(PDF) Restructuring and renewable energy developments in California:using Elfin to simulate the future California power market

LBNL-41569 Restructuring and Renewable Energy Developments in California: Using Elfin to Simulate the Future California Power Market Chris Marnay, Suzie Kito,† Dan Kirshner,* Osman Sezgen, Steve Pickle, Katja Schumacher, and Ryan Wiser Environmental Energy Technologies Division Ernest Orlando Lawrence Berkeley National Laboratory University of California Berkeley, California 94720 † currently with MRW & Associates Oakland, California *Environmental Defense Fund Oakland, California June 1998 The work described in this study was funded by the Assistant Secretary of Energy Efficiency and Renewable Energy, Office of Utility Technologies of the U.S. Department of Energy under Contract No. DE-AC03-76SF00098. Contents Acknowledgments . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . vii Acronyms and Abbreviations . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . ix 1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1 1.1 Approach . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2 1.2 Organization of the Report . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4 2 Background . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5 2.1 Overview . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5 2.2 PURPA and Qualifying Facilities in California . . . . . . . . . . . . . . . . . . . . 5 2.3 The California Restructuring Decision . . . . . . . . . . . . . . . . . . . . . . . . . . . 8 2.4 Legislative Action: AB 1890 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 10 2.5 Alternative Policies for Fostering Renewables . . . . . . . . . . . . . . . . . . . . 10 2.6 The Policy Question . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 11 3 Modeling the California Pool . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 13 3.1 Overview . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 13 3.2 Elfin’s Pool Price Payments and ITRE Logic . . . . . . . . . . . . . . . . . . . . . 13 3.3 Key Assumptions in the California Pool . . . . . . . . . . . . . . . . . . . . . . . . . 17 4 Results of the Elfin Runs of the California Pool . . . . . . . . . . . . . . . . . . . . . . . . . 25 4.1 Overview . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 25 4.2 Scenarios and Policies Modeled in Elfin . . . . . . . . . . . . . . . . . . . . . . . . . 25 4.3 Results . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 27 4.4 Comparing Scenario Results . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 34 5 Conclusions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 43 References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 45 Appendix A: Detailed Results . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 47 Appendix B: The Expansion Planning Logic of Elfin . . . . . . . . . . . . . . . . . . . . . . . . . . 89 B.1 Traditional Cost-Minimizing Capacity Expansion Planning . . . . . . . . . . 89 B.2 The MC-ITRE Algorithm . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 90 B.3 Towards a Competitive Expansion Logic . . . . . . . . . . . . . . . . . . . . . . . . 90 B.4 Market Dispatch Logic . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 91 B.5 How Much Entry Will Occur . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 92 B.6 Finding the Best Plan . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 95 i CONTENTS B.7 Revised Algorithm . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 95 B.8 Conclusion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 96 Appendix C: Resource Options . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 97 C.1 Overview . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 97 C.2 Ranges of Costs and Other Operational Parameters . . . . . . . . . . . . . . . . 97 C.3 Summary of Parameters Used in this Analysis . . . . . . . . . . . . . . . . . . . 108 C.4 Offsets . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 112 Appendix D: Extreme Search Test for Market Equilibrium Plans . . . . . . . . . . . . . . . . 113 D.1 Procedure . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 113 D.2 Findings . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 114 D.3 Conclusion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 116 Appendix E: Simulation Difficulties in the B Policy Simulation . . . . . . . . . . . . . . . . . 117 Appendix F: Carbon Tax: Policy H . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 121 Appendix G: Cost Duration Curves for the California Pool . . . . . . . . . . . . . . . . . . . . . 123 ii Tables Table 2-1. Dependable Capacity for California Utilities. . . . . . . . . . . . . . . . . . . . 6 Table 2-2. Dependable Capacity for California Utilities’ Non-Utility Generators 7 Table 3-1. Peak Week Capacity Included in California Data Set for 1995 . . . . 18 Table 3-2. Capacity, Sales, Imports, and Exports for LADWP, SMUD, and IID . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 18 Table 4-1. Assumptions for Neutral, Good, and Bad Renewable Environments 26 Table 4-2. Summary of Elfin Pool Results with No Renewables Policies . . . . . 28 Table 4-3. Summary of Elfin Pool Results with a Straight Purchase Requirements and a Neutral Environment . . . . . . . . . . . . . . . . . . . . . 30 Table 4-4. Summary of Elfin Pool Results with a Banded Purchase Requirement and a Neutral Environment . . . . . . . . . . . . . . . . . . . . . 31 Table 4-5. Summary of Elfin Pool Results with Surcharge Policy with a Neutral Environment . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 33 Table 4-6. Total Generation (GWh) for N Cases . . . . . . . . . . . . . . . . . . . . . . . . 35 Table 4-7. Costs for N Cases . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 35 Table 4-8. Benefits for N Cases in Terms of Reductions of Emissions and Thermal Dependency . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 37 Table A-1. Resource Mix Under Scenario ON (GWh/a) . . . . . . . . . . . . . . . . . . . 48 Tables A-2-6. Key Indicators for Scenario ON . . . . . . . . . . . . . . . . . . . . . . . . . . . . 49 Table A-7. Resource Mix Under Scenario OG (GWh) . . . . . . . . . . . . . . . . . . . . 51 Tables A-8-12. Key Indicators for Scenario OG . . . . . . . . . . . . . . . . . . . . . . . . . . . . 51 Table A-13. Resource Mix Under Scenario OB (GWh) . . . . . . . . . . . . . . . . . . . . 54 Table A-14-18. Key Indicators for Scenario OB . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 54 Table A-19. Resource Mix Under Scenario AN (GWh) . . . . . . . . . . . . . . . . . . . . 57 Tables A-20-24. Key Indicators for Scenario AN . . . . . . . . . . . . . . . . . . . . . . . . . . . . 57 Table A-25. Resource Mix Under Scenario AG (GWh) . . . . . . . . . . . . . . . . . . . . 60 Tables A-26-30. Key Indicators for Scenario AG . . . . . . . . . . . . . . . . . . . . . . . . . . . . 60 Table A-31. Resource Mix Under Scenario AB . . . . . . . . . . . . . . . . . . . . . . . . . . 63 Tables A-32-36. Key Indicators for Scenario AB . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 63 Table A-37. Resource Mix Under Scenario BN (GWh) . . . . . . . . . . . . . . . . . . . . 66 Tables A-38-42. Key Indicators for Scenario BN . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 66 Table A-43. Resource Mix Under Scenario BG (GWh) . . . . . . . . . . . . . . . . . . . . 69 Tables A-44-48. Key Indicators for Scenario BG . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 69 Table A-49. Resource Mix Under Scenario DN (GWh) . . . . . . . . . . . . . . . . . . . . 72 Tables A-50-54. Key Indicators for Scenario DN . . . . . . . . . . . . . . . . . . . . . . . . . . . . 72 Table A-55. Resource Mix Under Scenario DG (GWh) . . . . . . . . . . . . . . . . . . . . 75 Tables A-56-60. Key Indicators for Scenario DG . . . . . . . . . . . . . . . . . . . . . . . . . . . . 75 Table A-61. Resource Mix Under Scenario DB (GWh) . . . . . . . . . . . . . . . . . . . . 78 Tables A-62-66. Key Indicators for Scenario DB . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 78 iii TABLES Table A-67. Resource Mix Under Scenario HN (GWh) . . . . . . . . . . . . . . . . . . . . 81 Tables A-68-72. Key Indicators for Scenario HN . . . . . . . . . . . . . . . . . . . . . . . . . . . . 81 Table A-73. Resource Mix Under Scenario HG (GWh) . . . . . . . . . . . . . . . . . . . . 84 Tables A-74-78. Key Indicators for Scenario HG . . . . . . . . . . . . . . . . . . . . . . . . . . . . 84 Table A-79. Resource Mix Under Scenario HB (GWh) . . . . . . . . . . . . . . . . . . . . 87 Tables A-80-84. Key Indicators for Scenario HB . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 87 Table C-1. Gas Combined Cycle Costs and Other Parameters . . . . . . . . . . . . . . 98 Table C-2. Gas Combustion Turbine Costs and Other Parameters . . . . . . . . . . . 99 Table C-3. Wind Plant Costs and Other Parameters . . . . . . . . . . . . . . . . . . . . . 100 Table C-4. Geothermal Costs and Other Parameters . . . . . . . . . . . . . . . . . . . . . 101 Table C-5. Solar Thermal Costs and Other Parameters . . . . . . . . . . . . . . . . . . . 103 Table C-6. DOE Solar Thermal Plant Costs . . . . . . . . . . . . . . . . . . . . . . . . . . . 103 Table C-7. Plant Costs and Other Parameters . . . . . . . . . . . . . . . . . . . . . . . . . . 104 Table C-8. Nuclear Plant Costs and Other Parameters . . . . . . . . . . . . . . . . . . . 105 Table C-9. Coal Gasification Combined Cycle Plant Costs and Other Parameters . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 106 Table C-10. Advanced Coal Plant Costs and Other Parameters . . . . . . . . . . . . . 107 Table C-11. Biomass Plant Costs and Other Parameters . . . . . . . . . . . . . . . . . . . 108 Table C-12. Summary of Capital Costs Assumptions . . . . . . . . . . . . . . . . . . . . . 109 Table C-13. Other Characteristics of Generic Technologies . . . . . . . . . . . . . . . . 110 Table C-14. Emissions Characteristics of Generic Technologies . . . . . . . . . . . . 111 Table D-1. Start Plans for the MEP Search Test . . . . . . . . . . . . . . . . . . . . . . . . 114 Table D-2. MEPs Found by Each Search . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 115 Table F-1. Summary of Elfin Pool Results with a Low Carbon Tax and a Neutral Environment . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 121 iv Figures Figure 2-1. Overall Resource Mix for California Power Generation 1994 . . . . . . . . 11 Figure 3.1. Gas and Oil Plant Commission Dates . . . . . . . . . . . . . . . . . . . . . . . . . . . 20 Figure 4-1. Cumulative New Capacity Under Scenario ON . . . . . . . . . . . . . . . . . . . 27 Figure 4-2. Cumulative New Capacity Under Scenario OG . . . . . . . . . . . . . . . . . . . 29 Figure 4-3. Cumulative New Capacity Under Scenario AN . . . . . . . . . . . . . . . . . . . 32 Figure 4-4. Cumulative New Capacity Under Scenario DN . . . . . . . . . . . . . . . . . . . 31 Figure 4-5. Cumulative New Capacity Under Scenario BN . . . . . . . . . . . . . . . . . . . 34 Figure 4-6. Total Social Cost of Generation in Comparison to ON Case . . . . . . . . . 36 Figure 4-7. ON Pool Price in 2025 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 38 Figure 4-8. AN Pool Price in 2025 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 39 Figure 4-9. Percentage Deviation in Costs and Benefits from ON Case in 2025 . . . 41 Figure A-1. Resource Mix Under Scenario ON . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 48 Figure A-2. Resource Mix Under Scenario OG . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 48 Figure A-3. Resource Mix Under Scenario OB . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 53 Figure A-4. Resource Mix Under Scenario AN . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 56 Figure A-5. Resource Mix Under Scenario AG . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 59 Figure A-6. Resource Mix Under Scenario AB . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 62 Figure A-7. Resource Mix Under Scenario BN . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 65 Figure A-8. Resource Mix Under Scenario BG . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 68 Figure A-9. Resource Mix Under Scenario DN . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 71 Figure A-10. Resource Mix Under Scenario DG . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 74 Figure A-11. Resource Mix Under Scenario DB . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 77 Figure A-12. Resource Mix Under Scenario HN . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 80 Figure A-13. Resource Mix Under Scenario HG . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 83 Figure A-14. Resource Mix Under Scenario HB . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 86 Figure B-1. Technology Profit Function . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 93 Figure C-1. Forecasts of Offset Costs . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 112 Figure D-1. MEPs Found by Each Search . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 116 Figure E-1. Per-kWH Subsidy to Wind - BN . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 117 Figure E-2. Net Total Subsidy to Wind - BN . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 118 Figure G-1. Cost Duration Curve for 2010 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 124 Figure G-2. Cost Duration Curve for 2015 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 125 Figure G-3. Cost Duration Curve for 2020 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 126 Figure G-4. Cost Duration Curve for 2025 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 127 Figure G-5. Cost Duration Curve for 2030 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 128 v vi Acknowledgments The work described in this study was funded by the Assistant Secretary of Energy Efficiency and Renewable Energy, Office of Utility Technologies of the U.S. Department of Energy under Contract No. DE-AC03-76SF00098. The authors would also like to thank Joseph Eto of the Berkeley Lab; Diane Pirkey, Jack Cadogan, and Joe Galdo at the U.S. DOE; Pat MacAuliffe, Angela Tanghetti, and Joel Klein at the CEC; and Francis Chapman at EDF for their invaluable assistance with this work. vii viii Acronyms and Abbreviations AB 1890 California State Assembly Bill 1890 ANPC Adjusted Net Present Cost BRPU Biennial Resource Planning Update CEC California Energy Commission CPUC California Public Utilities Commission CTC Competition Transition Charge DISCO Distribution Company DOE United States Department of Energy DSM Demand Side Management EDF Environmental Defense Fund Elfin Electric Utility Financial & Production Cost Model EPRI Electric Power Research Institute ER94 1994 Electricity Report of the California Energy Commission FERC Federal Energy Regulatory Commission ICEM Iterative Cost-Effectiveness Method IID Imperial Irrigation District IOU Investor Owned Utility ISO Independent System Operator ITRE Iterative Test for Resource Evaluation LADWP Los Angeles Department of Water and Power MEP Market Equilibrium Plan NPC Net Present Cost O&M Operation and Maintenance PG&E Pacific Gas and Electric PURPA Public Utilities Regulatory Policy Act PX Power Exchange QF Qualifying Facility under the terms of PURPA RFP Request for Proposal RPS Renewable Portfolio Standard RWG Renewables Working Group SCE Southern California Edison SDG&E San Diego Gas and Electric SMUD Sacramento Municipal Utility District SO4 Standard Offer #4 QF Contract TSC Total Social Cost WEPEX Western Power Exchange ix x CHAPTER 1 Introduction The California Public Utilities Commission’s (CPUC) historic decision of December 20, 1995 signaled the beginning of a new era for the state’s electric utility industry. Prompted by some of the highest electricity rates in the nation, the Commission’s decision started the phase out of the regulated world of protected utility customer bases and guaranteed returns on investment and replaced it with plans for a competitive environment in which electricity generators will market their product to intermediaries as well as directly to end-use customers. With some modification, the CPUC’s vision for a competitive power market was adopted and endorsed by the California State Legislature with the passage of Assembly Bill 1890 (AB 1890), a bipartisan electric industry restructuring bill signed into law by Governor Wilson on September 23, 1996. The new, restructured market for electric power will likely be cutthroat, price-driven, and commodity-oriented; however, the sustainability and environmental impact of generation technologies will be important to some customers, and green marketing strategies are emerging. Nonetheless, restructuring appears to pose significant problems for renewable energy generators, who stand to lose the existing support mechanisms that have helped make their relatively higher-cost resources financially viable. To the extent that AB 1890 offers support to the renewables industry, it is temporary in nature and must be renewed by future legislatures. Should renewable power cease to be viable under restructuring, California would lose the sustainability, environmental, and fuel supply diversity benefits that renewable energy offers. In this study, we attempt to model the California power sector in the next century in order to assess the potential impact of restructuring on renewably generated electricity. Specifically, we use the Elfin production costing and capacity expansion planning model to address three broad research questions: (1) Are new renewable resources likely to be viable in California’s competitive electric industry of the next century? (2) What policies could foster the growth of renewable resources? And (3), what are the costs and benefits associated with these policies? These questions are important not just in California, but across the country and internationally as many other states, regions, and countries have begun to restructure their electricity industry. In assessing the impact of restructuring on renewables, we focus on several new policies designed to support renewables in a market environment. While competition may favor renewable resources that produce during peak periods (e.g., solar), competitive power markets are likely to be detrimental to renewable resources in as much as the benefits associated with renewables (fewer emissions, resource diversity, reduced fuel price risk exposure) will no longer be considered in a regulated resource planning system. With restructuring, existing renewables support mechanisms—including regulatory proceedings and a variety of state and federal policies—will be modified and in some cases eliminated. New policies already under consideration to support renewables include: minimum 1 CHAPTER 1 renewables purchase requirements (MRPR), also called renewable portfolio standards (RPS), and surcharge-funded renewables programs. Carbon tax policies, if enacted, would also have a major impact on the viability of renewable generation. We attempted to model variants of all three of these policy options in this work. Provisions to support green marketing have also been debated and facilitated, but are not modeled or assumed in this study. 