of technology development.
3. Quantify the impact of uncertainty and variability in critical process parameters to identify
the relative contributions of each to total performance.
4. Inform future research and development efforts to maximize relative performance and
economic competitiveness and identify the conditions that would provide the greatest value
proposition for adopting this technology.
Conclusions
This analysis yielded a feasible process design for using, recovering, and recycling the inorganic ion
exchanger hydrotalcite (HTC) for the purpose of removing silica from water produced during
enhanced oil recovery. Process design parameters such as the calcination and ion exchange residence
times and the number of HTC regeneration cycles were optimized based on experimental data to
minimize water treatment cost. Capture of CO2 from HTC calcination for use in pH adjustment of
the treated water stream was found to be uneconomical compared to use of sulfuric acid.
The results of the analysis indicated that the proposed HTC-based silica removal process was
uncompetitive with existing silica removal technologies under the current state of technology
development. The cost of silica removal via the HTC-based process is dominated by the cost of
HTC; efforts to improve the economic performance of the process should therefore focus on
tailoring HTC material properties to reduce HTC consumption. Performance of the process is
maximized where freshwater is costly and water treatment facilities are located large distances from
disposal sites. Such geographic factors will be crucial for assessing economic feasibility for
constructing treatment processes.
45
2.3. TEA of Solid-State Transformers
TEA Staff: Scott Paap and Steven Wiryadinata
Project Dates: April 2019 – September 2019
This work was funded by internal discretionary funding. The funding senior manager (Ben Wu) was
the primary customer.

The goal of this work was to understand the state of solid-state power transformer (SST) technology
relative to conventional technology, by pursuing the following objectives:
1. Understand power transformer technology and use in the modern electric grid.
2. Investigate the value proposition of solid-state transformer technology in terms of cost, grid
reliability and resilience, and additional functionality.
3. Identify potential “on-ramp” applications for solid-state power transformer technologies.
Conclusions
The initial findings of this analysis indicated that the large power transformer market presented
significant obstacles to entry for the proposed SST devices, due to (i) the risk-aversion of the utility
sector and the societal (vs. corporate) nature of the benefits derived from SSTs, and (ii) differences
in the inherent price-versus-capacity curves for conventional transformers and SST devices. These
findings led to a change in scope of the TEA away from large power transformers to application in
low-voltage distribution transformers without phase and frequency decoupling or reactive power
control.
The results of the re-scoped TEA suggested that the proposed SST devices could be competitive
with conventional distribution transformers on the basis of both price and performance in the near
term (within 5 years). The transformer efficiency standards promulgated by the U.S. DOE were
identified as an important external factor. The standards require higher efficiency as transformer
capacity (kVA) increases, defining the current upper capacity limit of SST devices. The DOE
standards also dictate the calculation of the overall device capacity rating based on the device’s
efficiency versus capacity curve. All other things being equal, an increase in SST efficiency thus leads
to higher device capacity and therefore lower SST price ($/kVA).
2.4. Microsystems-Enabled Photovoltaics (MEPV)

TEA Staff: Scott Paap and Vipin Gupta
46
This work was a task within the LDRD Grand Challenge “Microsystems-Enabled Photovoltaics”
(MEPV).

The goal of this work was to assess the potential cost of a novel photovoltaics (PV) technology and
inform research efforts to realize cost reductions, by pursuing the following objectives:
1. Construct models yielding the expected costs of producing, installing, and operating the
novel PV modules for utility-scale electricity production.
2. Identify a cost reduction pathway to achieve a total installed system cost of $1 per peak-Watt
($1/Wp) by 2020.
Conclusions
The MEPV architecture potentially enables PV systems with the high efficiency of concentrating
(CPV) and lower system-level costs of non-concentrating Si PV. Analysis of the MEPV concept
revealed a credible pathway for achieving the 2020 LCOE target of $0.06/kWh set by the US
DOE’s SunShot Initiative; however, achieving this target would require cost reductions across the
entire MEPV value chain from the production of components to the installation and operation of a
utility-scale power plant.
2.5. Potential Strategies for Integrating Solar Hydrogen Production and

Concentrating Solar Power
TEA Staff: Scott Paap

Project Dates: July 2014-November 2015
The funding for this work was provided by the Fuel Cell Technologies Office (FCTO) of the U.S.
Department of Energy (DOE) Office of Energy Efficiency and Renewable Energy (EERE).

The goal of this work was to investigate potential synergies that may be realized by integrating solar
hydrogen (H2) production and concentrating solar power (CSP) technologies, by pursuing the
following objectives:
1. Identify representative CSP-H2 production scenarios for analysis.
2. Construct simplified economic models representing the behavior of the selected scenarios.
47
3. Employ the simplified models in analyses to identify important performance drivers and
fundamental conditions that favor the integration of CSP and H2 production.
Conclusions
The collection of solar thermal energy is a significant cost for both CSP and solar H2 production.
The study identified several possible options for reducing overall system costs. One potential
strategy for improving the performance of both CSP and H2 production is heat integration. From
the perspective of H2 production, CSP-H2 integration is favored when CSP price is lower than
electricity price.
2.6. Pathways to Carbon Neutral Energy Systems at the University of

California, Davis
TEA Staff: Steven Wiryadinata, Josh Morejohn, Kurt Kornbluth

Project Dates: September 2014-August 2016
This work was funded under the NSF-PIRE (National Science Foundation – Partnerships in
International Research and Education): US-Denmark Cooperative Research and Education in
Intermittency, and by the University of California, Davis, Facilities Management - Energy and
Engineering (UCD FM-EE) office.

The goal of this work was to understand the techno-economic feasibility of transitioning the energy
system of the University of California, Davis campus to a carbon neutral system by 2025, by
pursuing the following objectives:
1. Assess the availability of renewable resources in the local context of UCD.
2. Investigate the cost-optimal phase-in times and system capacities of identified RES
technologies.
3. Compare and contrast distinct pathway scenarios for achieving carbon neutrality, along
with identification of major economic and technical drivers for implementation.
Conclusions
Transformation of UCD to a carbon neutral energy system is feasible using currently available PV,
ST and BM technologies, but alternatives are not expected to be financially attractive relative to
48
baseline unless fossil fuel rates and costs of carbon credits escalate beyond the low values evaluated
in the present work, and CAPEX of alternate systems are lower than evaluated here. High diversity
of the campus loads and supply-demand mismatch manifest as low generator CFs and highlight the
value of generators that can modulate especially under the power export restrictions. Intermittent
sources like solar require storage to fulfill this duty, which increases the cost of the system. Meeting
a diversified load solely through a large HP capacity with large storage is economically infeasible
relative to introducing backup generators. Complete electrification will also require an additional
carbon-neutral grid power purchase agreement (PPA) for supply-demand balance. Heavy reliance on
biomass or solar systems carries a larger risk in feedstock sourcing reliability and in power
imbalance, and thus, diversification of the campus energy system across a wider range of generator
types, is likely the best path forward.
2.7. Industrial process water treatment and reuse enabled by selective ion
exchange materials


The goal of this work was to understand and maximize the near-term potential of a novel silica
removal.
2. Determine the economic performance of this process to remove silica from a
representative water stream given the current state of technology development.
3. Quantify the impact of uncertainty and variability in critical process parameters to
identify the relative contributions of each to total performance.
economic competitiveness and identify the conditions that would provide the greatest
value proposition for adopting this technology.
Conclusions
49
This study identified the total levelized cost of service for a novel silica separation technology
compared to conventional technologies. In general, there is some cost advantage due largely to
maintenance cost savings and, depending on the climate and building type, lifetime capital cost
savings and energy cost savings. Smaller facilities situated in mild climates benefit from levelized
capital cost reductions, but minimal energy savings and energy cost savings. Larger facilities situated
in more extreme climates benefit from higher energy efficiency but lower levelized capital cost
savings. In all cases, the energy LCOS savings were estimated to be lower than the maintenance
LCOS savings. This new technology could also significantly benefit the utility during peak cooling
period.
2.8. Large Scale Geologic Storage of Hydrogen

Large-scale geologic storage of hydrogen may offer substantial storage cost reductions and buffer
supply disruptions to meet regional demands. A TEA study developed the scale, cost and scenario-
based market penetration needs by using underground geologic salt formations. The full suite of
capital, O&M, component, site costs and the levelized cost of Hydrogen appear in Lord, et al., 2014.
Figure 20, Figure 21 and Figure 22 represent the TEA conceptual framework, key results for the
levelized cost of Hydrogen and capital costs, and percent of the regional transportation demand met,
respectively.
50
Figure 20. The Geologic Hydrogen Storage model and assessment framework (Lord et al., 2014)
51
Figure 21. Geologic Hydrogen Storage Cost Model. (levelized cost of Hydrogen, $/kg; Lord et al.,
2014)
Figure 22. Total Capital costs, four city demand scenarios to meet 10, 25 or 100% of city Hydrogen
transportation demand (Lord et al., 2014)
52
2.9. CO2 Capture, Management and Water Treatment
Capturing and storing CO2 in geological formations is one of several proposed methods to manage
CO2 from the power sector. In the instance of thermoelectric generation, coal and natural gas-based
power plants in particular require notable amounts of water for cooling. Combining CO2 capture
and saline formation water extraction and treatment represents one proposed solution to both
manage CO2 emissions and water demands for these power systems.
A program for energy-water-power studies developed numerous TEA-like studies at Sandia and
elsewhere (Kobos et al., 2011, 2016, 2018; Tidwell et al., 2011, 2013, 2018; Heath et al., 2012, 2014;
Klise et al., 2013; Roach et al., 2014). Figure 23, Figure 24, Figure 25 and Figure 26 represent the
assessment conceptual framework, key single power plant results, national water demand profile for
all coal and natural gas power plants, and national CO2 stored for the coal and natural gas power
plant fleet, respectively.
Figure 23. Water, Energy and Carbon Sequestration Simulation Model (WECSsim©) Assessment
Framework (Klise et al., 2013; Kobos et al., 2016)
53
Figure 24. Base Case CO2 capture, storage and water treatment costs for a large, coal-fired power
station (Kobos et al., 2011)
Figure 25. United States National Power Plant water demand and extracted saline formation water
for all coal and natural gas fired power stations (Kobos et al., 2016)
54
Figure 26. United States national cost curve for CO2 capture and storage for all coal and natural
gas fired power station system (Kobos et al., 2016)
2.10. Economic Impacts of Nuclear Power Plants Accident Analysis
This section provides a brief summary of Sandia’s economic impacts estimation work and the
Regional Disruption Economic Impact Model (RDEIM), created for the Nuclear Regulatory
Commission (NRC), to assess the likely economic impacts for a wide range of nuclear incidents.
The MELCOR Accident Consequence Code System (MACCS) code is the NRC code used to
perform probabilistic health and economic consequence assessments for atmospheric releases of
radionuclides. MACCS is used by U.S. nuclear power plant license renewal applicants to support the
plant specific evaluation of severe accident mitigation alternatives (SAMA) analyses as part of an
applicant’s environmental report for license renewal. MACCS is also used in severe accident
mitigation design alternatives (SAMDA) and severe accident consequence analyses for
environmental impact statements (EISs) for both existing and new reactor license applications. The
NRC uses MACCS in its cost-benefit assessments supporting regulatory analyses that evaluate
potential new regulatory requirements for nuclear power plants. NRC regulatory analysis guidelines
recommend the use of MACCS to estimate the averted “offsite property damage” cost and the
averted offsite dose cost elements, which are both benefits in the cost/benefit analysis (NRC, 1997;
NRC, 2004).
55
The original cost-based MACCS economic model was published by Jow, et al. (1990) and is referred
to in this document as the cost-based model. This cost-based model is a generalization of the one in
CRAC2 (Ritchie, et al., 1983). Since the implementation of the cost-based economic model in
MACCS, government-sponsored economic data related to GDP have become readily available,
along with tools to gather and process these data. With the availability of government-produced,
standardized data, an alternative MACCS economic model can be employed to implement a gross-
domestic-product-based (GDP-based) estimation of offsite economic costs of a nuclear power plant
incident. To implement the GDP-based economic model, a variant of the Regional Economic
Accounting analysis tool (REAcct) created at Sandia National Laboratories, has been integrated into
MACCS. To signify that this model is significantly different than REAcct, it has been named
RDEIM, which stands for the Regional Disruption Economic Impact Model.
RDEIM achieves the following objectives:

• Consistency of identifying off-site costs for nuclear reactor accidents with state-of-practice
methods used to estimate impacts from other disruptions that have the potential for large-
scale economic impacts
• Developing estimates of the offsite cost impacts from business disruption using current
state-of-practice input-output (I-O) economics
• Estimating the impact on the regional communities, industries, and infrastructure
• Estimating the impacts of multi-year disruptions when the region cannot be remediated
quickly
• Estimating indirect effects on the national economy outside the directly affected region
• Estimating induced effects to the regional and national economies resulting from lost
income to workers
RDEIM calculates the indirect losses using “net total requirements” (NTR) multipliers based on the
Regional I-O Modeling System (RIMS II) data. It uses employment by county, value added 13 and
gross output by industry, total requirements tables, final demand value-added multipliers (RIMS II
model) provided by the U.S. Bureau of Economic Analysis (BEA), as well as other data provided by
the Bureau of Labor Statistics and other sources.
The total economic impact (loss) caused by a disruption is typically grouped into three categories
(BEA, 2012):
13Value added is defined as the sum of labor compensation, capital income, and net indirect taxes (producer taxes,
import tariffs minus subsidies).
56
• Direct 14 GDP impacts occur due to a loss of final demand, which occurs in the context of an
accident because production is stopped for a period in the affected area, which represents a
loss of the value added by the affected firms.
• Indirect GDP impacts occur because the loss of final demand also affects the supplier firms
as their input to the curtailed production is no longer required. In the context of an accident,
supplier firms are outside the affected area. GDP impacts represent value-added losses to
indirectly affected firms.
• Induced GDP impact relates to the spending of workers whose earnings are affected by the
disruption. 15 Induced GDP losses correspond to both workers inside and outside the directly
affected area.
The GDP loss calculated by RDEIM estimates the losses accrued over time at the regional scale of
the impacted area and at the national scale. It also allows the recovery schedules for regional and
national scales to be varied independently of each other with the proviso that regional recovery is
never faster than national recovery.
14 The notion of direct (and by extension indirect and induced) impacts in this application does not map directly to the
existing literature, due to the nature of disruption, where all industries are shut down in a region. Therefore, the impacts
in the directly affected area that would have been indirect if only one industry were shut down, are treated as direct given
that all industries are shut down. This is the reason for using the net value-added multipliers.
15 For example, employers may lay off workers to reduce their realized losses and that in turn creates an induced loss
from the reduced spending of their employees.

57
3. ILLUSTRATIVE TEA STRENGTHS AND COMPETENCIES AT OTHER
NATIONAL LABS
The goal of this section is to document resources developed by other national laboratories that
could be useful tools for TEA projects at Sandia.
3.1. Argonne National Laboratory (ANL)
Argonne’s website notes that “Argonne maintains a wide-ranging science and technology portfolio
that seeks to address complex challenges in interdisciplinary and innovative ways.” Economic
analysis is a key component of this portfolio. 16 Argonne’s Energy Systems Division has a large focus
on system assessment and analysis support in four main areas 17:
• Economic and Infrastructure Analysis

• Life cycle Analysis
• Prospective Benefits of Vehicle Technologies
• Water Technologies
A main product of their Life Cycle Analysis group is the Greenhouse Gases, Regulated Emissions,
and Energy Use in Transportation Model (GREET). GREET is a widely cited tool available in a
range of platforms that estimates well-to-wheel (WTW) emissions for a wide-variety of vehicle
technologies and pollutants. The online version provides good summaries of greenhouse gases,
energy usage, other pollutants, and water usage. Figure 27 (greenhouse gases) and Figure 28 (energy
usage) show the range of vehicle and fuel types considered. The Excel version of the model is
incredibly detailed and allows for comprehensive consideration of the tradeoffs associated with
various fuels and propulsion systems.
17 https://www.anl.gov/es/system-assessment-and-analysis
58
Figure 27. Sample output from the web-based version of GREET (Greenhouse Gases)
59
Figure 28. Sample output from the web-based version of GREET (Energy Usage)
Argonne’s Alternative Fuel Life-Cycle Environmental and Economic Transportation (AFLEET)

Tool builds on the GREET models and allows users to explore the economic and environmental
tradeoffs associated with a variety of vehicle options (from passenger vehicles to long-haul trucks)
and fueling options (gasoline, diesel, CNG, EV, etc.) Figure 29 shows an example of the type of
output available. This option compares costs and emissions for a typical gasoline, plug-in hybrid
electric vehicle (PHEV), compressed natural gas (CNG), and fully electric vehicle (EV).
60
Figure 29. Representative output from Argonne's AFLEET model
61
In addition to these various tools, Argonne is the go-to source for all transportation energy.
Argonne recently released the 38th annual version of the Transportation Energy Databook, a
comprehensive collection of transportation data for the U.S. Department of Energy’s Office of
Vehicle Technologies Office. 18
3.2. National Renewable Energy Laboratory (NREL)
NREL has a wide-range of techno-economic tools and publications available, mostly focused on
renewable energy technologies. 19 Some of the key tools with possible applicability to on-going TEA
efforts at Sandia include the following.
NREL’s Levelized Cost of Energy tool is an easy-to-use, web-based, levelized cost of energy tool,
Figure 30, requiring just a few inputs on capital costs, capacity factors, O&M costs, and fuel costs.
The example shown is for a new NGCC plant using assumptions from Assumptions to the Annual
Energy Outlook 2020 (EIA, 2020). The estimated LCOE for this plant using this calculator is just 2.9
cents/kWh.
18 https://tedb.ornl.gov/
19 The main tools are listed here: https://www.nrel.gov/analysis/data-tools.html. A comprehensive list of all their tools
is available here: https://www.nrel.gov/research/data-tools.html.
62
Figure 30. NREL's simplified levelized cost of energy calculator
NREL’s Cost of Renewable Energy Spreadsheet Tool (CREST) is a series of downloadable

Excel models for solar, wind, geothermal, anaerobic digestion, and fuel cell technologies. 20 The
models allow the user to choose whether to provide analysis on either a simplified, intermediate, or
complex level. For example, for the solar PV version, for the simplified version, the user provides
20
https://www.nrel.gov/analysis/crest.html
63
aggregate estimates for capital, O&M, etc. whereas for the complex method, the user provides as
much specificity on individual components as possible, such as “mounting hardware.” The tools
include consideration of various state-level and federal incentives, as well as different resource-
related considerations (e.g., insolation at state level). The output includes detailed cash-flow analysis
as well as more generic LCOE estimation, although the latter seems to change based on the level of
detail selected.
NREL’s System Advisor Model (SAM) is listed as a “free techno-economic software model that
facilitates decision-making for people in the renewable energy industry” 21 and includes a wide range
of renewable energy systems, including: 22
• Photovoltaic systems, from small residential rooftop to large utility-scale systems

• Battery storage with Lithium ion, lead acid, or flow batteries
• Concentrating Solar Power systems for electric power generation, including parabolic
trough, power tower, and linear Fresnel
• Industrial process heat from parabolic trough and linear Fresnel systems
• Wind power, from individual turbines to large wind farms
• Solar water heating
• Geothermal power generation
• Biomass combustion for power generation
• High concentration photovoltaic systems 23
SAM’s financial models are very detailed and provide analysis for a wide range of projects, ranging
from residential and commercial projects and include consideration of various types of ownership
and financing options from power purchase agreement projects (PPAs) to lease agreements.
NREL has several non-publicly available TEA tools as well. One of these tools is the Distributed
Generation Market Demand (dGen) model which “simulates customer adoption of distributed
energy resources (DERs) for residential, commercial, and industrial entities in the United States or
other countries through 2050.” 24 NREL notes they are currently developing an open-source version
of this model as part of the “Resilient Planning for Distributed Energy Resources project.” 25
NREL’s Jobs and Economic Development Impact (JEDI) models “estimate the economic
impacts of constructing and operating power generation and biofuel plants at the local and state
levels.” 26 This suite of tools is particularly useful to project teams seeking to quantify the economic
impact of a possible or proposed project.
21 https://sam.nrel.gov/
22 This bulleted list is taken from https://sam.nrel.gov/.
23 https://sam.nrel.gov/
24 https://www.nrel.gov/analysis/dgen/
25
https://www.nrel.gov/analysis/dgen/
26 https://www.nrel.gov/analysis/jedi/
64
There are several available models, covering a wide range of both renewable and non-renewable
resources. Each of these models relies on economic impacts data from a widely used input-output
economic model called the Economic Impact Analysis for Planning tool (IMPLAN) which provides
detailed economic impacts of any economic activity at the regional or state level. Typically, input-
output models capture the entire economic impact of a project, including direct and indirect jobs
(restaurants, hotels, travel, etc.), and the total GDP impact regionally and nationally.
Figure 31 shows an example of the type of output the JEDI models provide. This example considers
a 100 MWe concentrating solar power (CSP) trough plant constructed in NM in 2020. The model
estimates the plant would have a capacity factor of 31.4% based on user-defined input on solar
radiation 27. The project would create 1,958 jobs during the construction period; 882 of their jobs
would be onsite, with an additional 696 created elsewhere in the supply chain and 380 jobs created
through induced impacts, meaning jobs created due to increased household income related to the
direct impacts. In terms of economic impact, the project has a projected total value added
(compensation, proprietary income, rent and other income, and taxes) of $286 million during
construction and $7.7 million annually post-construction. For comparison, a similar project
constructed in NY would have a vastly lower capacity factor (19.3%) due to the much lower solar
resource 28, but similar economic impact (1,781 jobs and $279 million value added).
27 For this example, assumed direct solar normal radiation of 6.0 kwh/m2/day in NM at the project location.
28 Assumed 3.0 kwh/m2/day.
65
Figure 31. Sample output from JEDI model for a 100 MWe CSP project in New Mexico
3.3. National Energy Technology Laboratory (NETL)
NETL is the only national laboratory focused solely on fossil fuels. The Systems Engineering and
Analysis group has a dedicated group focused on techno-economic analysis and life cycle
66
assessment. Over the years, there has been significant collaboration and joint publications with
teams or individuals at Sandia, including with P. Kobos and T. Drennen. 29
NETL has several publicly available tools devoted to life cycle assessment, including economic
considerations, for a wide variety of energy technologies, including some of the key renewable
options. As NETL notes: “Life Cycle Analysis (LCA) is a comprehensive form of analysis that
utilizes the principles of Life Cycle Assessment, Life Cycle Cost Analysis, and various other
methods to evaluate the environmental, economic, and social attributes of energy systems
ranging from the extraction of raw materials from the ground to the use of the energy carrier to
perform work.” 30
NETL has released several Excel-based LCA models, such as the Lifecycle Analysis of Natural
Gas Extraction and Power Generation. 31 In addition, NETL contributes modules to openLCA,
an open source, free software for Life Cycle Assessment. 32 As part of this effort, NETL has
created an openLCA database to help researchers conduct LCA using the openLCA software.
NETL has created an Excel tool to take the openLCA results and to translate them into stacked
bar charts for presentation in reports and presentations. 33 This appears to be a very powerful and
useful tool for Sandia researchers interested in quantifying the energy use and associated
emissions with a range of energy generating technologies. For example, NETL’s Electricity LCI
Explorer allows one to quantify a wide range of emissions or other environmental considerations
associated with a pre-set or custom mix of generating facilities.
The following categories are used to group the emissions and other environmental
considerations:
• Eutrophication potential
• Acidification potential
• Particulate matter formation potential
• Photochemical smog formation potential
• Global warming potential
• Ozone depletion potential
• Water consumption
29 See for example: T. Drennen, J. Andruski, T. Skone, and J. Adder, “Power Systems Life Cycle Analysis Tool”,
Version 2.0, National Energy Technology Laboratory, June 2012, available at:
http://www.netl.doe.gov/research/energy-analysis/power-systems-lcat2-1.
30 For a more complete description, see netl.doe.gov/LCA.
31 The report and model (Littlefield et al., 2019) are available at: https://netl.doe.gov/energy-analysis/details?id=3198.
32 The openLCA software is available from: http://www.openlca.org/.
33
The various tools and training resources are available from: netl.doe.gov/LCA/CO2U
67
As an example of the output, for the default mix currently used in the southwest (Federal Energy
Regulatory Commission region), Figure 32 summarizes the associated greenhouse gas emissions
(kgCO2e/MWh) and Figure 33 shows the associated eutrophication potential (kgN-e/MWh)). One
can use this tool to then compare alternative mixes.
This NETL tool allows one to drill down to specific pollutants of interest and includes several
hundred known air and water pollutants. As an example, Figure 34 quantifies the airborne
mercury emissions associated with the default grid mix in the southwest.
Figure 32. Greenhouse gas emissions associated with the current grid mix in the southwest US
68
Figure 33. Eutrophication potential associated with the current grid mix in the southwest US
Figure 34. Mercury emissions associated with the current grid mix in the southwest US
69
Overall, the NETL tools are powerful tools for Sandia researchers interested in quantifying the
environmental aspects of a wide variety of technologies. To develop this comprehensive suite of
economic and environmental tools, NETL partners with researchers at both Carnegie Mellon
University (and in particular, the Engineering and Public Policy Program), Penn State, and the
University of Pittsburgh. 34
34A comprehensive listing of the various researchers is included in this report:

https://www.utoledo.edu/research/pdfs/Skone-netl.pdf.
70
4. REFERENCES
Beardsley, S. S.; Coker, S. D.; Whipple, S. S., 1994. The Economics of Reverse Osmosis and Ion
Exchange. In WATERTECH Expo 1994, Houston, TX.
Bernard, Michael, 2016. DYMATICA, Dynamic Multi Scale Assessment Tool for Integrated
Cognitive-behavioral Actions, Modeling, Assessment and Training. SAND2016-4349R, March.
Bernard, Michael, 2020. DYMATICA, Dynamic Multi-scale Assessment Tool for Integrated
Cognitive-behavioral Actions, SAND2020-3307M.
Bixler, N. E., A. V. Outkin, D. Osborn, N. C. Andrews and F. Walton, 2020. Economic Model for
Estimation of GDP Losses in the MACCS Offsite Consequence Analysis Code, Sandia National
Lab.(SNL-NM), Albuquerque, NM.
Defense Advanced Research Projects Agency (DARPA), 2018. Making Gray-Zone Activity more
Black and White, https://www.darpa.mil/news-events/2018-03-14.
Dewers, T., Eichhubl, P., Ganis, B., Gomez, S., Heath, J., Jammoul, M., Kobos, P., Liu, R., Major, J.,
Matteo, E., Newell, P., Rinehart, A., Sobolik, S., Stormont, J., Taha, M.R., Wheeler, M. and D.
White, 2018. Heterogeneity, pore pressure and injectate chemistry: Control measures for geologic
carbon storage, International Journal of Greenhouse Gas Control, 68, pp. 203 – 215.
Drennen, T., O. Paananen, A. Baker, and D. Harris, 2002. U.S. Energy and Greenhouse Gas Model:
Technical Description, SAND2000-2638, Sandia National Laboratories, January.
Drennen, T., A. Baker, S. Jones, W. Kamery, A. Lutz and J. Rosthal, 2004. The Hydrogen Futures
Simulation Model (H2Sim) Technical Description, SAND2004-4397, Sandia National Laboratories,
October.
Drennen, T., A. Baker, and W. Kamery, 2005. Electricity Generation Cost Simulation Model (GenSim),
SAND2002-2276, Sandia National Laboratories, March.
Drennen, T., R. Williams, and A. Baker, 2010. Alternative Liquid Fuels Simulation Model (AltSim),
Version 2.0, National Energy Technology Laboratories, March.
Drennen, T, J. Andruski, T. Skone, and J. Adder, 2012. Power Systems Life Cycle Analysis Tool, Version
2.0, National Energy Technology Laboratory, June.
Drennen, T., Andruski, J., Skone, T. and J. Adder, 2012. “Power Systems Life Cycle Analysis
Tool”, Version 2.0, National Energy Technology Laboratory, June, available at:
http://www.netl.doe.gov/research/energy-analysis/power-systems-lcat2-1.
71
Drennen, T., 2016. Simplified Power Systems Lifecycle Analysis Tool, SAND2016-56130.
Drennen, T., and B. Lance, 2019. sCO2 Brayton Cycle Economics Modeling, Level 3 Milestone Report,
M3ST-19SN0207011, March.
Drennen, T., 2020. sCO2 Brayton System Market Analysis, SAND2020-3248, March.
Electric Power Research Institute, 2003.Use of Degraded Water Sources as Cooling Water in Power
Plants, 1005359; Palo Alto, CA.
Heath, J., McKenna, S., Dewers, T., Roach, J. and P. Kobos, 2014. “Multi-well CO2 injectivity:
Impact of Boundary Conditions and Brine Extraction on Geologic CO2 Storage Efficiency and
Pressure Buildup,” Environmental Science & Technology, 48(2), pp. 1067 – 1074.
Heath, J.E., Kobos, P.H., Roach, J.D., Dewers, T.A. and S.A. McKenna, 2012. “Geologic
Heterogeneity and Economic Uncertainty of Subsurface Carbon Dioxide Storage,” SPE Economics
& Management, January.
Heimer, B., Paap, S., Sasan, K., Brady, P., Nenoff, T., 2019. “Industrial process water treatment and
reuse enabled by selective ion exchange materials,” Industrial & Engineering Chemistry Research,
58, 14873-14879.
Jeffers, R. F., M. M. Hightower, N. S. Brodsky, M. J. Baca, A. Wachtel, S. Walsh, M. S. Aamir, J.

