Large Scale Renewable Energy Integration: Identification of Optimal Implementation Plans
by Steven Burns
B.S. in Mechanical Engineering and Engineering and Public Policy, May 1997, Carnegie Mellon University M.S. in Mechanical Engineering, December 2003, University of Massachusetts
A Dissertation submitted to
The Faculty of The School of Engineering and Applied Science of The George Washington University in partial satisfaction of the requirements for the degree of Doctor of Philosophy
May 19, 2019
Dissertation directed by
Jonathan P. Deason Professor of Engineering and Applied Science
The School of Engineering and Applied Science of The George Washington University certifies that Steven Burns has passed the Final Examination for the degree of Doctor of
Philosophy as of April 1, 2019. This is the final and approved form of the dissertation.
Large Scale Renewable Energy Integration: Identification of Optimal Implementation Plans
Steven Burns
Dissertation Research Committee:
Jonathan P. Deason, Professor of Engineering and Applied Science, Dissertation Director
Hernan G. Abeledo, Associate Professor of Engineering and Applied Science, Committee Member
Robert W. Orttung, Associate Research Professor of International Affairs, Committee Member
Joost R. Santos, Associate Professor of Engineering Management and Systems Engineering, Committee Member
Ekundayo A. Shittu, Assistant Professor of Engineering Management and Systems Engineering, Committee Member
ii
Abstract
Large Scale Renewable Energy Integration: Identification of Optimal Implementation Plans
A significant potential for addressing climate change and fuel security is the transition to renewable energy from traditional fossil fuel-based power. However, the inherent technical limitations associated with renewable energy power generation will make this transition costly, especially if the necessary infrastructure upgrades occur in a piecemeal fashion that is subject to regional variances in environmental regulation and individual utility and transmission operator appetite and capability for renewable energy transmission and generation. Projections of increased electric power demand in the
United States necessitate the need for new generation sources, and renewable energy simultaneously offers security of reduced reliance on fossil fuels and the complication of integrating variable energy production. Renewable energy generation could be maximized if placed in higher resource areas, such as solar power being placed in the desert Southwest, but the higher relative capital costs of renewable energy sources and the need to address transfer of renewable energy-based electricity across multiple jurisdictions to reach demand centers could limit the buildout.
This paper describes an optimization model for determining the least cost approach to large-scale renewable energy development in the United States, considering several energy policy options and focused on renewable energy integration. Model outputs showed that there are scenarios under which the Southwest could supply reliable electric power to the United States. While the model was not intended to identify specific infrastructure projects, it yielded useful results based on the trends of predictions across
iii
the scenarios. Of note is that reduced transmission construction times and costs could enable supply of electric power from the Western United States to load centers in the
Mid-Atlantic and Northeast, but that the magnitude of the expansion likely would extend construction activities beyond 2040. In the meantime, capital cost reductions for solar power and growing trends towards distributed generation may obviate the need for such a buildout. In any case, without significant storage capability, wind and solar power might have difficultly achieving expanded market penetration because the expanded baseload capacity additions would be necessary to balance the variable resource. Expansion of the developed model would allow further exploration of specific infrastructure needs to make remote and/or local renewable generation more cost effective.
iv
Table of Contents
Abstract ...... iii
Table of Contents ...... v
List of Figures ...... vii
List of Tables ...... viii
Chapter 1 – Introduction ...... 1
1.1 Statement of the Problem ...... 3
1.2 Organization of the Document ...... 8
1.3 Background ...... 9
1.4 Purpose ...... 11
1.5 Significance...... 12
1.6 Scope ...... 12
1.7 Limitations ...... 13
Chapter 2 – Literature Review ...... 15
Chapter 3 – Research Goals and Approach ...... 25
3.1 Research Goals...... 25
3.2 Research Approach ...... 27
3.2.1 Model Structure ...... 27
3.2.2 Indices ...... 32
3.3 Model Formulation ...... 33
3.3.1 Trade-offs ...... 32
3.3.2 Decision Variables ...... 34
3.3.3 Constraints ...... 36
3.3.4 Additional Inputs ...... 37
v
3.3.5 Scenarios ...... 40
3.4 Data Requirements and Collection ...... 41
3.5 Analytical Procedure ...... 41
3.6 Modeling Simplifications...... 42
Chapter 4 – Model Description ...... 45
4.1 Objective Function ...... 47
4.2 Universal Constraints ...... 51
4.3 Scenario-Driven Constraints ...... 59
4.4 Parametric Analyses...... 60
Chapter 5 – Results, Conclusions, and Recommendations ...... 62
5.1 Outputs ...... 62
5.1.1 Scenario 1 – Base Case ...... 63
5.1.2 Scenario 2 – Increased Solar Power Focus ...... 67
5.1.3 Scenario 3 – Improved Development Processes ...... 72
5.2 Conclusions ...... 70
5.3 Future Research ...... 75
References ...... 78
Appendix 1 – Determination of Model Nodes ...... 81
Appendix 2 – Model Initial Conditions and Costs ...... 106
Appendix 3 – Model Text ...... 169
Appendix 4 – Model Outputs ...... 179
vi
List of Figures
Figure 1-1. Average Electric Generator Fuel Prices (1999-2010) ...... 4
Figure 1-2. Wind Resource and Transmission Line Locations...... 7
Figure 2-1. North American Balancing Authorities & Reliability Organizations ...... 16
Figure 2-2. Net Annual Power Flow Between Regions in 2010, million MWh ...... 18
Figure 2-3. Line Losses for Various Transmission Voltages ...... 20
Figure 3-1. Basic Concept Diagram for Regional Demand Balancing ...... 28
Figure 3-2. Nodal Model Representation (pathways excluded for simplicity) ...... 31
Figure 3-3. Sample Power Dispatch Stack Operation ...... 32
Figure 3-4. State Renewable Portfolio Standards and Goals ...... 39
Figure 5-1. Scenario 1 Predicted Generation Capacity Additions ...... 64
Figure 5-2. Scenario 1 Predicted LVAC Capacity Additions ...... 66
Figure 5-3. Scenario 1 Predicted HVDC Capacity Additions ...... 67
Figure 5-4. Scenario 2 Predicted Generation Capacity Additions ...... 68
Figure 5-5. Scenario 2 Predicted LVAC Capacity Additions ...... 70
Figure 5-6. Scenario 2 Predicted HVDC Capacity Additions ...... 71
Figure 5-7. Scenario 3 Predicted Generation Capacity Additions ...... 73
Figure 5-8. Scenario 3 Predicted LVAC Capacity Additions ...... 74
Figure 5-9. Scenario 3 Predicted HVDC Capacity Additions ...... 75
vii
List of Tables
Table 4-1. Model Components ...... 45
Table 5-1. Scenario 1 Predicted Generation Usage Statistics ...... 65
Table 5-2. Scenario 2 Predicted Generation Usage Statistics ...... 69
Table 5-3. Scenario 3 Predicted Generation Usage Statistics ...... 73
Table A-1. Combination of Balancing Authorities and Utilities into Model Nodes ...... 82
Table A-2. Initial Generation Capacities by Region ...... 110
Table A-3. Initial Low Voltage Interconnection Capacities ...... 112
Table A-4. Initial High Voltage Interconnection Capacities ...... 123
Table A-5. Summer Daytime Peak Instantaneous Demand by Region ...... 123
Table A-6. Summer Nighttime Peak Instantaneous Demand by Region...... 127
Table A-7. Winter Daytime Peak Instantaneous Demand by Region ...... 130
Table A-8. Summer Nighttime Peak Instantaneous Demand by Region...... 134
Table A-9. Summer Daytime Total Demand by Region ...... 137
Table A-10. Summer Nighttime Total Demand by Region ...... 143
Table A-11. Winter Daytime Total Demand by Region ...... 148
Table A-12. Winter Nighttime Total Demand by Region ...... 153
Table A-13. Balancing Authority Geographic Midpoints ...... 158
Table A-14. Generation Technology Costs ...... 160
Table A-15. Wind and Solar PV Capacity Factors by Region ...... 161
Table A-16. Fuel Costs ($2016)...... 164
Table A-17. Localized Natural Gas Price Adjustment ...... 165
Table A-18. Electric Transmission Costs ...... 166
Table A-19. Renewable Portfolio Standards for Scenario Analyses ...... 167
Table A-20. Annual New Generation Capacity Installed (Scenario 1) ...... 180
viii
Table A-21. Annual Low Voltage AC Transmission Capacity Installed (Scenario 1) ....186
Table A-22. Annual High Voltage DC Transmission Capacity Installed (Scenario 1) ...186
Table A-23. Annual Electricity Generation, Usage, Imports, and Exports by Region (Scenario 1) ...... 188
Table A-24. Annual Renewable Energy Usage (Scenario 1) ...... 209
Table A-25. Annual New Generation Capacity Installed (Scenario 2) ...... 230
Table A-26. Annual Low Voltage AC Transmission Capacity Installed (Scenario 2) ....236
Table A-27. Annual High Voltage DC Transmission Capacity Installed (Scenario 2) ...237
Table A-28. Annual Electricity Generation, Usage, Imports, and Exports by Region (Scenario 2) ...... 240
Table A-29. Annual Renewable Energy Usage (Scenario 2) ...... 261
Table A-30. Annual New Generation Capacity Installed (Scenario 3) ...... 282
Table A-31. Annual Low Voltage AC Transmission Capacity Installed (Scenario 3) ....288
Table A-32. Annual High Voltage DC Transmission Capacity Installed (Scenario 3) ...289
Table A-33. Annual Electricity Generation, Usage, Imports, and Exports by Region (Scenario 3) ...... 293
Table A-34. Annual Renewable Energy Usage (Scenario 3) ...... 314
ix Chapter 1 – Introduction
Global power consumption is continuing to grow at a rapid pace, leaving many nations dealing with supply shortfalls and uncertain economic futures. In the United
States, generation capacity reserve margins are decreasing, and by 2030, there is a projected 32% increase in power demand relative to 2008 (Energy Information
Administration, 2008). The investment needed to meet this challenge, however, will be complicated by the often competing demands of economic growth and environmental preservation, and the tension between these priorities will likely grow as the climate change debate continues and pressure mounts to reduce greenhouse gas (GHG) emissions, especially from the power sector.