1.1 Approach Although our general approach is applicable to other jurisdictions considering restructuring, our focus in this report is on the effects of restructuring on renewables in California. California is among the states furthest along in restructuring, and a large share of national renewable generation is in the state. Data sets with detailed operational parameters for generation resources are publicly available, although are rapidly becoming outdated. And, the Elfin model has been specifically adapted for California’s regulatory process and has been updated to reflect changing market conditions. In keeping with our goal to look beyond the uncertainties associated with California’s four- year transition period to full competition, we focus on the quarter century after 2005. By looking beyond the transition period, we are able to put aside some complex problems and uncertainties associated with the transition, notably the recovery of stranded assets through the competitive transition charge (CTC) and the renewal of subsidies. Under AB 1890, utilities must retire their traditional ratebase, and the traditional arrangement under which a plant is constructed by utility investors with cost recovery in rates guaranteed for prudent investments disappears. The net outstanding ratebase is to be collected during the transition through the CTC. We assume that this process, along with utility divestiture, will be completed on schedule and that at the beginning of our forecast period, 2006, all generators can be treated as independent competitors. Even though generation companies will likely hold multiple stations and local monopoly power may exist, we assume that each generator bids at its marginal cost and accepts the market share that such bidding provides. Moreover, that transmission constraints may create local market power is overlooked in this work because the computational demands and the limitations of our model argue against consideration of this issue; further, the importance and persistence of transmission constraints well into the next century is hard to gauge. To determine the effects of restructuring on renewable energy resources, we use Elfin, an expansion planning and production cost model developed by the Environmental Defense Fund (EDF 1997). Elfin was developed in the context of California regulatory proceedings, but EDF has since modified the model to reflect expected market conditions. Traditional production cost models have simulated operations solely on the basis of minimizing cost. Elfin now estimates prevailing pool prices, including an energy payment to generators, and iteratively estimated payments to committed generators with pool prices below their bid prices, and to generators dispatched out of order for purposes of maintaining target spinning 2 CHAPTER 1 reserve levels. The current model version builds only resources that will be profitable over the lifetime of the project; that is, the net present value of the revenue stream exceeds the net present value of all costs for the project lifetime. More detailed information regarding Elfin algorithms is found in Appendix B. Building a data set for the future California pool was the major task undertaken during this project. The pool is assumed to consist of the electricity demand and generation resources of the current three major California investor-owned utilities (IOU), Pacific Gas and Electric (PG&E), Southern California Edison (SCE), and San Diego Gas and Electric (SDG&E). The source of the data for these three companies is the ER94 filings by the companies with the California Energy Commission (CEC). The new data set is built on the assumption that the demand forecast by the three current vertically integrated IOUs can be simply summed to represent the future pool demand, and that all of the assets reported by these companies represent the full complement of generating assets currently available to meet this load. Implicit in this assumption is the belief that bilateral contracting will behave in the same way as the pool, overall. That is, the fact that a large segment of demand will be met through bilateral agreements and that the prices for these sales need not directly reflect pool prices and payments can be overlooked, at least initially, because, overall, these transactions will reflect pool prices. Therefore, treating the whole pool as a single competitive market represents a reasonable first approximation of ultimate conditions. Imports to the state are treated as large generators that are also available to meet this demand. Imports from the Southwest are treated as a large coal generator, and imports from the Northwest are treated as a partial hydro and partial coal station. The outcome of an Elfin run is a profit-maximizing market equilibrium plan (MEP) that details when and how many units of possible generic generating technologies will be constructed. An MEP is a construction program for all available generating technologies that guarantees all entrants are profitable but that no additional generator can be built profitably. That is, no additional economic entry is possible and the industry remains in a sustainable equilibrium. Further, of the market equilibrium plans found by Elfin’s search algorithm, the one under which entrants make the most overall profit is deemed the best overall plan. In other words, investors in technologies cannot make more total profit with any other combination of new construction. Of course, like any economic concept of equilibrium, this approach represents an idealized outcome, but one that we believe approximates the likely real-world result. Having established a base plan, we run several scenarios with different combinations of assumptions and renewable energy support policies. For example, we vary the growth rate of gas prices as this will significantly affect the types of resources that Elfin finds profitable. That is, with high gas prices, fewer combustion turbines and combined-cycle plants will make money in the new market environment and this would favor wind, biomass, and other technologies. We also vary capital costs of renewable and gas-fired technologies and assess the costs and benefits of renewable policy options, including carbon taxes, renewable 3 CHAPTER 1 purchase requirements, and renewable subsidies. Costs are defined as the costs of renewable resources in excess of the less expensive resources that Elfin would have chosen. The benefits are measured in terms of reduced emissions, fuel diversity, and energy independence. We quantify these benefits to the extent possible. 1.2 Organization of the Report The rest of this report proceeds as follows: & In Chapter 2, we provide some basic background information on support for renewables in California, on the expected operation of the power pool and bilateral markets, and on the three key policy types modeled here. & In Chapter 3, we discuss the Elfin production cost and expansion planning model as well as key assumptions that we made to model the future California pool. & In Chapter 4, we present results from the successful Elfin models runs. & In Chapter 5 we discuss the implications of the study, as well as key areas for future research. & Additional information on results, Elfin’s expansion planning logic, and resource options can be found in the appendices. 4 CHAPTER 2 Background 2.1 Overview This chapter provides background information on the development of renewables in California, on the CPUC’s restructuring decision, and on key provisions of California’s restructuring legislation, AB 1890. We discuss the development of the renewables industry to highlight California’s historic commitment to the development and growth of the renewables industry and to illustrate the importance of government policies in fostering this growth. We consider three types of policies: renewable purchase requirements, surcharge- funding subsidies, and carbon taxes. We pay particular attention to the role of the independent system operator (ISO), the power exchange, and the utilities in the restructured environment, and we discuss how market participants will be paid for energy and other ancillary services they provide. 2.2 Public Utilities Regulatory Policy Act and Qualifying Facilities in California In response to the oil crisis of 1973 and in recognition of the United States’ dependence on fossil fuels, Congress enacted the Public Utilities Regulatory Policy Act of 1978 (PURPA). The purpose of PURPA was to open the market to non-traditional electricity supply options in an effort to: • increase the supply of electricity, leading to lower rates over time; • reduce reliance on oil and gas, with their high price volatility; and • increase system reliability by the presence of a large number of smaller facilities, since the probability that a number of facilities will fail at the same time is much smaller than the probability that one large facility will fail. In response to the passage of PURPA and the adoption of Federal Energy Regulatory Commission (FERC) regulations in 1978, the CPUC established standards for the purchase of power from qualifying independent power facilities (QF). On January 21, 1982, the Commission issued Decision 82-01-103, which developed “standard offer” contracts describing the terms and conditions associated with utilities’ obligations to purchase power from a QF at avoided cost. The standard offers represents a complete transaction, with prices, interconnection requirements, and other relevant terms. These offers were to be available to all QFs without exception or conditions. The commission developed four standard offers—three short-run standard offers based on shortage and running costs of existing utility resources, and a long-run offer based on the 5 CHAPTER 2 costs of a new utility resource that could be avoided by purchasing power from QFs. Because the short-run offers did not appear to provide the desired stimulus to the growth of the QF industry, the CPUC developed an interim long-run offer to be used while it developed the final long-run offer. The interim long-run standard offer contracts guarantee fixed-priced payments over long time periods (up to 10 years) to provide QFs with some certainty in the return on their investments. Primarily as a result of the fixed-price contracts, the QF industry has grown from a handful of projects to a mature industry representing well over 10,000 MW of installed capacity, although less than this amount in terms of dependable capacity (Kito 1992). Table 2-1 shows the dependable capacity of Pacific Gas & Electric (PG&E), Southern California Edison (SCE), San Diego Gas & Electric (SDG&E), and the total for California in 1994. The non- utility category can be further broken down and it shows QF capacity as well as some self generation (see Table 2-2). Table 2-1. Dependable Capacity for California Utilities PG&E SCE SDG&E State Total (MW) (%) (MW) (%) (MW) (%) (MW) (%) Oil & Gas 7,080 35% 8,746 43% 1,951 63% 22,579 40% Coal 0 0% 1,938 10% 0 0% 4,239 7% Geothermal 756 4% 0 0% 0 0% 1,036 2% Nuclear 2,160 11% 2,327 12% 430 14% 5,326 9% Hydro 4,540 22% 785 4% 0 0% 6,830 12% Pumped 1,188 6% 217 1% 0 0% 3,222 6% Storage Non-Utility 3,648 18% 4,177 21% 236 8% 8,297 15% Imports 852 4% 1,913 9% 486 15% 5,136 9% Total 20,224 11% 20,103 100% 3,103 100% 56,665 100% Source: “Electricity Report,” California Energy Commission, November 1995, Table 7-1. 