Gibson, W. E. Fogleman and W. J. Peplinski, 2017. A Grid Modernization Approach for
Community Resilience: Application to New Orleans, LA, Sandia National Lab, Albuquerque, NM.
Jeffers, R. F., A. Staid, M. J. Baca, F. M. Currie, W. E. Fogleman, S. DeRosa, A. Wachtel and A. V.

Outkin, 2018. Analysis of Microgrid Locations Benefitting Community Resilience for Puerto Rico,
Sandia National Lab, Albuquerque, NM.
Jeffers, R., A.M. Wachtel, A. Zhivov, A. Srivastava and P. W. Daniels, 2020. "Integration of
Resilience Goals into Energy Master Planning Framework for Communities." ASHRAE
Transactions 126: 803-823.
Klise, G.T., Roach, J.D., Kobos, P.H., Heath, J., Gutierrez, K., 2013. “A framework for analyzing
the cost to utilize non-traditional waters from geologic saline formations to meet energy demands in
a CO2 capture and storage regime,” Hydrogeology Journal, Issue 21:3.
Kobos, P., J. Erickson, and Drennen, T., 2006. Technological learning and renewable energy costs:
implications for US renewable energy policy, Energy Policy 34.
72
Kobos, P.H., Malczynski, L.A., Walker, L.N., Borns, D.J. and G.T. Klise, 2018. Timing is
Everything: A Technology Transition Framework for Regulatory and Market Readiness Levels,
Technological Forecasting and Social Change, 137, pp. 211 – 225.
Kobos, P.H., Klise, G.T., Malczynski, L.A. and L.N. Walker, 2016. Parametric Analysis of
Technology Costs for CO2 Storage in Saline Formations, Special Issue, Nexus of Water and Carbon
Capture and Storage, International Journal of Greenhouse Gas Control, 54, pp. 574 – 587.
Kobos, P.H., Cappelle, M.A., Krumhansl, J.L, Dewers, T.A., McNeamar, A. and D.J. Borns, 2011.
“Combining power plant water needs and carbon dioxide storage using saline formations:
Implications for carbon dioxide and water management policies,” International Journal of
Greenhouse Gas Control, Volume 5, Issue 4, July, pp. 899 − 910.
Lord, Anna S., Kobos, Peter H. and David J. Borns, 2014. Geologic storage of hydrogen: Scaling up
to meet city transportation demands, International Journal of Hydrogen Energy, 39, pp. 15570-
15582.
National Energy Technology Lab (NETL), 2013. Technology Learning Curve (FOAK to NOAK),
DOE/NETL-341/081213, August.
National Energy Technology Laboratory (NETL), 2020, netl.doe.gov/LCA/CO2U.
Littlefield, J., Roman-White, S., Augustine, D. Pegallapati, A., Zaimes, G.G., Rai, S., Cooney, G and
T.J. Skone, 2019. Life Cycle Analysis of Natural Gas Extraction and Power Generation,
DOE/NETL-2019/2039.
National Renewable Energy Laboratory (NREL), 2020. Available at:

https://www.nrel.gov/analysis/jedi/.
OpenLCA, 2020. Available at: http://www.openlca.org/.
Paap, S., West, T., Manley, D., Steen, E., Beller, H., Keasling, J., Dibble, D., Chang, S., Simmons, B.,
2013. “Guiding Optimal Biofuels: A Comparative Analysis of the Biochemical Production of
Ethanol and Fatty Acid Ethyl Esters from Switchgrass,” SAND2013-0208.
Paap, S., West, T., Manley, D., Steen, E., Beller, H., Keasling, J., Dibble, D., Chang, S., Simmons, B.,
2013. “Biochemical production of ethanol and fatty acid ethyl esters from switchgrass: A
comparative analysis of environmental and economic performance,” Biomass and Bioenergy, 49, 49-
62.
Paap, S., Wiryadinata, S., “TEA for Solid-State Transformers,” 2019. Presented on September 23.
73
Paap, S., 2012. “Biochemical production of ethanol and water-immiscible biofuel from
lignocellulosic biomass: A comparative analysis of environmental and economic performance,”
American Institute of Chemical Engineers 2012 Annual Spring Meeting.
Paap, S., Gupta, V., Tauke-Pedretti, A., Resnick, P., Sanchez, C., Nielson, G., Cruz-Campa J. L.,
Jared, B., Nelson, J., Okandan, M., Sweatt, W., 2014. “Cost analysis of flat-plate concentrators
employing microscale photovoltaic cells for high energy per unit area applications,” 2014 IEEE
Photovoltaic Specialist Conference (conference paper and poster).
Paap, S., Gupta, V., Cruz-Campa J. L., Okandan, M., Sweatt, W., Jared, B., Anderson, B., Nielson,
G., Tauke-Pedretti, A., Nelson, J., 2013. “Cost analysis for flat-plate concentrators employing
microscale photovoltaic cells,” 2013 IEEE Photovoltaic Specialist Conference (conference paper
and presentation).
Paap, S., 2016.“Potential strategies for integrating solar hydrogen production and concentrating solar
power: A systems analysis,” US DOE Fuel Cell Technologies Office Webinar.
Pasch, J., 2016.“User Manual: RCBC Evaluation, Trades, and Performance Tool”, version 160827,
SCR #1573.0, Sandia National Laboratories
Rhodes, J., King, C., Gulen, G., Olmstead, S, Dyer, J., Hebner, R., Beach, F, Edgar, T., and Webber,
M., 2016. “A Geographically Resolved Method to Estimate Levelized Power Plant Costs with
Environmental; Externalities”, University of Texas Energy Institute, UTEI/2016-04-1, available at:
https://live-energy-institute.pantheonsite.io/sites/default/files/UTAustin_FCe_LCOE_2016.pdf
Roach, J.D., Heath, J.E., Kobos, P.H. and G.T. Klise, 2014. System-Level Benefits of Extracting and
Treating Saline Water from Geologic Formations during National-Scale Carbon Capture and
Storage, International Journal of Greenhouse Gas Control, 25, pp. 186 – 197.
Shandiz, S. C., G. Foliente, B. Rismanchi, A. Wachtel and R. F. Jeffers, 2020. "Resilience framework
and metrics for energy master planning of communities." Energy: 117856.
Skone, T.J., 2019. Life Cycle Analysis @ NETL, Presentation, University of Toledo, October 10.
Tidwell, V.C., Malczynski, L.A., Kobos, P.H., Klise, G.T., Shuster, E.P., 2013. “Carbon Capture and
Sequestration: Potential Impacts on U.S. Water Resources,” Environmental Science & Technology,
April.
Tidwell, V.C., Moreland, B.D., Shaneyfelt, C.R. and P. Kobos, 2018. Water Availability, Cost and
Projected Consumptive Use in the Eastern United States with Comparisons to the West,
Environmental Research Letters, 13, January.
74
Tidwell, V.C., Kobos, P.H., Malczynski, L.A., Klise, G. and C.R. Castillo, 2011. “Exploring the
Water-Thermoelectric Power Nexus,” Journal of Water Resources Planning and Management,
December, 138 (5), pp. 491–501.
U.S. Department of Energy, Energy Information Administration (EIA), 2013. Updated Capital Cost
Estimates for Utility Scale Electricity Generating Plants, U.S. Department of Energy, April.
U.S. Department of Energy, Energy Information Administration (EIA), 2016. Capital Cost Estimates
for Utility Scale Electricity Generating Plants, U.S. Department of Energy, November.
U.S. Department of Energy, Energy Information Administration, 2020a. Assumptions to the Annual
Energy Outlook, January 2020.
U.S. Department of Energy, Energy Information Administration (EIA), 2020b. Capital Cost and
Performance Characteristic Estimates for Utility Scale Electric Power Generating Technologies, February.
U.S. Department of Energy, Energy Information Administration (EIA), 2020c. Assumptions to the
Annual Energy Outlook 2020: Electricity Market Module, January.
U.S. Department of Energy, Energy Information Administration (EIA), 2020d. Levelized cost and
levelized avoided cost of new generation resources in the Annual Energy Outlook 2020, February.
Wachtel, A., R. F. Jeffers and C. R. Lawton, 2018. Overview of Resilience Planning Framework,
Sandia National Lab, Albuquerque, NM.
Wachtel, A. M., 2013. A scalable methodology for modeling cities as systems of systems,
Massachusetts Institute of Technology.
Wiryadinata, S, Morejohn, J & Kornbluth, K., 2018. Pathways to carbon neutral energy systems at
the University of California, Davis. Renewable Energy. 130. 10.1016/j.renene.2018.06.100.
Wiryadinata, S, Modera, M, Jenkins, B & Kornbluth, K., 2016. Technical and Economic Feasibility
of Unitary, Horizontal Ground-Loop Geothermal Heat Pumps for Space Conditioning in Selected
California Climate Zones. Energy and Buildings. 119. 10.1016/j.enbuild.2016.03.032.
75
APPENDIX A. ADDITIONAL DETAILS ON TEA STUDIES SUMMARIZED
IN MAIN REPORT
This section provides additional details for several of the TEAs performed by salient Sandia subject
matter experts (SMEs) summarized in section 3 of this report. These examples are primarily in
energy technology, but offer key lessons learned and illustrate the approaches taken to meet the
goals of their respective TEAs.
A.1. Guiding Optimal Biofuels (TEA of Advanced Biofuels)

TEA Staff: Scott Paap (PI)
Project Dates: February 2010-February 2012
This work was funded under the (subsequently discontinued) Early Career LDRD program to
support new staff members.
Project Structure and Interaction with Customer/Technical Team

This was a stand-alone TEA project based on on-going but separately funded work at Sandia
National Laboratories (SNL) and the Joint BioEnergy Institute (JBEI). Funding supported the PI,
mentorship from a senior staff member (Todd West), and interactions with subject matter experts
(SMEs) as needed. The PI reported to the LDRD office; reporting requirements were similar to
conventional LDRD projects.

The goal of this work was to understand the near-term state of advanced biofuels technology
relative to biomass-based ethanol, by pursuing the following objectives:
1. Conduct an assessment of the relative economic and environmental performance of
representative processes to produce ethanol and a water-immiscible biofuel from
lignocellulosic material under the current state of technology development.
2. Investigate the effect of uncertainty and variability in process parameters and the relative
importance of these parameters in determining the process performance.
3. Explore the long-term prospects for improving the performance of the fermentation and
product recovery sections of the water-immiscible biofuel production processes, and the
implications for their relative competitiveness versus the pathways for producing cellulosic
ethanol.
Deliverables
76
- SAND report: Paap, S., West, T., Manley, D., Steen, E., Beller, H., Keasling, J., Dibble, D.,
Chang, S., Simmons, B. (2013). “Guiding Optimal Biofuels: A Comparative Analysis of the
Biochemical Production of Ethanol and Fatty Acid Ethyl Esters from Switchgrass,”
SAND2013-0208.
- Journal article: Paap, S., West, T., Manley, D., Steen, E., Beller, H., Keasling, J., Dibble, D.,
Chang, S., Simmons, B. (2013). “Biochemical production of ethanol and fatty acid ethyl
esters from switchgrass: A comparative analysis of environmental and economic
performance,” Biomass and Bioenergy, 49, 49-62.
- Conference presentation: Paap, S. (2012). “Biochemical production of ethanol and water-
immiscible biofuel from lignocellulosic biomass: A comparative analysis of environmental
and economic performance,” American Institute of Chemical Engineers 2012 Annual Spring
Meeting.
Background
The evolution of this project began with discussions between the PI, management and other staff in
the SNL/CA Systems Analysis group, and subject matter experts pursuing biofuels research at
SNL/CA and JBEI. Based on these discussions and an extensive literature search by the PI, TEA of
novel biofuels production pathways was identified as an important gap and a promising area for
collaboration between the Systems Analysis and biofuels research communities. Although the
SNL/CA Systems Analysis group had previously conducted studies of the biofuels industry, this
project represented the first TEA effort undertaken by its staff. The TEA methodology leveraged by
the SNL/CA Systems Analysis group in subsequent analyses was developed during the course of
this work.
The biofuels research at SNL/CA and JBEI focused on developing economically-viable, carbon-
neutral fuels derived from bioenergy crops. However, JBEI did not conduct research on biomass-
derived ethanol, the only commercially relevant fuel derived from plant sugars. Despite ethanol’s
process technology maturity and efficient production in microbes, it suffers from a relatively low
energy density (23.5 MJ/L versus 34.7 MJ/L for gasoline), and its corrosivity and miscibility with
water presents challenges for the use of existing gasoline distribution infrastructure for ethanol-rich
fuel blends. These well-known shortcomings drove interest in the production of alternative high
energy density, infrastructure-compatible fuel molecules from plant-derived sugars, commonly
referred to as “drop-in fuels”, “advanced biofuels”, or “fungible fuels”, which became possible with
advances in metabolic engineering. Researchers at JBEI had demonstrated the production of C12-
C18 fatty acid ethyl esters (FAEE) including ethyl hexadecanoate from glucose and xylose in
recombinant Escherichia coli (E. coli) strains, driving the selection of FAEE as a representative
water-immiscible fuel molecule for comparison with ethanol. Many of the conclusions derived from
the analysis are more broadly applicable to other water-immiscible biofuels; however, the fact that
FAEEs are included in the broader definition of biodiesel is an important advantage, as it obviates
the need to undergo potentially lengthy fuel certification processes.
77
The water-immiscibility of advanced biofuels eliminates the need for distillation operations for
product recovery, and potential reductions in the energy requirements of the fuel production process
were touted by some researchers as an important benefit of these fuels. However, a TEA of
advanced biofuels production to quantify these projected benefits in a process-level comparison to
“cellulosic” ethanol (ethanol derived from bioenergy crops) had not yet been published in the
literature, and the study described here was proposed to address this gap.
Scope
The scope of the TEA was limited to the FAEE production process. Switchgrass (Panicum
virgatum) was selected as a representative biomass feedstock due to its prominence in the literature;
however, the production and transportation of switchgrass was outside the scope of this work. The
TEA analysis encompassed biofuel production pathways that were recognized in the literature as
leading candidates for commercialization, and that were the subject of intensive research efforts
within government, industry, and academia. The key process steps and their primary mass and
energy inputs and outputs are depicted schematically in Figure 35. The biochemical processes
considered in the analysis include four major operations: biomass pretreatment to disrupt the
structure of the lignocellulosic material; hydrolysis of cellulose, xylan, and other sugar polymers via
the action of enzymes to yield their respective monomeric sugars (saccharification); microbial fuel
production; and fuel recovery and purification. In addition, industrial processes will require
operations to recover valuable residual streams, treat and dispose of waste, and generate process
steam and electricity. Four biofuel production pathways were examined in the analysis, based on acid
and alkaline pretreatment technologies to produce ethanol and FAEE. The reader is referred to
SAND2013-0208 for a more detailed discussion of the TEA approach, results, and technical
conclusions.
78
Figure 35. Process pathways evaluated in this study for the conversion of cellulosic feedstocks to
ethanol and biodiesel (FAEE)
TEA Approach
The primary goal of the analysis was to assess the projected performance of a process being
developed at SNL and JBEI to produce FAEE relative to a more established process to produce
ethanol. To this end, a deterministic process model representing the biochemical production of
ethanol and a FAEE from acid- and alkaline-pretreated switchgrass was constructed in Microsoft®
Excel, based largely on detailed models of cellulosic ethanol production summarized in design
reports published by researchers at NREL and Michigan State University. The process model in the
SNL study comprised equipment-level mass and energy balances as well as estimates of project
investment costs, operating expenses, and revenues. The model yielded an estimate of the minimum
fuel selling price based on a 20-year discounted cash flow analysis, as well as overall mass and
thermal conversion efficiency, electricity production or consumption, process water consumption,
and net greenhouse gas (GHG) emissions.
The detailed NREL model that served as the starting point for this work was implemented in the
ASPEN chemical process simulation software package. Construction of a simplified model in Excel
enabled the incorporation of multiple pathways from biomass to biofuel within a unified framework,
and facilitated comprehensive Monte Carlo analysis in a fraction of the time that would be required
using detailed models in process simulation software. The Monte Carlo analyses in turn provided
79
valuable insight into the relative impacts of process parameters based on both their intrinsic
influence on the model as well as the variability and uncertainty in their values. The ability to
conduct this type of sensitivity analysis is crucial for identifying key parameters and process steps in
early-stage technologies such as those currently under consideration, for which performance at scale
is uncertain. Upon identification of the parameters driving performance, it is possible to explore the
potential for improving the performance of the technology with directed research and development
efforts.
Conversion Process Model Description

The Excel-based model developed for the study incorporates many simplifications to the ASPEN-
based process simulation models upon which it is based, yet retains the important physical
relationships that govern the behavior of the system. Material and energy balances are calculated
across all major equipment, with detailed accounting of 22 chemical components. A
parameterization approach was employed in place of detailed kinetic and thermodynamic modeling
of reaction systems (e.g., hydrolysis and fermentation), in which the reaction conditions, yield, and
rate are input as independent parameters to the model. Mass balances for separation operations (e.g.
distillation, flash separation, filtration, and centrifugation) are obtained by specifying separation
efficiencies and/or outlet concentrations for key components over the relevant mixture composition
ranges.
Heating and cooling requirements associated with heats of reaction, phase and temperature changes,
and power dissipation are explicitly accounted for, and sensible heat effects are calculated using
constant average heat capacity values or previously published empirical relationships. Power
requirements were estimated using equipment-specific published correlations or vendor
specifications.
80
Equipment costs for the ethanol processes are estimated according to the following exponential
scaling expression, using parameters taken from published design reports:
𝑵𝑵𝑵𝑵𝑵𝑵 𝑺𝑺𝑺𝑺𝑺𝑺𝑺𝑺 𝒂𝒂
𝑵𝑵𝑵𝑵𝑵𝑵 𝑪𝑪𝑪𝑪𝑪𝑪𝑪𝑪 = 𝑩𝑩𝑩𝑩𝑩𝑩𝑩𝑩 𝑪𝑪𝑪𝑪𝑪𝑪𝑪𝑪 �
𝑩𝑩𝑩𝑩𝑩𝑩𝑩𝑩 𝑺𝑺𝑺𝑺𝑺𝑺𝑺𝑺
� (8)
where a is a scaling exponent associated with a specific piece of equipment. Installation costs are
accounted for using an equipment-specific installation factor that is applied to the uninstalled
equipment cost. Equipment costs for FAEE fermentation and recovery operations were estimated
based on a combination of vendor quotes and the built-in cost models of the SuperPro Designer®
process simulation software package. The cost-year for the analysis was 2009, and all equipment
costs were adjusted to this year using Chemical Engineering Magazine’s Plant Cost Index. The total
capital investment (TCI) for the project was estimated by applying factors to the total installed
equipment costs to account for direct and indirect costs associated with the construction of the
facility, as described in the NREL design report. Fixed operating costs associated with various
overhead items were also estimated in this manner.
Raw materials and waste handling charges and labor costs were taken from the NREL design report.
Raw materials and waste disposal costs were adjusted to the cost-year using the Industrial Chemicals
Producer Price Index published by the U.S. Bureau of Labor Statistics, with annual costs calculated
based on 8406 plant operating hours per year. Labor costs were adjusted to the cost-year using the
Employment Cost Index for Wages and Salaries in Manufacturing published by the U.S. Bureau of
Labor Statistics.
This TEA approach adopts “nth-plant” economics in which it was assumed that n other plants have
been previously constructed at other sites and are operating using the same technology. This is often
referred to as the “Nth of a kind” (NOAK) assumption, versus a plant that is the “first of a kind”
(FOAK). Neglecting one-time R&D costs and the risk associated with starting-up a pioneer plant
provides a fairer basis of comparison to mature technologies. A plant size capable of processing
2,000 dry metric tons of switchgrass per day was selected for all pathways, as proposed in a previous
analysis by researchers at NREL for initial biorefineries utilizing cellulosic materials.
The project TCI, operating expenses and revenues were utilized in a discounted cash flow analysis in
order to determine the minimum fuel selling price that would be required to obtain a net present
value of zero for the overall project. Financial parameters - discount rate, debt/equity ratio, income
tax rates, plant life, and construction start-up duration - and depreciation methodology were
identical to those employed in the NREL design report, assumed to be appropriate for an “nth
plant” among a hypothetical fleet of similar biorefineries.
Greenhouse gas emissions attributed to the conversion process were estimated by tabulating the
emissions associated with the main process inputs as well as the avoided emissions associated with
the co-generation of electricity. As no location is specified for the biofuels production facility, the
emissions associated with electricity production are based on the average generation mix for the U.S.
81
national grid. Emissions associated with the biorefinery infrastructure and land-use changes are
neglected.
The use of a simplified spreadsheet-based model enabled a full Monte Carlo analysis based on input
probability distributions for a subset of technical and economic model parameters, and reduced the
time required for such an analysis by several orders of magnitude compared to those employing
detailed models implemented in process simulation software packages. The ability to conduct rapid,
credible Monte Carlo analyses is particularly valuable when evaluating pre-commercial processes and
technologies for which performance at scale is uncertain. Additionally, the flexible and expandable
nature of the model framework makes it possible to incorporate promising new process
technologies as they are developed, such that the performance of all technology options is analyzed
on a consistent basis.
Monte Carlo Analysis

Many of the technologies envisioned for biochemical conversion of biomass to biofuels were still
under development at the time of the study, and their performance in commercial-scale facilities was
unproven. Monte Carlo analysis techniques were employed in order to explicitly incorporate this
uncertainty and variability in estimates of overall process performance, Figure 36. Probability
distributions based on literature data were constructed for a subset of input parameters to the
process model; where data were sparse or unavailable (e.g., FAEE fermentation performance), the
subject matter experts at SNL and JBEI employed engineering judgment based on prior research
experience to construct input parameter distributions. The Crystal Ball (Oracle) add-in for Excel was
utilized to perform 10,000 Monte Carlo trials for each combination of pretreatment technology and
fuel, yielding stochastic distributions for each metric of interest. Input parameters were represented
using the beta-PERT probability distribution, requiring only estimates of the minimum, maximum
and most likely values.
Figure 36. Monte Carlo analysis to explore the impact of uncertainty in conversion process model
parameters
The performance of the biomass conversion processes was evaluated based on the following
economic and environmental metrics: (1) fuel production per metric ton of dry biomass, (2)
82
minimum fuel selling price, (3) net electricity production, (4) process water consumption, and (5)
process GHG emissions. Process water consumption does not include water associated with the
production of feedstocks, other chemical inputs, or grid electricity used in the process.
Results
This TEA explored both near-term process performance and the potential impacts of future process
improvements.
Near-Term Process Performance