Increasing concerns about greenhouse gas emissions and, to some extent, future fuel supply security, will leave renewable energy sources poised to become a significant contributor to generation capacity. In the United States, regulatory support has grown for renewable energy sources to the point that 39 states mandate or have voluntary goals for a percentage of renewable energy supply (Federal Energy Regulatory Commission,
2011). In the future, where carbon emissions constraints or fuel supply scarcity could significantly increase fossil fuel-based power prices, renewable energy sources would be in even greater demand. At present, however, traditional renewable energy limitations such as high capital costs and the large distance between areas of strong renewable resource and electric load demand centers limit large-scale renewable energy deployment.
Many nations will confront issues of power grid resiliency and the need to lessen impacts on consumers in the face of this energy transition. For example, without
1
development of cost-effective local supply options or significantly improved operating efficiencies in low resource environments by wind, solar or other renewable energy technologies, the transition away from fossil fuels to renewable energy to provide power likely will require a significant electric transmission infrastructure buildout: a buildout that some liken to that required of the Eisenhower Interstate System (Heyeck and Wilcox,
2008). With varying mandates between states and lack of a true cost advantage for renewable energy in most locations, the necessary infrastructure buildout is under risk of occurring within limited, disparate contexts. An optimal buildout scenario would determine time and place infrastructure upgrades as needed, accounting for energy demand across the nation as opposed to within individual regions. This optimized buildout would limit waste and therefore lessen consumer costs and improve system reliability.
This document address issues related to moving from a theoretical construct to an implementation plan for large-scale renewable energy development. A test case is examined through the development of a tool to compile optimal renewable energy buildout scenarios for the United States, based on scenarios provided by the Solar Grand
Plan (Zweibel, Mason, and Fthenakis, 2008) and under varying regulatory, demand, and technical inputs. The scenarios use optimization modeling, beginning with existing generating capacity and transmission infrastructure as available from utilities and regional transmission operators, and allowing for infrastructure additions that most effectively meet energy demand under different energy demand and regulatory scenarios over the next 30 years. Regulators and the development community will be able to use
2
the results of this effort to more cost effectively plan for large-scale renewable energy deployment.
1.1 Statement of the Problem
A significant potential for addressing climate change and fuel security is the transition to renewable energy from traditional fossil fuel-based power; however, the inherent technical limitations associated with renewable energy power generation will make this transition costly, especially if the necessary infrastructure upgrades occur in a piecemeal fashion that is subject to regional variances in environmental regulation and individual utility and transmission operator appetite and capability for renewable energy transmission and generation.
From a financial standpoint, a strong case can be made for the path away from fossil fuels by considering the volatility of fossil fuel prices, the potential for supply interruptions, and the association of climate change with fossil fuel consumption. The current national generation portfolio is nearly 40% coal, and many markets rely on natural gas-based generation for peaking support, thus leaving marginal power prices subject to natural gas market volatility – a risk not duly considered in the current low- price environment for natural gas. However, as illustrated in Figure 1-1 (per data from
Energy Information Administration, 2011), fuel prices for all non-renewable power sources exhibited significant volatility over the past decade, with maximum prices ranging from 30% to 50% of average prices, depending on fuel source.
3
80
60
Average Fuel Price 40 ($/MWh)
20
0 1999 2001 2003 2005 2007 2009 Year
Nuclear Fossil Steam Gas Turbine and Small
Figure 1-1. Average Electric Generator Fuel Prices (1999-2010)
These fuel supply costs have led, in part, to an average 44% power price increase between 2000 and 2010 (Ibid), and the Energy Information Administration projects continued power price increases throughout the next twenty years in a business as usual scenario (Energy Information Administration 2008). If a scenario emerges where demand increases for the electrification of the transportation network, there is potential to need for even more generation capacity, which could drive additional power price increases. In either case, with or without increased power demand, elevated power prices will have the potential to limit economic growth and make certain American industries, such as the primary aluminum industry (United States International Trade Commission,
2010) less competitive in the global marketplace.
Some of the price volatility and ultimate increases can be attributed to the supply interruptions seen in the summer of 2008. The natural gas distribution network was damaged from Hurricane Katrina. Transportation problems limited rail movement of
4
coal from the Powder River Basin and the Midwest. Beyond these national concerns, international interests can also affect U.S. prices; prior to the 2008 Recession, construction material costs were on a steady increase as U.S. growing demand for materials with China and other rising economies drove up prices. The extrapolation of these instances to future energy fuel imports highlights the concern of energy security, especially in consideration of political volatility in many of the prime oil-bearing regions of the world, including the Middle East, Nigeria and North Africa, and Venezuela.
Near term steps in securing national energy supply might include increased domestic coal production and/or biofuel use, but without technological advances to limit greenhouse gas emissions or reduce competition with food crops, these options will not be singular answers to satisfy growing energy demand. Anthropogenic carbon dioxide emissions are continuing to increase, especially from the power generation sector.
Energy demand is expected to grow at least 1% annually over the next 20 years (Energy
Information Administration, 2008), and if coal remains the primary energy source in the
United States, carbon dioxide emissions will continue to increase along with economic growth.
Also, the deliverance from oil once promised by biofuels has failed to materialize.
No breakthroughs have yet occurred to allow commercial-scale production of algae or cellulose-based biofuels that can be grown on marginal lands. The corn price run-up in
2008 further serves as a cautionary tale of significant biofuel use.
Given renewable energy's limited carbon footprint, combined with the fact that it can be installed in a dual land use capacity (i.e., wind turbines in farm land) or in marginal lands (i.e., solar fields in unoccupied desert spaces), the concerns above point to
5
renewable energy as a pathway to meeting energy needs while limiting concerns of fuel price volatility and climate changes. Unfortunately, renewable energy-based power generation has significant drawbacks, especially when installed on a large scale. One of these drawbacks is the increased capital cost and intermittent operation of renewable energy capacity relative to fossil fuel-based power plants (Energy Information
Administration, 2008) or the need for storage technology to save power generated in periods of low demand. Unfortunately, cost effective storage technology is not available.