6 CHAPTER 2 Table 2-2. Dependable Capacity for California Utilities’ Non-Utility Generation PG&E SCE SDG&E State Total (MW) (%) (MW) (%) (MW) (%) (MW) (%) Fossil 1862 51% 2,090 50% 162 69% 4,114 50% Cogeneration Biomass 627 17% 302 7% 7 3% 936 11% Geothermal 154 4% 701 17% 0 0% 855 10% Hydro 36 1% 13 0% 3 1% 73 1% Wind 167 5% 165 4% 0 0% 333 4% Solar 1 0% 364 9% 0 0% 365 4% Self- 801 22% 542 13% 64 27% 1621 20% Generation Total 3648 100% 4,177 100% 236 100% 8,297 100% Source: California Energy Commission 1995. “Electricity Report.” November. Table 7-1. Note: Dependable capacity for wind and hydroelectric is usually reduced by 80% to reflect the “firm” or “effective” capacity. This occurs because wind and hydroelectric power projects generally have low capacity factors compared to other resources. Because the ten-year fixed-price portion of the final long-term Standard Offer #4 (SO4) contracts was based on predictions of high future fossil fuel prices, the terms of the contract were very lucrative for QF projects. At its peak, the net statewide subsidy to renewable QFs under SO4 contracts probably exceeded a billion dollars annually. Once QFs reach the end of the ten-year, fixed-price portion of the SO4 disappears, however, the contract payment becomes variable and is based on the pool price plus an approximately 2.5 ¢/kWh capacity payment. Given that the earlier predictions of high fossil fuel prices were incorrect, and that the market price for electric power is now quite low, QFs with SO4 contracts face a significant revenue drop as they move from the fixed to the variable-price portions of their contracts. In general, revenues to renewable QFs are halved in this shift, and the sudden drop in revenues to renewable QFs is referred to as the “cliff.” By the end of the restructuring transition period, the beginning of our forecast period, virtually all renewable QFs holding SO4 contracts will have exhausted the long-run fixed-price portion of their contracts, and will have fallen off the “cliff.” In the base case of this study, we assume that renewables compete entirely without any subsidy; that is, all resources are left to compete on economics alone. 7 CHAPTER 2 2.3 The California Restructuring Decision 2.3.1 Reorganization of the Industry Independent System Operator In its December 1995 decision, the CPUC indicated that it intended to establish an independent system operator (ISO) to operate the utilities’ transmission systems. Previously, the investor-owned utilities (i.e., PG&E, SCE, and SDG&E) owned and operated their transmission systems. The primary reason for transferring control from the utilities to an ISO is that the utilities could use the transmission system strategically to their benefit and to the detriment of other market participants. The Commission, in its decision, enumerated four immediate and lasting advantages of an ISO: 1. The state will achieve a permanent and functional resolutions of transmission access disputes between the transmission-owning utilities and those dependent upon access to the system. 2. There will be a lasting efficiency gain resulting in cost savings due to combining the now distinct control function of many entities under the auspices of a statewide independent system operator. 3. There will be an operational efficiency inherent in a transmission network which has no economic interest other than fostering open access and the facilitation of supply from generators irrespective of their ownership. 4. There will be a consistent pricing system for the use of the common network facilities that prevents cost shifting and supports the competitive market. (CPUC 1995, p. 30). Power Exchange As outlined by the CPUC, the Power Exchange (PX) will be separate from the ISO and will “function as a clearinghouse by providing a transparent market for generation with hourly or half-hourly price signals evident to users and long-term investors” and thus “provide critical information vital to informed market decisions by generators, wholesale buyers, and end users.” (CPUC 1995, p.47). The PX will be open to all generators, including municipalities and out-of-state generators, and will work by accepting supply and demand bids and through this process determine time-differentiated market-clearing prices. The utilities are required to bid their generating capacity into the PX and to buy electricity from the PX to serve their customers. 8 CHAPTER 2 Utilities Utilities will continue operate their generation and distribution facilities and will be responsible for procuring energy for their full-service customers. Timetable The California ISO and the PX became operational in April 1998. WEPEX and Payment Provisions The CPUC left many implementation issues unresolved. In particular, the CPUC ordered the investor-owned utilities to submit proposals to the Federal Energy Regulatory Commission (FERC) for the design and operation of the ISO and the PX. The investor-owned utilities formed a working group, WEPEX, and submitted their initial proposal on April 29, 1996 to FERC. On November 29, 1996 FERC approved key parts of the WEPEX filing, conditionally authorizing the establishment of the ISO and PX, subject to a “Phase II” filing of March 31, 1997. For the purposes of this analysis, the WEPEX filings were used to delineate the responsibilities of the ISO and the PX and how bids would be submitted and market clearing prices for energy and ancillary services would be established. The PX will accept both supply and demand bids. The supply bids will include three parts: the energy bid, the no-load bid, and the start-up bid. The PX will determine the market clearing price, develop a preferred schedule, and submit this schedule to the ISO. The ISO will recommend changes to reduce transmission congestion and, based on this information, the PX will resubmit its schedule. After the final schedule is chosen by the ISO, the PX will provide the supply, demand, and price schedules to the buyers and sellers. This final price schedule is considered a financial commitment. Any unexpected changes in demand will be met with supply from the shorter-term hour-ahead market, at the prices of the hour-ahead market. 2.3.2 Proposed Minimum Renewable Purchase Requirement In its December 1995 decision, the CPUC indicated its support for a minimum renewable purchase requirement (MRPR). The MRPR would require that a certain percentage of a state’s annual electric use (or capacity) come from renewable energy. To implement the policy, a renewables purchase requirement (as a percent of energy or capacity sales) would be applied and enforced upon retail electric suppliers in the state. Individual obligations would be tradeable through a system of renewable energy credits (RECs), which is designed to add flexibility in meeting the MRPR. In advocating such a policy, the Commission left 9 CHAPTER 2 many issues unresolved: what is the appropriate level for such a MRPR, should requirements be equally applicable to all distribution companies (DISCO), what type of noncompliance penalty should be established, should the MRPR be established on a MW or MWh basis, is a transition strategy necessary, are floors for certain technologies appropriate, etc.? A Renewables Working Group (RWG) was authorized by the Commission to assess and provide recommendations on many of these issues. The RWG Report was submitted to the CPUC on August 23, 1996 (Renewables Working Group 1996). 2.4 Legislative Action: AB 1890 Although the CPUC initiated the process of electric power industry restructuring, it was the State Legislature’s passage of AB 1890 which will actually bring about restructuring in California because changes beyond the jurisdiction of the CPUC are required. In contrast to the CPUC’s MRPR approach, AB 1890 established a surcharge-funded program to partially support existing, new, and emerging renewables in the state between January 1998 and December 2001. The policy will sunset on December 31, 2001, and no additional long- term renewables policy is proposed. Total renewables funding over this four-year period will apparently equal $540 million. These funds are to be collected by the three largest investor- owned utilities (IOUs) through distribution surcharges. The California Energy Commission is to administer the distribution of these funds, with the provision that not less than 40 percent of the monies collected by the IOUs go either to existing or new and emerging renewable projects. AB 1890 directed the CEC to issue a report to the legislature outlining its recommendations on fund distribution. Finalized in late March, 1997, the CEC report calls for a variety of different approaches including: technology-specific production incentives for existing technologies, competitively auctioned production incentives for new technologies, multiple requests for proposals (RFP) using different mechanisms for emerging technologies, customer incentives and customer education (CEC 1998). 