Stochastic distributions of the chosen economic and environmental performance metrics based on
Monte Carlo analysis were generated for each of the four modeled pathways from switchgrass to
fuel (see Figure 37 and Figure 38). The input parameter ranges employed to generate these
stochastic distributions were selected to be representative of the current state of technology for the
respective processes for an “nth plant” among a hypothetical fleet of similar biorefineries. It is
important to note that the resulting distributions capture only the variability and uncertainty in the
feedstock compositions and process parameters represented with input distributions, and do not
represent the complete range of possible performance for a given process. For example, financial
and operational parameters – which were fixed for the current analysis – can be expected to have a
significant impact on estimated fuel production costs. The value of analyses such as presented here
is in comparing similar processes and investigating the relative importance of input parameters to
the process model, rather than providing absolute values of performance metrics.
83
Figure 37. (a) Minimum fuel selling price ($ per ethanol-equivalent liter), (b) Fuel yield (ethanol-
equivalent liters per metric ton of dry biomass), (c) Net electricity generation (MW) for a 2,000
metric ton/day conversion facility, and (d) process water consumption (liters of water per liter of
ethanol-equivalent fuel produced) for all pathways to ethanol and FAEE.
(Stochastic distributions are represented as box plots. The top and bottom edges of each box represent
the 75th and 25th percentiles, respectively, the middle line in each box represents the 50th percentile,
and the top and bottom whiskers represent the 97.5th and 2.5th percentiles, respectively.)
84
Figure 38. Net GHG emissions (g CO2 equivalents per MJ of fuel produced) for all pathways to
ethanol and FAEE
(Results in (a) include credits due to co-production of electricity; results for ethanol pathways are
presented in (b) both with and without electricity.)
Analysis of the raw Monte Carlo results revealed the respective contributions from key input
parameters for the pathways employing dilute acid pretreatment (see Figure 39). It is important to
note that the results depicted in Figure 39 are based on relative contributions, and that the total
variance in ethanol MFSP ($0.03) is significantly lower than that for FAEE ($1.62). Not surprisingly,
fermentation parameters dominate the variance in FAEE production costs, as this is the novel
portion of the process. This comparison highlights the relatively narrow distribution on the expected
fermentation parameters for ethanol, a result of thousands of years of experience in producing this
molecule.
85
a b
Figure 39. Contributions to the variance in minimum fuel selling price for process to produce (a)
ethanol and (b) FAEE from switchgrass using dilute acid pretreatment.
(Variance in ethanol and FAEE selling price is $0.03 and $1.62, respectively.)
The total fuel selling price was also broken down into contributions from different unit operations
and raw materials streams for pathways employing dilute acid pretreatment (see Figure 40), for the
scenario in which the ‘most likely’ values of each parameter is input to the conversion process
model. The relative contributions are similar across both pathways, reflecting the fact that the
increased cost for FAEE is primarily a result of lower fuel yield. There is also a significant increase
in the relative cost of the fermentation unit operation for FAEE production, attributable to
increased capital costs associated with aerobic fermentation and longer fermentation times, as well as
a dramatic swing from net electricity production to net consumption. It is interesting to note that
the product and solids recovery section represents only a minor contribution to the overall cost of
fuel production in both cases.
86
Figure 40. Comparison of costs for unit operations and major inputs for the ‘most likely’ input
parameters for ethanol and FAEE production from switchgrass using dilute acid pretreatment
(Unit operation costs include capital recovery charges, fixed costs, and raw materials and electricity costs
(excluding biomass and enzymes).)
Potential improvements to the FAEE process
At the time of this study, microbial production of FAEE had been demonstrated at bench-scale only
recently, and the fermentation performance could be expected to improve as resources were devoted
to this goal. To inform future research efforts, the study investigated the specific improvements that
would be necessary to approach or exceed the performance of cellulosic ethanol processes. The
obvious starting point for such an investigation was the fermentation step itself, as poor yields and
the use of aerobic fermentation were shown to negatively affect the performance of the FAEE
processes. There may also be possibilities for enhancing the separation of FAEE following
fermentation, which was modeled as occurring via centrifugation. Use of a less capital- and energy-
intensive unit operation such as a gravity settler may be possible, given the immiscibility of FAEE
with water. The potential impacts of improvements in these areas were analyzed through
comparisons of the following scenarios:
- Ethanol baseline scenario: Production of ethanol from switchgrass employing dilute acid
pretreatment, based on baseline parameter ranges.
87
- FAEE baseline scenario: Production of FAEE from switchgrass employing dilute acid
pretreatment, based on baseline parameter ranges.
- FAEE high yield scenario: FAEE baseline scenario with ethanol fermentation parameter
ranges assumed for FAEE production.
- FAEE anaerobic high yield scenario: FAEE high yield scenario with anaerobic fermentation
of sugars to FAEE.
- FAEE best case scenario: FAEE anaerobic high yield scenario with zero cost and zero
energy use for FAEE recovery operations.
Stochastic distributions of fuel cost for each of the five scenarios are compared in Figure 41. An
improvement in the FAEE fermentation parameters - yield and residence time - to match those of
ethanol leads to a 65% reduction in the median fuel cost, whereas the switch to anaerobic
fermentation and elimination of centrifugation costs yield more modest reductions. The
improvement of fermentation yields is the main potential driver of cost reduction, as it would enable
increased fuel production - and hence greater revenue - for essentially the same total production
cost. Switching from aerobic to anaerobic fermentation reduces capital costs due to less stringent
sterilization requirements and the elimination of aeration equipment, and also reduces electricity
demand for agitation. Eliminating the costs of separating FAEE yields the smallest incremental cost
reduction.
Figure 41. Minimum fuel selling price ($/ethanol-equivalent liter) for the EtOH and FAEE scenarios
Although the advances that would be necessary to attain the FAEE best case scenario in practice
present an enormous challenge, even the combination of these three changes does not result in a
median fuel cost that is lower than that for ethanol. Because the only difference between the ethanol
baseline process and the FAEE best case scenario - other than the fuel being produced - is the lack
88
of separation costs in the FAEE scenario, one might expect the FAEE costs to be lower; however,
the cost reductions that accompany the “free” separation of FAEE are offset by reduced absolute
sugar conversion efficiencies that are intrinsic to the metabolic pathways utilized by the microbes to
produce FAEE. This conclusion is emphasized in Figure 42, showing a comparison of the costs
associated with the ethanol baseline scenario and the FAEE best case scenario. A significant
negative contribution due to reduced FAEE separation costs is offset by smaller increases across all
other process areas and inputs due to the reduced fuel output. Thus, even in the most optimistic
case of technology development for the FAEE production process, the best that can be expected is
to match the performance of cellulosic ethanol technology. It is assumed that all improvements in
feedstock production practices and pretreatment and hydrolysis technology will apply equally to the
production of both fuels.
Figure 42. Comparison of costs for the ‘most likely’ parameter values of the ethanol baseline
scenario and FAEE best case scenario by unit operation and major inputs
Conclusions
The results of this analysis indicated that biochemical processes to produce FAEE from switchgrass
were uncompetitive with similar processes to produce ethanol under the existing state of technology
development, in terms of both economic and environmental metrics. Beyond this initial assessment,
the study identified the important drivers of performance for both process types and explored
options for improving the performance of the FAEE processes through research and development.
The study found that significant challenges confront potential efforts to close the gap in
performance with cellulosic ethanol processes.
89
The two most promising areas for potential improvement to the FAEE processes investigated are 1)
an enhancement in fermentation parameters (yield on sugars and fermentation time) and 2) a switch
from aerobic to anaerobic fermentation. The main path to process improvement is to increase the
fermentation yield, and thus the production of fuel from a given amount of biomass. By improving
the fermentation parameters to be on par with those for ethanol, the FAEE processes can make up
most of the differences in cost, fuel production, GHG emissions, and water use relative to the
ethanol pathways. The switch to anaerobic fermentation will offer modest gains in cost and water
use but will significantly decrease electricity consumption such that the processes become net
electricity producers rather than net electricity consumers. Further improvements in FAEE
separation efficiency offer only minor cost and energy reductions. Ultimately, FAEE production
processes will approach intrinsic limitations in the metabolic pathways utilized by the host
microorganisms, which place FAEE at a disadvantage versus ethanol in terms of the maximum
potential fuel production from sugars (15.2 MJ of ethanol per kg of sugar vs. 13.8 MJ of FAEE per
kg of sugar). This disadvantage in maximum energy conversion efficiency appears to offset any
potential gains resulting from the elimination of distillation and dehydration steps in the cellulosic
ethanol processes.
While the feedstock and processing technologies selected for analysis in the current study represent
only a small subset of the many potential biochemical pathways from biomass to liquid
transportation fuels, the conclusions that have been reached regarding the relative economic and
environmental performance of the selected processes to produce ethanol and FAEE can be readily
extended to any combination of feedstock and pretreatment method, provided they do not exhibit
an influence on the fermentation step that would favor the production of one fuel over the other.
While the comparison between processes to produce ethanol and FAEE is an imperfect one – as
ethanol is a gasoline additive/replacement and FAEE is appropriate for diesel engines – the results
of our analysis are broadly applicable to other water-immiscible biofuels, most if not all of which
suffer from similarly low maximum theoretical fermentation yields. This point also underscores the
need to extend the scope of such comparisons beyond the biorefinery to the distribution of the fuel
to end users and ultimately to combustion in engines. Further study is needed to explore the
question of whether potential advantages in infrastructure compatibility and combustion efficiency
will be sufficient to overcome the apparent process limitations of producing FAEE for fuel via
biochemical pathways.
Lessons Learned
• Leverage previous analyses whenever possible. This includes previously published
process and economic models. For example, the current study highly leveraged previous
work at NREL focused on cellulosic ethanol production.
90
analysis, and consider the impact on raw materials costs (either increasing or decreasing) and
potential secondary effects if the commercial-scale process would have a significant impact
on existing markets.
• For process-dominated TEAs, mass and energy accounting make it relatively easy to
extend the analysis to include environmental metrics. Energy and water consumption
may already be calculated in the course of the TEA; GHG emissions may be calculated
based on emissions of materials and energy inputs taken from published databases.
efforts.
• Develop a system-level understanding of the technology. It is necessary to understand
how interactions between the elements of a process or technology impact the overall
performance. The current study revealed system-level trade-offs between the fermentation
step and the product separation step that were central to the conclusions.
biology, chemistry, etc. This particularly important when exploring pathways for
improving the performance of the technology of interest. In this study, fundamental
limitations on the metabolic pathways to produce FAEE in microorganisms had a significant
impact on the conclusions.
91
A.2. Industrial process water treatment and reuse enabled by selective ion
exchange materials


This TEA task supported a technology development LDRD. The TEA team reported to the LDRD
PI and worked closely with the technical team, participating in weekly meetings and all project
reviews. The TEA team was engaged from the proposal stage through the conclusion of the project,
and was encouraged to provide input into the technical direction of the project.

The goal of this work was to understand and maximize the near-term potential of a novel silica
removal.
2. Determine the economic performance of this process to remove silica from a representative
water stream given the current state of technology development.
3. Quantify the impact of uncertainty and variability in critical process parameters to identify
the relative contributions of each to total performance.
economic competitiveness, and identify the conditions that would provide the greatest value
proposition for adopting this technology.
Deliverables
- Journal article: Heimer, B., Paap, S., Sasan, K., Brady, P., Nenoff, T. (2019). “Industrial
process water treatment and reuse enabled by selective ion exchange materials,” Industrial &
Engineering Chemistry Research, 58, 14873-14879.
Background
SNL has extensive expertise and capabilities to address both energy security and water security, and
efforts at the Laboratories and at the Federal level to address the energy-water nexus have gained
urgency as water supplies are stressed due to drought and population growth. Cooling operations for
92
thermoelectric power generation represent a prominent example of this nexus, accounting for
approximately 40% of water withdrawals in the United States.
The use of so-called “produced water” from enhanced oil recovery (EOR) operations as cooling
water for thermoelectric power plants has the potential to reduce reliance on freshwater withdrawals
in drought-prone areas. Produced water is a byproduct of the EOR techniques employed to increase
domestic crude oil production, and thus its use in cooling operations would eliminate a waste stream
in addition to supplanting the use of fresh water for this application. However, the high cost of
removing contaminants from produced water and other degraded water resources represents a
significant barrier to their use in place of fresh water. Removing these contaminants using
conventional water treatment technologies is 1.5- to 2.5-fold costlier than simply disposing of the
produced water, due largely to the difficulty of removing silica.
This LDRD leveraged SNL capabilities in the areas of water treatment and chemistry and the
synthesis of inorganic ion exchange materials to develop novel hydrotalcite (HTC) materials for the
removal of silica from produced water. These materials have the potential to remove silica directly
from a variety of waters and lower the energy penalty for treatment without generating large waste
streams. Because the primary obstacle facing conventional technologies for silica removal is cost, the
LDRD PI defined a separate TEA task to guide the development and selection of HTCs and to
assess the projected costs of silica removal processes based on the most promising HTC candidates.
Figure 43. Novel silica getter materials potentially enable the reuse of produced water for cooling
applications in thermoelectric power generation.
Scope
The scope of the TEA included the design of an appropriate commercial-scale process to utilize the
novel silica getter materials, construction of a cost model representing the proposed process, and
93
employment of the model to assess the expected economic performance of the process for
treatment of a representative degraded water stream.
The feedstock selected for analysis was degraded water produced during oil and gas recovery in
California’s Central Valley. The silica loading was varied across the range of concentrations 10 – 200
mg/L typically encountered in produced water, with 120 mg/L selected as the baseline value. The
silica removal process was designed to condition this water for use in thermoelectric power
generation cooling operations where the “exceedingly high” silica concentration greatly limits the
cycles of concentration achievable in the cooling tower; the plant was scaled to 2.8 million gallons
per day (MGD) (4.4x105 kg/h) to meet the demand for make-up water in a cooling tower operating
with side-stream softening at a 500 MW generating facility.
TEA Approach
The techno-economic analysis (TEA) here serves two purposes: (1) determine the minimum water
treatment price (MWTP) to serve as a barometer of commercial viability for the HTC-based process,
and (2) identify the cost drivers for operating the process to determine cost/benefit for future
research and development efforts.
The thermodynamic relations, equipment-level material and energy balance calculations, and capital
and operating expenses were implemented in a deterministic model. The process was subdivided
into the three-unit operations, and calculations were performed separately for each and combined
for the aggregate analyses. Monte Carlo analyses was employed to provide insight into the relative
impacts of process parameters based on both their intrinsic influence on the model as well as the
variability and uncertainty in their values.
Conversion Process Model Description

Material and energy balances were calculated across all major equipment in the proposed silica
removal process. Capital costs for common plant equipment (e.g., pumps and pressure vessels) were
estimated from the process flow sheets using established chemical engineering correlations of the
form:
𝑪𝑪𝒆𝒆 = 𝒂𝒂 + 𝒃𝒃𝑺𝑺𝒏𝒏 (9)
where Ce is the purchased equipment cost, S is the equipment size, and a, b, and n are scaling
constants. Specialized equipment (e.g., rotary kilns and carbon dioxide distribution manifold) costs
were obtained from vendor quotations. Equipment costs were adjusted to the base analysis year,
2015, using the Chemical Engineering Plant Cost Index (CEPCI). Labor costs were similarly
adjusted using the U.S. Bureau of Labor and Statistics employment cost index data for
manufacturing. Raw material and projected utility (e.g., electricity and natural gas) prices used for
variable operating cost calculations were obtained from industry and government sources.
Produced water was assumed to be available to the process in sufficient quantity at no cost for the
baseline case in this analysis. Alternatively, some authors have suggested that produced water,
94
despite being a waste stream, has an “inherent value” approximated as 0.15 2003USD/1,000 gal
(0.055 2015USD/1,000 kg). Other authors in the techno-economic analysis literature suggest a
tipping fee for waste stream feedstocks should incur a negative cost (i.e., a revenue). Such a tipping
fee for produced water would likely fall in the range of underground injection disposal costs (0.07 -
1.60 2012USD/bbl) reported by the U.S. Government Accountability Office (GAO).
The TEA approach adopted “nth-plant” economics in which it was assumed that n other plants
have been previously constructed at other sites and are operating using the same technology.
Neglecting one-time R&D costs and the risk associated with starting-up a pioneer plant provides a
fairer basis of comparison to mature technologies.
The analysis was conducted assuming a twenty-year plant economic life. The total cash flow,
including both capital and operating costs, for each year of operation was discounted to the base
analysis year to calculate the net present value (NPV) of the project. Setting the NPV equal to zero
enabled calculation of the break-even point corresponding to the MWTP.
Monte Carlo Analysis

Process and economic modeling were supported by laboratory experiments. Given the lack of pilot-
scale validation, the TEA team used Monte Carlo analysis methods to formally include uncertainty
and variability in the water treatment cost estimate. Maximum and minimum values for a subset of
the model’s input parameters were estimated from the literature, market data, experimental results,
or when none of these were available, best engineering judgment. The distribution of probable
values within each range was chosen to best reflect the basis of the natural uncertainty in the
associated input parameter. A triangular distribution was used when only the maximum, minimum,
and the baseline (or “most likely”) values were known. The Crystal Ball (Oracle) application was
used to execute 10,000 deterministic model trials each using a separate set of parameter values
randomly sampled from the corresponding probability distribution.
Results
The TEA team designed a commercial-scale process to utilize the novel HTC materials to remove
silica from wastewater, and investigated the near-term and potential future economic performance
of the process.
Process Description for Silica Removal

The process flow diagram in Figure 44 shows all major steps required to treat produced water using
the novel HTC materials. The three major operations are calcining/regenerating HTC, silica
removal, and post-treatment conditioning.
95
Figure 44. Process flow diagram for HTC calcination, silica removal, and post-treatment
conditioning
The calcination step is a thermal treatment to prepare or regenerate the HTC for use; calcined HTC
removes silica from water approximately 9-fold more effectively than does crystalline (non-calcined)
material. HTC is calcined using a rotary kiln in the process described here. Based on experimental
data, the residence time for HTC in the kiln was determined to balance greater silica removal
capacity with increased process and operating costs at longer times. Incorporating a recycle stream
allows the HTC material to be re-calcined and used multiple times before being discarded. Data
gathered by the technical team show that HTC can be regenerated three times before it is unable to
remove >95% of silica from the influent. Therefore, spent HTC needs to be replaced with fresh
calcined HTC material.
The removal of silica from the feed water occurs in parallel ion exchange columns. Two such units
are shown in Figure 44 because it is expected that one vessel will operate on-stream while the HTC
from the other vessel is being regenerated. Produced water with high ionic silica concentration is
pumped over the packed bed of HTC, and treated water is discharged as alkaline solution at
approximately pH 11. The increase to pH 11 is a result of hydroxide anions generated during HTC
recrystallization.
Treated water is neutralized to pH 7.5 to meet cooling tower requirements for pH (7.5 – 8.5) and
help remove the solution sulfides by shifting the equilibrium toward the volatile species. Addition of
sulfuric acid (H2SO4), carbon dioxide (CO2, captured as a byproduct from HTC calcination), or a
combination thereof are the preferred methods analyzed. If CO2 is used, the flue gases from the kiln
96
are passed through a condenser to remove water; the CO2 is then subsequently compressed for
delivery to the aerator vessel via a distribution manifold. The option of capturing water vapor and
CO2 would require using an indirectly fired rotary kiln to avoid carbon dioxide dilution from air flow
in an open, directly fired rotary kiln. This concept as well as the alternative of using a directly fired
rotary kiln with no CO2 capture were evaluated and compared for overall economics.
Near-term Process Performance

The process shown in Figure 44 was the basis for both mass and energy calculations, as well as
capital and operating expense calculations. The MWTP for the baseline case was $0.023/gal (23
2015USD/1,000 gal or 6.08 2015USD/1,000 L); the components contributing to the total are
shown in Table 2. Variable operating costs (e.g., raw materials and utilities) comprise 84% of the
total, or $0.019 for every gallon ($0.005/L) of water treated and are driven principally by HTC
consumption. Capital depreciation is a minor expense equivalent to only 1.6% of the annual
operating cost.
Table 2. Cost comparison of silica removal technologies.
† This work
The break-even price for water treatment is shown on a component basis in Table 2 for three
conventional technologies – membrane based ion exchange, double-pass reverse osmosis, and lime
softening – and the HTC based ion exchange method described here. These mature technologies are
expected to incur only incremental changes over time; as such, the cost index for purchased
equipment (Chemical Engineering), industrial chemicals producer price index (U.S. Bureau of Labor
Statistics), and average hourly earnings, chemical manufacturing (U.S. Bureau of Labor Statistics)
were used to adjust each cost component for these technologies from the basis year to the year of
analysis to facilitate a more accurate comparison. The analysis shows that due to the high cost of the
97
raw material, the HTC method is the costliest to operate. The median and mean values from the
Monte Carlo analysis represent a cost 2- to 4-fold greater than these existing methods.
Capture of CO2 that was liberated during HTC calcination was anticipated as necessary for both the
reduction of greenhouse gas emissions and sulfuric acid consumption for neutralizing the basic
treated water. However, the modeling results indicated it was not economical due to the cost of
additional required capital equipment. The cost of additional installed equipment required for carbon
dioxide capture and bypass (i.e., indirect instead of direct fired rotary kiln, condenser, compressor,
and aerator tank) amounts to $820,000, equivalent to the cost of purchasing sulfuric acid for
approximately thirty-eight years of operation; this cost negates any potential economic advantage.
For this reason, H2SO4 was modeled as an addition to the process for pH adjustment, thereby
eliminating the need for CO2 bypass.
The results of a single parameter sensitivity analysis are depicted in Figure 45; the lines with the
steepest slope correspond to the variables with the strongest influence on MWTP. As such, HTC
material properties (silica binding capacity and number of regeneration cycles) and cost are the most
significant contributors. Electricity utility usage and capital equipment costs are very minor
contributors to MWTP which represent real advantages over some existing silica removal
technologies.
Figure 45. Single parameter sensitivity analysis on minimum water treatment price (MWTP).
98
Projected improvements to the HTC process
HTC was only recently demonstrated for specific silica sorption in the laboratory. The optimized
performance of this material will improve with process and materials optimization toward its
commercial deployment. Therefore, Monte Carlo analysis was used to evaluate improvements in
HTC material properties (ion exchange capacity and reusability) as well as HTC cost reductions.
Furthermore, the modeling was used to account for the other remaining uncertainties in the
scenarios described below.
- Baseline produced water scenario: Current state of technology for silica removal from
produced water using HTC.
- High reusability scenario: Follows the same parameter ranges as baseline produced water
scenario but assumes HTC can be recycled seven times possibly due to future material
preparation method refinements.
- High binding capacity scenario: Follows the same parameter ranges as the high reusability
scenario but assumes HTC binding capacity is 120 mg/g (versus a baseline value of 50
mg/g).
- Low HTC Price scenario: Follows the same parameter ranges as the high binding capacity
scenario but assumes that HTC costs $2.50/kg. Potential HTC price reductions were
determined in consultation with HTC suppliers based on the notion that increases in
demand for the raw material (as more HTC-based water treatment processes are deployed)
would allow them to achieve economies of scale in production thereby reducing costs
toward the gross margin for producing HTC.
The MWTP for each of the stochastic scenarios are compared in Figure 46. Improving HTC
reusability yields the greatest effect and causes the median cost to decrease 40% from 15 to 9
2015USD/1,000 gal (3.96 to 2.38 2015USD/1,000 L). The reductions from improving silica binding
capacity and reduced HTC cost are more modest but yield reductions of 13% and 20%, respectively.
Together they bring the best-case cost down to 4 2015USD/1,000 gal (1.06 2015USD/1,000 L), or
27% of the baseline value. This MWTP would clearly be competitive with the technologies shown in
Table 2.
99
Figure 46. Stochastic minimum water treatment price distributions for four scenarios given
potential improvements on HTC materials
Additional considerations
Silica removal technologies based on HTC ion exchange methods will be economical in certain key
applications. Co-location or near-location of silica removal facilities to produced water generation
sites will dramatically decrease or eliminate produced water disposal transportation costs. Currently,
disposal costs vary over an order of magnitude ($1.67/1,000 gal – $38.10/1,000 gal or $0.44/1,000 L
– $10.06/1,000 L) based on where the water is generated. This cost is primarily determined by the
distance produced water must be transported – typically by truck – to an injection well for disposal.
Produced water generators with many nearby injection wells, therefore, pay dramatically lower
disposal prices due to lower transportation costs. This dynamic exists because oil field operators are
required to dispose of the produced water at the closest injection well. Furthermore, the injection
wells’ locations are determined by both geography and geology, not proximity to oil fields. The
HTC-based treatment process reported here can be co-located with produced water generators and
water consumers to provide substantial water transportation cost savings.
Additional external factors favoring HTC silica removal process are at locations which are
experiencing severe drought conditions. For example, in 2014, Kern County in California’s Central
Valley had astronomically high freshwater prices due to an extended drought. At a maximum, the
high open market for freshwater was priced at $3.99/1,000 gal ($1.05/1,000 L). The ability to reduce
the need for this costly freshwater for use in local industrial and agricultural processes by using
100
treated produced water will dramatically reduce the county’s need for buying freshwater.
Furthermore, the sum of revenue saved from disposing produced water and providing treated water
will move processes into the economically sound arena.
Conclusions
This analysis yielded a feasible process design for using, recovering, and recycling the inorganic ion
exchanger hydrotalcite (HTC) for the purpose of removing silica from water produced during
enhanced oil recovery. Process design parameters such as the calcination and ion exchange residence
times and the number of HTC regeneration cycles were optimized based on experimental data to
minimize water treatment cost. Capture of CO2 from HTC calcination for use in pH adjustment of
the treated water stream was found to be uneconomical compared to use of sulfuric acid.
The results of the analysis indicated that the proposed HTC-based silica removal process was
uncompetitive with existing silica removal technologies under the current state of technology
development. The cost of silica removal via the HTC-based process is dominated by the cost of
HTC; efforts to improve the economic performance of the process should therefore focus on
tailoring HTC material properties to reduce HTC consumption. Specifically, increases in both silica
binding capacity and the number of viable HTC regeneration cycles will be required. Monte Carlo
based scenario analysis showed that, despite having a current treatment cost of approximately 17±9
2015USD/1,000 gal (4.49±2.38 2015USD/1,000 L), reasonable levels of improvement in these
parameters, along with lower HTC costs that may be achievable through economies of scale, could
enable the proposed HTC-based process to compete with incumbent technologies.
Given the criticality of both energy and water, materials such as these HTCs may satisfy a
compelling need to upcycle wastewater into a value stream and further sustainable practices in these
sectors. HTC offers both durability and stability over existing technologies as a result of being an
inorganic ion exchanger. Additionally, the HTC process benefits from a low capital cost in which
capital depreciation accounts for only 1% of the MWTP. Performance of the process is maximized
where freshwater is costly and water treatment facilities are located large distances from disposal
sites. Such geographic factors will be crucial for assessing economic feasibility for constructing
treatment processes.
Lessons Learned
estimate for a point design. TEA can reveal obstacles and opportunities that significantly
impact the direction of research and development efforts.
analysis, and consider the impact on raw materials costs (either increasing or decreasing) and
potential secondary effects if the commercial-scale process would have a significant impact
on existing markets.
101
efforts.
promising applications of the technology, especially when combined with knowledge of
constraints and opportunities in potential markets. In this case, opportunities were identified
based on specific conditions in the EOR industry, and targets were identified to place these
opportunities within reach.
chemistry, etc. This is particularly important when exploring pathways for improving the
performance of the technology of interest.
A.3. TEA of Solid-State Transformers

TEA Staff: Scott Paap and Steven Wiryadinata
Project Dates: April 2019 – September 2019
This work was funded by internal discretionary funding. The funding senior manager (Ben Wu) was
the primary customer.

This project supported an ongoing LDRD to develop solid-state transformer (SST) technology to
realize grid-level resilience and reliability improvements. At the onset of the TEA project, the
LDRD was in the 2nd year of funding. Funding supported the TEA staff.