Therefore, for renewable energy sources to be cost competitive, the installed power plants must have high capacity factors, which is currently achievable only in high resource areas such as the desert for solar power and the Great Plains for wind power: areas that are remote from electric demand centers. Figure 1-2 (National Renewable Energy
Laboratory, 2012) illustrates the distance between renewable energy sources and primary electric demand centers in the United States along with the relative lack of electric transmission capacity between these areas. 1
1 Specifically, Figure 1-2 shows the relative lack of high voltage transmission between the high wind resource areas in the country and the demand centers along the East and West Coasts. Similarly, the figure indicates how additional transmission in capacity The Southwest would allow for greater power flow into California.
6
Figure 1-2. Wind Resource and Transmission Line Locations
These distances and the associated transmission losses that would occur over the existing transmission network point to the need for an extensive transmission infrastructure upgrade and buildout to support cost effective large-scale renewable energy integration. This buildout would have to consider regional load balancing to support power grid stability; the optimal transfer of power among regions; and the relative need of the regions to obtain renewable energy from remote sources. It is this need to coordinate activities across regions that may be problematic. For example, as transmission lines are constructed across different jurisdictions, differing environmental and permitting requirements as well as cost allocation and siting methods must be considered. While the development of regional transmission organizations and regional
7
development plans is a positive step, more benefits could be likely with planning on a larger scale.
Plans such as the Solar Grand Plan that bridge these regional boundaries will be necessary to move forward with a transition to renewable energy. Even these initial plans, however, require some refinement to illustrate the need for cooperative action to optimize renewable energy deployment. Without an understanding of the efficiencies achievable in renewable energy deployment, planners and regulators will be at risk of making sub-optimal decisions and increasing costs to consumers.
1.2 Organization of the Document
This document contains five chapters. The Introduction details the background of the need for renewable energy deployment and the associated limitations of transitioning the electric generation system to focus on intermittent resources. The following Literature
Review provides an overview of the literature or relevant to the state of transmission planning in the United States and the need for creation of a model for determining pathways to transition to renewable energy. The third chapter contains the research methodology, detailing the primary research goals and elaborating the primary research methods used and expected results from this study. The fourth chapter documents a model developed to project optimal electric transmission buildouts for deploying large- scale renewable energy across the United States, and the final chapter describes model outputs under varying scenarios, illustrating respective timelines of electric transmission buildout for optimal transition to large-scale renewable energy supply.
8
1.3 Background
The publication of the Solar Grand Plan was timely given that renewable energy power generation can become a significant contributor to U.S. generation capacity. The
Solar Grand Plan addresses energy security and climate change concerns and offers a path away from fossil fuel-based generation and associated greenhouse gas and other pollutant emissions. In addition, the increase in the solar energy and associated technology markets has the potential to set a global precedent, ultimately leading to reduced worldwide carbon dioxide emissions.
The specifics of the Solar Grand Plan include the installation of large photovoltaic solar arrays in the desert Southwest, using high voltage direct current electric transmission lines to transfer the energy to demand centers around the country. The plan’s authors claim that solar power plants could supply 69% of the nation’s electricity by 2050. Achieving this supply would necessitate the use of compressed air storage in underground caverns near population centers – specifically, excess daytime electricity would be used to run air compressors to build pressure in the caverns, and the pressure would be slowly released to drive turbines for power at night. The authors project that implementing the plan would require subsidies of approximately $400 billion over 40 years and would cut national greenhouse gas emissions to 62% below 2005 levels by
2050, using solar cells with slightly greater efficiency than available at the time of publication.
The Solar Grand Plan is a conceptual step in addressing concerns over increasing energy prices and environmental degradation resulting from the burning of fossil fuels for electric power and the transportation network. Taking this plan from concept to implementation requires that several technological barriers be overcome such as the
9
improvement of solar cell efficiencies and the development of compressed air storage.
Early deployment planning will ease implementation and limit costs once these barriers are overcome. Therefore, an initial step-by-step project timeline, scope, and cost estimate is proposed for transmission grid development, allowing for identification and examination of Solar Grand Plan roadblocks and weaknesses. The following potential weaknesses to renewable energy development are considered in the construction of the implementation plan under this proposed research:
• Land – renewable energy sources typically are land intensive and any
transmission upgrades will require right-of-way procurement, which can lead to
significant project expense;
• Resource intermittency/variability – solar and wind resource are variable,
requiring that backup generation or energy storage technology be incorporated
into the implementation plan;
• Cost – though renewable energy uses “free” fuel, renewable energy facility capital
costs are still greater than for fossil fuel generation facilities, and without a carbon
tax or significant government subsidies, renewable energy sources will have
difficulty competing;
• Distance from load – peak renewable energy resource (solar and wind) typically
are located remote from primary load centers (e.g., wind in the Dakotas and solar
in the desert Southwest), requiring that significant transmission upgrades occur to
bring power to areas where needed; and
• Production capacity – large-scale buildout of renewable energy generation
capacity will require ramp-up of solar and wind power generation, storage
10
technology, and transmission infrastructure component production capacity,
which could lead to near-term component price increases.
A project plan detailing a phased buildout of renewable energy based generation, along with associated transmission and storage technologies, will identify gaps in the
Grand Plan, bring forth any limitations preventing full plan execution, and highlight areas necessary for further research.
1.4 Purpose
The purpose of the research is to further the concept presented in the Solar Grand
Plan and its updates by developing an optimized buildout schedule to transition the
United States to a renewable energy-based electric generating system. The research focuses on the construction of a linear programming model that will output a buildout schedule based on a least-cost approach, balancing consumer electric costs with resource constraints, technology and development costs, and applicable regulatory requirements and limitations.
This project moves beyond typical market penetration analyses performed for renewable energy systems. Rather than focusing on technology penetration levels that would be achieved in a free market under a given set of economic and regulatory conditions, the project instead focuses on determining the buildout steps and timing to most efficiently achieve desired renewable energy supply. Consequently, this project can inform the debate regarding large-scale renewable energy deployment with detailed optimized deployment scenarios that account for cost and resource limitations, providing insight into the technical and economic achievability of renewable energy and greenhouse gas reduction targets. The derived deployment schedules focus on solar
11
power as described in the Solar Grand Plan, but the results can also be generalized to large-scale renewable energy development activities across the nation.
1.5 Significance
The significance of this research lies in the need for continuing power development in the United States in light of increasing environmental and energy security limitations.
The Department of Energy projects significant growth in U.S. electric power demand over the next 20 years, a period in which fuel supply might be at greater risk due to socio- political events. Also, the United States could ultimately face greenhouse gas emission reduction targets, with the electric power industry as a focus for required cuts.
Consequently, power generation and transmission planners will be challenged to balance significant buildout requirements with cost, security, and environmental concerns. The nature of renewable energy sources is such that energy security and environmental concerns are limited relative to fossil fuels due to comparatively low fuel transport requirements and carbon emissions. Unfortunately, higher relative capital costs of renewable energy sources and the need to address transfer of renewable energy-based electricity across multiple jurisdictions to reach demand centers could limit how planners address these other concerns. As such, this project’s optimized buildout schedule for renewable energy provides a useful tool to address these concerns while also limiting extra costs associated with renewable energy-based electric power.
1.6 Scope
The research focuses on the creation of a model that to be used for the determination of optimal buildout scenarios for transitioning the United States to a renewable-energy based power generation portfolio. The research scope includes the development of a
12
model for determining renewable energy buildout scenarios, focused on determination of the following:
1. The expected costs and required timeframes for transitioning the continental
United States to a renewable energy-based power generation portfolio under
different energy demand and renewable energy regulatory scenarios; and
2. the efficiencies that could be achieved in the transition by enacting a national
renewable energy development plan as opposed to relying on individual regions
to coordinate development activities.
The expected timeframes, costs, and benefits are based on optimized buildout scenarios as determined from a regional model of the continental United States, accounting for electricity demand, renewable resource availability, existing and expected generation capability, and electric transmission constraints.
1.7 Limitations
The research led to a model for predicting general buildout scenarios of generation and transmission capacity throughout the United States over the next 30 years. The presented buildout models are conceptual and should not be considered as recommendations for specific infrastructure projects. Even though the model is intended to be conceptual, the primary research limitation derives from this simplified view. As described later in this document, the model predicts infrastructure construction based on electricity demand and construction and operational costs, and the model assumes that infrastructure will be built without opposition. Congestion fees, distribution costs, and the differences between localized market conditions and policy environments are not considered. When these factors aren’t considered, there isn’t a market signal for users to
13
reduce usage and/or for transmission system operators to prioritize system upgrades.