2.5 Alternative Policies for Fostering Renewables In addition to an MRPR and surcharge-based renewables support mechanisms, there are numerous other alternative policies to foster the development of renewable resources. These policies include: imposing criteria pollutant and/or carbon externality taxes on traditional generation sources (e.g., oil, natural gas, coal) and fostering the development of voluntary green markets. Here we examine only three policy types: an MRPR, a surcharge-based production credit, and a carbon-oriented externality tax. 10 CHAPTER 2 Figure 2-1. Overall Resource Mix for California Power Generation 1994 Coal Utility 19% Natural Gas 22% Nuclear Natural Gas 15% Cogen (CHP) Other 15% Renewables 2% Imported In-state Hydro Geothermal Hydro Biomass 8% 6% 3% 10% 2.6 The Policy Question The overall fuel mix of power generated for use in California is shown in Figure 2-1. The state’s approximately 250 TWh of electricity consumption is derived from a surprisingly diverse fuel mix. This is partially fortuitous, since there is considerable hydro in the state and nearby, because the state’s nuclear stations have operated successfully, and because there is a considerable geothermal resource. There are no coal stations in California, but coal is a factor both because some of the imported electricity is generated from coal and because California utilities own coal-generating capacity out-of-state. California has also actively sought diversity through its policies to foster renewable generation and cogeneration. In fact, California has more than 90 percent of the total U.S. generation of each of geothermal, wind, and solar, and about 15 percent of biomass. Looking into the next century, however, most of the nuclear capacity was commissioned in the 1980s and so will be nearing retirement by the 2000 teens, the hydro resource is fully developed and output is unlikely to increase significantly, and, most importantly, the pie is growing significantly. Population growth in the state is expected to be around 1.5 percent per year for some time, raising electricity demand. The net effect of these processes will be an inevitable increase in the thermal share in generation, making the state more dependent on imported fuel and less likely to meet Kyoto greenhouse gas emission reduction goals. The 11 CHAPTER 2 challenge is, in the face of falling nuclear and hydro shares in generation, to ensure that the share of the other carbon-free generation does not decline, and, preferably, increases. The key policy question under consideration in this study then is how can the tools we have available to simulate the future California market be used to evaluate the effectiveness of alternative policy instruments to enhance the renewable share of generation, and how can these simulation tools be improved? 12 CHAPTER 3 Modeling the California Pool 3.1 Overview In this chapter, we briefly discuss Elfin’s logic for calculating pool prices, including energy, commitment, and spin payments, and discuss key assumptions that were used to create the California pool data set. For the pool data set, we relied primarily on the Elfin data sets that were created for the 1994 Electricity Report (CEC 1995), with some modifications. The expansion planning options were developed using a variety of sources, including the Elfin data sets, EPRI (1993), and DOE (1994). 3.2 Elfin’s Pool Price Payments and Iterative Test for Resource Evaluation (ITRE) Logic 3.2.1 Elfin’s Pool Price Payments Energy Payments The energy payment concept in Elfin works on the assumption that the bid price of the most expensive unit operating in a particular time period (e.g., hourly) sets the pool price for all operators. Elfin does not currently calculate hour-by-hour marginal energy payments, but does calculate (a) the average marginal energy payment over as many as 99 subperiods from a typical week from up to 13 seasons, and (b) the payments that each resource block will receive during each time period. Because the computational requirements of calculating so many subperiods during the search for an MEP are prohibitive, only three subperiods and four seasons were used during the search, although, once the MEP has been found, subsequent estimations of pool prices were made using more subperiods and seasons. The average marginal energy payments in each subperiod are calculated by multiplying the marginal energy costs by the time marginal and summing these blocks. For example, if during the peak period one unit was marginal for 75 percent of the time at a cost of 3 ¢/kWh and one unit was marginal for 25 percent of the time at a cost of 5 ¢/kWh, the average marginal cost during the peak period would be 3.5 ¢/kWh [(0.75 × 3.0) + (0.25 × 5.0)]. This final amount would be reported in the Elfin results. Elfin also disaggregates these results by resource block—essentially providing the amount of time each resource block would be operating during each subperiod and the average payment this unit would receive for the time that it runs. While nothing in the algorithm requires it, in this work we assume that all generators bid their marginal cost. This is equivalent to assuming that no market power exists and no generator can gain by strategic bidding. 13 CHAPTER 3 Because generators are not perfectly reliable and there is always a positive probability that load cannot be met, the social cost of outage is properly included as an ultimate high cost supply resource. Therefore, in some instances, the energy payment will be greater than the cost of running existing units. This occurs when the system is expected to be unable to meet load because of unexpected outages, for example. In these instances, we have specified the value of energy not served as $1.00/kWh and this drives up the pool price considerably during peak periods. Another way to think of this is that as the pool develops, some customers may submit demand-side bids, indicating that they will curtail their energy at a price of say, of $1.00/kWh. If the ISO has no other options, they would curtail this load and the $1.00/kWh would be on the margin for a short period of time. While this assumption has a powerful effect on the peak pool price and, therefore, seems to be of great significance and worthy of a sensitivity analysis, in practice prices are so peaky that the overall effect on the year-round pool price is not great. Commitment Payments If power systems operated smoothly with generators started, stopped, and dispatched instantaneously, then the energy payment alone would ensure generators covered their fuel costs. However, operations in practice are complex and many other services than simple energy inflow must be provided by generators. Commit and spin payments reflect rewards for two such services. Because plants cannot be started and stopped instantaneously, generators are forced to run at times when the pool price is below their marginal operating costs. Elfin estimates a commit payment that ensures that during any subperiod, the last generator committed during the subperiod is made whole for its full operating costs. This commit payment is distributed to all generators. This approach has proven to be a poor representation of actual practice as it is emerging in California, in which generators self-commit, numerous generators are kept running to meet reliability requirements under private ISO must-run contracts, and the ISO purchases four ancillary services in open markets (Klein 1997). Simulating this system is not feasible for long-range planning, so we accept commitment and spin payments as a reasonable proxy. Spin Payments At times of high load it becomes more difficult for systems to maintain their spinning reserve requirement, which is a small margin of generating capacity kept fully functional and ready to replace any failing generator. When combustion turbines are started, thermal generators are backed down to provide the spinning requirement. As such, these generators are not 14 CHAPTER 3 creating as much revenue as they potentially could. Elfin makes these generators financially whole by providing them with a spin payment equivalent to their sacrificed revenues. 3.2.2 Elfin’s ITRE Logic For the purpose of this work, the expansion planning logic of Elfin has been considerably restructured. All expansion planning has solves two problems, an upper one in which new capacity is chosen, and a lower one in which existing plant is operated and the pool prices calculated. For the lower problem, Elfin uses a load duration curve model to dispatch resources. Elfin first creates a load duration curve for each specified subperiod and then dispatches the least expensive resources, subject to user-specified constraints (e.g., a unit must operate throughout the week if its committed to meet load any time during the week). For the upper problem, Elfin uses the Iterative Cost Effectiveness Method and the Iterative Test of Resource Effectiveness (ICEM-ITRE). ICEM tests the cost effectiveness of resources in each year of the optimization. ICEM will calculate the cost effectiveness of all resource options and build the most cost effective, as long as it is cost effective in its first year. ICEM continues to build the most cost-effective resources until no further resources are cost effective and then moves on to the next year. With ICEM, once a resource is built, it cannot be taken out of the resource mix. ITRE tests the cost effectiveness of resources in each year that resources are eligible for inclusion, and adds the resources in years in which the benefits are greatest. Net benefits might be greater in later years because technology costs could have decreased. ITRE’s advantage over ICEM is that ITRE might wait for a few years to build a new technology in order to capture larger benefits, whereas ICEM might build in the early years and thereby foreclose more cost-effective opportunities in the future. Also, ITRE can eliminate chosen resources and can find better solutions as a result. During this project, EDF has modified Elfin’s ITRE logic considerably in order to reflect new market conditions. Under old ITRE logic, Elfin added new plants to the resource mix if they reduced the total net present value of costs of running the entire system during the time period under consideration. In the pool, actual system operations are assumed to be identical to current approaches. That is, the ISO runs the system through time no differently than an existing IOU. What is quite different in the pool is the decision-making process for capacity expansion. Under the new ITRE logic, Elfin only adds new plants if they are able to cover their fixed and variable costs (i.e., if they break even). The most profitable projects are built first and ultimately all projects that break even, or better, are built. An example illustrating how the old and new logic could result in different resource plans is helpful. Imagine an instance where building a new combined-cycle unit would displace an inefficient unit and that these “production” benefits were more than offset by the annualized capital and operation and maintenance (O&M) costs of the new plant. In this case, overall costs are 15 CHAPTER 3 reduced and Elfin builds the new plant. However, imagine now that this new plant does indeed reduce production costs, but the revenues this plant receives are not sufficient to cover its annualized capital expenditures. Thus, under the new logic, even