The goal of this work was to understand the state of solid-state power transformer technology
relative to conventional technology, by pursuing the following objectives:
1. Understand power transformer technology and use in the modern electric grid.
2. Investigate the value preposition of solid-state transformer technology in terms of cost, grid
reliability and resilience, and additional functionality.
3. Identify potential “on-ramp” applications for solid-state power transformer technologies.
102
Deliverables
- Customer briefing: Paap, S., Wiryadinata, S., “TEA for Solid-State Transformers,” (2019).
Presented on September 23, 2019.
Background
The electric power grid is fundamental to the operation and security of modern society. The grid
connects power generating plants to consumers through a network of transmission lines,
distribution lines and power conditioning systems. One critical component of these power
conditioning systems is transformers whose primary function is to convert power among voltage
levels in the grid. This is done to minimize heating losses (I2R losses) during transmission across
long distances and to ensure proper voltage levels are delivered at the end-use for proper operation
of electrical equipment. In the United States, transmission-level large power transformers operate in
the 115 – 765 kV range, while distribution transformers service the 120V – 24 kV loads (Figure 47).
Large power transformers in the US have capacity ratings greater than 60 MVA (mega volt-amps of
apparent power) across various voltage ratings and have an average life of 40 years. Distribution
transformers on the other hand are typically less than 2.5 MVA in capacity.
Figure 47: Transformer voltage levels

[Source: DOE, 2014. Large Power Transformers and the U.S. Electric Grid.]
While current transformer technology is efficient and cost effective, its simplicity is paired with an
inherent weakness: transformers are passive systems that readily transmit input disturbances and can
introduce harmonics through the grid, potentially leading to transformer and other cascading
failures. The risk to an electric grid escalates when power load previously handled by a system with a
failed transformer is diverted to, and increases the load on, other parts of the grid. Should large
power transformer failures occur, recovery efforts can be costly and time-consuming due to the fact
that these are large, expensive, custom-built devices with low or non-existent inventory and long
lead times (~months) for manufacturing and shipping.
103
Concerns of grid instability caused by cascading failures, primarily at the transmission level with
large power transformers, are driving the development of solid-state transformers (SSTs) at Sandia
and elsewhere. SSTs combine traditional voltage regulation with power electronics components to
decouple phase and frequency and enable reactive power (VAR) control. The decoupling and VAR
control minimize the transmission of disturbances and introduction of harmonics, strengthening the
resiliency of the grid. Furthermore, as power electronics enable voltage transformation at very high
frequency relative to the incumbent technology, SSTs can offer the same capabilities in a much
smaller package (smaller transformer core). Much smaller size, combined with modular designs that
reduce the need for custom-built systems, would significantly reduce deployment burden and
inventory overhead and thus shorten the recovery time for grid disruptions.
The development of this project began with a request for the TEA team to support an LDRD-
funded effort to develop SST technology at Sandia. The LDRD team was focused on demonstrating
a novel SST device architecture and investigating its potential impact on grid resilience through
modeling and simulation. The TEA team was asked to (i) assess the projected costs of producing the
novel SST devices compared to those for conventional transformers and to (ii) quantify the
reliability and resilience benefits offered by the SST technology.
Scope
SST devices across all voltage levels (distribution transformers to large power transformers) were
considered for the present work, and economic impacts up to and including potential
macroeconomic effects of a large-scale grid disruption were within scope. The TEA was initially
focused on large power transformers, in alignment with the LDRD work, with a secondary emphasis
on identifying potential smaller scale “on-ramp” applications that could provide entry markets for
the novel SST technology. However, insights gathered during the course of the work later prompted
a TEA refocus to low voltage distribution transformers and analysis of the purchase cost of the
corresponding SST devices. The insights behind the rescope are discussed in Results section, and
relate to the cost scaling relationships of conventional transformers and SSTs as well as non-cost
drivers in the large power transformer market.
TEA Approach
The TEA team initially proposed a two-phase approach, with a focus on large power transformers:
- In Phase 1, the cost and functionality impacts of SSTs for power transmission would be
explored under normal operation of the grid. Cost models for SST device production and
for utility-scale electricity generation/transmission would be constructed and employed in
this phase.
- In Phase 2, the potential economic impact of SSTs on the mitigation of and recovery from
large grid disruption events would be explored based on modeling and simulation results
from the LDRD team.
The TEA began with an effort to understand conventional transformers technology, costs, and
markets. Resources included government, industry, and academic reports, as well as market analysis
104
reports provided by the SNL Business Development organization. As described below in the Results
section, the initial data collection effort revealed factors that present significant obstacles to the
adoption of SST technology for large power transformers by the electric power industry. In addition,
as the project progressed it became clear that the timeline for the modeling and simulation effort of
the LDRD did not align with the TEA duration.
For these reasons, and based on consultations with the LDRD team, the decision was made to
rescope the TEA to focus on a “drop-in” replacement for distribution transformers. These
transformers would not decouple phase and frequency or enable reactive power (VAR) control but
would provide the same functionality and efficiency as conventional dry-type transformers typically
used by industrial or commercial users; the only notable difference to the end user would be the
smaller size of the SST device. Because the SST devices would be drop-in replacements with no
performance benefits (or drawbacks) for the end user, the device purchase price was selected as the
relevant metric for comparison to conventional transformers.
The SST device developed by the LDRD team is a combination of a conventional MnZn
transformer with power electronics; at the center of the SST power electronics are arrays of silicon
carbide (SiC) metal oxide semiconductor field effect transistors (MOSFETs) that switch rapidly to
enable SST operation at high frequency. High frequency operation allows for the use of smaller
transformer cores but necessitates the use of thin Litz wire conductors to reduce resistance in the
windings with increasing frequency. Additional SST components include MOSFET drivers and
power supplies, electromagnetic interference (EMI) filter components, star point synthesis (to
provide 4-wire three-phase power and allow series voltage cascading of SSTs), and a two-state
electronic tap changer. An optimized SST is the result of a performance-to-cost maximization effort
considering tradeoffs among the various elements of the bill of materials; however, such an
optimization was outside the scope of the current analysis. The TEA team analyzed two device
architectures proposed by the LDRD team as an initial gauge of competitiveness.
The purchase price calculations began with cost estimates for each of the SST device components.
For most components, relationships of unit cost versus order volume were developed using scraped
data from wholesale sources such as digikey.com and Alibaba.com to estimate costs for bulk
materials to feed a high-volume production process. Such relationships can be derived for mature
components for which economies of scale are the primary driver for cost reductions. In the case of
SiC MOSFETs, however, technology development is expected to yield significant cost reductions in
the next several years. For this component, near-term cost ($/Amp) was estimated based on
projections from the PowerAmerica National Manufacturing Innovation Institute and a leading
MOSFET manufacturer.
Leveraging available cost structure information for the manufacture of conventional transformers
and power electronics, cost factors as a function of materials-only cost were established to account
for manufacturing labor and overhead, as well as non-production costs including profit, selling,
general & administrative costs, interest, warranty and shipping. The decision to use multiplicative
factors to estimate the non-materials components of the selling price - rather than develop a
105
bottom-up model of the production process - was made based on the relatively short timeframe for
the study and the availability of data indicating that the selling price for similar devices is dominated
by materials costs.
The LDRD team provided the TEA team with efficiency versus capacity (% vs load) curves for two
unoptimized representative SST architectures appropriate for low-voltage dry-type distribution
transformers. U.S. Department of Energy (DOE) Transformer Efficiency Standards (Technical
Support Document for Energy Conservation Standards for Distribution Transformers, 10 CFR 431,
2011) call for the nameplate rating of dry-type transformers to be quoted at 35% load, representing a
region of peak efficiency. Thus, to facilitate a fair comparison to conventional transformers, the
TEA team determined the appropriate efficiency rating of the SST architectures based on the DOE
Standards. The DOE Standards specify the minimum efficiency for transformers as a function of
size (kVA), with higher required efficiencies for larger transformers. For the smaller of the two
proposed designs, the DOE Standards specified a minimum efficiency of 98.4%, corresponding to
19-kVA on the efficiency-capacity curve. As depicted in Figure 48, this resulted in a rating of 54-
kVA (19 kVA divided by 35%). Using the same approach, the second (larger capacity) design was
rated at 106 kVA based on a minimum efficiency of 98.8%. The TEA team estimated the potential
price points for the two SST architectures and compared them to the representative vendor prices
for conventional transformers at the same rated capacity.
Figure 48: Example calculation of dry-type transformer capacity rating based on the device
efficiency curve
Results
The decision to focus on the 54-kVA and 106-kVA designs for low voltage distribution service was
heavily informed by an understanding of the challenges facing the SST technology in the large power
transformer market. The first set of challenges is related to the structure and incentives of the U.S.
electric utility sector. Utilities tend to be risk averse and hesitant to adopt unproven technologies,
especially for transmission infrastructure. Additionally, both the benefits – and potential avoided
106
costs – afforded by SSTs in the transmission system accrue to society at large rather than to the
utility, weakening the business case for the technology. For example, a utility will suffer lost revenue
during a large-scale grid disruption, but it will not bear the full cost to the broader economy. The
reader is referred to the 2009 National Academies of Science report, “America’s Energy Future:
Technology and Transformation”, for further discussion of these challenges. It should also be noted
that these conclusions generally do not apply to the distribution transformers market. The lack of
cascading effects, relatively high growth, and more frequent equipment upgrades in the distribution
system enable the small-scale testing of new technologies with low risk to utilities and other users.
The second challenge facing SSTs in the large power transformer market is economic, stemming
from the scaling relationships for both conventional transformers and SST devices. Large power
transformers follow a typical cost versus scale relationship, in which the cost per unit capacity
($/kVA) decreases as the capacity increases. Thus, large power transformers are less expensive (per
kVA) than small distribution transformers ($7-$13/kVA in 2011 dollars for large power
transformers based on a 2014 DOE report, versus $86/kVA in 2019 dollars for a 20kV distribution
transformer). Conversely, the SST devices under development by the LDRD team consist of
multiple identical devices connected in series and/or parallel to achieve larger capacities. Although
this approach offers significant advantages stemming from standardization and modularity, the
scaling relationship becomes linear as capacity increases. That is, the cost per unit capacity ($/kVA)
approaches a constant value at higher values of transformer capacity. Based on these scaling
relationships, lower capacity transformers would be a more attractive entry market for SST devices.
The two aforementioned insights steered the TEA to an assessment of SST devices without phase
decoupling and VAR support for the distribution transformers market. The comparison against low-
voltage, dry-type conventional transformers was based on the purchase cost per unit rated capacity
($/kVA). The estimated purchase price of the 54-kVA and 106-kVA devices was $22/kVA (+4/-5,
$/kVA) and $31/kVA (+5/-8, $/kVA), respectively; the components of the purchase price are
depicted in
Figure 49. For the prescribed architecture at high production volume, the MOSFETs comprise
~30% of the total potential sale price of the SST architecture and ~49% of the materials-only cost
(lightest blue bar in Figure 49). The second-highest cost component is the EMI filter (darkest blue
bar in figure) consisting of inductors and film capacitors. The Copper (Cu) Litz winding represents
the third highest cost of materials (dark green bar in figure). Due to relatively low production
capacity, Aluminum (Al) Litz wire is currently 50-80% more expensive than Cu Litz wire for the
same strand # and wire size. However, the use of Al Litz wire should be considered in the future,
since Al is one-third the cost of Cu and ~60% the conductivity of Cu on a mass basis.
107
Figure 49: Components of total purchase price for the representative SST architectures
Aside from being the largest single contributor to the total purchase price of the proposed SST
devices, the SiC MOSFETs also drive the performance in terms of overall device efficiency. The
proposed SSTs are subject to core and winding losses as in a conventional transformer (on a smaller
scale), with additional losses in the power electronics components. Given the current state of
MOSFET technology, the largest source of power electronics-related losses – and the largest source
of losses overall – is the total drain-source resistance of the MOSFET arrays (Rdson). As noted above,
the capacity ratings of the representative, unoptimized SST architectures were determined based on
the 35% operating point specified by the DOE Efficiency Standards. Reductions in Rdson shift the
efficiency curve (see Figure 48) upward and thus lead to an increased kVA rating for an otherwise
identical device. Assuming the cost of the MOSFET remains constant, this reduces the purchase
price per unit rated capacity ($/kVA). Furthermore, larger transformers are subject to higher
efficiency standards, and the proposed SST architectures are unable to meet the standards for
devices above ~300 kVA capacity given current MOSFET performance. The implication is thus:
decreases in Rdson of SiC MOSFETs can unlock SST use at higher capacities.
Currently, only a limited number of commercial off the shelf (COTS) MOSFETs feature values of
Rdson that are low enough (i.e., yield sufficiently high efficiency) for use in the SST devices. These
MOSFETs tend to be more expensive than those with higher Rdson values; Figure 50 depicts price
data for Cree SiC MOSFETs at two relevant voltage levels. However, increasing demand for high-
voltage MOSFETs in EV and PV inverters, power distribution (low voltage), data centers,
communications equipment and power supplies is expected to drive continued improvements to the
technology and reductions in MOSFET costs. As the technology advances, MOSFETs with lower
Rdson are expected to enter the market and the cost curves in Figure 50 are expected to shift lower.
At this point, an optimization will be required for each device to balance trade-offs in efficiency
(driven by MOSFET Rdson) versus MOSFET cost.
108
Figure 50. Price vs Rdson for Cree SiC MOSFETS
(The red box indicates the region in which Rdson values are sufficiently low for use in the proposed
SST devices. Data gathered from Mouser.com in September 2019.)
Estimated purchase prices for the 54-kVA and 106 kVA SST architectures are plotted in Figure 51
along with price data from a representative conventional 480V dry-type transformer. The
conventional transformer technology follows a power law-style cost reduction with increasing
capacity. In contrast, the present SST architectures follow an increasing cost trend with increasing
capacity due to the aforementioned factors: more stringent efficiency standard and limitations of
current COTS SiC MOSFETs. Thus, the 106 kVA device has a potential price of $31/kVA (+5/-8,
$/kVA), placing it on par with the baseline transformer at $32/kVA. In the notional sense, the larger
SST design defines the upper limit on cost-competitiveness, while the smaller of the SST
architectures offers potentially significant value in terms of lower first-cost to industrial and
commercial consumers. These results indicate that the low-voltage dry-type transformer market is a
promising entry market to demonstrate SSTs for workhorse transformer duty, and expected
improvements in high-voltage SiC MOSFET technology will only drive the SST price curve
downward.
109
Figure 51: Comparison of SST and conventional transformer retail price versus capacity.
Conventional transformer price gathered from manufacturer’s website in July, 2019
Conclusions
The initial findings of this analysis indicated that the large power transformer market presented
significant obstacles to entry for the proposed SST devices, due to (i) the risk-aversion of the utility
sector and the societal (vs. corporate) nature of the benefits derived from SSTs, and (ii) differences
in the inherent price-versus-capacity curves for conventional transformers and SST devices. These
findings led to a change in scope of the TEA away from large power transformers to application in
low-voltage distribution transformers without phase and frequency decoupling or reactive power
control.
The results of the re-scoped TEA suggested that the proposed SST devices could be competitive
with conventional distribution transformers on the basis of both price and performance in the near
term (within 5 years). The SiC MOSFETs used in the SST architectures were found to be the
primary driver of both price and performance (efficiency), and the drain-source resistance of the
MOSFET, Rdson, was identified as the key parameter governing MOSFET – and SST – efficiency.
Current market and technology maturation trajectories also pointed to MOSFETs as the largest
potential source of cost and performance improvements for SSTs.
The transformer efficiency standards promulgated by the U.S. DOE were identified as an important
external factor. The standards require higher efficiency as transformer capacity (kVA) increases,
defining the current upper capacity limit of SST devices. The DOE standards also dictate the
calculation of the overall device capacity rating based on the device’s efficiency versus capacity
110
curve. All other things being equal, an increase in SST efficiency thus leads to higher device capacity
and therefore lower SST price ($/kVA).
The structure of the efficiency standards and the diverging nature of the cost-versus-capacity curves
for conventional transformers and SSTs (see Figure 51) led to the conclusion that dry-type, low-
voltage, three-phase transformers of less than 100 kVA represented the most promising entry
market for SSTs. Looking to the future, existing market drivers are expected to sustain MOSFET
technology development efforts, leading to decreasing prices and Rdson values. These trends will only
make SSTs more attractive for the distribution transformers market. They will also open other
markets to SSTs, as the SST price curve in Figure 51 shifts downward and extends to higher
capacities. In this fashion, the SST technology could move along the conventional transformer price
curve to become competitive at progressively higher capacities. By the time the technology is
inexpensive enough to compete with large power transformers on price, the SST technology will
have been demonstrated in commercial devices, potentially making their adoption by the
traditionally risk-averse utility industry more likely.
Lessons Learned
• The initial scope may need to be adjusted. Plan for frequent interactions with the
customer early in the project to define the analysis scope. Be open to changes in scope and
project direction based on initial results.
• Identify the desired outputs and the inputs (data, modeling/simulation results)
required to obtain them as soon as possible. If the required inputs are unavailable,
consider re-scoping or conducting a scenario-based analysis.
estimate for a point design. TEA can reveal critically important obstacles and opportunities
that significantly impact the direction of research and development efforts.
• Account for non-economic factors, as they may dominate the TEA conclusions. As
discussed in the present work, risk aversion is a major factor in the electric power sector, and
government efficiency standards are a significant external constraint within the distribution
transformers market.
• Start with simple, narrowly-scoped models. In this work, a full levelized cost calculation
was deemed to be unnecessary, and the cost model was limited to the production and sale of
the device.
economic models. For example, the current study leveraged previous work at DOE for costs
of both large power transformers and distribution transformers.
111
promising applications of the technology and reveal potential advantages in comparison to
incumbent technologies.
• Funding TEAs can be a challenge. Even if a PI recognizes the benefits of TEA, it may
not be possible to fund TEA work under the existing funding structure.
A.4. Microsystems-Enabled Photovoltaics (MEPV)

TEA Staff: Scott Paap and Vipin Gupta
This work was a task within the LDRD Grand Challenge “Microsystems-Enabled Photovoltaics”
(MEPV).

This TEA task supported a technology development LDRD Grand Challenge. The TEA team
reported to the Grand Challenge PI and worked closely with the technical team, participating in
weekly meetings and all LDRD and External Advisory Board reviews. The TEA team was engaged
throughout the project and was encouraged to provide input into the technical direction of the
project.

The goal of this work was to assess the potential cost of a novel photovoltaics (PV) technology and
inform research efforts to realize cost reductions, by pursuing the following objectives:
1. Construct models yielding the expected costs of producing, installing, and operating the
novel PV modules for utility-scale electricity production.
2. Identify a cost reduction pathway to achieve a total installed system cost of $1 per peak-Watt
($1/Wp) by 2020.
Deliverables
- Paap, S., Gupta, V., Tauke-Pedretti, A., Resnick, P., Sanchez, C., Nielson, G., Cruz-Campa J.
L., Jared, B., Nelson, J., Okandan, M., Sweatt, W. (2014). “Cost analysis of flat-plate
concentrators employing microscale photovoltaic cells for high energy per unit area
applications,” 2014 IEEE Photovoltaic Specialist Conference (conference paper and poster).
- Paap, S., Gupta, V., Cruz-Campa J. L., Okandan, M., Sweatt, W., Jared, B., Anderson, B.,
Nielson, G., Tauke-Pedretti, A., Nelson, J. (2013). “Cost analysis for flat-plate concentrators
employing microscale photovoltaic cells,” 2013 IEEE Photovoltaic Specialist Conference
(conference paper and presentation).
- Briefings and reports to LDRD office and External Advisory Board.
112
Background
The US Department of Energy Solar Energy Technologies Office launched the SunShot Initiative in
2011 with the goal of reducing costs for solar energy by approximately 75% by 2020. This translated
to a levelized cost of electricity (LCOE) target of $0.06 per kilowatt hour for utility-scale solar, and a
total installed system cost target of $1 per peak-watt for photovoltaics (PV). While conventional
non-concentrating PV and concentrating PV (CPV) were moving towards this target, significant
obstacles remained for both technologies. Researchers at SNL proposed an alternative approach to
achieve the SunShot target, dubbed “Microsystems-Enabled Photovoltaics” (MEPV), which
leveraged deep SNL expertise in semiconductor processing and optical systems.
The MEPV module architecture is based on arrays of millimeter-scale polymer lenses positioned to
focus light on hexagonal photovoltaic cells with maximum vertex-to-vertex distances between 100
μm and 1000 μm (Figure 52). Concentrating optics enable the use of high efficiency PV cells to
increase the energy output of the system, while the miniaturization of PV cells and concentrating
optics significantly reduces the amount of lens material that is required, eliminates the need for heat
sinks or other heat dissipation strategies, and enable a module form factor that is similar to
conventional (e.g., silicon-based) non-concentrating PV modules.
Figure 52. Schematic representation of a microsystems-enabled photovoltaics (MEPV) concept.
The potential for the MEPV architecture to enable cost reductions compared to incumbent
technologies (non-concentrating PV and CPV) can be understood by examining the components of
the LCOE equation (Figure 53). Here, the goal is to reduce the costs in the numerator while
increasing the electricity generation in the denominator.
113
Figure 53. Mathematical equation for calculating the Levelized Cost of Electricity (LCOE). NPV
refers to the net present value of the quantities in brackets.
(The design goals are to increase energy generation while lowering costs.)
As indicated in Table 3, the MEPV architecture has system-level impacts on cost and performance
that extend beyond the module, providing a pathway to leverage the lower system-level costs of
non-concentrating PV while generating electricity at the high conversion efficiencies of macroscale
CPV systems. However, the attractiveness of MEPV as an alternative to traditional PV and
macroscale CPV depends on the development of low-cost module designs that offer high
conversion efficiency. The TEA described here attempted to quantify all contributions to the LCOE
equation – with an emphasis on MEPV module cost – in order to assess the potential for the MEPV
architecture to meet the 2020 SunShot targets.
Table 3. Comparison of the components of LCOE for non-concentrating PV, CPV, and MEPV
technologies. (O&M: Operations and Maintenance.)
Scope
The scope of the TEA included all MEPV design concepts proposed by the Grand Challenge team.
The TEA team addressed all elements of the LCOE equation (Figure 53) for a hypothetical utility-
scale power generation facility; the team also responded to requests for analysis support to inform
114
design decisions. Energy generation estimates were generated for representative high-insolation (e.g.,
Las Vegas, NV, Albuquerque, NM) and low-insolation (e.g., Wilmington, DE) locations.
TEA Approach
Cost models were developed spanning the entire MEPV value chain from the production of
components to the installation and operation of a utility-scale power plant (Figure 54); models
yielding estimates of cell fabrication cost and module assembly cost were at the core of the overall
cost modeling framework. The key metrics of interest were the cost of the system in dollars per
peak-Watt ($/Wp) and the levelized cost of electricity (LCOE, $/kWh) produced by a utility-scale
MEPV power plant.
Figure 54. MEPV cost modeling framework and components
The early-stage nature of the MEPV technology presented a significant challenge, as many aspects of
MEPV module design and production were initially unknown or ill-defined, and iteratively evolved
over the course of the Grand Challenge. To address this challenge, the cost modeling structure was
designed to be flexible and adaptable to design changes. In addition, cost estimates were augmented
with sensitivity studies and trade-off analyses to identify parameters governing system cost and
compare proposed designs. The insights gained from these analyses were employed to guide the
design of successive prototypes, identify areas for further research and development, and inform the
selection of important system parameters such as concentration ratio and cell size.
Cell cost model

The cell cost model yields an estimate of the cost ($/wafer or $/cm2) to fabricate cells based on an
industrial-scale cell fabrication process flow. The cost of each of the more than 200 steps in the cell
fabrication process was estimated based on contributions from raw materials, capital costs, labor,
facilities overhead, and consumables. Model input parameters, including tool cost and performance
parameters (e.g., throughput, capital costs, labor requirements, materials and energy consumption,
and footprint) and materials costs were obtained through direct inquiries with tool vendors.
Module cost model
115
The cell fabrication cost serves as an input to the module cost model, which calculates the expected
module cost ($/WP or $/m2) for a given optical concentration ratio and cell size. The module cost
model also calculates the cost of the optics system as a component of the overall module cost, based
on scaling relationships tailored to each prototype design. The overall efficiency of the module is
calculated based on estimates of the cell efficiency (conversion of light to electricity) and the optical
efficiency of the module (fraction of the light striking the module that reaches the active surface of
the cells) provided by the technical team.
The MEPV architecture purposefully leverages many of the same materials and processes (e.g.,
screen printing, placement of optics, module sealing, junction box installation) used to manufacture
conventional Si PV modules; 2013 costs and projections to 2020 for these components and
processes were taken from reports by the National Renewable Energy Laboratory (NREL). Costs
for other materials were obtained directly from vendors, and costs for other processes were modeled
based on analogous processes in related industries. Cost contributions from facility overhead, capital
costs, labor, and consumables were assumed to be similar to those for Si PV production and were
taken from NREL report.
Installed system cost model

The module cost and efficiency estimates are provided as inputs to the total installed system cost
model, which also accounts for tracking systems, other balance-of-system components (e.g.,
inverters, wiring), and installation labor. Because the MEPV modules were explicitly designed to be
robust and have similar form factors as Si PV modules, many of the system-level costs for MEPV
are similar to those for conventional non-concentrating PV; current costs and projections to 2020
for relevant components and installation operations were taken from reports by the National
Renewable Energy Laboratory (NREL). A recent report by researchers at NREL describes system-
level PV costs for utility-scale plants with one-axis tracking and outlines a cost-reduction pathway
through the year 2020; these projected costs were used directly in the current analysis. Two-axis
trackers were assumed to be twice as expensive than one-axis trackers, based on an article in an
industry publication.
Although LCOE is the most appropriate metric for comparing different PV technologies, the Grand
Challenge focused on the total installed system cost as the primary metric. This enabled the team to
pursue the target of $1/WP for the installed MEPV system without addressing assumptions
regarding financing and the location-specific solar resource that are necessary for LCOE
calculations. In certain instances, the total installed cost ($/WP) was adjusted using a factor to
account for the additional energy generation (kWh/WP-year) that would be expected from an MEPV
system.
Calculation of LCOE
Although the Grand Challenge focused on the metric of total installed system cost, the TEA team
generated LCOE estimates throughout the project for representative locations using the System
Advisor Model (SAM) as well as the Cost of Renewable Energy Spreadsheet Tool (CREST), a
Microsoft Excel-based cash flow model, both provided by NREL. Assumptions regarding
116
operations and maintenance (O&M) costs were taken from NREL reports and industry trade
publications; MEPV O&M costs were assumed to be similar to those for conventional non-
concentrating systems with dual-axis tracking.
Estimates of annual electricity production of installed MEPV systems (kWh/WP-year) utilized data
obtained from NREL’s National Solar Radiation Data Base (NSRDB), accounting for the various
losses within the system; Figure 55 depicts the primary energy flows and losses for a representative
MEPV system.
Figure 55. Energy flows and losses in an MEPV system (FOV: Field of View)
Results
This TEA generated both near-term (2013) estimates and 2020 projections for the total installed
system cost for four MEPV successive prototype designs (Protypes 1-4). The TEA team also
conducted analysis to inform design decisions for each prototype and to identify potential pathways
for cost reductions.
Optimization of cell size and concentration ratio