Incorporation of these elements is therefore an avenue for further exploration.
Technical limitations include modeling capacity and quality of data. The model was developed in consideration of computational tractability considerations, dictating that a relatively high-level granularity of the power grid be used. For example, whereas a true model of the U.S. power sector would use software such as PSS-E® and/or include elements to account for all power plants and load centers and the associated operating characteristics at all times, the model in this research instead aggregated data on power plants, transmission capacities, and electricity demands. While this data aggregation led to some loss of accuracy in the model results, the model described herein aggregates power plants and regions in such a way that electric transmission limitations are adequately captured – an acceptable aggregation given that the model is focused on electric transmission buildout scenarios. In addition, the model takes a conservative approach to temporal considerations by balancing available electricity supply and demand only during hours of peak diurnal and seasonal electric demands and hours of minimum available generation capacities. Chapter 3 contains a detailed discussion of limitations associated with the modeling approach.
Finally, the quality of the data is limited to publicly available data on generation facilities, electric transmission interconnections between regions, and projected seasonal electricity demands. The model outputs depend on the quality of these data, so where possible, several sources have been consulted, including utility reports to the Federal
Energy Regulatory Commission and aggregate data as compiled by the Energy
Information Administration.
14
Chapter 2 – Literature Review
The technical limitations associated with a large scale transition to renewable energy such as intermittency and distance from demand centers are relatively easy to understand within the context of the structure of the electric power system. The interface with the electric power system in the United States for most individuals is their local electric utility. In 2010, there were more than 3,000 electric utilities in the United States (Energy
Information Administration, 2012). Deregulation of the electric power industry resulted in varying responsibilities for electric utilities in different markets, but in simple terms, electric utilities are responsible for providing reliable, affordable electricity to their customers.
Balancing authorities cover the service territories of a single or several utilities that are closely interconnected. The balancing authorities are responsible for maintaining power grid stability in their regions through balancing generation, demand, and power imports and exports with neighboring authorities. Because electricity cannot be stored on a large scale, the amount of generated electricity must match demand within an allowable tolerance. This balance of generation and demand is an on-going process that requires constant dispatching or taking generators offline; modifying power plant output as allowable; and adjusting import and export quantities to and from the region. See
Balancing and Frequency Control (National Electric Reliability Corporation, 2011) for a detailed description of the procedures for balancing energy demand and generation.
Figure 2-1 displays the approximately 130 existing balancing authorities in the United
States and Canada along with the regions into which the balancing authorities are registered.
15
Figure 2-1. North American Balancing Authorities & Reliability Organizations (National Electric Reliability Corporation, 2012)
The National Electric Reliability Corporation (NERC) and the Regional Entities noted in Figure 2-1 set reliability standards and work with the balancing authorities and utilities to ensure provision of adequate, reliable power to customers. Together, these regions form the major interconnections of the power grid: the Eastern Interconnection that covers the East Coast, the Midwest, and much of the Plains states; the Western
Interconnection that covers the West Coast and the Rocky Mountain States; ERCOT, which covers much of Texas; and the Quebec Interconnection. These interconnections operate independently of each other, though limited interties do exist.
In considering the structure of the electric power industry, several barriers to large- scale renewable energy deployment become apparent, as follows below.
16
Capital Costs. The installed cost of renewable energy generators is greater than fossil fuel-based systems. Based on industry data, the Department of Energy assumes renewable energy capital costs to be between 1.9 (wind) and 2.6 (solar photovoltaic) times the capital costs of natural gas-based power generation facilities (Energy
Information Administration, 2016). Without fossil fuel costs significantly greater than current levels, renewable energy technologies will be at a competitive disadvantage.
Over large well-populated regions, the impact can be reduced through optimized locations in areas of high resource and by spreading costs over several consumers.
However, cost decisions are made at the local level, and less-populated or resource-poor regions do not have an incentive to install generation that would be higher cost.
Transmission. The strongest solar resource in the continental United States is located in the Southwest, and the strongest wind resource is largely located in the Plains States
(see Figure 1-2) – locations remote from demand centers such as California, the Mid-
Atlantic, and the Northeast. Making large-scale renewable energy supply problematic with respect to geography is the fact that the United States power grid is structured largely around local supply. Most electric power demand is served by local generators.
Net interregional trade accounted for less than 1% of delivered power in 2010.
Figure 2-2 illustrates net interregional power flows, indicating that California is the largest net importer of electricity, consuming power produced in the Northwest and
Southwest. These two regions provide about 25% of California's electricity supply, already stressing available transmission infrastructure. The other large electric power movement is from the far western part of the PJM Interconnection (northern Illinois) to
17
the rest of PJM 2. This transit, which would limit additional power flow into the region from the Plains States, provides low-cost electricity generated with nuclear and coal-fired capacity.
Figure 2-2. Net Annual Power Flow Between Regions in 2010, million MWh (Energy Information Administration, 2011b)
Existing transmission capacity is limited in the high import areas of the Mid-Atlantic
States, New York, and New England as well as in moving power from the Southwest into
California. Several transmission projects are underway to alleviate this congestion, and there were plans for nearly $50 billion in new transmission investments through 2015
(Edison Electric Institute, 2012).
Therefore, transmission access is a key variable for large-scale renewable energy development, but transmission access presents a series of problems including cost allocation, multi-jurisdictional project approval requirements, and the existing strain on the national electric transmission system. Cost allocation is problematic from the
2 The PJM Interconnection is the regional transmission organization that coordinates electricity supply in in all or parts of Delaware, Illinois, Indiana, Kentucky, Maryland, Michigan, New Jersey, North Carolina, Ohio, Pennsylvania, Tennessee, Virginia, West Virginia and the District of Columbia.
18
standpoint of the intermittency of renewable resources (discussed below) and the typical remote location of strong resource availability. The remoteness of the renewable resource typically means that new transmission infrastructure will need to be built to support renewable energy development, and there is debate over who should pay for this buildout, for example whether the cost should be allocated to equally across consumers or whether costs should be allocated to the generators using the new line, thus affecting their bid price point in a competitive market. The choice of allocation method can affect renewable energy penetration levels. If generators are required to absorb all transmission costs into their basis, project financing will be more difficult, leaving only projects in the highest resource areas.
Regardless of allocation method, costs will ultimately be borne by consumers, and transmission across long distances can result in higher capital costs and large transmission losses. Transmission capital costs in the Western United States range anywhere from $650,000 per mile to over $2 million per mile, with additional costs for line terminations, substations, and other ancillary equipment, based on jurisdiction
(California Energy Commission, 2007). Figure 2-3 indicates expected losses for long distance transmission.
19
Figure 2-3. Line Losses for Various Transmission Voltages (Heyeck and Wilcox, 2008)
An additional complication in transmission buildout to support transfer of renewable energy throughout the nation is the issue associated with cross-jurisdictional transmission projects. Whereas typical utility-scale solar and wind projects can be completed in less than two years, transmission lines often take several years to complete due to the need to procure large quantities of land as well as to satisfy varying environmental and other permitting requirements. In 2010, the Federal Regulatory Energy Commission (FERC) established Order 1000 to improve transmission planning, but the timeframe for transmission projects is still extended. This difference in development timelines will therefore continue to serve as a limiter on renewable energy development as projects in remote locations will continue to have difficulty securing financing due to uncertain transmission access.
20
Intermittency. Renewable energy generator output is subject to the available resource such as wind speed or solar output, which vary based on season and time of day as well as year-to-year. A twist of irony is that while renewable energy may be a useful climate change mitigation tool, it is perhaps an energy source most at risk from climate effects in that changing weather patterns causes uncertainty in predictions of future output.
As noted above, energy storage technology is not available for utility-scale projects, and generation and demand must therefore be balanced at all times. Therefore, these variations limit the ability to predict and schedule output from renewable sources, and system operators cannot count renewable generators as firm capacity, requiring that additional balancing reserve be made available. With respect to transmission, the intermittent output creates problems in allocating capacity on transmission lines and determining which transmission projects should be prioritized. Specifically, the lack of certainty of energy transfer, and therefore revenue generation, on a given transmission line makes it difficult to justify transmission to service renewable energy generators.