though this plant reduces overall system costs during the time period under consideration, the plant is not profitable and, therefore, will not be built. The paradigm adopted in the new ITRE logic is that of an investment community which invests in the most profitable resources first. This community will continue to invest until no more profitable opportunities exist. Unfortunately, unlike the simple cost-minimization algorithm of the previous ITRE, the search for plans in which no more profitable entry is possible is unstable and time consuming, and no unique solution exists. All MEPs must be found and the best chosen from among them. All MEPs represent sustainable market equilibria in which all entrants earn positive net present profits on constructed units, but no additional unlimited-entry generation units can be profitably built. Since all existing capacity will, in general, become less profitable as more capacity enters, its profitability cannot be considered in choosing the best MEP. Therefore, the choice of best MEP must be based on the profitability of unlimited-entry generation only. The paradigm, therefore, is one where investors in the unlimited-entry technologies (e.g., gas combined cycle) will find the MEP which maximizes their profits. As explained above, we adjusted Elfin’s iterative test for resource evaluation (ITRE) logic to select the most profitable resource options to be built, rather than those that minimize overall net present cost of a centrally planned utility. In practice, the search for a combination of new capacity construction is completed in two steps. In the first step, the traditional minimum-cost solution is found. This is the expansion plan that Elfin would select if it were run for the previous centrally planned, cost minimizing, regime. We start the search for a market-driven construction plan from this point because we know that in a perfect world, the cost-minimizing and profit-maximizing plans should be the same, or, at least very similar. In the second step, with this given starting plan, Elfin switches to its market simulation mode and calculates projected pool prices, including energy, commitment, and spin payments. Given these projected payments, Elfin evaluates resources based upon their profitability and builds the most profitable resource in its most profitable year. In this search, Elfin is restricted to adding one resource at a time. Elfin then recalculates projected pool prices, and again evaluates potential resource additions. If a previously added resource becomes unprofitable at any time during the search, it is deleted from the expansion plan. Elfin continues this iterative process until no additional resources can be profitably added to the system, and yet all the resources that are in the plan can be profitably built. Unfortunately, there may be several plans that meet this criteria, although, in practice, they tend to be similar plans. To select among these “market equilibrium plans” (MEPs), Elfin selects the one that is most profitable to the new entrants as the “best” plan. 16 CHAPTER 3 3.2.3 Limitations of Elfin There are several limitations associated with using Elfin to model the California pool. The computational demands of searching for MEPs is daunting, often taking days of computer time. A tradeoff of accuracy and computer time is inevitable. One of the major sacrifices was the need to run Elfin with only four seasons and three subperiods, meaning the year is represented by 12 load duration curves. While we ran the expansion plan with fewer time periods in order to use the program more efficiently, we then often, in separate production cost runs, developed expected pool prices using a greater number of time periods. Another limitation associated with Elfin is that the model does not take into consideration transmission constraints. In the near term, these constraints are important. Limits on transmission lines will prevent complete trading between areas and thus the system as whole will not be operated as efficiently as it could in the absence of these constraints and pool prices will differ across the state. While these constraints are important in the near term, we assume that these constraints will be largely eliminated by the year 2005. A third limitation of the Elfin model is that it does not simulate strategic bidding. For example, we expect that hydro resource owners will strategically bid into the pool in order to maximize revenues. This problem arises because Elfin bases its logic on actual costs, whereas the pool will operate based upon bids submitted by resource owners, who will likely bid strategically. If an external source of strategic bids were available, they could simply replace the cost data in Elfin and all could proceed as before. Unfortunately, there is no easy solution and thus we assume for the purposes of this project that no strategic bidding occurs. Bushnell and Borenstein (1996) look at market power and strategic bidding in the California pool using a Cournot mode and find that the California utilities may be able to exert considerable market power in the pool. 3.3 Key Assumptions in the California Pool 3.3.1 Resources Included in Data Set We include generating resources from the ER-94 data sets of Southern California Edison (SCE), Pacific Gas & Electric (PG&E), and San Diego Gas & Electric (SDG&E) in the data set (see Table 3-1). The CEC will not release the ER-96 data sets into the public domain. We do not include resources from Los Angeles Department of Water & Power (LADWP), Sacramento Municipal Utilities District (SMUD), or Imperial Irrigation District (IID) (see Table 3-2). The rationale for the exclusion of the municipalities and irrigation districts is that it would be difficult to determine the current and future amounts that they might bid and/or buy from the pool and that these amounts are likely to be relatively small compared to the size of the entire pool. While LADWP was expected to import roughly 19 percent of its energy for 1995, this energy is primarily from the Northwest and Southwest and not from 17 CHAPTER 3 other California utilities. In addition, LADWP is expected to export only a small amount of energy (approximately 2%). SMUD, by contrast, is expected to import nearly 75 percent of its energy, with 2,978 GWh to come from SCE and PG&E. While this represents a large amount of SMUD’s power (31%), this is only a small fraction of expected pool sales. Table 3-1. Peak Week Capacity Included in California Data Set for 1995 (MW) PG&E SCE SDG&E TOTAL Utility-Owned Nuclear 2,160 2,324 430 4,914 Oil & Gas 6,740 9,856 1,879 19,607 Oil & Gas‡ 874 (347 in) 517 (225,† in) Coal 0 1,943 0 1,943 Geothermal 677 0 0 677 Hydroelectric 5,015 1,580 0 6,595 Pumped Storage 1,158 217 0 1,375 Subtotal 2,309 QFs/Self Generation* 4,054 4,159 315 8,528 Imports 6,600 ‡ Long-term reserve † Huntington, Unit 4 *Includes non-firm QF/self-generation capacity Table 3-2. Capacity, Sales, Imports, and Exports for LADWP, SMUD, and IID LADWP SMUD IID 1995 Annual Peak (MW) 5,692 2,362 624 1995 Sales (GWh) 27,911 9,430 2,654 Imports (GWh) 5,223 ~6,983 na Exports (GWh) 626 0 na Source: ER 1994 files. In addition, we have represented all of the electricity inflows as generic imports from the Southwest and Northwest. These imports behave as monolithic units. The Southwest is entirely coal, while the Northwest has a hydro lower block and a coal upper block. A more complete analysis would estimate out-of-state resources and associated loads and allow utilities and other private power producers to bid in only excess energy into the California pool. 18 CHAPTER 3 In terms of actual resources, we included all existing utility assets, including nuclear, coal oil/gas, geothermal, hydroelectric, and pumped and battery storage units, although we did make some modifications. Notably, we took out all penalty factors to ensure that plants were dispatched economically.1 We included only nuclear, wind, and solar PV units as must runs, and, finally, we did not implement local reliability constraints. We have also included the qualifying facilities found in the three utilities’ data sets. We assume that the marginal costs for all of these units is zero or, in other words, that all energy that the QFs produce will be accepted into the pool. This assumption is least plausible for cogeneration units because these units will incur variable costs when they operate. However, in the absence of detailed information on the steam load of each generator, a zero marginal cost assumption is reasonable. There are, however, some resources that were in the utilities ER94 data sets that we excluded for this analysis. In particular, we exclude the biennial resource planning update (BRPU) resources because the majority of these contracts will not be implemented, DSM resources because the savings estimates are potentially uncertain, and utility-specific contracts because we chose to model these as blocks of energy being imported from the Northwest and Southwest. 3.3.2 Load Forecast and the Annual Load Profile We combine the load forecasts that were found in the ER94 data set for the planning areas of SCE, PG&E, and SDG&E, taking into consideration that the utilities had different peak days. The combined load for 1995 was approximately 38 GW, growing annually at a rate ranging from 1.3 to 2.2 percent. To simplify, we use a base case of 1.5 percent load growth and specify the high and the low load growth rates at three and zero percent, respectively. 3.3.3 Retirements Because the ER94 data sets only extend through 2013 and provide little information on expected retirements, and because we expect substantial numbers of retirements within the time frame of this study, we chose to set retirement dates for the nuclear facilities and oil and gas plants. For other types of plants (e.g., hydroelectric, coal, and QFs), we assume no retirement within the time frame of this study on the theory that this capacity will likely be replaced in kind upon retirement. For the nuclear units, we assume that they will be retired after 30 years of operation. This means that the nuclear plants that supply California will retire between 2013 and 2018. We 1 These penalty factors are usually added to account for aspects of operations not represented in the model. 