For each potential module architecture, the costs of the PV cells and the optics must be balanced by
adjusting the concentration ratio and cell size, taking into account the desired performance of the
module in terms of thickness, weight, and optical efficiency.
In general, the thickness of the optical system – and the amount of material required for the lenses –
increases with both the concentration ratio and cell size. From a cost perspective, these trends
would tend to favor lower concentration ratios and smaller cell sizes. However, the total cell area
(and thus the cost of cells) required to provide a given amount of power decreases with increasing
concentration ratio, which favors designs featuring higher concentration ratios.
Additional constraints on both concentration ratio and cell size include the following.
- Alignment tolerances in the manufacture of lenses and the placement of cells limit the
maximum concentration ratio to approximately 500X, depending on the optical design.
- During the cell fabrication process, individual cells are etched from a larger wafer and
material of a defined thickness is lost at the border of each cell during the etching process.
As the cell size decreases below approximately 150μm, the material lost to etching becomes
significant.
117
- Module thickness and weight must not significantly exceed those of conventional PV
modules to ensure that MEPV can leverage existing components and installation practices.
For each module prototype, the TEA team conducted an analysis based on cost trends for PV cells
and the optical system to identify the optimal parameter space for the concentration ratio and cell
size. Note that the optimal parameter space strongly depends on the assumed cell cost and the
specific optical design. An example of the outputs generated in such an analysis of an early MEPV
module concept is depicted in Figure 56. For all prototypes, the TEA team recommended a cell size
of approximately 200-250µm.
Figure 56. Identification of the optimal region of concentration ratio and cell size for an early
MEPV module concept
Optimization of optical system design

In addition to facilitating the optimization of cell size and concentration ratio for each MEPV
prototype, the TEA informed the optical design of each successive prototype.
As an example, MEPV Prototype 2 featured two optically aligned polycarbonate (PC) lens arrays,
with a polydimethylsiloxane (PDMS) fill material between the two arrays to improve optical
performance (Figure 58). A cost analysis indicated that the lens and fill material were the main
contributors to the optical system cost (Figure 57), representing a barrier to cost reductions.
Figure 57. Optical system costs as a function of cell size and concentration ratio.
118
Based on input from the TEA team, the optics team pursued designs which employed less – and less
expensive – material. The final MEPV design (Prototype 4) concentrates sunlight onto PV cells
using two optically aligned polydimethylsiloxane (PDMS) lens arrays, with an air gap between the
arrays (Figure 58). The lens design specifically minimizes the use of plastics (PDMS and PC) for the
lenses and accepts slightly reduced optical efficiency with the removal of the fill material. The
amount of PDMS increases only slightly with concentration ratio; therefore, the Prototype 4
concentration ratio is 200X, the highest practical value given constraints on the ability to maintain
alignment of the optics within expected manufacturing tolerances.
Figure 58. Schematic representations of (a) an early MEPV module design and (b) the final module
design (Prototype 4).
During the development of the Prototype 4 design, the optics team considered the use of either
PDMS or ethylene-vinyl acetate (EVA) for the lens material. EVA was a more established
commodity with a lower (and stable) price, while PDMS had a higher price but offered better optical
performance in the MEPV system. The TEA team conducted a break-even cost analysis to
understand the conditions under which the higher optical efficiency of PDMS outweighed its higher
cost (Figure 59). Based on this analysis, the optics team selected PDMS as the lens material.
119
Figure 59. Output of a break-even analysis comparing EVA and PDMS as lens material
Analysis of cell placement strategies

The PV cells used in MEPV modules are fabricated in closely packed arrays on semiconductor
wafers. These cells must then be transferred to the MEPV module substrate in a sparse array. So-
called “pick-and-place” tools are commonly used in the microelectronics industry to assemble
microscale components; however, the use of these tools to place microscale PV cells was found to
be cost prohibitive due to the large number of cell placements required per module at the
concentration ratios of interest (best-case cost for placing 250 µm cells at 200X concentration ratio
is approximately $0.40/Wp, as shown in Figure 60).
Figure 60. Cost of MEPV cell placement with "pick-and-place" tools.
120
In order to reduce the cost per cell transferred, several concepts were developed for massively
parallel transfer of cells to sparse arrays. The primary factors determining the cost of this step are the
amount of time required for one transfer cycle (cycle time, T), the number of cells transferred per
cycle, P, and the total capital and operating costs, CT, over the lifetime, L, of the tool performing
the transfer operation. These factors determine the cost to place a single cell, CU, according to the
following relationship.
𝑪𝑪𝑪𝑪 = 𝑪𝑪𝑪𝑪/(𝑳𝑳 × 𝑷𝑷 ⁄ 𝑻𝑻) (10)
The proposed parallel placement concepts are at an early stage of development, and thus significant
uncertainty is associated with the specific cost and performance factors that would determine cell
placement cost. Rather than estimate cell placement cost for a specific concept, a generalized
analysis was conducted in order to understand the parameter space that would yield cost reductions
due to parallel placement compared to conventional pick-and-place techniques. A cell size of 250
μm was selected based on previous analysis results, and tool life, L, was assumed to be five years; the
number of cells per placement, P, was set at 500, which is a conservative estimate based on the
lowest value of P among the parallel placement concepts considered. The total cost, CT, and
placement time, T, are subject to the greatest uncertainty, and were varied in order to provide insight
into their impact on cell placement cost in terms of $/Wp; the results are plotted in Figure 61, with
total costs given in terms of costs for a pick-and-place process.
Figure 61. Contour plot of cell placement costs, based on 250 µm cells, 200X concentration ratio,
5-year tool life, and 500 cells per placement cycle
(Colors represent total cost of placing cells in units of $/Wp.)
The expected cycle time (placement time) for the parallel techniques under consideration is on the
order of seconds to tens of seconds. This is substantially longer than placement times for
conventional pick-and-place tools, which are able to place several cells per second. However, the
121
large numbers of cells placed per cycle more than compensate for this difference and translate to
significant improvements in the throughput of parallel placement approaches. An order of
magnitude reduction in cell placement cost compared to pick-and-place operations (from $0.40/WP
to ~$0.05/WP) would be achievable even for cycle times in the tens of seconds and total costs that
exceed those for pick-and-place by several times.
Cost reductions for cell fabrication

The high-efficiency III-V semiconductor PV cells used in the MEPV modules were found to be a
significant driver of the overall module cost. The cell fabrication cost model identified the process
steps that were the largest contributors to the cost of cells (Figure 62). The largest single cost is the
epitaxial growth of the III-V semiconductor. The TEA team engaged directly with a prominent tool
vendor to quantify expected improvements in epitaxy performance (e.g., faster growth rates,
enhanced materials utilization efficiency) over the subsequent 5-year period; these improvements
were incorporated in the cell fabrication model to generate the 2020 cell cost estimate. The GaAs
substrate on which the III-V semiconductor is grown was also a major cost in early estimates,
prompting the MEPV cell team to focus on substrate re-use in order to reduce costs.
Figure 62. Components of estimated (2013) and projected (2020) cell fabrication cost for the final
MEPV design.
Bonding and de-bonding, deposition of metals, and dry etch steps also represent significant costs in
both the 2013 and 2020 estimates. It should be noted that the costs of these steps were estimated
based on the use of conventional tools currently used in the microelectronics industry; this is in
contrast to the furnace, PECVD, and resist pattern steps, which are performed using tools
developed specifically for the Si PV industry. Whereas the microelectronics industry demands tools
that deliver high quality and small feature sizes, the focus of the PV industry is on high throughput,
large surface area, and low cost. The replacement of conventional lithography (“Litho track” and
“stepper” costs in the 2013 estimate) with screen printing or laser ablation (“Pattern resist” in the
2020 estimate) is an example of the significant cost reductions that may be achieved by tailoring
tools and processes to the needs of the PV industry, and the MEPV concept is able to capture these
benefits. It is expected that similar cost reductions may be possible for steps involving bonding/de-
bonding, metal deposition, and dry etching, although these were not explored in the current work.
122
As noted above, the loss of material in the border etching process becomes significant at the PV cell
sizes that are relevant for MEPV. Additional area on the cell is lost in order to make electrical
connections. After the TEA highlighted the importance of these inactive cell areas, the technical
team focused on reducing them, resulting in significant cost reductions at the module level.
Hybrid module trade-off analysis
Multiple module concepts were proposed during the course of the Grand Challenge, including a
“hybrid” module that placed III-V semiconductor cells on a wide-area silicon PV substrate to collect
diffuse light that could not be concentrated by the MEPV lens arrays. The increased power output
of the hybrid modules came at an increased cost due to the use of silicon substrates in place of a less
expensive substrate. The TEA team conducted an analysis to determine whether this was a
worthwhile tradeoff, generating LCOE estimates for a representative high solar resource location
(Albuquerque, NM) and a location with a higher fraction of diffuse radiation (Wilmington, DE) over
a range of MEPV (non-hybrid) module cost values (Figure 63). These preliminary results indicated
that the hybrid module concept had the potential to reduce LCOE in locations with more diffuse
light. Because the hybrid module architecture did not confer a significant advantage in locations with
less diffuse light (and lower LCOE), it was not seen as a more attractive path to achieve the SunShot
targets of $1/WP and $0.06/kWh.
Figure 63. LCOE estimates for MEPV and "hybrid" modules in two representative locations.
Module cost and Installed system cost estimates

The final Prototype 4 module design concentrates sunlight onto microscale PV cells using two
optically aligned polydimethylsiloxane (PDMS) lens arrays (see Figure 58). One lens array is cast on
the rear of a solar-grade low-iron front glass (3.2mm thickness), and an identical lens array is cast
directly over PV cells on a lower-cost commercial glass that serves as a rear substrate. Silver
123
interconnects screen-printed on the rear substrate collect the generated electricity from the cells and
direct it to a junction box installed on the rear of the module. A glass spacer maintains the required
separation between the front and rear glass, and a hermetic edge seal prevents moisture ingress to
the air gap within the module. The module design requires no polymer backsheet (e.g., Tedlar) or
encapsulant (e.g., EVA).
The proposed module assembly process utilizes proven tools and methods with minimal
modifications. The process begins with screen printing of interconnects on the rear glass; the use of
solar glass with dimensions that are common to the Si PV industry enables the use of currently
available tools for handling glass sheets and performing screen printing operations. Deposition of
thin conducting lines (<50µm) by screen printing has been demonstrated for PV modules. The PV
cells from the cell fabrication process are transferred to the rear glass in a sparse array; the sparse-
array transfer is a novel operation that will require a new tool. The lens arrays are cast on the front
and rear glass substrates using standard techniques, although custom machinery may be required for
casting PDMS on meter-scale substrates. Finally, the front and rear glass sheets are aligned and
assembled using commercial adhesives and edge sealants, and a junction box is installed on the rear
of the module. The final 2020 module cost estimate for Prototype 4 was $0.36/WP, a significant
reduction over the 2013 Prototype 3 estimate of $1.44/WP (Figure 64).
Figure 64. 2013 (Prototype 3) and 2020 (Prototype 4) MEPV module cost estimates
Because MEPV modules are similar in thickness and weight to conventional non-concentrating
silicon PV modules, large-scale MEPV systems can directly leverage the components and installation
methods that have been developed for the conventional PV market. The primary non-module costs
for a utility-scale PV plant are land acquisition and site preparation (including permitting fees);
purchase of tracker systems, inverters, wiring, and other electrical components; supply chain costs to
account for project overhead (including interest during construction, inventory, and project-
contingency costs); system installation (including general and electrical labor and installer profit and
overhead); and sales tax. The use of two-axis trackers increases costs for land and wiring, as two-axis
trackers require wider spacing in order to avoid shading of adjacent modules. The current analysis
124
does not account for potential cost reductions in wiring and other electrical components that may be
possible by connecting many small cells in parallel within a single MEPV module. Inverters
represent the only system-level cost component that does not depend on module efficiency. Because
module efficiency for MEPV is substantially higher than that for conventional silicon-based PV,
MEPV system-level costs will always be significantly lower on a per-Watt basis than those associated
with non-concentrating PV, even as costs are reduced across the PV industry. After the addition of
module costs, the total MEPV system price estimate for 2020 is $1.16/WP (Figure 65).
Figure 65. The contributions of non-module system costs as a function of module efficiency, with
estimates of MEPV module cost for multiple prototypes
Conclusions
The MEPV architecture potentially enables PV systems with the high efficiency of CPV and lower
system-level costs of non-concentrating Si PV. Analysis of the MEPV concept revealed a credible
pathway for achieving the 2020 LCOE target of $0.06/kWh set by the US DOE’s SunShot
Initiative; however, achieving this target would require cost reductions across the entire MEPV value
chain from the production of components to the installation and operation of a utility-scale power
plant:
PV cell fabrication
The PV cells are both a significant contributor to module cost and the primary driver of module
efficiency, thus impacting the effective cost per watt of the entire system. System efficiency must
approach or exceed 40%, and cell fabrication costs must be reduced. Projected cost reductions in
epitaxial growth of III-V semiconductor must be realized, and semiconductor processing techniques
for high-value integrated circuit components must be replaced by low-cost, wide-area processing
approaches such as those being adopted by the Si PV sector. The width of the inactive border
125
around each PV cell was also found to be an important parameter, effectively limiting the minimum
size of the cells.
Optics system design and fabrication

Optics system designs featuring a PDMS fill material between lens arrays were found to be too
expensive due to high materials cost; an air gap between the lens arrays is preferred despite reduced
optical efficiency. In general, MEPV optics should minimize the amount of plastic lens material
within constraints of manufacturability. The optical design should be optimized in response to the
cost of the cell material, with higher expected cell costs driving designs to higher concentration
ratios even if it results in costlier optics. Large-format casting or molding of lenses is necessary to
minimize fabrication costs.
Module assembly
MEPV modules should leverage components and processes from Si PV manufacturing to the
greatest extent possible. Development of methods for massively parallel transfer of PV cells to the
module substrate in sparse arrays will be crucial to reduce module assembly costs; analysis showed
that an order of magnitude reduction in cell placement cost compared to conventional pick-and-
place operations would be achievable even for conservative assumptions of tool cost and cycle time.
System components and installation

As with MEPV modules, installed systems should fully leverage components and processes from the
Si PV sector. This constrains the optical system design and PV cell size to allow similar form factor
and weight compared to Si PV. Because the concentration ratios relevant for MEPV are lower than
those for CPV, the tolerances of dual-axis trackers may be eased with concomitant reductions in
cost. High-voltage operation of MEPV modules should be explored for its potential cost benefits.
Finally, light-to-electricity conversion efficiency improvements should be pursued at every level of
the system to reduce the $/WP cost of the system components.
Lessons Learned
economic models. For example, the current study leveraged previous work at NREL for
costs at all stages of the value chain, from raw materials costs to LCOE calculators.
• Start with simple, narrowly-scoped models. In this work, models for cell fabrication cost,
module cost, and total installed system cost were constructed to be as simple as possible
while capturing the relevant physical and cost relationships. This enables more of the project
time to be spent on employing the cost models to conduct analysis.
• Focus on effectively communicating results to support the technical team. Complex
models that accurately represent system behavior are of little use if the technical team is
unable to extract useful insights.
126
estimate for a point design. TEA can reveal critically important obstacles and opportunities
that significantly impact the direction of research and development efforts. To realize these
benefits, the TEA team must be closely engaged with the technical team throughout the
project and be flexible in responding to the analysis needs of the technical team.
promising applications of the technology and reveal potential advantages in comparison to
incumbent technologies.
• Develop a system-level understanding of the technology. It is necessary to understand
how interactions between the elements of a process or technology impact the overall
performance. The current study revealed that a focus on module costs alone could cause one
to overlook the system-level benefits of the MEPV concept.
A.5. Potential Strategies for Integrating Solar Hydrogen Production and

Concentrating Solar Power
TEA Staff: Scott Paap

Project Dates: July 2014-November 2015
The funding for this work was provided by the Fuel Cell Technologies Office (FCTO) of the U.S.
Department of Energy (DOE) Office of Energy Efficiency and Renewable Energy (EERE).

This was a stand-alone TEA project intended to identify promising areas for collaboration and
integration across two program areas of interest to SNL and DOE. Funding supported the TEA PI
and support from technical SMEs. The PI reported directly to FCTO program managers, providing
regular status updates via teleconference meetings and one in-person meeting.

The goal of this work was to investigate potential synergies that may be realized by integrating solar
hydrogen (H2) production and concentrating solar power (CSP) technologies, by pursuing the
following objectives:
1. Identify representative CSP-H2 production scenarios for analysis.
2. Construct simplified economic models representing the behavior of the selected scenarios.
127
3. Employ the simplified models in analyses to identify important performance drivers and
fundamental conditions that favor the integration of CSP and H2 production.
Deliverables
- Webinar: Paap, S. (2016) “Potential strategies for integrating solar hydrogen production and
concentrating solar power: A systems analysis,” US DOE Fuel Cell Technologies Office
Webinar.
Background
Concentrating solar power (CSP) and solar hydrogen (H2) production technologies both convert
thermal energy derived from concentrated sunlight into other forms of useful energy. In the case of
CSP, this solar thermal energy is converted to electrical energy; in the case of solar H2 production, it
is converted to chemical energy stored in H2 molecules. Multiple technical approaches exist for both
CSP and H2 production, and SNL has extensive expertise and capabilities in both areas, primarily
funded by the EERE Solar Energy Technologies Office (SETO) in the case of CSP and FCTO in
the case of H2 production.
Although CSP and solar H2 production leverage similar physical systems to harvest sunlight and
convert it to thermal energy, development of the technologies had progressed in relative isolation
with little consideration of potential synergies. To address this gap, the study described here
explored whether integration of the two solar thermal technologies could yield process advantages
that reduce the cost of both CSP electricity and solar H2.
Scope
This analysis explored potential benefits resulting from process-level integration of CSP and solar H2
production systems. Consideration of the potential benefits of employing H2 for energy storage at a
CSP plant were outside the scope of the analysis, as were potential transportation costs or savings
related to the geographical location of solar H2 production plants.
The specific CSP and H2 production technologies considered for analysis were limited to those of
greatest interest to SETO and FCTO, respectively. The DOE SunShot initiative has identified
power towers (also referred to as central receivers) as the technology with the greatest potential for
achieving its 2020 electricity cost target of $0.06/kWh. Thus, a 150 MW power tower with molten
salt heat transfer media and thermal storage was selected as the representative CSP technology for
the current study, in alignment with SNL’s power tower roadmap (SAND2011-2419) and DOE’s
SunShot Vision Study. In consultation with FCTO, the following H2 production technologies were
128
selected for analysis: polymer-electrolyte membrane (PEM) water electrolysis, high-temperature
(high-T) water electrolysis, and high-temperature thermochemical cycles based on the reduction and
re-oxidation of metal oxides. Hybrid H2 production cycles were not considered in the current study.
Figure 66. Flow schematic of a molten-salt central receiver system. EPG: Electric power
generation system. Source: SAND2011-9320.
The purpose of the current work was to draw general conclusions regarding the potential for
synergies resulting from the integration of CSP and H2 production, and to identify promising areas
for collaborative research at the intersection of these technologies. Thus, optimization of specific
integrated CSP-H2 production scenarios and the generation of detailed cost estimates (e.g., a design
case) were outside the scope of the study.
TEA Approach
The TEA began with an effort to understand existing and proposed CSP and H2 production
processes as described in reports and models published by SNL and DOE, with the objective of
defining a set of integrated processes for analysis. The relevant CSP and H2 production processes
were examined at the level of major process equipment and material and energy streams, with a
focus on waste heat and material (byproducts). Following a preliminary analysis, a total of seven
distinct processes (“Cases”) were identified for inclusion in the study, each associated with one of
the three H2 production technologies listed in the Scope section above.
Leveraging previous analyses of CSP and H2 production, cost models representing the selected
Cases were constructed in Microsoft® Excel to investigate the potential benefits of process
integration. In the case of CSP, DOE SunShot and SNL reports provided power conversion
calculations, capital and operation and maintenance (O&M) costs, levelized cost of electricity
129
(LCOE) calculations, and cost and performance targets. In the case of H2 production, process
configurations and costs were taken directly from cost models produced for the DOE Hydrogen
Analysis (H2A) Project. The H2A models provide calculations of H2 cost ($/kg) using a discounted
cash flow analysis based on conversion efficiency and capital, O&M, and materials costs. For both
H2 production and CSP, the technical and cost assumptions represent future systems: the “Future
Central Hydrogen Production” H2A models used for this study represent a start-up year of between
2025 and 2030, and 2020 SunShot target costs were assumed for CSP.
As H2 was the primary product in all Cases, the cost models constructed in the TEA yielded the
break-even cost of H2 ($/kg) as the primary output, with electricity revenue accounted for in the H2
cost according to.
𝑨𝑨𝑨𝑨𝑨𝑨𝑨𝑨𝑨𝑨𝑨𝑨𝑨𝑨𝑨𝑨𝑨𝑨𝑨𝑨 𝒄𝒄𝒄𝒄𝒄𝒄𝒄𝒄𝒄𝒄𝒄𝒄𝒄𝒄 𝒄𝒄𝒄𝒄𝒄𝒄𝒄𝒄𝒄𝒄+𝑨𝑨𝑨𝑨𝑨𝑨𝑨𝑨𝑨𝑨𝑨𝑨 𝑶𝑶&𝑴𝑴 𝒄𝒄𝒄𝒄𝒄𝒄𝒄𝒄𝒄𝒄−𝑬𝑬𝑬𝑬𝑬𝑬𝑬𝑬𝑬𝑬𝑬𝑬𝑬𝑬𝑬𝑬𝑬𝑬𝑬𝑬𝑬𝑬 𝒓𝒓𝒓𝒓𝒓𝒓𝒓𝒓𝒓𝒓𝒓𝒓𝒓𝒓
𝑪𝑪𝑪𝑪𝑪𝑪𝑪𝑪 𝒐𝒐𝒐𝒐 𝑯𝑯𝑯𝑯𝑯𝑯𝑯𝑯𝑯𝑯𝑯𝑯𝑯𝑯𝑯𝑯 =
𝑨𝑨𝑨𝑨𝑨𝑨𝑨𝑨𝑨𝑨𝑨𝑨 𝑯𝑯𝑯𝑯𝑯𝑯𝑯𝑯𝑯𝑯𝑯𝑯𝑯𝑯𝑯𝑯 𝒑𝒑𝒑𝒑𝒑𝒑𝒑𝒑𝒑𝒑𝒑𝒑𝒑𝒑𝒑𝒑𝒑𝒑𝒑𝒑
(11)
A full discounted cash flow analysis was not performed; instead, annualized capital costs are
calculated according to:
𝑨𝑨𝑨𝑨𝑨𝑨𝑨𝑨𝑨𝑨𝑨𝑨𝑨𝑨𝑨𝑨𝑨𝑨𝑨𝑨 𝒄𝒄𝒄𝒄𝒄𝒄𝒄𝒄𝒄𝒄𝒄𝒄𝒄𝒄 𝒄𝒄𝒄𝒄𝒄𝒄𝒄𝒄𝒄𝒄 = (𝑭𝑭𝑭𝑭𝑭𝑭)(𝑻𝑻𝑻𝑻𝑻𝑻𝑻𝑻𝑻𝑻 𝒅𝒅𝒅𝒅𝒅𝒅𝒅𝒅𝒅𝒅𝒅𝒅 𝒄𝒄𝒄𝒄𝒄𝒄𝒄𝒄𝒄𝒄)(𝟏𝟏 + 𝑰𝑰𝒏𝒏𝒏𝒏𝒏𝒏𝒏𝒏𝒏𝒏𝒏𝒏𝒏𝒏 𝒄𝒄𝒄𝒄𝒄𝒄𝒄𝒄𝒄𝒄𝒄𝒄𝒄𝒄)(𝟏𝟏 + 𝑨𝑨𝑨𝑨𝑨𝑨𝑨𝑨𝑨𝑨) (12)
where FCR is the constant-dollar fixed charge rate and AFUDC is the allowed funds during
construction to cover interest charges. The value of FCR, AFUDC, and a multiplier for indirect
charges were taken from published reports. Direct costs were estimated for each process section
according to the following exponential scaling expression, using parameters taken from published
design reports and models:
𝑵𝑵𝑵𝑵𝑵𝑵 𝑺𝑺𝑺𝑺𝑺𝑺𝑺𝑺 𝒂𝒂
𝑵𝑵𝑵𝑵𝑵𝑵 𝑪𝑪𝑪𝑪𝑪𝑪𝑪𝑪 = 𝑩𝑩𝑩𝑩𝑩𝑩𝑩𝑩 𝑪𝑪𝑪𝑪𝑪𝑪𝑪𝑪 �
𝑩𝑩𝑩𝑩𝑩𝑩𝑩𝑩 𝑺𝑺𝑺𝑺𝑺𝑺𝑺𝑺
� (13)
where a is a scaling exponent associated with a specific process unit. The price of electricity was
parameterized in the study due to large uncertainties in future values (2020-2030).
Results
The information-gathering and preliminary analysis phase of the TEA focused on understanding the
material and energy flows within the relevant processes, with an emphasis on the collection of solar
energy as heat. In the case of CSP, the analysis was limited to power tower concepts with molten salt
heat transfer media, in which an array of rotating mirrors (heliostats) focuses sunlight on a solar
receiver that transfers heat to flowing molten salt. Solid particle receivers, in which heat is
transferred to a stream of falling solid particles inside the solar receiver, were also considered for
collection of thermal energy at higher temperatures that are relevant for certain H2 production
approaches. Regardless of the heat transfer media and application (CSP or H2 production),
collection of solar thermal energy represents a large portion of the overall process cost, and in both
concepts heat losses in the receiver – and therefore the cost of thermal energy – increase with
130
temperature (Figure 67). Higher temperatures generally enable higher efficiencies for a given
downstream process (electricity generation or H2 production), but these benefits must be balanced
with the increased cost of solar energy collection. Previous analysis at SNL determined that the
optimal temperature for heat collection in a power tower CSP plant is 565°C.
Figure 67. Energy flow diagram of solar energy capture and conversion to heat in a representative
power tower with molten salt heat transfer fluid and nine hours of thermal storage
Because thermal energy is a large cost, and because neither CSP nor H2 production processes yield
large material byproduct streams, integration strategies for CSP and H2 production focused on
thermal energy management. In the case of CSP systems, electricity generation cycles feature
efficient internal heat integration such that waste heat exiting the system is of low quality (i.e., low
temperature). Thus, energy flows in an integrated process are expected to go from H2 production to
electricity generation.
Analysis Cases
Based on the preliminary findings, seven Cases were identified for further analysis, each one
associated with a H2 production Scenario:
- Scenario 1: PEM water electrolysis using grid-delivered CSP (1 Case)
- Scenario 2: High-temperature electrolysis integrated with CSP (3 Cases)
- Scenario 3: High-temperature thermochemical hydrogen production integrated with CSP (3
Cases)
Existing H2A models were available for each of the three Scenarios considered for the analysis, the
results of which reveal large variation in the amount of thermal energy used by each technology
(Figure 68). While PEM electrolysis uses no thermal energy, high-T electrolysis leverages a relatively
small amount of thermal energy to significantly increase efficiency of H2 production, and
thermochemical metal oxide (TC) cycles convert larger amounts of thermal energy directly to
chemical energy. Note that the H2A model for High-T electrolysis assume that thermal energy is
supplied by a nuclear reactor.
131
Figure 68. Estimated future hydrogen production costs for four representative hydrogen
production technologies, from H2A models.
Scenario 1: PEM water electrolysis using grid-delivered CSP