Typically, renewable energy penetration is limited within an electric power system due to problems with addressing intermittency. States such as Hawaii and California have ambitious renewable energy targets (Federal Energy Regulatory Commission,
2011), and renewable energy advocates point to Denmark, which produces nearly 20% of its electricity from wind turbines, as an example of successful wind power integration. It should be noted however, that Denmark has several days per year where power must be imported and that it relies on its interconnection with the larger European power system to balance its wind power production.
21
Larger, more diverse systems can handle more intermittent resources due to the broad geographic range of renewable energy generation sites, that is, due to lesser risk of all renewable energy generators depowering at the same time due to the same weather event; larger availability of balancing generators (i.e., more power plants are available to offset drops in renewable energy output), some of which are kept running as “spinning reserve” to quickly meet demand if necessary; and a more stable demand profile (i.e., demand variations tend to balance out over a larger pool of consumers). In spite of these benefits, there are still limits on the amount of intermittent load that a power system can absorb.
Baseload coal and nuclear power plants cannot adjust load beyond a few percent of nameplate capacity; significant cost is required to maintain power plants in reserve; and there is still a potential for localized voltage drops and loss of service even in large power grids.
Considering the breadth of generation types needed to ensure reliable supply and the need for transmission buildout, a transition to large-scale renewable energy supply would be most efficient and cost effective under a national strategic planning effort. Several tools and models are available to inform this buildout, some of which are described below.
MARKAL/TIMES. Developed by the International Energy Agency’s Energy
Technology Systems Analysis Programme, the MARKAL/TIMES models can be configured to present a least cost energy system over several decades based on projected end-use energy consumption such as lighting, heating, and transport (Energy Technology
Systems Analysis Programme, 2005). Based on projected energy use in multiple time periods, the model will make decisions on trade-offs between new generation, imports,
22
and alternative operating regimes for existing generation. Constraints can be applied in several ways such as based on end-use technology consumer choices, power plant emissions control requirements, required capacity or power import/export additions and restrictions.
The National Energy Modeling System (NEMS). NEMS is an energy-economy modeling system of the United States, used by the Energy Information Administration of the Department of Energy to predict the production, import, use and price of energy through 2030 under different technical, economic and policy scenarios (Department of
Energy, 2009). NEMS uses modules that predict energy demand in several sectors; determine the generation fleet and energy trades necessary to meet the required demand over several time periods, accounting for environmental constraints and revenue requirements; and calculate the cost of power.
Regional Energy Deployment System (ReEDS). The National Renewable Energy
Laboratory developed ReEDS to predict buildout of electric power generation and transmission throughout the United States (National Renewable Energy Laboratory,
2011). The model can predict an optimized buildout to meet power demand based on varying demand, emissions, or generation-type limits. ReEDS was the model primarily used for the Department of Energy’s 20% Wind Energy by 2030 report.
PSS-E®. This software, offered by Siemens, simulates flows in electrical transmission networks in steady-state conditions. The software can be used to model transmission networks and to identify areas of power grid congestion under varying loading conditions.
23
Application of any of these models to implementing the Solar Grand Plan would provide insight into optimal deployment strategies. Except for PSS-E®, these models do not account for transmission infrastructure buildout requirements. Use of PSS-E® would provide insight into needed transmission lines to reduce power grid congestion, but the model does not provide insight into optimal timing of these investments.
The research documented herein does not identify detailed market impacts or specific infrastructure requirements to allow for large-scale renewable energy integration.
Instead, this research takes a different approach by optimizing solar power deployment timelines along with associated transmission requirements. Specifically, the research projects the optimal timing and general locations of the buildout as opposed to investigating the optimum end states under a variety of economic and policy scenarios.
Therefore, the information detailed in the following chapter describes the construction of a simplified model to analyze the buildout scenarios specific to the development of a renewable energy infrastructure plan. The research does not suggest methods for reducing costs, but instead presents optimal buildout approaches under varying scenarios.
24
Chapter 3 – Research Goals and Approach
The research described herein focuses on the design and development of a model for determining the least cost approach to large-scale renewable energy development in the
United States, considering several energy policy options and focused on renewable energy integration. Through application of this model, decision makers will be able to gain a greater understanding of the cost and timing implications associated with various buildout scenarios, using variants such as differing renewable energy incentive structures as compared to requiring large-scale coordination of technology deployment on state, regional, and national levels. This role will highlight efficiencies that can be achieved in the buildout, giving decision makers information necessary for easing the transition to a renewable energy-based electric power generation system.
3.1 Research Goals
The research has two primary goals: (1) to develop a decision support tool that can predict costs and timeframes for large-scale renewable energy deployment in the United
States, accounting for the regional energy demand balancing and varying energy policies; and (2) to apply the tool to several development scenarios to illustrate changes in development timelines and costs for large-scale renewable energy deployment under varying conditions.
A linear programming model that accounts for a 30-year timeframe, covering multiple generation technologies and regions, is complex by necessity. Therefore, achieving the first goal requires that the model address the following sub-goals:
25
• Developing decision variables that allow for generation and transmission capacity
to be added or subtracted annually as regional energy demands fluctuate or
operating costs outweigh the usefulness of the components;
• Creating an objective function that efficiently calculates project costs and
timelines by relating costs of the renewable energy buildout to a “no-action” or
“business-as-usual” scenario and by determining the timeframes necessary for
these this buildout to be cost effective; and
• Finding and constructing constraints that accurately model energy demand and
pre-existing generation and transmission capacity in each region in addition to
interregional transmission losses and variable regional energy policies and
construction and operating costs.
Achieving the second research goal entails an exploratory study of the levels of incentives necessary to bring large-scale renewable energy to grid parity or better under differing project planning, energy demand, and energy pricing scenarios. Variants in implementation cost and timeframe will be determined for differing levels of policy and construction coordination across the United States in addition to alternate carbon tax structures or other renewable energy incentives. The investigation of the predicted buildout costs and times will serve to verify the cost estimates envisioned in the Solar
Grand Plan while also highlighting key project limitations.
Another aspect of the research is the exploration of policy options for enacting a transition to renewable energy. Specifically, incentives considered through reduced capital costs or development timeframes were considered. A key note is that this research was not designed to set policy, but rather to inform policy makers of the
26
potential efficiencies that can be achieved through coordinated action on the national transition to renewable energy. Individual policy instruments were not evaluated beyond determinations of the specific time and cost savings that could be achieved through coordinated action given various renewable energy targets and economic scenarios.
3.2 Research Approach
The research approach utilized a liner programming model to identify annual electric generation and transmission installation required and rate necessary to meet energy demand under different scenarios that vary energy demand, construction costs, or specific targets for generation from a specific source. This chapter describes the structure of the model in as well as to the primary model decision variables, model data inputs, and the scenarios chosen to illustrate model outputs.
3.2.1 Model Structure
The model objective function calculates the total energy system cost, including construction costs, annual operating costs, fuel costs, and any environmental costs, adjusted for based on the discount rate and the year of the expenditure. The model is structured to minimize this cost while achieving the desired renewable energy buildout.
Therefore, the model identifies the electric generation capacity and transmission additions necessary in each year to yield the overall lowest energy system cost over the
30-year timeframe in each scenario. The model balances electricity supply and demand in all regions, selecting to import or export electricity between regions as necessary to minimize total energy system cost. Figure 3-1 illustrates a basic concept diagram for the balancing that occurs in each region, with elements explained later in this chapter.
27
Total Electricity Demand (including spinning reserve)in Region X
Imported LVAC electricity from other regions (reduced by Exported LVAC electricity to losses based on distance) other regions Region X – Local Power Generation Used Imported HVDC electricity Exported HVDC electricity to from other regions (reduced other regions by losses based on distance)
Electricity Generated from Power Plants in Region X
Limits Limits
Must be able to meet Installed transmission Installed generation peak demand, capacity between regions capacity in Region X considering reserve margin
Limits
Pre-installed Annual Regions viable for Pre-installed Annual capacity construction limits connection capacity construction limits
Figure 3-1. Basic Concept Diagram for Regional Demand Balancing
As noted in Chapter 2, balancing authorities are responsible for maintaining load- resource balance within a given area, matching generation to electricity demand and maintaining power grid stability in each region, and the planning for larger transmission projects between balancing authorities that often occurs within regional entities charged with maintaining the power grid. Consequently, the linear model is constructed as a nodal model, with each node representing a balancing authority in the contiguous 48 states and connected regions of Canada. To limit model size, certain balancing authorities are grouped in cases of strong interconnection and where alternative connections to other balancing authorities are not realistic, such as when a municipal or
28
cooperative entity is encompassed within a significantly larger region. Electricity imported from Mexico is modeled as being contained within the balancing authority that imports the given power. Figure 3-1 displays the modeled balancing authorities, including 67 nodes to model the approximate 160 balancing authorities in the United
States and Canada. These nodes include a set aside area for solar development in the
Southwest and also allow solar development to occur in areas of least cost. Appendix 1 describes how the balancing authorities are combined into these nodes.