19 CHAPTER 3 Figure 3-1. Gas and Oil Plant Commission Dates 1800 1600 1400 1200 1000 MW 800 600 400 200 0 1948 1951 1954 1957 1960 1963 1966 1969 1973 1976 1989 also assume a fairly ambitious retirement schedule for the oil and gas plants in the next century, in large part because many of these plants were built in the 1950s and 1960s (see Figure 3-1) and, therefore, will be quite old in 2010 and beyond. Assuming that these plants have engineering lives of 40 to 50 years, many of these plants will retire between 2000 and 2020. Exactly when these units retire will have important implications for the pace of new construction of generation in California. We considered a number of scenarios for retirements, including (1) imposing retirement at the end of the engineering life of the plants (2) using the utilities’ retirement scenarios through 2013 and imposing mandatory retirements at the end of the engineering lives for the remaining plants, (3) imposing no retirements, but tying repower to retirements and allowing Elfin to retire units economically, and (4) retiring units economically. Ultimately, we selected option 2. The ER94 data sets for PG&E and SDG&E provided data on retirements and plants put on short-term reserve, which we interpreted as early retirement. The ER94 data set for SCE did not provide retirement dates, but retirement dates for some of the plants were obtained from Elfin runs for the ER94 report. For the remaining plants, we implemented the following retirement scheme: 20 CHAPTER 3 100 MW or less: retired in 2014 if greater than 40 years old retired in 40th year for remaining plants 101 to 300 MW: retired in 2014 if greater than 45 years old retired in 45th year for remaining plants 301 MW or greater: retired in 2014 if greater than 50 years old retired in 50th year for remaining plants Combustion Turbines: retired in 2014 or in 30th year, whichever is earlier No retirement dates were specified for hydroelectric units, coal plants, or QFs. We assume that no hydroelectric plants retire during the time frame of this study. The ER94 data set does not provide retirement dates for SCE’s nearly 2,000 MW of coal capacity, roughly three quarters of which came on line between 1969 and 1971. In addition, SCE added 50 MW in 1979, 320 MW in 1986, and 50 MW in 1993. We assume that this capacity does not retire because it will be profitable and would be replaced when it is retired. We use the utilities’ estimates of QF retirement through 2013, after which we assume no units retire. We assume that cogeneration plants are tied to an industrial facility and would be replaced upon retirement. We treat geothermal capacity slightly differently. ER94 indicates that geothermal capacity degrades over time, and we assume that this degradation continues, with the steam reserves depleted by 2026. 3.3.4 Cost of Existing Resources We obtained much of the cost information for existing resources from the ER94 data sets, although we greatly simplified the fuel costs and developed fixed O&M costs for oil and gas plants. The variable O&M costs for existing plants varied considerably across plants and utilities. SCE used $0.00098/kWh for its oil and gas plants; SDG&E used $0.00014/kWh; and PG&E used values varying from $0.000004/kWh to $0.005/kWh. PG&E also specified variable costs for its geothermal units of about $0.001/kWh and for its coal plants, varying from $0.0014/kWh to $0.0041/kWh. The utilities generally used different gas price forecasts. We simplified all of the fuel costs and used base case, high, and low cost scenarios in our analysis. For natural gas, we assume that the price in 1995 varies from the summertime low of $1.78 to $2.25/GJ in December with real growth rates of zero, 1.5, and three percent. 2 For SCE’s existing coal units, we simplified the utility’s estimates of future coal prices, which ranged from $1.01 to $1.73/GJ 2 Recent forecasts predict gas price increases in the 1% per year range 21 CHAPTER 3 with real growth rates ranging from zero to 1.7 percent. We assume that the coal fuel costs represent contract prices and will not change during our study period. For new coal units, we assume that the coal price is $1.42/GJ in 1995 and escalates at 1.5 percent. In addition, the Northwest and Southwest imports are also tied to this generic coal price. For geothermal, we assume that geothermal operators pay for the steam at a rate of about $0.60/GJ, with real growth rates of zero, 1.5, and three percent. For nuclear plants, we assume fuel costs of approximately $0.005/kWh with a real growth rate of zero percent. Finally, we assume distillate fuel costs $4.58/GJ with real growth of 1.3 percent, and no fuel costs for cogeneration facilities, that is, cogenerators bid a zero price into the pool. Fixed O&M for existing plants was not contained in the ER94 data sets. We assume that existing oil and gas plants (excluding combustion turbines) have fixed O&M costs of $50/kW, which is roughly double that of new plants. We recognize that the fixed O&M costs will be variable and higher for progressively older plants, but did not want to add this additional level of complexity to our analysis. Note that the fixed O&M costs do not affect the pool price, only the retirement decision. In addition, we did not include fixed O&M data for any other existing plants other than the oil and gas plants. Resource Options Some of most important assumptions in any capacity expansion exercise are the specifications of generic resource options. Computational constraints require that the number of options be small, while the desire to represent the true range of technical choices available argues strongly for a large number of options. The ITRE method used by Elfin is, fortunately, much less limiting in this regard than dynamic programming based methods, which quickly reach the limits of computational feasibility when numerous options are considered. For this analysis, we included the following 12 potential resource options: • NG combined cycle • NG combined cycle repower • NG combustion turbine • advanced coal • coal gasification • biomass • geothermal • solar thermal • solar photovoltaic • wind farm • wind with combustion turbine • nuclear 22 CHAPTER 3 For each option, we specify low, medium, and high capital costs and a variety of operational parameters (e.g., variable and fixed O&M, heat rates, etc.). Appendix C provides a more complete description of the current cost ranges for these various technologies and explains how we chose the capital cost ranges for the technologies examined in this report. This appendix also presents emissions and offset values that are used. In later work, simulations containing many more options have been conducted, notably ones in which the wind resource has been represented by 36 specific sites, rather than generically (Sezgen, Marnay, and Bretz 1998). 23 24 CHAPTER 4 Results of the Elfin Runs of the California Pool 4.1 Overview In this chapter we present key results of our Elfin runs of the California market. A more expansive set of tables and figures detailing the results of each scenario modeled is presented in Appendix A. In the next section, we discuss the assumptions used for the policies modeled. These policies include imposing a minimum renewable purchase requirement, imposing carbon emission taxes, and providing subsidies to encourage the development of the renewable technology that appears closest to economic viability, namely wind. In the following section, we present the results of these runs. We report results for the construction of new renewable capacity. We also attempt to compare the benefits and social costs of renewables by reporting the emissions resulting from the various scenarios, thermal dependence, and natural gas usage, and the overall net present cost of running the system through the forecast period. We begin by presenting results of runs in which no policy is assumed and the effects of our three renewables environments, “neutral,” “good,” and “bad,” can be plainly seen. Each of our policy cases is modeled under each of these three environments, but here we focus on the results of a “no policy” case under each environment and on other policies modeled in only the “neutral” environment. Results of policies modeled in the “good” and “bad” environments are shown in Appendix A, and serve primarily as sensitivity results for equivalent runs conducted under neutral environment assumptions. In the subsequent section, we present more comparative details of the simulations. 4.2 Scenarios and Policies Modeled in Elfin Our approach is to conduct full Elfin ITRE runs for a limited set of policy cases under each of three economic environments. The three different future environments are: 1. conventional wisdom or best guess, called “neutral.” 2. favorable to renewable technologies (i.e., high gas prices, etc.), called “good” 3. unfavorable to renewables (i.e., low gas prices, etc.), called “bad” Growth in electricity demand in all three environments is assumed to be 1.5 percent per year. Table 4-1 specifies other aspects of these three economic climates. The actual dollar values of “high,” “low,” and “base” costs can be found in Table C-12 of Appendix C. In the good environment, coal and nuclear additions are not allowed in order to represent a future in which these resources are excluded on environmental or political grounds. 