Although PEM electrolysis utilizes no thermal energy, it was included as a baseline for the analysis
due to its maturity relative to the other Scenarios. Hydrogen production via PEM electrolysis
requires large amounts of electricity, and the price of electricity is the primary driver of H2 cost. For
this Scenario, electricity is provided by a CSP plant; assumptions for CSP electricity costs – 2010
SNL estimate of $0.15/kWh; 2020 SunShot target of $0.06/kWh – are identified in Figure 69.
Figure 69. Estimated current and future hydrogen production cost for PEM electrolysis as a
function of electricity cost
Scenario 2: High-temperature electrolysis integrated with CSP

High-T electrolysis leverages a relatively small amount of thermal energy to significantly increase
efficiency of H2 production. Both electrolysis (H2 production) and electricity generation feature
132
efficient heat integration, resulting in low-quality waste heat. Thus, Scenario 2 does not involve any
process-level integration of CSP and H2 production; in all Cases, electricity is produced by CSP and
transferred to the co-located H2 production facility.
Case 1: Single tower dedicated to providing thermal energy, multiple additional CSP towers
to provide electricity
In this Case, a single high-T (~850°C) tower provides thermal energy for the H2 production process,
while additional lower-T (565°C) CSP towers generate electricity for electrolysis (Figure 70). Because
high-T electrolysis requires more electricity than heat, eleven additional CSP towers would be
necessary to supply electricity for each tower supplying heat exclusively for H2 production. The CSP
and H2 production facilities do not need to be co-located in this configuration. Previous studies have
identified optimal size ranges for towers; therefore, if more heat is needed, there are few economies
of scale such that adding more towers of optimal size is preferred.
Figure 70. Schematic representation of Case 1 of Scenario 2
Case 2: Single tower dedicated to hydrogen production

For Case 2, 9% of the thermal energy from a single tower at ~850°C is used directly for H2
production, and the remaining 91% is utilized to generate electricity for electrolysis. Here, the
efficiency of electricity generation is higher due to the higher temperature. However, increased
temperature also leads to higher costs for thermal energy collection.
133
Figure 71. Schematic representation of Case 2 of Scenario 2.
Case 3: Two towers for hydrogen production, each providing thermal energy at a different
temperature
In this Case, one tower is operated at ~850°C and a second below 565°C. Eighteen percent of the
thermal energy at 850°C is utilized to raise the temperature of electrolysis, and the excess thermal
energy from the first tower along with all thermal energy from the second tower is utilized for
electricity generation (Figure 72). This configuration leads to more cost-effective solar thermal
energy collection in the lower-temperature tower while still enabling more efficient electricity
generation that an optimized CSP plant (48% efficiency at >600°C vs. 42% for a CSP plant at
565°C).
Comparison of Scenario 2 Cases
The cost of H2 production and the contributions by component for each Scenario 2 Case are
depicted in Figure 73. Lower capital costs due to economies of scale and more efficient thermal
energy collection favor Case 1, despite its lower electricity generation efficiency. Case 3 is preferred
over Case 2 due to lower costs for thermal energy collection. It should be noted that consideration
of different tower sizes was outside the scope of the TEA; it may be advantageous to use a smaller
134
tower to generate thermal energy for H2 production as in Case 3, such that there is just enough heat
at 850°C to drive H2 production and also raise the temperature (and efficiency) of electricity
production.
Figure 73. Comparison of estimated hydrogen costs for Scenario 2 Cases
Scenario 3: High-temperature thermochemical hydrogen production integrated with CSP

Thermochemical metal oxide cycles convert thermal energy directly to chemical energy via the
thermal reduction of metal oxide particles at high temperature and subsequent splitting of water to
yield H2, Figure 74. Thermal energy is exchanged at high temperature between the solid particle
streams cycling between the reduction chamber and the water splitting chamber; inefficiencies in this
heat recovery step result in the generation of “waste” heat at high temperature (>1150°C). The
optimal size of the thermochemical H2 production towers is 4.24 MWT, compared with a single large
1000 MWT tower for CSP; thus, H2 production of 90,000 kg/day would require 231 towers.
135
Figure 74. Schematic representation of a high-temperature thermochemical metal oxide hydrogen
production cycle.
Case 1: Electricity purchased from the grid
This is a baseline Case for Scenario 3, similar to that represented by the SNL H2A model. The waste
heat from the process is not utilized, and electricity – accounting for approximately 10% of the total
cost of H2 – was assumed to be purchased from the grid. The “Hydrogen production” unit depicted
in Figure 75 corresponds to the dashed box in Figure 74.
Case 2: Internal power generation from waste heat

Here, the waste heat from thermochemical H2 production is utilized to generate electricity for
internal use. Electricity generation is sufficient to meet process power needs, with a small excess
potentially available for sale to the grid. In this Case, there is no need to purchase grid electricity, but
the smaller scale of power generation equipment reduces efficiency and increases cost compared to
full-scale CSP.
136
Case 3: Integration with CSP

In Case 3, the excess thermal energy from the metal oxide thermochemical cycle is combined with
the thermal energy of a full-scale CSP plant. By doing so, the temperature of CSP electricity
generation is raised from 565°C to over 600°C, increasing the efficiency of power generation from
42% to 48%. At the same time, the solar thermal energy collection in the CSP tower is more
efficient at a lower temperature. Due to this efficiency-enhancing effect, waste heat from metal oxide
thermochemical H2 production can be viewed as a potential high-value “feedstock” for CSP.
Comparison of Scenario 3 Cases

The cost of H2 production and the contributions from each of the components for each Scenario 3
Case are depicted in Figure 78 as a function of electricity price. From the perspective of H2
production, integration with CSP (Case 3) is attractive at electricity prices above $0.08/kWh, and H2
cost approaches zero as electricity prices climb above $0.14/kWh. Note that the Case 3 curve (blue)
in the plot at the bottom-right of Figure 78 will shift to the left as CSP costs are reduced beyond the
137
SunShot target. At very low electricity prices, the baseline case with no integration with CSP (Case 1)
yields the lowest H2 cost.
Figure 78. Hydrogen production costs versus electricity price for Scenario 3 Cases
Scenario 3 Cases at lower hydrogen production efficiency

The Scenario 3 Cases described above assume a future metal oxide thermochemical cycle that has
been optimized to maximize H2 production. This “optimal” process assumes more efficient heat
recovery and a higher H2 production temperature (1150°C) than the state of the art at the time of
the TEA. A second set of assumptions reflecting the 2015 state of the art was also examined,
primarily a lower H2 production temperature (800°C) and less efficient heat recovery. For this
assumption set, more “waste” heat is available for electricity generation. The results depicted in
Figure 79 and Figure 80 indicate that although capital and O&M costs are higher in Case 2 and Case
3 under the assumptions yielding more waste heat, these additional costs are more than
compensated for by increased electricity revenue as electricity price exceeds $0.05/kWh.
138
Figure 79. Contributions to hydrogen production costs versus electricity price for Scenario 3
Cases reflecting an “optimal” future hydrogen production process and the 2015 state of the art
yielding more waste heat
Figure 80. Hydrogen production costs versus electricity price for Scenario 3 Cases reflecting an
optimized process (solid lines) and the 2015 state of the art (dashed)
139
Conclusions
General
Collection of solar thermal energy is a significant cost for both CSP and solar H2 production
Heat integration is a potential strategy for improving the performance of
both CSP and H2 production
Optimal temperature of CSP is lower than that for H2 production
CSP yields no high-T waste heat or significant material byproducts
Necessary to look for potential heat flows from H2 production to CSP
Electricity prices have a significant impact on the analysis results
From the perspective of H2 production, CSP-H2 integration is favored when CSP
price is lower than electricity price
High-T Electrolysis
A relatively small input of heat is required compared to electricity needs
No high-T waste heat is available from H2 production
Integration of multiple towers for combined H2 + electricity production is potentially
attractive
More efficient collection and conversion of thermal energy
Excess heat from high-T tower can be diverted to raise the efficiency of electricity
production by 15% (relative)
Diverting high-T heat to power generation will decrease thermal energy
collection efficiency
Case-by-case optimization will be required to determine lowest-cost configuration
Metal oxide TC cycles
For metal oxide TC cycles, high-quality “waste” heat may be available in larger quantities
than is needed for internal electricity generation
Electricity demand of MO TC cycles is relatively small
Internal electricity generation using waste heat has minimal impact for low to
moderate electricity prices
Integration of MO TC cycles and separate CSP tower is potentially attractive
Impact of high-T waste heat is amplified by integration with CSP
Efficiency of electricity generation could be increased by 15% (relative)
Waste heat from H2 production has high potential value as CSP feedstock
Future metal oxide TC cycles assume reductions in inert material, high recuperation of high-
T heat
Current metal oxide TC cycles may generate significantly more waste heat
Increased potential for electricity revenue as a bridge to future development
Uncertainty and Sensitivity

Solar H2 technologies are at an early stage of development
140
Costs and performance are highly uncertain
Detailed optimizations are premature
The key analysis results are the set of insights regarding favorable conditions for CSP-H2
integration
These results (insights) are robust
Insights are driven by inherent characteristics of processes
Insights are unaffected by absolute H2 production costs (excluding electricity costs)
A.6. Pathways to Carbon Neutral Energy Systems at the University of

California, Davis
TEA Staff: Steven Wiryadinata, Josh Morejohn, Kurt Kornbluth

Intermittency, and by the University of California, Davis, Facilities Management - Energy and
Engineering (UCD FM-EE) office.

This was an NSF funded project, cost-shared by internal university funding due to aligned research
interest. Funding supported the TEA staff of the University of California, Davis. The TEA team
worked with the UCD FM-EE to scope and develop a roadmap for RES implementation for the
UCD campus.

The goal of this work was to understand the techno-economic feasibility of transitioning the energy
system of the University of California, Davis campus to a carbon neutral system by 2025, by
pursuing the following objectives:
1. Assess the availability of renewable resources in the local context of UCD.
2. Investigate the cost-optimal phase-in times and system capacities of identified RES
technologies.
3. Compare and contrast distinct pathway scenarios for achieving carbon neutrality, along with
identification of major economic and technical drivers for implementation.
141
Deliverables & Awards
Journal article: Wiryadinata, S, Morejohn, J & Kornbluth, K. (2018). Pathways to carbon neutral
energy systems at the University of California, Davis. Renewable Energy. 130.
10.1016/j.renene.2018.06.100.
Internal communication: Presentation to UCD FM-EE staff
Award: 2019 AASHE Sustainability Award in Campus Sustainability Research
Background
This effort was driven by the UC-wide Carbon Neutrality Initiative (CNI) in light of recognized
adverse climate effects of carbon emissions, and to spur creative solutions in implementing carbon
reductions measures. The UC CNI called for net zero CO2 emissions from stationary energy uses
(buildings and facilities) of all 10 UC campuses by 2025, achieved through the combination of
energy efficiency (technological and behavioral) to reduce carbon-emitting energy uses, and
implementation of RES to provide the balance of energy demand.
The University of California, Davis comprises of the main campus at Davis, CA, a medical center at
Sacramento, CA and various satellite labs throughout the state. To date, UCD has implemented
renewable energy based on organic waste and solar, and aggressive modernization of facilities
through energy efficiency. However, significant work remains as a majority of buildings was built pre
1970s (53% of the campus total building footprint) and continue to operate on outdated
technologies such as constant air volume systems, high air pressure terminal units, uninsulated
building envelop and single glazing windows without thermal breaks. The campus is largely served
by a natural gas-fueled district system encompassing 42-km of underground steam piping that have
degraded thermal performance due to age and wear heating. Since steam heating produces high
grade heat for medium (process) and primarily low grade (space heating) end-uses, the system's
energy and exergy efficiency is low. A conversion project to transition the steam district heating
system to a hot water system is underway to reduce heating loads and the associated energy and
maintenance costs.
With the limitations of how far cost-effective energy efficiency can reduce demand, there was
pressing need to identify the techno-economic feasibility of RES to balance the energy needs of the
campus. Implementation of RES comes with the challenge of high first cost relative to fossil
technologies and mismatch of energy supply and demand if storage technologies are not
implemented. In making decisions to implement RES, a techno-economic analysis (TEA) is
indispensable in evaluating the systems' cost-effectiveness. TEA of energy production systems
simulates the potential electricity and thermal generation of the systems, usually in hourly scale over
an entire year, and evaluates these against their estimated capital, operation and maintenance costs
142
over the system lifetime. TEA provides an analytical indication of the technical and economic
feasibility of these systems.
Scope
The scope of the study is a techno-economic assessment of relatively mature RES, limited to the
main UCD campus comprising of property and activities such as office, laboratory and residential
buildings, waste and wastewater treatment, infrastructure, maintenance and transportation. Specific
objectives are to:
Assess local renewable resource availability to down select classes of RES that have the potential for
significant energy generation
Compose energy system electrical and thermal load profiles for the UCD campus.
Analyze and if necessary, develop the tools for analyzing, the hourly output of RES to investigate
supply-demand relationships.
Perform a parametric TEA of RES options and combination of RES options to reveal system
capacities and phasing timelines with lowest potential levelized cost.
Highlight opportunities and tradeoffs of certain RES implementation pathways.
TEA Approach
RES Down Select

The selection of renewable generator types appropriate for UCD, located in Yolo County, California
was based on an evaluation of readily available local resources, namely: biomass (BM), solar and
wind. While hydropower constitutes a major portion of the UCD's power mix (18% of the average
annual electricity load), it is a result of the utility portfolio, rather than a locally operated hydropower
plant. In California, hydropower exceeding 30MW is not considered a renewable resource because
of detrimental impact on wildlife and water resources, and the reliance on large hydro reduces
innovation in other technologies. Despite this issue, the present study considered the contribution of
hydropower in the grid electricity purchase as carbon neutral.
UCD is advantageously located in a sector with abundant BM availability compatible with technically
proven thermochemical conversion processes such as direct combustion, or gasification. These BM
are available in the form of forest and agricultural residues, and some diverted biosolids and
municipal solid wastes. To place in context, the feedstocks in Yolo county can be converted into
approximately 3 PJ/yr of electrical energy, assuming a 28-MWe direct combustion power plant at
143
30% thermal efficiency and 100% capacity factor (CF). Considering availability of the same biomass
types in the neighboring counties, an additional 25 PJ/yr. can be available. Despite the great
abundance, collection and transport of these resources remain reliant on fossil-fueled equipment
which make them costly and carbon intensive endeavors and is captured in this analysis.
Solar is a highly attractive resource to tap into at UCD which receives approximately 1.8 MWh/m2
of total horizontal insolation annually with a non-zero average of 388 W/m2 and a maximum of 1
kW/m2. Assuming a fixed collector tilted at the latitude of 38.5° and no maintenance downtime, the
system capacity factor (CF) is estimated at 18%. With 1-axis azimuth and 2-axis tracking capabilities,
the CF increases to 25% and 27%, respectively, which are on par with the average 2015 CF of 28%
for utility scale solar PV and 22% for utility scale solar thermal. As of 2016, the 13.4-MWe 1-axis
tracking PV array at UCD demonstrated a 29% CF.
Wind in Davis is classified as a low potential resource with hourly wind speed averaging at 3.8 m/s,
a non-zero minimum of 1 m/s and a maximum of 19.5 m/s. Assuming a 3.5 m/s cut-in speed, a 15
m/s cut-out speed and an average power coefficient of 0.49 (based on Siemens SWT-2.3 to SWT-3.4
series on-shore turbines), the wind turbine capacity factor is estimated at 4.7%. If the cut-out speed
is decreased to 10 m/s (lower wind turbine capacity), the CF increases to 14.9%. Both figures are
remarkably lower than the average 2015 utility-scale wind CF of 32%.
Due to the above reason, BM and solar are the two renewable resources selected in this work.
Pathway Analysis
Following a bottom-up approach, inputs including hourly demand profiles, equipment performance
parameters and costs were first defined. Hourly energy output and fuel use were then calculated, and
the net present cost (NPC), levelized cost of energy (LCOE) and rate of return (ROR) were
calculated over the 20-yr analysis period.
Four pathways scenarios were considered (Table 4): Baseline (Base); Alternate #1 - Biomass (Alt_1);
Alternate #2 - Biomass and Electrification (Alt_2); and Alternate #3 - Electrification (Alt_3).
The Base case represents the business-as-usual campus energy system consisting of:
Three NG steam boilers with a combined capacity of 82-MWt and standalone gas burners of various
capacities for heating demand.
Eight x 8.8-MWt electric chillers, and standalone electric air-conditioners of various capacities for
cooling demand.
A combination of a 13.4 MWe one-axis tracking solar PV array, 0.6 MWe fixed-tilt rooftop and
canopy PV systems, an anaerobic digester coupled to a microturbine and organic Rankine cycle
system (biodigester, BG), and purchased grid electricity for electrical demand.
144
Table 4. Energy System for Each Scenario
The grid electricity includes contribution by hydro (36.3 MWh/yr.) and an 18-MWe grid PV PPA
(estimated 46.4 MWh/yr.). By 2020, the BG system is expected to operate at 400 kW, ramping by
200 kW in 2021 and by 325kW in 2022. Additionally, the campus is limited from exporting power to
the grid due to a non-export agreement with the local utility provider (Pacific Gas & Electric
Company). While a net energy metering agreement can be adopted, export capacity is limited to only
1MW (approximately 6% of the campus minimum demand). The UCD facilities management has no
plans to pursue this avenue before investigating other options such as on-site storage and operation
of electric equipment to coincide with PV production.
Hourly heating and cooling demand and electrical demand profiles were developed based on 2013-
2015 data. These profiles were then scaled annually based on load projections developed from
historical trends. The Base case projection represents decreases in annual loads due to on-going
aggressive energy efficiency which reduces the heating, cooling and electrical demand at year 2035 to
75, 70 and 80% of the 2016 values, respectively, despite a projected 26% growth in occupied space
and activities in the same time frame. The baseline assumes continued operation of the existing
steam heating system. The Alt cases includes the influence of phased steam to hot water (SHW)
conversion assumed to begin in 2018 and anticipated to result in 30% lower district heating demand
by the 2025 due to reduction in distribution losses.
Parametric Analysis
Analysis was done in Python 2.7.10 with system performance and cost analysis codes built from the
literature. The analysis space encompasses a 20-year period (2016-2035) with a 6-year phasing period
(2020-2025) and considers electric generator capacities of up to 40-MWe and thermal generator
capacities of up to 80-MWe (2015 campus peak electrical and heating loads, respectively) in 1MW
increments. Additionally, each generator type can be broken into up to 3 stages of implementation
(e.g. 5MW in 2020, 5MW in 2021, 10MW in 2025 for a total of 20 MW).
145
Component Modeling
Solar photovoltaic (PV). The output of a south-facing, 1-axis tracking PV system tilted at the latitude
with a 15% efficiency at nominal temperature (NOCT) of 25 °C was calculated. A transient energy
balance of a tilted flat plate subject to insolation, convection heat loss and power output was solved
to account for efficiency changes beyond NOCT. Direct normal and diffuse insolation data were
based on measurements at UCD. The Liu-Jordan model with Hottel adjustment was used to
calculate the total insolation on the tilted panel. Costs were compiled from various references shown
in the publication.
Solar Thermal (ST). The output of a fixed, south-facing, single-glazed flat plate collector tilted at the
latitude was calculated. The ST efficiency was defined using the average performance figures from
SRCC. Reduction in output due to refraction of light through the ST glazing was considered. Supply
and return hot water (HW) temperature was set to 70 °C and 30 °C, respectively, and used to
estimate the system pump power draw at 70% pump efficiency. Costs were compiled from various
references shown in the publication.
Integrated gasification combined cycle (IGCC) plant. Thermal and electrical energy outputs of the
IGCC were calculated. The system consists of a gasification and upgrading system to convert BM to
methane at 60% efficiency (lower heating value), followed by gas turbine (GT), heat recovery steam
generation (HRSG) and steam turbine (STM) operated in back pressure configuration to enable
process heat output. This system was chosen based on three main considerations: the IGCC is more
complex with more conversion steps relative to more straightforward biomass-to-electricity (BTE)
technologies (such as a direct biomass combustion powering a steam cycle, or a gasifier with
reciprocating engine) and therefore acts as a conservative upper estimate in capital and operational
cost for this class of technology; the combined cycle offers a higher electrical efficiency which is
favorable since grid electricity costs 3x times higher than thermal (natural gas) energy; and the
presence of the gasifier and upgrading steps in the system affords some flexibility in syngas use for
future transportation fuel production via thermochemical processes, which can be applied to address
the transportation aspect of the CNI. The GT is modeled using the air-standard assumption while
accounting for part-load efficiency penalty. An air precooler with coefficient of performance (COP)
of 3 was modeled to maintain a maximum GT inlet air temperature of 15 °C which corresponds to
27% full load electrical efficiency. The operating pressure of the steam system is set to 20 MPa to
reflect realistic pressures for a Rankine cycle of the scale considered in this work. The combined
cycle heat and power efficiency was capped at 80% of the fuel heating value. The IGCC prioritizes
electricity production over thermal due to higher electricity rates. Costs of the system and biomass
feed (18 kJ/kg LHV) were compiled from various sources. The biomass cost is a tiered structure
where every 30 k tons of biomass is subject to a linear multiple of the base biomass rate, i.e. the first
30 k tons is priced at $20/ton, the next 30 k tons is priced at $40/tons. The price tiers were assumed
based on an upper limit of $120/ ton for sourcing all of Yolo's county's technical availability. This
146
tier structure roughly simulates the increasing costs of sourcing more biomass from further
locations.
Electric heat pumps (HP). Cooling and heating output and electricity input for water-to-water HP
were calculated. With this system, cogeneration of cooling and heating thermal energy is enabled by
extracting heat from both reservoirs of the vapor-compression cycle. The overall system COP is
consequently higher than standalone operation. Two operating strategies were specified: for Alt_2,
HP follows the lower of the hourly heating and cooling loads, and for Alt_3, HP utilizes all available
excess electricity regardless of heating and cooling load and charges thermal storage tanks during
excess production. System COP (in terms of cooling capacity) as a function of water temperature
through the condenser (30 °C in, 70 °C out) and evaporator (20 °C in, 7 °C out) ranges from 1.7 to
5.5 based on manufacturer data. Costs were compiled from various references shown in the
publication.
Electric water chiller (eChiller). eChiller power draw was calculated for district cooling loads not
served by the HP in Alt_2 and Alt_3, based on averaged COP trends developed using measured data
of the campus chillers. These trends are described by 2nd order fit curves for various ranges of
outside air temperatures.
Boiler (NG boiler, BM boiler, eBoiler). Fuel inputs for two types of load-following boilers were
calculated: condensing, forced draft boiler based on NG or BM, and electric resistance. Boiler
efficiency trends as a function of load factor (ratio of heat output to capacity) were developed based
on the average of manufacturer data. Costs were compiled from various references shown in the
publication.
Thermal storage. A peak shaving thermal storage control was implemented in the present work, in
which the storage system prioritizes reduction of the hourly load peaks based on available energy.
Convection heat loss (or gain) of the storage was calculated hourly based on a 1.5:1 (length to
diameter) aspect ratio steel cylinder with R-30 (170 W/m2-K) insulation, along with historical wind
and ambient temperature data. In order to simplify analysis, heat loss or gain manifested as reduced
heating or cooling capacity as opposed to changes in thermal stratification in a physical system.
Costs for seasonal thermal storage were compiled from various references shown in the publication.
Energy efficiency (EE). EE was modeled as incurred cost as a function of annual load reduction,
develop using historical data of EE projects at UCD. The reduction of district heating load in the
2018-2020 period is treated separately and captured as incremental cost of replacing 42 km of high
pressure steam (HPS) piping to HW piping.
Capital, operation and maintenance cost. Capital (CAPEX) and operation & maintenance (OM)
costs were calculated for the replacement renewable systems and the baseline systems, as well as OM
for grid energy purchases ($0.08/kWh electricity and $0.03/kWh NG, 2016$), ongoing EE, load
shift, district system maintenance (HW OM is 21.6% the steam OM) and emissions offset. Costs
147
were obtained from vendor information, cost manuals, case studies and UCD expense data. Costs
were varied in terms of 1) system capacity in order to account for the influence of economy of scale
and 2) learning curve projection to account for cost reduction as technology matures. Power law
relationships between system capacity and cost per capacity were developed for each system. Capital
cost projection is flat for fossil fuel systems as it is a mature technology, and fossil OM increases
linearly according to the consumer price index. The Base case has added emissions offset cost of
$10/ton.
Cost analysis. Constant dollar net present cost (NPC) analysis accounts for a discount rate of 6%,
and individual escalations rates for OM based on the CPI and historical campus energy bills. These
escalation rates range from 0.3% to 7% with an average of 2%. The NPC is then normalized per
unit total energy to yield LCOE. Equipment salvage values were calculated using MARCS
depreciation according to remaining equipment life.
Results
Phasing trends
Within the space of system capacities and phasing periods investigated in this work, Alt_1 is
anticipated to have the lowest NPC for combined electricity and thermal service, while Alt_3 the
highest (Table 5). Only Alt_1 has a positive ROR of 4% and >1 benefit/cost (B/C) ratio over the
20-yr analysis period (2015-2035). No salvage values were accounted for in these figures since the
systems are assumed to continue operation past the analysis period.
148
Table 5. Phasing and Economic Summary of Alternatives
149
The following phasing trends were identified:
1) Most systems should be phased in as late as possible (as close to 2025) to take advantage of lower
projected installation costs due to learning curve. This is particularly true for Alt_3, where the HP,
ST and associated seasonal storage, and a large fraction of additional PV needed to power the HP,
are most cost effective if implemented in 2025. The thermal storage in this case is significant at
almost 19 GWh of combined annual capacity. An exception is the earlier ST implementation at 2023
for Alt_1 and Alt_2, due to rising natural gas costs and relatively low thermal storage capacity.
Another exception is 8-MWe PV in 2020 due to PV cost sitting at the lowest limit of its learning
curve at $1/W, which yields favorable LCOE across the three cases. For Alt_2, the implementation
of HP can accommodate a 9 MWe PV in 2020 before over-generation occurs. Deferring system
phasing will incur increasing carbon emissions among the three Alt cases with a difference as much
as 90.3 kton CO2e (Alt_3 vs Alt_1). This is because carbon cost is not a factor pre-2025.
2) Even though rising grid energy rates drives the above earlier phasing of ST and PV,
implementation of electric boilers would be delayed until 2025 since electricity is 3x the natural gas
rate over equivalent energy basis.
3) Implementation of thermal generators in 2020 can reduce the 20-yr NPC by avoiding the
scheduled eChiller and NG boiler replacement. A 30-MWt BM boiler and a 20-MWt heat pump
should be implemented in 2020 for Alt_1 and Alt_2, respectively. An exception to the trend is Alt_3
where the storage costs associated with the HP and ST are so large that it will be more cost effective
to still implement the eChiller and NG Boiler replacement.
4) Implementation of any particular system should be done once at the largest capacity needed in
order to take advantage of the economy of scale. A main example is HP in Alt_3 where deferring
everything till 2025 despite taking the additional expense of eChiller and NG boiler replacement
(which is avoided inAlt_2) would be more cost effective. However, there may be an instance where
the benefit of economy of scale is superseded by more rapid cost reductions due to the learning
curve. This is the case for PV where up to 8 MWe at 2020, followed by 20 MWe in 2025 would be
more cost effective than delaying the entire system till 2025.
5) Significant supply-demand mismatch (solar PV to HP cooling/ heating) requires not only large
storage capacities for Alt_3, but also additional grid PPA approximately 150% the 2017 grid PV's
annual energy generation (69 MWh) to fully meet the electrical and thermal demand of the campus.
Without this arrangement, the needed HP, onsite-PV and storage capacities will be so large as to
push NPC over $2 billion. In the case of Alt_1 and Alt_2, the biomass functions as valuable storage
for on-demand dispatch.
The 20-year projected annual cash flow for the three cases (defined as the difference: Base - Alt) is
marked by large negative spikes (payment) in the 2018-20 period and in 2025 (Figure 81). Except for
the obvious avoided cost of the baseline steam system replacement in 2019, net positive cash flow
150
(income) is not realized until 2021 for Alt_1 due to BM boiler displacing NG use. Alt_2 and Alt_3
will not yield income until 2026 when all capital expenditures to achieve a carbon neutral energy
system are complete. Since alternate systems implementation is largely stacked at the midpoint of the
analysis period (2025), all three cases (and Alt_3 in particular) could use a longer analysis period in
order to yield more favorable ROR, but is beyond the scope of this work.
Even though Alt_1 appears to be the best economic choice for UC Davis, this option relies heavily
upon biomass resources whose sourcing may be unreliable. While the chemical storage capability of
biomass cannot be understated, it requires fossil fueled transportation from source locations and
thus carries a carbon cost which needs to be offset. On the other extreme, complete electrification
by way of heat pumps as in Alt_3 necessitates large seasonal thermal storage capacities whose
footprint demand may be difficult to meet despite the large land ownership of UCD. Additionally,
the Alt_3 energy system relies heavily on a grid-based PPA to balance its supply-demand mismatch.
As the grid itself introduces more renewable systems and are relied upon more heavily as energy
storage, such a grid-based PPA may become more costly to adopt. Therefore, the most sensible
option could be one that attempts to diversify the campus load across larger numbers of generator
types as in Alt_2 which sits in the middle of projected cost among the three options and minimizes
the risk associated with either extreme. The reader is referred to the publication for a detailed
discussion of supply-demand balance for each scenario.
Figure 81. Projected annual cash flow over the analysis period
Alternatives are generally not expected to be more financially attractive than the baseline unless 1)
fossil fuel rates and carbon offset costs escalate beyond the values evaluated here, and 2) CAPEX of
alternate system drops more rapidly than the declining trend modeled here. In terms of constant
151
dollar LCOE, Alt_1, Alt_2 and Alt_3 were estimated at 8.8¢/kWh, 9.2¢/kWh and 9.6¢/kWh,
respectively, which are greater than the baseline 8.6¢/kWh. Even though the Alt_1 NPC is lower
than the Base case, the former's LCOE is slightly higher due to increased overall system efficiency
which reduces the 20-year projected energy needs. Extension of the analysis period beyond 2035
could present a different financial picture but is beyond the scope of this work. The Base case NPC
is relatively insensitive to -100 to 100% changes in either the base cost of carbon credits (0-20
$/ton) or in the cost escalation rate (0-6.4%), resulting in NPC ranging from $630.1 to only $638.6M
(Figure 82). Sensitivity to base cost of carbon credits follows a linear trend and requires a 1640%
change ($174/ton CO2) before the Base NPC matches Alt_3's NPC. The sensitivity to escalation
rate follows a 3rd order polynomial and requires a 24% carbon cost escalation rate before Alt_3's
NPC is matched.
Figure 82. Sensitivity of Baseline to changes in carbon cost and escalation rate
Relative to carbon offset cost which amounts to a negligible 0.6% of the Base case NPC, grid
electricity and NG as well as the district heating OM represent the top expenses at 17%, 19% and
18%, respectively (Figure 83). Increases in carbon credits, despite its low value in the present market,
will increase the ROR of the Alt case and push the phasing of Alt systems earlier. The largest single
year total expenditure across the three Alt cases occurs in 2018 due to CAPEX of the first phase of
the S-HW conversion in the 2018e20 period. Additional cost spikes occur in years 2020 and 2025
due to CAPEX of renewable systems. This highlights the need to reserve capital prior to the
identified phasing periods.
152
Figure 83. Base case cumulative present cost (PC)
For Alt_1 (Figure 84) and Alt_2 (Figure 85), on-going energy efficiency in the electrical and thermal
system and the IGCC form the top three CAPEX. The unit capital cost ($2016) for IGCC in Alt_1
is higher than in Alt_2 since the system is larger in Alt_2. For Alt_3 (Figure 85), HP takes the place
of the IGCC as the third largest CAPEX which is larger than Alt_2's HP, due to the former's
advantage both in economy of scale and learning curve. Any decreases to these unit costs will
improve the economics of earlier phasing. For Alt_1, the IGCC, PV, ST, BM boiler, thermal storage
and Sta. eBoiler (in order of descending CAPEX) amount to $51M. For Alt_2, the IGCC, HP, PV,
district eBoiler, ST, thermal storage and Sta. eBoiler (in descending order) amount to $79M. In
similar order for Alt_3, the HP, PV, thermal storage, ST, district eBoiler and Sta. eBoiler amount to
$73M.
153
Figure 84. Alt_1 cumulative present cost (PC)
154
Conclusions
Transformation of UCD to a carbon neutral energy system is feasible using currently available PV,
ST and BM technologies, but alternatives are not expected to be financially attractive relative to
baseline unless fossil fuel rates and costs of carbon credits escalate beyond the low values evaluated
in the present work, and CAPEX of alternate system are lower than evaluated here. High diversity of
the campus loads and supply-demand mismatch manifest as low generator CFs and highlights the
value of generators that can modulate especially under the power export restrictions. Intermittent
sources like solar require storage to fulfill this duty, which increases the cost of the system. Meeting
a diversified load solely through a large HP capacity with large storage is economically infeasible
relative to introducing backup generators as is the case for Alt_3. Complete electrification will also
require an additional carbon-neutral grid PPA for supply-demand balance. Heavy reliance on
biomass or solar systems such as the case of Alt_1 and Alt_3 carries a larger risk in feedstock
sourcing reliability and in power imbalance, and thus, diversification of the campus energy system
across a wider range of generator types, as in the case of Alt_2, is likely best path forward.
Lessons Learned
Scope in consultation with customer. Engage with the customer early and frequently to identify
main questions and needed capabilities, and to arrive at a scope of work in a timely manner.
155
Leverage existing tools and databases. Whenever possible, utilize available tools to address analysis
and data needs. In this study, campus energy data and statistical analysis tools already exist.
Be flexible enough to implement a custom solution. Take the initiative to develop a customized
analysis tool based on the analysis need and assessed gaps in currently available tools. The
customized analysis can be more sophisticated or simpler. For example, a customized, but simplified
tool was developed using SRCC’s methodology in lieu of a detailed thermal simulation of ST using a
commercial software. On the other hand, no commercial software at the time captured the
temperature effect on PV operating efficiency. Thus, a customized tool was developed to
incorporate this phenomenon in the hourly PV output simulation.
A.7. Technical and economic feasibility of unitary, horizontal ground-loop