Interconnections among the balancing authorities are modeled as pathways between the nodes with distances being calculated based on the geographic center of each node.
Two pathways are allowed between each node: a low voltage alternating current (LVAC) and high voltage direct current (HVDC) transmission pathway. The initial transmission capacity between regions is based on data supplied by the balancing authorities in their
FERC Form 714 filings, which list interconnections and capacity resources. The model balances the relative capital and operating costs and efficiencies of the two forms of transmission to determine the optimal interconnections to be constructed in each future year.
The model also balances the energy demand and available generation in each balancing authority, allowing for energy imports and exports between the authorities subject to transmission capacity limitations and other constraints. This balancing will occur four times a year for 30 years, modeling seasonal and diurnal peak demands. The available generation and transmission capacity are constrained to require power transfer to meet peak and total energy demand during both day and night in the summer and winter.
29
1. Allete (Minnesota Power) - 2. Alliant Energy-East - ALE 3. Alliant Energy-West - ALW MP 4. Ameren (Illinois Power Co. 5. Ameren Corporation Control 6. Arizona Public Service Control Area) - IP Area - AMR Company - APS 7. Associated Electric 8. Avista Corporation - AVA 9. Bonneville Power Cooperative, Inc. - ASC Administration, USDOE - BPA 10. California Independent 11. Cleco Corporation - CLC 12. Dairyland Power Cooperative System Operator - CIS - DLP 13. Duke Energy Carolinas, 14. Duke Energy Corp. - DEC 15. El Paso Electric Company - LLC - DUK EPE 16. Empire District Electric 17. Entergy Corporation/Services 18. ERCOT - ECT Company (the) - EDE (Entergy System) - EES 19. FirstEnergy Corporation - 20. Florida Power & Light 21. Hoosier Energy REC, Inc. - FE Company - FPL HE 22. Idaho Power Company - IPC 23. Imperial Irrigation District - 24. ISO New England Inc. - NE IID 25. Kansas City Power & Light 26. Los Angeles Department of 27. Louisville Gas & Electric and Company - KCP Water and Power - LW Kentucky Utilities - LGE 28. Michigan Electric Power 29. MidAmerican Energy 30. Nebraska Public Power Coordinated Center - MIP Company - MEC District - NPP 31. Nevada Power Company - 32. New York Independent 33. Northern Indiana Public NPP System Operator, Inc. - NY Service Company - NIP 34. Northern States Power 35. Northwestern Energy - NWE 36. Oklahoma Gas & Electric Company - NSP Company - OGE 37. Omaha Public Power 38. Otter Tail Power Company - 39. PacifiCorp - East - PE District - OPD OTP 40. PacifiCorp - West - PW 41. PJM Interconnection LLC - 42. Portland General Electric PJM Company - PGE
30
43. Progress Energy (Carolina 44. Progress Energy (Florida 45. Public Service Company of Power & Light Company) - Power Corp.) - FPC Colorado - PSC CPL 46. Public Service Company of 47. Public Service Company of 48. Puget Sound Energy, Inc. - New Mexico - PNM Oklahoma - PSO PSE 49. Sacramento Municipal 50. Salt River Project - SRP 51. Sierra Pacific Resources - Utility District (& City of SPP Redding Electric Utility) - SUD 52. South Carolina Electric & 53. Southern Company - SOC 54. Southwestern Power Gas - SCE Administration (DOE) - SPA 55. Southwestern Public Service 56. Sunflower Electric Power 57. Tennessee Valley Authority - Company (Xcel) - SPC Corporation - SEP TVA 58. Tucson Electric Power 59. Turlock Irrigation District - 60. Upper Peninsula Power Company - TEP TID Company - UPP 61. Westar Energy (KPL) - KPL 62. Western Area Power Admin - 63. Western Area Power Upper Missouri-East (Upper Administration - Colorado- Great Plains Region operat - Missouri Control Area WAE (Rocky Mtn Re - WR 64. Western Area Power 65. Western Area Power 66. Wisconsin Electric Power Administration - Lower Administration - Upper Company - WEP Colorado control area Missouri West (Upper Great (Desert Southwe - WDS Plains Regi - WW 67. Wisconsin Public Service Corporation - WPS Figure 3-2. Nodal Model Representation (pathways excluded for simplicity)
As required, the model calculates additional generating capacity and LVAC and
HVDC transmission capacity to meet energy demand or other requirements. These additions occur with the intent of allowing renewable energy generation and distribution while minimizing cost to the consumer. The intent of this structure is to model an open- access market, where the lowest cost of generation is achieved in each region.
Effectively, the model incorporates a typical power dispatch stack as illustrated in
Figure 3-2.
Based on this dispatch stack, the installed renewable energy generators always operate. Lower-cost coal-fired and hydro power will constitute the bulk of the power supply, and natural gas facilities will provide generation for peak demand periods. This dynamic may change over time as fuel prices change, solar and wind generator output becomes more predictable, and energy storage technology becomes affordable.
31
120
100
1. Lower cost coal, hydro, and nuclear 80 units supply baseload power
60
40 2. As system demand increases, 3. In times of peak demand, the most higher cost units are dispatched to expensive units (typically gas peaking meet the additional load, units) are dispatched, significantly increasing the average power price. elevating power price.
Operational Cost of Generation ($/MWh) of Generation Cost Operational 20
0 0 20,000 40,000 60,000 80,000 100,000 120,000 Total System Power Demand (MW) Notes: 1. The model assumes that power will be dispatched as in a deregulated market (no bilateral contracts). 2. In this market, solar and wind generators receive precedence and always operate. 3. The remaining power demand is met via this dispatch stack, and lower cost generators will be favored. 4. Regional power price will be the weighted average of the renewable power price and the clearing price from the dispatch stack. 5. Ultimately, the usage of storage mechanisms will also be incorporated into this stack. Stored power will be treated as other generators. Due to the expense, stored power will only be dispatched in critical situations because enough cheaper fossil fuel baseload power will be available.
Figure 3-3. Sample Power Dispatch Stack Operation
3.2.2 Indices
As stated earlier, the model projects infrastructure buildout over the course of 30 years, taking into account several variables, noted by the following indices:
• Year. The model accounts for years 0 (initial condition) to 15, in two-year
increments, leading to 30 years total.
32
• Season. The model balances electricity demand and generation and also ensures
that adequate generation and transmission capacity is available to meet demand in
all hours in summer and winter.
• Time of Day. The model balances electricity demand and generation to ensure
that adequate generation and transmission capacity is available to meet demand in
all hours in summer and winter.
• Generation. The model considers five generation types: wind power, solar PV,
other renewable energy (modeled as hydro power), coal, and natural gas
combined cycle.
• Regions. The model balances electric generation and demand by ensuring
adequate generation capacity is available in each balancing authority or that
adequate transmission is available for power transfer between balancing
authorities.
3.3 Model Formulation
3.3.1 Trade-offs
To identify the timing, geographical distribution, and physical make-up (i.e., type of generation and transmission) necessary to meet the requirements of the scenarios, the model incorporates several trade-offs:
• Local vs. remote power generation – the model balances the lower delivered
power cost associated with local generation versus the improved capacity factors
and therefore lower unit cost of energy associated with generation sourced from
regions with higher renewable resource.
33
• LVAC vs. HVDC transmission – the model balances the higher capital costs
associated with HVDC transmission with the lower cost and also lower efficiency
LVAC transmission.
• Near term capital cost expenditure vs. staged capacity installation – the model
accounts for annual limitations in total capacity installation and pending
regulatory requirements for renewable energy to determine the ideal installation
timeframe for new capacity.
• Renewable energy vs. fossil fuel-based generation – the model accounts for
Renewable Portfolio Standard (RPS) requirements, capacity factors, fuel prices,
and relative capital and operating costs in determining which generation
technology to dispatch to meet energy demand.