geothermal heat pumps for space conditioning in selected California
climate zones
TEA Staff: Steven Wiryadinata, Mark Modera, Bryan Jenkins, Kurt Kornbluth
Intermittency.
This was an NSF funded project supporting the TEA staff of the University of California, Davis.
The TEA team worked in consultation with the technology developer and implementation
contractor of the analyzed system.

The goal of this work was to understand the techno-economic feasibility of large-scale
implementation of a prototype ground-source terminal heat pump system (GTHP) in selected target
markets, by pursuing the following objectives:
1. Analyze the performance of technology in relevant applications.
2. Compare and contrast the technology of interest against a reference baseline.
3. Identify major drivers affecting the techno-economic performance metrics.
156
Deliverables & Awards
Journal article: Wiryadinata, S, Modera, M, Jenkins, B & Kornbluth, K. (2016). Technical and
Economic Feasibility of Unitary, Horizontal Ground-Loop Geothermal Heat Pumps for Space
Conditioning in Selected California Climate Zones. Energy and Buildings. 119.
10.1016/j.enbuild.2016.03.032.
Internal communications: Presentation to technology developer and implementation contractor.
Background
HVAC energy use in the lodging (hotels and motels) building sector in California accounts for
approximately 38% of total building electrical energy end-use and 17% of total building natural gas
end-use. HVAC energy use in the multifamily (apartments) building sector accounts for
approximately 27% of total building electricity and natural gas use. Measures targeted at reducing
HVAC energy use can therefore reduce the total energy use of a building.
Relative to air-source heat pump systems whose performance degrades during extreme ambient
temperatures, geothermal heat pumps (GHP) maintain operating performance because they
exchange heat with the ground through a ground-loop heat exchanger (GHX). The ground (i.e. soil)
experiences smaller temperature variations over the year, particularly at increasing depth. The lower
operating temperature swings manifest as lower stress which confers benefit in lower maintenance
requirements. Despite the energy, maintenance and long-term operational benefits, GHPs incur high
capital costs, which have limited their adoption in the United States. As of 2011, GHPs accounted
for only 2 % of the value of all shipments of HVAC equipment while air-source heat pumps
accounted for over 10%.
In the residential and commercial sectors, the main market barriers for the technology are attributed
to the GHX design complexities and limited experience among installers. Complexities in design
arise from variability in ground conditions such as non-uniform conductivity which directly
influences heat exchange performance, from the presence of existing below grade infrastructure
such as utility lines (typically buried at 1 m depth in California), and from regulatory limitations on
boring primarily due to concerns on potential contamination of ground water. These factors
manifests in longer installation times and increased costs.
This study was motivated by the drive to develop a lower cost GHP system. As a subgroup of
GHPs, ground source terminal heat pumps (GTHPs) are self-contained terminal units coupled to a
U-tube, horizontally drilled (HD) ground loop through an exterior wall to deliver space
cooling/heating service without the use of an interior ducting system. HD GHX have the lowest
installed cost compared to other closed loop GHX, such as vertical bores that require large and
expensive specialized drilling equipment that are often hindered by site obstructions and add to
157
logistic costs, or horizontal trenches that require extensive and time-consuming excavation. The HD
GHX discussed here utilizes a compact directional driller, which is not hindered by the above issues.
Scope
The study is scoped to compare the performance of the HD GTHPs against a baseline unitary air-
source packaged terminal heat pump (PTHP) systems which are in use prevalently in the low-rise
lodging and multifamily facilities. The space of analysis was limited to the two building types in two
distinct climate zones, coastal and inland to assess the impact of different weather profiles on system
and economic performance. The coastal climate, represented by the city of Oakland, California, is
characterized by a mild outside air temperature (Toa) profile, while the latter, represented by the city
of Fresno, California, is characterized by more extreme annual temperature swings. Specific
objectives are to:
Develop load profiles for representative buildings in the selected climate zones
Design and analyze a GTHP system for each building
Perform an economic assessment of a 20-year analysis period to quantify a levelized cost of service
(LCOS)
Quantify sensitivities of LCOS to input parameters
TEA Approach
Building load profile and schematic
The lodging and multifamily building models in the study leveraged a previous PG&E (Pacific Gas
& Electric) study as representative of consumers in their service area. The physical schematic of the
modeled buildings directly influences the arrangement of GHX distribution piping network and
pumping energy use. The dimensions for the modeled lodging facility were estimated based on total
floor area of a typical low-density lodging facility in California (Figure 87). This is in contrast to a
high-density lodging facility, which would typically be represented by single high-rise building. Hb,
Wb and Lb represent the facility height, width and length, respectively. Wb1 and Wb2 represent
individual building width and the separation distance between buildings. The dimensions of the
modeled multifamily building were extracted directly from DEER (Figure 88). The lengths of the
distribution and header piping which influence the pump energy use are based on the facility
dimensions.
158
Figure 87. Schematic of the modeled lodging facility consisting of four 4-story buildings
(Fine dotted lines represent piping running along the outside of the building. Dashed lines represent
the header connected to multiple GHX bores. Only a single bore is shown.)
Figure 88. Schematic of the modeled 2-story multifamily apartment
159
(Fine dotted lines represent distribution piping. Dashed lines represent the header. The GHX bores
are connected to the two headers on the sides of the building.)
The major components of a GHP system include the heat pump unit, the air delivery system and the
hydronic system, which comprises of the GHX and distribution piping. The arrangement of these
components within a building can vary widely. In the case of the GTHP, the air delivery system is
self-contained within the heat pump unit, which is typically installed through an exterior wall for
connection with the GHX. The GHX consists of multiple horizontally bored U-tube high-density
polyethylene pipe surrounded by grout. The bore enters the ground at 30◦and levels out at the
typical design burial depth of 4.6 m after traveling 7.9 m horizontally. Since a typical bore length is
76 m long, the bore is exposed primarily to the ground temperature at the burial depth.
At depths shallower than 16 m, the temperature of the ground (Tg) can vary from one season to the
next depending on the soil type and climate. At a depth of 4.6 m in the California inland (CZ3) and
coastal climates (CZ13), Tg can vary by 8◦C and 3◦C, respectively. During cooling period, Tg is
higher than it would be during heating period. This temperature swing reduces the performance
(efficiency and capacity) of the GTHP as it would have to operate at a higher condensing
temperature during cooling mode and lower evaporating temperature during heating mode relative
to the fairly constant undisturbed Tg of deeper soil.
The design of the GTHP system depends on the space heating and cooling load profiles of each of
these buildings. Developed using California Commercial End-Use Survey (CEUS) lodging data and
the Database for Energy Efficient Resources (DEER) prototype multifamily data, the profiles are
driven not only by the occupancy type but also the ambient weather conditions (Table 6). The
coastal climate is characterized by a mild outside air temperature (Toa) with an annual mean,
maximum and minimum of 14◦C, 33◦C and 2◦C, respectively. By contrast, the inland climate is more
extreme with an annual mean, maximum and minimum Toa of 18◦C, 42◦C and −2◦C, respectively.
Peak hourly cooling and heating loads are both higher in the inland climate. The reader is referred to
the publication for graphical hourly load profiles of the two buildings.
160
Table 6. Summary of loads and facility area
System performance data

One factor influencing the performance of any GHP is the temperature of water entering the heat
pump (Tew). Tew is influenced by the heat transfer performance of the GHX, which is based on the
temperature differences between the bore and the ground and the thermal conductivities of the
borehole and soil. A relationship between COPs in heating and cooling and Tew were developed
empirically for the GTHP based on 15-min data sampling of two units operated at a test site at the
University of California, Davis. The COP is defined as the ratio of the cooling or heating capacity
over the sum of the compressor power, control systems power draw, and the pumping power
needed to overcome the pressure drop only in the water-refrigerant heat exchanger. Pump power
calculations for the distribution lines, headers and GHX are evaluated separately based on the piping
scheme of the modeled buildings. Manufacturer data for two brands of comparable commercial
water-source heat pumps were compared to the developed GTHP COP curves as a way for sanity
check. Similarly, COP trends developed for the baseline system was vetted against two brands of
comparable commercial air-source heat pumps. For the same condensing/evaporating temperatures,
the COP of the test unit was practically the same as the two commercial units.
Sizing and simulation of baseline and alternate systems
The design of the horizontal GHX in this study is based on the analytical approach presented by
Bernier with slight modification according to IGSHPA. The reader is referred to the publication for
discussion on the modified approach and input assumptions. The design COP recommended by
ASHRAE was first applied across all hours to estimate the mean GHX bore temperature (Tm).
Changes in Tm affect changes in Tew. The COP value based on the revised Tew was then used to
recalculate the hourly heat transfer. This process was iterated until the COP value converged. The
resulting annual Tew approached Tg based on the magnitude of load and somewhat traced the
sinusoidal profile of Tg. Estimations of the hourly Tm are based on the modeling assumption that
161
only a fraction of the design bore length is active, based on the required ground heat transfer to
satisfy the space load at any particular hour. Thus, pumping energy use is minimized.
Economic evaluation
The levelized cost of service, LCOS, ($/Wht) represents the dollar value to supply each unit of
thermal load to produce a net present value of zero, and was calculated using:
(14)
Where:
(15)
And where P is the net present value of all costs of the GTHP and PTHP systems over their
lifetime, expressed in dollars; Co is the cost at the 1st year of analysis; n and s are the annual interest
of 3% and annual price escalation rates (2% for capital cost, and between 0.4% and 5.7% for the
various components of energy based on the utility historical 2003–2014 trends), respectively; M is 20
years; the y subscript denotes the various components of total cost; and j is the analysis year (1–20).
Capital cost, including the PTHP, GTHP, HDPE piping and insulation, pumps, variables peed drive
controller and GHX drilling and grouting, was evaluated at j = 1 for the PTHP and GTHP, and at
the replacement year j = 16 for the PTHP. Since the baseline system was retired early at the 20th
year of analysis (5th year of service for the replacement unit installed at year 16), salvage value was
taken as the book value calculation from the U.S Federal General Depreciation System (MACRS)
depreciation. Only the GTHP and pumps are considered at replacement of the GTHP system.
Published cost data and contractor pricing were used to establish capital costs. The annual
maintenance costs for an air-source heat pump and ground-source heat pumps were reported to
decrease as a function of time. To project these costs to the years of analysis, power curve fits was
assumed. Total facility energy cost was calculated from the time-of-use utility rate schedules for low-
density residential and medium density commercial.
Factors that affected the LCOS of the GTHP include overall system efficiency, capital cost,
maintenance cost, equipment life, energy cost rates and interest rate. For each facility in each climate
zone, sensitivity of the LCOS was investigated by varying these factors in 10% increments.
162
Results
System configuration
Due to differences in occupancy, thermal load and building dimensions, the lodging facility required
significantly more unitary heat pump units, longer distribution piping and GHX, and larger water
pumps than the multifamily facility. In order to meet pumping power benchmarks, more numerous
parallel GHX bores were specified for the inland buildings relative to their coastal counterparts
(Table 7). The GTHP COP varies from 3.7 to 4.5 depending on Tew. The pumping power
requirement is greatest through the GHX and ranges from 50% to 70% of the pump capacity, and
between 3% to 10% of the GTHP system energy consumption. The contribution of the GHX to
the total system cost increased significantly as cooling load increased. The larger of the space loads
in the inland and coastal climates dictated the total number of heat pumps for the facility due to the
assumption of identical interior building configuration. The GHX length for the facilities in the
coastal climate is approximately 40% the length required in the inland climate because the coastal
space cooling loads are smaller and the mean ground temperatures are lower.
Table 7. Summary of System Configuration
Energy use and emissions reductions

Over the analysis period, energy use reductions (EUS) of operating the GTHP over the baseline
PTHP ranges from 5 to 15 kWh/Wht except for the coastal multifamily facility (Table 8). In the
latter case, the pump energy use cancels any savings realized from utilizing a ground loop, which is
minimal due to the mild coastal weather. Conversely, large facilities(lodging) situated in more
extreme climates (inland) benefit from significantly lower demand at peak loads and lower overall
energy use from the use of the GTHP. Larger differences between Tg and Toa during these peak
hours resulted in higher demand reductions. Inland (both lodging and multifamily) facilities
consistently see energy savings from GTHP. Thus, those inland facilities that do not have the
infrastructure to support internally centralized systems (such as a central heat pump) can be targeted
for implementation of this technology.
163
Table 8. Summary of Energy Use and Emissions Reductions
For lodging, the reductions in LCOS were estimated at 7.7% and 8.7% in the coastal and inland
climates, respectively (Figure 89). For multifamily, the reductions were 5.5% and 3.8% in the coastal
(CZ3) and inland climates (CZ13), respectively (Figure 90).
Figure 89. LCOS for modeled lodging facility
164
Figure 90. LCOS for modeled multifamily facility
In California, the estimated floor areas of the low-rise lodging and multifamily market were 880,000
m2 and 12,900,000 m2 respectively in both climate zones. Based on this and the estimated cost
savings, the application of the GTHP has the potential to achieve annual cost savings of
approximately $100 million annually. The energy cost reduction between the baseline and alternate
cases appears to be greater for larger facilities situated in more extreme climates, even though these
facilities required large GHX bore fields, which increased the implementation cost. The majority of
levelized cost savings are due to projected lower maintenance requirements and longer operating life
of the GTHP. Neither factor was validated with this particular study but is well-documented for
GHP systems in general. A long-term study on maintenance cost and equipment life for the GTHP
would give more insights into the total savings potential.
Application of the GTHP in both climates zones can result in cost reductions, but for different
reasons. In the coastal climate where loads are smaller than in the inland climate, energy cost savings
are negligible or negative, but this reduced savings is offset by capital cost reductions. In the inland
climate, GTHPs can make more effective use of the heat exchange with the ground for improved
system efficiency compared to the GTHPs in coastal climate, but the larger loads result in larger
system size, which offsets the capital cost savings due to longevity.
Furthermore, the economic viability of GTHP is dependent on the electricity rate structure.
Compared to the results under TOU rates, the LCOS savings under non-TOU rates are lower in the
inland climate (5.4%for lodging and 3.2% for multifamily) but higher in the coastal climate (8.1% for
lodging and 5.9% for multifamily). Thus, facilities subject to time-of-use rates will further benefit
from operating efficiency gains of the GTHP during cooling periods because lower electricity
demand occurs during the periods of high utility rates. From the utility point of view, the GTHP can
help reduce demand during peak period and mitigate the need to supply more power by building
new power plants.
Sensitivity
165
Generally, LCOS is most sensitive to changes in implementation cost. Reduction of GTHP
installation costs is critical, either by added incentives (e.g. tax or carbon credits), enhanced training
of installers, improved installation efficiency as increasing numbers of systems are deployed, or by
improved technology. The sensitivity of the LCOS depends on the size of the GHX. The inland
LCOS is more sensitive (Figure 91) to increases in capital cost than to equipment life, due to the
need for a larger GHX. The reverse is true for the coastal climate (Figure 92). The fact that this cost
occurs only once through the 20-year analysis period shows the significance of the GHX component
to the total system cost, especially in larger systems. The reader is referred to the publication for
sensitivity curves for the multifamily building.
Figure 91. Inland lodging sensitivity
Figure 92. Coastal lodging sensitivity
As the ratio of HVAC energy use to total facility energy use increases (as in the case of inland
lodging), system efficiency pro-duces increasingly significant changes in LCOS. The overall system
166
efficiency is comprised of multiple factors including but not limited to: GTHP COP, pumping
system efficiency (affected by piping design and pump mechanical efficiency), and GHX thermal
performance (affected by ground conditions, loop size and borehole performance). While an
identical ground type was assumed in this work, ground properties can vary significantly between
climates and geographical locations. In general, soils with lower conductivities necessitate the
installation of longer GHX. Moisture content or the presence of ground water typically enhances
GHX performance beyond heat diffusion in dry soil. As equipment life is decreased, LCOS
increases dramatically since replacement systems were installed with either the same or increased
efficiency, but at the expense of higher total capital costs that do not justify the energy savings.
These two competing factors produced a relatively constant LCOS as equipment life increases, but a
rapid increase in LCOS as equipment life decreases.
Conclusions
The total levelized cost of service appears to be comparable for the GTHP and PTHP, with some
advantage for the GTHP due largely to maintenance cost savings and, depending on the climate and
building type, lifetime capital cost savings and energy cost savings. Smaller facilities situated in mild
climates benefit from levelized capital cost reductions, but minimal energy savings and energy cost
savings. Larger facilities situated in more extreme climates benefit from higher energy efficiency but
lower levelized capital cost savings due to longer GHX requirements to meet peak loads. In all cases,
the energy LCOS savings were estimated to be lower than the maintenance LCOS savings. From the
magnitude of estimated demand and cost reduction, the GTHP technology can significantly benefit
the utility during peak cooling period.
Sensitivity analyses highlight the importance of reducing system cost and ensuring long equipment
service live. As HVAC energy use increases relative to the total facility energy use, system efficiency
becomes an increasingly important factor. For the multifamily facilities, as equipment life increases,
LCOS increases slightly from the nominal case (0% change) due to the tiered utility rate structure
imposing a higher cost to increasing electrical energy use.
Lessons Learned
Scope in consultation with customer. Engage with the project sponsor early and frequently to
identify main questions and needed capabilities, and to arrive at a scope of work that is manageable
in a timely manner.
Leverage expertise of project partners. While self-education to overcome learning curve and develop
new skills is desired, existing capabilities should be employed to facilitate progress. This project
benefited from the decades-long experience of the technology developer and system installer in the
design and analysis of ground heat exchangers and heat pumps.
167
A.8. DYMATICA
The Dynamic Multi-Scale Assessment Tool for Integrated Cognitive-behavioral Actions
(DYMATICA) is a tool to help decision makers better understand and anticipate the effects of
societal changes due to select geopolitical, economic and other stimuli. DYMATICA developed
across multiple years to address a suite of scenarios that can generate assessments that analyze
regional conflict and the military, political and economic vulnerabilities of countries over time
(Bernard, 2016, 2020).
Figure 93 illustrates the graphical user interphase of DYMATICA in a user-influence scenario to test
‘what if’ future states.
168
Figure 93. Illustrative DYMATICA assessment tailored to fit the need of the customer.
169
A.9. Economic Impacts of Nuclear Power Plants Accidents Analysis
This section provides a brief summary of Sandia’s economic impacts estimation work and a tool,
called RDEIM, created for the Nuclear Regulatory Commission (NRC). A more detailed account
can be found in Bixler et al. (2020).
The TEA-related work at Sandia includes Probabilistic Risk Assessment and modeling of nuclear
accidents for the NRC.
The MELCOR Accident Consequence Code System (MACCS) code is the NRC code used to
perform probabilistic health and economic consequence assessments for atmospheric releases of
radionuclides. MACCS is used by U.S. nuclear power plant license renewal applicants to support the
plant specific evaluation of severe accident mitigation alternatives (SAMA) analyses as part of an
applicant’s environmental report for license renewal. MACCS is also used in severe accident
mitigation design alternatives (SAMDA) and severe accident consequence analyses for
environmental impact statements (EISs) for both existing and new reactor license applications. The
NRC uses MACCS in its cost-benefit assessments supporting regulatory analyses that evaluate
potential new regulatory requirements for nuclear power plants. NRC regulatory analysis guidelines
recommend the use of MACCS to estimate the averted “offsite property damage” cost and the
averted offsite dose cost elements, which are both benefits in the cost/benefit analysis (NRC, 1997;
NRC, 2004).
The original cost-based MACCS economic model was published by Jow, et al. (1990) and is referred
to in this document as the cost-based model. This cost-based model is a generalization of the one in
CRAC2 (Ritchie, et al., 1983). Since the implementation of the cost-based economic model in
MACCS, government-sponsored economic data related to GDP have become readily available,
along with tools to gather and process these data. With the availability of government-produced,
standardized data, an alternative MACCS economic model can be employed to implement a gross-
domestic-product-based (GDP-based) estimation of offsite economic costs of a nuclear power plant
incident. To implement the GDP-based economic model, a variant of the Regional Economic
Accounting analysis tool (REAcct) created at Sandia National Laboratories, has been integrated into
MACCS. To signify that this model is significantly different than REAcct, it has been named
RDEIM, which stands for the Regional Disruption Economic Impact Model. In this document, the
terms RDEIM model and GDP-based model are used interchangeably.
The GDP-based economic model achieves the following objectives:

• Consistency of identifying off-site costs for nuclear reactor accidents with state-of-practice
methods used to estimate impacts from other disruptions that have the potential for large-
scale economic impacts
170
• Developing estimates of the offsite cost impacts from business disruption using current
state-of-practice input-output (I-O) economics
• Estimating the impact on the regional communities, industries, and infrastructure
• Estimating the impacts of multi-year disruptions when the region cannot be remediated
quickly
• Estimating indirect effects on the national economy outside the directly affected region
• Estimating induced effects to the regional and national economies resulting from lost
income to workers
This document provides an overview of both the cost-based, original MACCS economic model and
the newer, GDP-based economic loss model. Following a description of each model, the
implementation of the GDP-based model into the MACCS framework is discussed. Verification
exercises and benchmarking of the GDP-based model are then covered in significant detail for a
variety of consequence scenarios.
A.9.1. Cost-Based Model Overview

The original MACCS economic model calculates offsite consequences of nuclear power plant
accidents that release radioactive materials into the atmosphere using a cost-based approach. The
MACCS cost-estimation model is described in the MACCS Model Description document written by
Jow, et al. (1990). The underlying economic methodology is described in an earlier document by
Burke, et al. (1984). Specifically, the costs calculated in the original MACCS economic model
include:
• Temporary evacuation and relocation costs, including food, lodging, and lost income for the
displaced population during the emergency and intermediate phases of the accident
• One-time relocation costs during the long-term phase
• Cost of decontaminating land and property during the long-term phase
• Lost return on investments from properties that are temporarily interdicted
• Depreciation of temporarily interdicted property that cannot be maintained
• The value of lost crops in the first year of the accident
• The value of farmland and of private, commercial, public, and supporting infrastructure that is
condemned during the long-term phase
MACCS costs are calculated based on protective measures and emergency response actions
undertaken during and after the accident. The unit costs associated with protective actions and
emergency responses are defined in the user input. The emergency responses include evacuation,
sheltering, and relocation. The long-term protective actions include decontamination, temporary land
interdiction and associated relocation of population, crop disposal, control/prohibition of food
production, and condemnation of property.
171
A.9.2. GDP Impact Model (RDEIM) Overview
RDEIM calculates the indirect losses using “net total requirements” (NTR) multipliers based on the
Regional I-O Modeling System (RIMS II) data. It uses employment by county, value added 35 and
gross output by industry, total requirements tables, final demand value-added multipliers (RIMS II
model) provided by the U.S. Bureau of Economic Analysis (BEA), as well as other data provided by
the Bureau of Labor Statistics and other sources.
The total economic impact (loss) caused by a disruption is typically grouped into three categories
(BEA, 2012):
• Direct 36 GDP impacts occur due to a loss of final demand, which occurs in the context of an
accident because production is stopped for a period in the affected area, which represents a
loss of the value added by the affected firms.
• Indirect GDP impacts occur because the loss of final demand also affects the supplier firms
as their input to the curtailed production is no longer required. In the context of an accident,
supplier firms are outside the affected area. GDP impacts represent value-added losses to
indirectly affected firms.
• Induced GDP impact relates to the spending of workers whose earnings are affected by the
disruption. 37 Induced GDP losses correspond to both workers inside and outside the directly
affected area.
The GDP loss calculated by RDEIM estimates the losses accrued over time at the regional scale of
the impacted area and at the national scale. It also allows the recovery schedules for regional and
national scales to be varied independently of each other with the proviso that regional recovery is
never faster than national recovery.
The RDEIM model includes indirect impacts to other sectors of the economy that are additional to
but causally related to the direct impacts. However, the indirect impacts in RDEIM are restricted to
the geographic areas not directly affected by the disruption. Induced impacts account for the effect
of lost income on purchases (sales), which in turn affect the overall economy.
35 Value added is defined as the sum of labor compensation, capital income, and net indirect taxes (producer taxes,
import tariffs minus subsidies).
36 The notion of direct (and by extension indirect and induced) impacts in this application does not map directly to the
existing literature, due to the nature of disruption, where all industries are shut down in a region. Therefore, the impacts
in the directly affected area that would have been indirect if only one industry were shut down, are treated as direct given
that all industries are shut down. This is the reason for using the net value-added multipliers.

172
While direct economic impacts occur to known regions of the country, the same is not true for
indirect impacts. Some, but not all, of the intermediate industries that sell to or buy from the
industries in the directly impacted region are also located in the directly impacted region but the
remainder, possibly the majority, are located outside of the directly impacted region; likewise, not all
of the workers that potentially lose income from the directly or indirectly impacted industries spend
all their income regionally or even nationally. Induced impacts are included as part of the values
reported as total impacts at the national level.
The spatial extent of disruption is represented in the model by two regions: “Intraregional” – the
area directly affected by contamination to the extent that land is interdicted38, and “Extra regional” –
the area not affected by contamination, representing the rest of the nation (excluding Alaska and
Hawaii). The intraregional/affected region is the region reporting the direct losses. All the
intraregional losses are treated as direct even though some of the losses are to suppliers to other
economic sectors. Intraregional and extra regional losses sum to the national value, where national
refers to the 48 contiguous states.
The indirect and induced losses are assigned to the extraregional losses (because all the intraregional
losses are considered direct). In this treatment of direct, indirect, and induced losses, the model
departs from the conventional RIMS II treatment of a single industry disruption and instead
represents a multi-industry disruption over a region.
A.9.3. NRC Economic Impacts Estimation and Cost-Benefit Analysis
This section provides an example analysis using the Sandia’s economic impacts estimation tool
created for the NRC. A more detailed account can be found in Bixler et al. (2020).
This section describes the methods employed in RDEIM to estimate the total GDP impact. The
total cost impact includes additional elements that are estimated in MACCS. These include the cost
of evacuation and relocation of the public and the cost of decontamination. The integrated model 39
results represent the overall cost impact and are provided as output from the integrated model.
RDEIM performs the following steps to estimate economic impact:
• An analysis area is defined. SecPop is an auxiliary code that is used to develop the site-
specific land-use, population, and economic data into a site file for MACCS. SecPop version
4.0 and newer creates a file containing the counties or fractions of counties contained in each
MACCS grid element. Fractions of counties are estimated both by area fraction and
39 The original economic impact estimation model was envisioned as REAcct working as a preprocessor to MACCS.
Because of the changes to the economic methodology, the current model largely uses the REAcct data, a modified
version of REAcct called RDEIM, and algorithms internal to MACCS for calculating the impacts.
173
population fraction. RDEIM uses this information to estimate GDP losses for each industry
within each grid element.
• RDEIM computes total GDP losses (direct, indirect, and induced) for each MACCS grid
element. This information is stored in a file that is used by MACCS.
o The number of employees for each industry within a MACCS grid element is
calculated and this information is used to estimate direct GDP losses. An estimate of
the impacts to other industries that are indirectly affected by the disruption is
performed using I-O multipliers.
o All economic activities within a MACCS grid element are disrupted for the same
duration of time 40, except for farmland, which may have a different recovery
schedule.
• For a specific source term and weather trial, MACCS determines the affected area and the
duration of the disruption for each grid element. MACCS aggregates the GDP losses over
the region and over the duration of disruption.
• RDEIM estimates the base-year value of future year GDP losses by accounting for an annual
GDP growth rate and an annual social discount rate. All dollars are reported in base-year
(currently 2011) dollars for an accident that is assumed to occur in the accident year specified
by the user. The user can adjust the value of the dollar to another year as a post processing
step, if desired.
• MACCS sums the GDP losses. 41 A suggestion for how this information can be used in a
cost-benefit analysis is provided in Section 5.
• MACCS repeats the process for a set of weather trials and provides statistical results to
characterize the variability from uncertain weather. The footprint of the affected area, the
degree of contamination, and the duration of economic losses can be different for each
weather trial; thus, the direct, indirect, and induced economic losses are generally different
for each weather trial.
For some scenarios, the extent of contamination may cause the land to be interdicted for a short
period of time (e.g., a few years) or condemned (i.e., not recoverable within the Maximum Duration
of Economic Impact) in the model. The user specifies the number of years of direct GDP loss
(Maximum Duration of Economic Impact) that are evaluated for an area that is condemned while
MACCS estimates the required interdiction period based on the extent of contamination. In most
cases, the interdiction period estimated by MACCS is less than the default value for Maximum
Duration of Economic Impact (10 years). When this is true, the GDP of the affected area is only
considered a loss for the interdiction period estimated by MACCS, not the full 10 years. For
40 The current framework is sufficiently flexible to allow differential recovery times by industry. However, it is not done
in the current version of the model.
41 The new model is fine-grained enough to represent the losses at the regional and national levels as they are projected
to occur over time. Such data can be used to analyze possible accident impacts in detail or to investigate tradeoffs
between different restoration policies.
174
agricultural land use, the minimum interdiction period is assumed to be one year because of the
seasonal nature of this industry.
A.9.4. Simple Example
This section describes a simple example to illustrate the model.
For simplicity, the affected area is composed of three grid elements, 𝑅𝑅 = {𝐴𝐴, 𝐵𝐵, 𝐶𝐶}, and four
industries, 𝐼𝐼 = {𝑈𝑈𝑈𝑈𝑈𝑈𝑈𝑈𝑈𝑈𝑈𝑈𝑈𝑈𝑈𝑈𝑈𝑈, 𝑀𝑀𝑀𝑀𝑀𝑀ℎ𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑖, 𝐴𝐴𝐴𝐴𝐴𝐴𝐴𝐴𝐴𝐴𝐴𝐴𝐴𝐴𝐴𝐴𝐴𝐴𝐴𝐴, 𝐹𝐹𝐹𝐹𝐹𝐹𝐹𝐹𝐹𝐹𝐹𝐹𝐹𝐹𝐹𝐹𝐹𝐹𝐹𝐹𝐹𝐹}. In this example, grid
element A is a partial county, and the grid elements B and C are complete individual counties. The
counties are also called A, B, and C, corresponding to the grid element that contains the county. The
region in the following discussion represents a 50-mile radius surrounding the reactor site. The
region is made up of the disrupted counties, A, B, C, and several other counties that are not
disrupted.
Other scenario parameters are as follows 42
Maximum duration of regional disruption, 𝑇𝑇𝑅𝑅 = 10 years.

The time needed for national recovery, 𝑇𝑇𝑁𝑁 = 4 years.
GDP growth rate, 𝑔𝑔 = 2.4%.
Social discount rate, 𝜌𝜌 = 3.0%.
Base year = database year = 2011.
GDP of the region is $3 billion, and national GDP is assumed to be $100 billion 43 in 2011. The
employment by industry and county is described in Table 9.
Table 9. Employment by industry in affected counties

County Employment
Industry
A B C
Utilities 100 45 55
Machinery 995 4000 30
Adm. Serv. 10 15 20
Food Serv. 50 300 5
The fraction of each county affected is represented in Table 10.
42 These are the default values of parameters, as provided in the model.

43 These numbers are made for illustration purposes and do not intend to represent any real geographic data.
175
Table 10. Fraction of each county affected
County Fraction in Grid Element
A B C
By population 0.5 1 1
By area 0.7 1 1
Given the weighting indicated for each industry, the lost employment for each grid element (each
containing all or part of the county of the same name) is estimated as shown in Table 11.
Table 11. Lost employment by grid element and industry

Lost Employment in Grid Element
Industry
A B C
Utilities 70 45 55
Machinery 498 4000 30
Adm. Serv. 5 15 20
Food Service 25 300 5
The grid element recovery schedule is shown in the Table 12.
Table 12. Grid element recovery schedule

Grid Element
Recovery Time A B C
(yr) 3.5 6 11
Table 13 describes the value added per worker for each industry.
Table 13. Value added per worker for each industry

Industry
Utilities Machinery Adm. Serv. Food Service
Value added per
worker/year 150,000 170,000 120,000 100,000
($/yr)
The value-added Type I and Type II multipliers are presented in Table 14.
176
Table 14. Value added multipliers of Type I and Type II
Value Added Multipliers
Industry Type I Type II Type I Type II
National National Regional Regional
Utilities 1.40 1.81 1.20 1.55
Machinery 1.87 2.41 1.35 1.74
Adm. Serv. 1.48 1.91 1.40 1.81
Food Serv. 1.87 2.41 1.60 2.06
These are adjusted for double counting by taking the difference between the national and regional
values and adding one. The resulting multipliers are shown in Table 15.
Table 15. Regional Type I and Type II multipliers adjusted for double counting
Adjusted Regional
Industry Multipliers
Type I Type II
Utilities 1.20 1.26
Machinery 1.52 1.67
Adm. Serv. 1.08 1.10
Food Serv. 1.27 1.35
Applying Equations (18), (20), and (22), various losses are calculated. Table 16 represents the
summary of direct losses.
Table 16. Summary of direct losses for the region. All values are discounted to 2011 (year 1 in
the table) using the social discount rate.
Annual Baseline Baseline Regional GDP
Cum. Dir. Direct Cum. Annual Percent Loss after
Loss GDP Loss GDP GDP of Regional Disruption
Year ($M) ($M) ($M) ($M) GDP (%) ($M)
1 830 830 2,991 2,991 27.8 2,161
2 1,656 826 5,964 2,973 27.8 2,148
3 2,477 821 8,919 2,955 27.8 2,135
4 3,244 768 11,857 2,938 26.1 2,170
5 3,959 715 14,777 2,920 24.5 2,205
6 4,670 711 17,680 2,903 24.5 2,192
7 4,686 16 20,565 2,885 0.5 2,870
8 4,702 16 23,433 2,868 0.5 2,852
9 4,717 15 26,284 2,851 0.5 2,835
177
10 4,732 15 29,118 2,834 0.5 2,818
11 4,732 - 31,935 2,817 0.0 2,817
The annual direct losses ($M) are represented in Figure 94 and as a percent of the regional GDP in
Figure 95.
Figure 94. Annual direct loss in GDP ($M) to regional economy. All values are discounted to
2011 (year 1) using the social discount rate.
178
Figure 95. Direct annual losses as percent of regional GDP.
The projected regional GDP without the disruption and estimated regional GDP accounting for the
disruption are shown in Figure 96.
179
Figure 96. Regional annual GDP with and without the disruption. All values are discounted
to 2011 (year 1) using the social discount rate.
The Table 17 represents the loss estimates on the national level.
180
Table 17. GDP losses at the national level.
All values are discounted to 2011 (year 1) using the social discount rate.
Cum. Annual
Cum. Annual Dir. + Dir. + Annual Annual
Total Total Indir. Indir. Indir. Induc. Baseline Baseline Annual Total
GDP GDP GDP GDP GDP GDP Cum. Annual GDP GDP after
Loss Loss Loss Loss Loss Loss GDP GDP Loss Disruption
Year ($M) ($M) ($M) ($M) ($M) ($M) ($M) ($M) (%) ($M)
1 1,193 1,193 1,089 1,089 258 105 99,701 99,701 1.20 98,507
2 2,041 847 1,861 773 (53) 74 198,805 99,104 0.86 98,257
3 2,546 505 2,322 461 (360) 44 297,316 98,511 0.51 98,006
4 2,672 126 2,437 115 (653) 11 395,238 97,922 0.13 97,796
5 2,672 - 2,437 - (715) - 492,574 97,336 0.00 97,336
6 2,672 - 2,437 - (711) - 589,328 96,754 0.00 96,754
7 2,672 - 2,437 - (16) - 685,504 96,175 0.00 96,175
8 2,672 - 2,437 - (16) - 781,104 95,600 0.00 95,600
9 2,672 - 2,437 - (15) - 876,132 95,028 0.00 95,028
10 2,672 - 2,437 - (15) - 970,591 94,460 0.00 94,460
11 2,672 - 2,437 - - - 1,064,486 93,894 0.00 93,894
Figure 97 represents the trajectory of National GDP with and without the disruption.
181
Figure 97. National GDP with and without disruption. All values are discounted to 2011
(year 1 in the plot) using the social discount rate.
The notable decline in the baseline future regional and national GDP over time occurs because the
projected GDP growth rate of 2.4% is less than the social discounting rate of 3.0% used in the
calculations. If those parameters were reversed, the baseline trends would show an upward slope.
The users of the model could make such changes. However, this analysis was conducted with the
default values of the parameters.
The total national GDP losses as a percentage of the unaffected national GDP are represented in
Figure 98.
182
Figure 98. GDP losses as a % of national GDP
A summary of annual direct, total, indirect, and induced losses is represented in Table 18 and in
Figure 99.
Table 18. Annual losses summary at national level. All values are discounted to 2011 (year 1)
using the social discount rate.
Annual Direct Annual Total Annual Indirect Annual Induced
GDP Loss GDP Loss GDP Loss GDP Loss
Year ($M) ($M) ($M) ($M)
1 830 1193 258 105
2 826 847 -53 74
3 821 505 -360 44
4 768 126 -653 11
5 715 0 -715 0
6 711 0 -711 0
7 16 0 -16 0
8 16 0 -16 0
9 15 0 -15 0
10 15 0 -15 0
11 0 0 0 0
183
Figure 99. Direct, total, indirect, and induced annual losses at the national level. All values
are discounted to 2011 (year 1 in the plot) using the social discount rate.
The temporal representation of the losses is valuable for understanding the effects of different
parameters, such as restoration schedules and can be used for optimizing the decontamination and
recovery schedules. For convenience, this example was implemented in an Excel worksheet, which
allows further experimentation with different parameters.
A.9.5. Presentation and Use of Results

The output of the model can be interpreted as shown in Table 19.
184
Table 19. GDP losses in the first accident year
GDP
Impact
Direct
Type Indirect ($) Induced ($) 44 Total ($)
($)
Region
Intraregional 𝜟𝜟𝑽𝑽𝑫𝑫 0 𝑥𝑥(𝜟𝜟𝑽𝑽𝑻𝑻 − 𝜟𝜟𝑽𝑽𝑫𝑫+𝑰𝑰 ) Row sum
(1 − 𝑥𝑥)(𝜟𝜟𝑽𝑽𝑻𝑻
Extra Regional 0 𝜟𝜟𝑽𝑽𝑫𝑫+𝑰𝑰 − 𝜟𝜟𝑽𝑽𝑫𝑫 Row sum
− 𝜟𝜟𝑽𝑽𝑫𝑫+𝑰𝑰 )
National 𝜟𝜟𝑽𝑽𝑫𝑫 𝜟𝜟𝑽𝑽𝑫𝑫+𝑰𝑰 − 𝜟𝜟𝑽𝑽𝑫𝑫 𝜟𝜟𝑽𝑽𝑻𝑻 − 𝜟𝜟𝑽𝑽𝑫𝑫+𝑰𝑰 Row sum
Table 20. GDP losses in year 4, assuming TN = 3.

GDP losses
Impact
Direct Indirect
Type Induced ($) Total ($)
($) ($)
Region
Intraregional 𝜟𝜟𝑽𝑽𝑫𝑫 0 0 𝜟𝜟𝑽𝑽𝑫𝑫
Extra Regional 0 -𝜟𝜟𝑽𝑽𝑫𝑫 0 -𝜟𝜟𝑽𝑽𝑫𝑫
National 𝜟𝜟𝑽𝑽𝑫𝑫 -𝜟𝜟𝑽𝑽𝑫𝑫 0 0
In addition to the GDP losses shown in Table 19 and Table 20, the implementation in MACCS
reports costs from evacuation and relocation of members of the public, for both short and long
term, and decontamination costs. Capital losses are also reported corresponding to condemned
property and depreciation of property improvements that cannot be maintained during periods of
interdiction. This represents a reasonably complete accounting of the types of losses that can be
attributed to the occurrence of a nuclear reactor accident or other release of radioactive material into
the atmosphere. However, it does not account for other potential types of losses, such as legal,
health, and stigma costs.
For the purposes of a cost-benefit analysis, the authors suggest using a sum of the direct GDP losses
plus evacuation costs, relocation costs, decontamination costs, depreciation losses, and condemned
property values. This may entail some degree of double counting as well as summing up
fundamentally different cost types, such as GDP losses and losses in tangible wealth. However, the
combination of these values represents a reasonable estimate of the total impact.
44The variable x is the ratio of 𝜟𝜟𝑽𝑽𝑫𝑫 to 𝜟𝜟𝑽𝑽𝑫𝑫+𝑰𝑰 . It approximates the ratio of the induced losses attributable to the directly
affected area and to the entire economy based on the ratio of economic impacts, excluding induced losses, to those same
areas.
185
APPENDIX B. COLLABORATIVE WORKSHOP AND PARTICIPATION
Colleague Organization
Peter Kobos 8822
Thomas Drennen Hobart and William Smith Colleges, Galisteo Consulting Group.
Alexander ‘Sasha’ Outkin 8724
Ben Knueven 1464
Bobby Middleton 8841
Serafina Lawles 8841
Molly Wilson 8841
Bobby Jeffers 8812
Ricky Concepcion 8813
John Eddy 8723
Scott Paap 8712
Stanley Atcitty 8811
Joseph Ulibarri 8161
Brian Naughton 8821
Vanessa Vargas 8724
Howard Passell 8811
Erik Webb 8860
Zach Bends 9310
Thushara Gunda 8825
Cameron Proctor 8751
Carmen Mendez 8841
Jessie Bonfield 8149
Taylor McKenzie 2835
Vincent Neary 8822
Steven Wiryadinata 8712
Ray Byrne 8813
186
Colleague Organization
Michael Bernard 5496, *Contributed materials after the workshop.
Greg Frye-Mason 1100 (was in 1900 at the time of the workshop).
187
DISTRIBUTION
Name Org. Sandia Email Address
Frye-Mason, Greg 01100 gcfrye@sandia.gov
Hillman, Elizabeth 01983 elucero@sandia.gov
Kistin, David 01983 dkistin@sandia.gov
Bernard, Michael 05496 mlberna@sandia.gov
Webb, Erik K. 06160 ekwebb@sandia.gov
Paap, Scott 08712 smpaap@sandia.gov
Wiryadinata, Steven 08712 swiryad@sandia.gov
Outkin, Alexander ‘Sasha’ 08724 avoutki@sandia.gov
Adkins, Carol 08800 cladkin@sandia.gov
Hanley, Charlie 08810 cjhanle@sandia.gov
Byrne, Raymond H. 08813 rhbyrne@sandia.gov
Halloran, Amy Randolph 08820 arhallo@sandia.gov
Kobos, Peter H. 08822 phkobos@sandia.gov
Tidwell, Vince 08825 vctidwe@sandia.gov
Bonano, Tito 08840 ejbonan@sandia.gov
Mendez, Carmen 08841 cmmende@sandia.gov
Benz, Zach 09310 zobenz@sandia.gov
Technical Library 01177 libref@sandia.gov
Company Email
Name Address Company Name
Drennen, Thomas E. drennen@hws.edu Hobart and William Smith Colleges
188
This page left blank
189
Sandia National Laboratories
is a multimission laboratory
managed and operated by
National Technology &
Engineering Solutions of
Sandia LLC, a wholly owned
subsidiary of Honeywell
International Inc. for the U.S.
Department of Energy’s
National Nuclear Security
Administration under contract
DE-NA0003525.

Techno-Economic Analysis: Best Practices and Assessment Tools

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

Techno-Economic Analysis: Best Practices and Assessment Tools

Uploaded by

Copyright:

Available Formats

SANDIA REPORT

Techno-Economic Analysis: Best

Available to DOE and DOE contractors from

Telephone: (865) 576-8401

Available to the public from

Telephone: (800) 553-6847

Figure 1. Lazard's estimated LCOE (Source: Lazard, 2019) ....................................................................18

This page left blank

The goals of this project include:

Our review of the projects includes several key lessons learned:

• TEA should be incorporated at an early stage to guide technology development. TEA

• TEA should assist the customer in recognizing/constructing/modifying the value

• Consider including life cycle assessment (LCA) as part of a comprehensive TEA.

• Be aware of fundamental constraints on the system of interest imposed by physics,

From these lessons, several application areas identified include:

• Apply these lessons for internal business operations;

• Utilizing a more targeted quantitative approach across programs to highlight higher-

GREET greenhouse gases, regulated emissions, and energy use in transportation

MACRS modified accelerated cost recovery system

1.1. Evaluating the Economics of New Technologies

The next section documents the levelized cost of energy methodology.

1.2. Levelized Cost of Energy Methodology

The levelized COE calculation is given by:

where: 𝐼𝐼 = total financed capital costs

2Rhodes (2016) provides a good overview of the LCOE methodology.

The FCR is calculated using:

𝐶𝐶𝐶𝐶𝐶𝐶[1 − 𝑏𝑏𝑏𝑏 ∑𝑀𝑀 𝑛𝑛

where: 𝐶𝐶𝐶𝐶𝐶𝐶 = capital recovery factor

The 𝑟𝑟𝑤𝑤𝑤𝑤𝑤𝑤𝑤𝑤 is calculated by:

where: 𝐸𝐸/𝑉𝑉 = percent of total project equity financed

The CRF is calculated using:

(1 + 𝑟𝑟𝑊𝑊𝑊𝑊𝑊𝑊𝑊𝑊 )𝑛𝑛 (4)

5 Drennen et al., 2002.

1.4. Estimating the LCOE for a new technology

1.4.1. Mechanical System Costs

Levelized Cost of Energy ($/kWhe) FOAK $ 0.055 NOAK $ 0.052

1.4.2. Scaling equations

This methodology includes the following categories:

• Civil and construction costs: 10.5% of mechanical + electrical

1.5. First- and Nth-of-a-kind Methodology

Figure 9. Theoretical Learning Curve (NETL, 2013).

𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝑁𝑁𝑁𝑁𝑁𝑁𝑁𝑁 = 𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐹𝐹𝐹𝐹𝐹𝐹𝐹𝐹 ∗ 𝑋𝑋 −𝑏𝑏 (6)

1.6.1. The sCO2 Brayton system

Table 1. Operating assumptions for 100 MWe reference case.

Figure 11. Estimated system configuration and efficiency from sBET

1.7. TEA, Resilience, and Planning

To adequately integrate resilience goals within a TEA framework, an understanding of how

1.7.1. Community Critical Service Availability and Burden Metrics

1.7.2. Economic resilience metrics

Figure 19. Blue-sky Performance versus Resilience (Jeffers et al. 2020)

2.1. Guiding Optimal Biofuels (TEA of Advanced Biofuels)

Goals and Objectives

TEA Staff: Brandon Heimer and Scott Paap

Goals and Objectives

Goals and Objectives

2.4. Microsystems-Enabled Photovoltaics (MEPV)

Goals and Objectives

2.5. Potential Strategies for Integrating Solar Hydrogen Production and

TEA Staff: Scott Paap

Goals and Objectives

2.6. Pathways to Carbon Neutral Energy Systems at the University of

TEA Staff: Steven Wiryadinata, Josh Morejohn, Kurt Kornbluth

Goals and Objectives