3.3.2 Decision Variables
These trade-offs were considered through the following decision variables:
• Generation Used. The output used in year (y), season (s), and time of day (t) in
region (r) from generator type (g), calculated in megawatt-hours (MWh).
• Generation Stored (not used but included in case future technology becomes
available). The output stored in year (y), season (s), and time of day (t) in region
(r) from generator type (g), calculated in MWh. The model is constrained such
that output stored during any time of day in any year and season must be used in
the coincident time of day in the same year and season.
• Spinning Reserve. The output used as spinning reserve in year (y), season (s), and
time of day (t) in region (r) from generator type (g), calculated in MWh. Spinning
reserve is set to 5% of total electricity demand and only generation derived from
34
non-renewable energy-based generators are allowed to fulfill this requirement,
serving as a proxy for needed ancillary services. The model is constrained such
that spinning reserve must be generated in the region in which it will be used or
through LVAC transmission lines with nearby regions.
• Generation Transferred via LVAC from One Region to Another. The output
generated in year (y), season (s), and time of day (t) in region (r) from generator
type (g) but transferred via LVAC to region (e), calculated in MWh.
• Generation Transferred via HVDC from One Region to Another. The output
generated in year (y), season (s), and time of day (t) in region (r) from generator
type (g) but transferred via HVDC to region (e), calculated in MWh.
• New Generation Capacity. The new generation capacity installed in each year (y)
and region (r), classified per generating technology (g), calculated in megawatts
(MW). First year capacity is determined per Energy Information Administration
Form 860.
• Existing Generation Capacity. The preexisting generation capacity from year (y-
1) and region (r), classified per generating technology (g), calculated in MW.
• New LVAC Transmission Capacity. The new LVAC transmission capacity
installed in year (y) between region (r) and region (e), calculated in MW. First
year capacity is derived per FERC Form 714 filings. To account for the difficulty
in obtaining right-of-way for local transmission, new LVAC connections is not
permitted between balancing authorities, but instead only upgrades of existing
connections are allowed.
35
• Existing LVAC Transmission Capacity. The preexisting LVAC transmission
capacity from year (y-1) between region (r) and region (e), calculated in MW.
• New HVDC Transmission Capacity. The new HVDC transmission capacity
installed in year (y) between region (r) and region (e), calculated in MW. First
year capacity is derived per FERC Form 714 filings.
• Existing HVDC Transmission Capacity. The preexisting HVDC transmission
capacity from year (y-1) between region (r) and region (e), calculated in MW.
3.3.3 Constraints
The model contains the following constraints:
• Total demand constraint – in each region in each year, the total energy generated
plus the total energy imported must equal the sum of the total exported energy and
the total regional energy demand.
• Peak demand constraint – in each region in each year, the total installed capacity
adjusted for expected capacity factor must be greater than or equal to the greatest
15-minute peak demand, accounting for required reserve margin.
• Generation capacity constraint – in each region in each year, the total generation
must be less than or equal to that able to be generated by the total installed
capacity adjusted for expected capacity factor.
• Transmission capacity constraint – in each region in each year, the transmission to
other regions must be less than or equal to that allowable by installed LVAC and
HVDC transmission.
• Flow constraints – the installed transmission and generating capacity in each year
must be additive to the installed capacity in the previous year.
36
• Construction limitations – in each region in each year, installed generation
capacity must not exceed capability for the market to supply materials. The
maximum installation capacity across all regions for all generation and
transmission types will initially be set to the maximum annual installation
capacity for similar technologies in the United States over the past 10 years. To
account for market response to demand for construction materials, in each year
the installation amount will be allowed to increase by 10% over the previous
maximum.
• Renewable portfolio standard (RPS) constraint – in each region in each year,
generation supplied from renewable sources must be greater than or equal to
regulatory required renewable energy amounts as defined by state renewable
portfolio standards.
• Non-zero constraints – all capacity and transmission additions must be greater
than or equal to zero.
3.3.4 Additional Inputs
Additional inputs to the model include the following:
• Electricity Demand. The electricity demand in year (y), season (s), and time of
day (t) in region (r), calculated in MWh.
• Peak Demand. The peak electricity demand in year (y), season (s), and time of
day (t) in region (r), calculated in MW.
• Required Reserve Margin. The amount of unused generation capacity in year (y)
in each region (r), defined as a percentage of overall capability. This input is used
37
as a proxy for required reserve capacity and is set equal to current reserve margin
in all regions.
• Expected Capacity Factor. The expected output as a percentage of maximum
generation capability in year (y), season (s), and time of day (t) in region (r) from
generator type (g). Capacity factors will be set to 0.9 for all generation types
except wind and solar due to availability of fuel supply. The 90% capacity factor
accounts for unexpected outages and maintenance shutdowns. Wind and solar
capacity factors are based on local weather data and available resource – output
performance data from leading equipment manufacturers.
• Projected Fuel Cost. The projected cost of fuel for electric power generation in
year (y) for generator type (g), calculated in $/MWh, based on energy projections
from the Energy Information Administration.
• Operating Cost, Variable and Fixed. The industry average variable and fixed
operating costs for generator type (g), calculated in $/MWh. Costs are adjusted in
year (y) to account for the expected inflation rate, which is set to the average
inflation rate over the past 20 years.
• Construction Cost. The capital cost, incurred as an overnight cost, for generator
type (g) or HVDC or LVAC transmission, calculated in $/MW. Costs are
adjusted in year (y) for inflation and discount rates.
• Construction Time. The time required for installation of generator or transmission
type. Costs are incurred as overnight costs in the year that construction begins, but
the generation or transmission are not modeled as operational until after the
construction time has elapsed. Construction times are based on industry averages.
38
• Renewable Portfolio Standard. The percentage of generation used and generation
imported to a region (r) in each year (y) that must be derived from renewable
energy resources, based on state regulatory requirements. Figure 3-3 illustrates
these constraints. If a balancing authority covers multiple states, the renewable
portfolio standard from the state in which most of the balancing authority lies is
applied.
Figure 3-4. State Renewable Portfolio Standards and Goals (Federal Energy Regulatory Commission, 2011)
• Transmission Losses. The losses in MWh, applied as a percentage of generation
transmitted. Losses are adjusted based on distance transmitted and type of
transmission used. Loss factors are based on industry data.
39
• Inflation and Discount Rates. Inflation and discount rates are applied to model
costs as base year costs. Inflation rates are set to the average inflation rate over
the past 20 years, and the discount rate will be assumed to be 5%.
3.3.5 Scenarios
The research project goal is to determine implementation plans for multiple scenarios, each envisioning a different regulatory and energy demand conditions to provide insight into potential renewable energy development pathways.
• Base Case. Initial generation and transmission capacity are set per the 2008
version of EIA Form 860 and FERC Form 714 filings. No RPS requirements are
incorporated. Annual energy demand and fuel costs are per Department of
Energy projections with trending into future years. No renewable energy capital
incentives or storage technologies are considered. Parametric analyses are run on
solar photovoltaic capital cost, fuel costs, and annual transmission construction
capacity.
• Increased Solar Power Focus. Initial generation and transmission capacity are set
per the 2008 version of EIA Form 860 and FERC Form 714 filings. Existing RPS
requirements are left in effect. Annual energy demand and fuel costs are per
Department of Energy projections with trending into future years. A 25% capital
cost reduction is put in place for generation capacity installed in the Southwest (a
designated solar-focused project area) due to expected economies of scale, but no
other renewable energy incentives are considered. Energy storage technologies
are not included. Parametric analyses are run on solar photovoltaic capital cost,
fuel costs, RPS levels, and annual transmission construction capacity.
40
• Improved Development Processes. Initial generation and transmission capacity
are set per the 2008 version of EIA Form 860 and FERC Form 714 filings.
Existing RPS requirements are left in effect. Annual energy demand and fuel
costs are per Department of Energy projections with trending into future years. A
25% capital cost reduction is put into place for generation capacity installed in the
Southwest (a designated solar-focused project area) due to expected economies of
scale, but no other renewable energy incentives are considered. Energy storage
technologies are not included. Initial construction time for transmission is
reduced from four years to two years. Parametric analyses are run on solar
photovoltaic capital cost, RPS levels, fuel costs, and annual transmission
construction capacity.
3.4 Data Requirements and Collection
The research requires the collection of data from multiple utilities and balancing entities regarding existing infrastructure, regulatory requirements, and electricity demand.
Utility demand data and Department of Energy, Energy Information Administration demand projections are also required. Installed capacities, interconnections, and demand data are available per the filings in Energy Information Administration Form 860 and
Federal Energy Regulatory Commission Form 714, which are available online. The
Energy Information Administration also publishes construction cost and fuel price projections for the power sector, which are used in this project.
3.5 Analytical Procedure
The collected data were compiled in a Microsoft Excel spreadsheet and transferred into the Mosel Xpress optimization software package. Chapter 4 describes the
41
optimization model, and Appendix 3 lists the model text as used in Mosel Xpress.
Appendices 1 and 2 contain a detailed description of model nodes and the model inputs, respectively.
Appendix 4 contains the optimization software outputs for all scenarios and parametric analyses. Specific attention was paid to the annual new build requirement for generation and transmission capacity under various scenarios.
3.6 Modeling Simplifications
The model described in Chapter 4 contains several simplifications to reduce computing time. These simplifications did not have significant impact on the results:
• Aggregation of control areas to a limited number of model nodes. To limit
computing time, the model aggregated small neighboring utilities into larger
groups (see Figure 3-1). This simplification had a large impact on predicted
overall power costs throughout the United States, primarily because intraregional
transmission costs and supply agreements between neighboring regions aren’t
included due to this simplification. The model, however, seeks to optimize
transmission buildout scenarios, and because the largest power flows occur
between larger utilities, clustering of local utilities did not impact projected
transmission line timing and locations. Decreasing the model granularity
therefore only led to the omission of smaller intra-regional transmission lines
• Use of regional geographic centerpoints to calculate transmission capital costs and
losses between regions. The use of geographic centerpoints for transmission line
pathways does introduce some inaccuracy in capital cost and operational loss
projections (i.e., because line distances are longer or shorter than would be
42
required), but this inaccuracy is acceptable for general optimization. Identifying
specific construction projects would require a more detailed U.S. power system
model.
• Limiting construction of low voltage transmission lines to those areas where lines
already exist. The model did not allow new transmission pathways to be
constructed but instead only allowed the transfer capacity of existing pathways to
be increased. The rationale for this simplification was that the large-scale
transmission buildout would require high-voltage projects rather than hundreds of
smaller low-voltage projects that would have difficulty obtaining environmental
and construction permits. The effect of this simplification on the results was
minimal. Some low voltage projects were eliminated, but the large scale projects
were unaffected.
• Limiting cross-border trade with Canada/Mexico. The model limits cross-border
trade with Canada and Mexico due to a lack of available data on their respective
power systems. Due to the relative magnitude of solar generation capacity
installation in the United States compared to the other countries, this
simplification had a minor effect on the projected buildout scenarios.
• Simplifying generation technologies. The model only considered wind, solar
photovoltaic, natural gas, coal and biomass generation technologies. This minor
simplification had a minor impact on transmission buildout because the projected
transmission needs were based on variable versus firm power and not on the
specific generation technologies available in those categories.
43
• Continuous function for block size. Projected generation and transmission
additions were not required to be in standard block sizes – for example, coal
power plants capacity additions could be less than 1 MW as opposed to standard
500 MW or larger sizes. Removing a lower limit on installation size would have
made model results unrealistic if it were seeking to identify specific projects,
however, because the model performed macro-scale additions over large
geographic areas and timeframes, this simplification did not skew the results of
the analysis.
44
Chapter 4 – Model Description
This chapter expands the general description of the model, detailing the optimization equations and the simplified representation of the U.S. electric power grid used in the research. The model attempts to minimize the total energy system cost in the United
States based on the regional power generation mix and several constraints. Table 4-1 lists the components of the model as described below, and Appendix 3 contains the full model text as input into the XPRESS-IVE ® optimization software.
Table 4-1. Model Components Model Details Component Objective TCost: Summation of the cost to generate and transmit electric power, including building of associated infrastructure to satisfy demand for the next 30 years Decision • GENERATED: Power generated in each region in each time period and year Variables • GENUSED: Power used in the region it is generated • SPINRESV: Power used to satisfy spinning reserve requirements in the region it is generated • TRANSGENAC: Power transmitted from one region to another via low voltage AC lines • TRANSGENDC: Power transmitted from one region to another via high voltage DC lines • TRANSRSVAC: Power transmitted from one region to another via low voltage AC lines, satisfying spinning reserve requirements • EXISTCAP: Existing power generation capacity • NEWCAPACITY: New power generation capacity installed in each year • LVAC: Existing low voltage AC transmission capacity between regions • HVDC: Existing high voltage DC transmission capacity between regions • NEWLVAC: new low voltage AC transmission capacity between regions installed each year • NEWHVDC: new high voltage DC transmission capacity between regions installed each year Parameters • ACLOSS: Losses for AC transmission between regions • DCLOSS: Losses for DC transmission between regions • PROXIMITY: Binary function, noting if regions are “close enough together” to merit building of transmission between them. All regions in Southwest allowed to connect to all other regions, and regions allowed to connect if geographic centerpoints within 1000 miles • LVACCOST: Cost to build AC transmission between regions • HVDCCOST: Cost to build DC transmission between regions • TRANFOM: Annual fixed operating and maintenance cost for transmission lines • PEAKDMND: Peak expected electricity demand in each time period in each region • TOTDMND: Total expected electricity demand in each time period in each region • CAPCOST: Capital cost to install new electric generation capacity • VOM: Variable operating and maintenance costs for different electric generation
45
technologies
Table 4-1 (cont’d) Model Details Component Parameters • FOM: Fixed operating and maintenance costs for different electric generation (cont’d) technologies • FUELCOST: Natural gas, coal, and biomass fuel costs • CAPFACT: Capacity factor for different electric generation technologies • RPS: Renewable portfolio standard – a requirement of percentage of electric demand satisfied by renewable energy. The model also uses WRPS SRPS and ORPS variables to cover wind, solar, and other technology-specific elements of the renewable portfolio standards • RESMARGIN: Reserve margin – a requirement that generation capacity exist in excess of that needed to generate peak demand Indices • REGIONS: 1..67, regions in the United States • YEARS: 0..15, model time steps from initial condition (Year 0) to Year 30, in two- year increments • GENTECHS: 1..5, respectively accounting for each generation type considered in the model – solar, wind, biomass, natural gas, and coal-based power • SEASONS: 1..2, accounting for summer and winter • TOD: 1..2, accounting for day and night Constraints • Electric transmission lines can only be constructed between regions with geographic (qualitative centerpoints within 750 miles on one another descriptions) • Low voltage electric transmission lines can only be constructed between regions already connected by low voltage transmission lines • Newly installed low voltage transmission line capacity is limited to twice the initial capacity • Transmission lines can only be constructed between regions (disallowing the model from assigning transmission from a region back to itself) • Power generated in a region plus power imported must equal demand • Local generation plus imports must also account for spinning reserve, which is 5% of total demand • Intermittent renewable energy sources (wind and solar power) cannot be used for spinning reserve • Generation for use, export, and spinning reserve cannot exceed the capacity of local power plants • Generation capacity of power plants plus import capacity must be sufficient to satisfy peak instantaneous demand • Transmission quantities between regions cannot exceed transmission line capacities • Capacity in each year is the sum of existing capacity plus newly installed capacities, accounting for construction times • Installation of new infrastructure cannot occur before construction time periods are completed • A limited amount of new construction can occur in each year • Decision variables must be non-negative • Renewable portfolio standards must be observed in select scenarios
46
4.1 Objective Function
The objective function is defined as follows with several subcomponents identified and broken out for illustration purposes:
Minimize Total_Power_Cost,
(4.1) = + + 1 + 2 where,
GENCAPCOST = the sum of the cost over all years (y) for building and
maintaining generation capacity for each generator (g) in
region (r)
GENCOST = the sum of the cost over all years (y), generators (g), and
regions (r) for producing electricity used in region “r”,
including costs for spinning reserve
TRANCOST1 = the sum of the cost over all years (y) for building and
maintaining high voltage DC transmission lines from one
region (r) to another region (e)
TRANCOST2 = the sum of the cost over all years (y) for building and
maintaining low voltage AC transmission lines from one
region (r) to another region (e)
Each of the subcomponents is defined below.