25 CHAPTER 4 Table 4-1. Assumptions for Neutral, Good, and Bad Renewables Environments Neutral (N) Good (G) Bad (B) Renewable Capital Costs Base Low High Gas-Fired Capital Costs Base Base Low Nuclear & Coal Included Yes No Yes Gas Price Growth (% per year) 1.5 3.0 0.0 Under each of the renewable environments, we modeled a number of policy cases in which efforts to promote the development of renewable technologies are estimated. We report here on the base case plus four example policy cases, as follows: • Policy 0: no policy (i.e., no new or existing policies present) • Policy A: non-hydro renewable purchase requirement (15% generation) • Policy B: surcharge policy ($620M/year to lowest cost renewable) • Policy D: 15 percent MRPR with technology bands (set at current market shares) Note that our focus is on the quarter century beginning in 2005. As noted above, we assume no special preferences designed to benefit renewables are in place, other than the policies explicitly being modeled. The three environments and five policy types combine for a total of 15 scenarios, although we were not able to simulate all cases satisfactorily, as is explained below. Using the adjusted Elfin logic described in Appendix B, we attempted and completed full Elfin runs for most of these cases. Not all of these scenarios are within the bounds of credibility, however. A wide range of improbable cases has been attempted due to our desire to (a) demonstrate the viability and potential pitfalls of our approach, and (b) identify some of the limiting or boundary cases. All cases are named with two-character names, for example, AN implies policy A, the purchase requirement is implemented in the (N)eutral renewables environment. In this chapter, results are reported for all three Policy 0 scenarios, and for the AN, BN, and DN scenarios. Simulating Policy B created some special simulation problems that are described in Appendix E. Policy H was particularly troublesome and has been relegated to Appendix F. However, summary results are presented for all the N scenarios. 26 CHAPTER 4 4.3 Results 4.3.1 Policy 0: No Policy In our Policy 0 cases, renewable generation competes head-to-head with all other technology options. In this first set of runs, we sought to determine whether unsubsidized renewable generation would be viable in the California market under our best estimate of current costs and trends, and given other circumstances that are both favorable and unfavorable to renewables as represented in the good and bad environments. Results for the final forecast year, 2030, are summarized in Table 4-2, and cumulative new capacity construction is shown in Figure 4-1. Despite large new total capacity additions of approximately 40 GW, no renewable capacity is added. In fact, all new capacity is of just three types, repowers of existing units, new combined cycles (CC), and new combustion turbines (CT). This result reflects conventional wisdom that only new gas-fired capacity can compete in the future California market, absent subsidies, taxes, or significant increases in the gas price. The steady increase in thermal dependency results in a steady increase of carbon emissions. Per kWh carbon emissions rise from 113 g in 2010 to 126 in 2030 and because of increasing generation, total carbon emissions are 50 percent higher. Here then is the heart of the California dilemma. Absent policy intervention, while in many ways clean compared to electricity generation elsewhere, the power sector will become increasingly thermal dependent and increasingly undesirable from a greenhouse gas perspective. Figure 4-1. Cumulative New Capacity Under Scenario ON MW 60000 Rep ow er 50000 CT CC 40000 30000 20000 10000 0 2010 2015 2020 2025 2030 Ye ar 27 CHAPTER 4 Table 4-2. Summary of Elfin Pool Results with No Renewables Policies Neutral (ON) Good (OG) Bad (OB) 2030 Cumulative New 0 85 0 Renewable (GW) 2030 NOx Emissions (kt) 224 188 (84%) 152 (68%) 2030 Carbon Emissions (Mt) 40 21 (53%) 37 (94%) 2030 % Thermal 81% 36% 81% 2030 Gas Consumption (EJ) 1.6 0.5 (32%) 2.0 (125%) NPV System Costs (109 $) 132 153 (116%) 102 (78%) Table 4-2 indicates that renewable technologies would be built in California only under the favorable circumstances of the good environment (i.e., high gas prices and low renewable costs), but in this environment, a remarkable 85 GW of wind capacity is chosen, as can be seen in Figure 4-2. Because of the low capacity factor of wind generation, more capacity of this technology than thermal generation must be built to meet energy requirements. Wind generation is first built in 2016, by which time wind capital costs have fallen to $792/kW.3 Furthermore, wind is the only renewable technology adopted, a result that derives directly from the fact that no artificial limits are imposed on the availability of wind generation at the assumed cost.4 In terms of NOx emissions, the good environment results in a 16 percent reduction from the neutral case, while the bad environment produces a 32 percent reduction. The bad environment results in greater NOx emissions reductions because low gas price growth (i.e., 0%) leads to greater gas generation displacing coal which has significantly higher NOx emissions per kWh generated. In contrast, the good environment results in a 47 percent reduction in carbon emissions, whereas the bad environment delivers only a six percent reduction, surprisingly little benefit compared to the reduction in NOx emissions. The good environment’s sharp drop in carbon emissions is due to the increase in wind generation. In the neutral and the bad environments, California remains dependent upon gas and other thermal technologies, such as coal, while the good case shows a marked reduction in thermal dependence and gas consumption. In both the neutral and bad environments, 81 percent of 3 Note that in Elfin, technical change is represented by predetermined declines in costs and/or improvements in operating characteristics. There is no internal logic that relates costs and performance of technologies to indicators of technical experience, such as installed capacity. 4 In new work (Sezgen, Marnay, and Bretz 1998), the availability of new wind generation to the California pool relative to its cost is being studied to better understand constraints on the supply of wind. Of course, other constraints exist on the construction of generic resources, which should also be studied. 28 CHAPTER 4 Figure 4-2. Cumulative New Capacity Under Scenario OG MW 120000 Win d Rep ow er 100000 CC 80000 60000 40000 20000 0 2010 2015 2020 2025 2030 Ye ar the electricity generated comes from thermal technologies, while in the good case, only 36 percent of the energy comes from thermal technologies. Gas consumption in 2030 rises well above today’s level of about 0.43 EJ in all three environments. It reaches 1.6 EJ in the neutral case, climbs to nearly 2.0 EJ in the bad case, and to 0.5 EJ in the good case. System costs are driven mostly by gas price growth. The bad environment has lower costs because we assume a gas price growth of zero percent for this scenario, whereas we assume 1.5 percent gas price growth in the neutral case. The good environment has higher costs because we assume gas price growth of three percent, and this assumption leads to the construction of a technology with higher capital costs, namely wind. The 0, or no policy case under the neutral environment (0N) serves as our base case for subsequent policy comparisons. The good and bad environments serve primarily to demonstrate that the model responds to test stimuli and shows how results might vary under diverse economic conditions. As with all simulations run, detailed results can be found in Appendix A. 29 CHAPTER 4 4.3.2 Policies A and D: Purchase Requirements With and Without Technology Bands In policy case A, we modeled a minimum renewable purchase requirement. Under this policy, we assume that 15 percent of energy must come from the cheapest non-hydroelectric renewable technology—wind. This compares with a current non-hydro renewables California energy share of about 11 percent (Renewables Working Group 1996). In general, existing renewable energy projects are assumed to persist into the indefinite future because we were unable to determine whether these technologies would fold for economic reasons or because they had reached the end of their engineering lives. The one exception is geothermal capacity owned by PG&E, which gradually degrades over time, and for which data were available in the PG&E data set. As a result of these assumptions, existing renewables meet a slowly declining percentage of the 15 percent purchase requirement, and new wind generation, clearly the least expensive technology under our assumptions, meets the remaining portion of the requirement. In Policy D, we modify the MRPR by adding purchase requirements for specific renewable technologies. These technology bands are designed to ensure purchases of less competitive and more expensive renewable technologies, such as biomass and PV. We set the technology bands in Policy D at levels consistent with current market share. The results of these runs are summarized in Tables 4-3 and 4-4, and cumulative capacity construction is shown in Figures 4-3 and 4-4. Table 4-3. Summary of Elfin Pool Results with a Straight Purchase Requirement and a Neutral Environment 15% Purchase Requirement No Policy (0N) Without Technology Bands (AN) Cumulative New Renewable (GW) 0 17 2030 NOx Emissions (kt) 224 224 (100%) 2030 Carbon Emissions (Mt) 40 36 (91%) 2030 % Thermal 81% 71% 2030 Gas Consumption (EJ) 1.6 1.3 (84%) NPV System Costs (109 $) 132 135 (103%) 30 CHAPTER 4 Table 4-4. Summary of Elfin Pool Results with a Banded Purchase Requirement and a Neutral Environment 15% Purchase Requirement No Policy (0N) with Technology Bands (DN) Cumulative New Renewable 0 7 (GW) 2030 NOx Emissions (kt) 224 224 (100%) 2030 Carbon Emissions (Mt) 40 37 (93%) 2030 % Thermal 81% 73% 2030 Gas Consumption (EJ) 1.6 1.3 (86%) NPV System Costs (109 $) 132 148 (112%) Figure 4-3. Cumulative New Capacity Under Scenario AN MW 60000 W in d Rep ow er 50000 CT CC 40000 30000 20000 10000 0 2010 2015 2020 2025 2030 Ye ar In both runs, NOx emissions are unchanged. This is a disappointing result and one not easily explained. One would certainly expect the fall in thermal generation to result in measurable NOx emissions reductions, but this is not the case. It could be that the generation being displaced is from clean new plants, and, at the same time, generation is shifting among generators in such a way that emissions from existing resources are increasing. This could also happen if the share of generation is shifting from air basins with more stringent regulations towards ones with less stringent regulations. However, this is a puzzle that cannot be resolved here. In future work, a more careful accounting should be made of the 31 CHAPTER 4 Figure 4-4. Cumulative New Capacity Under Scenario DN MW Bio 60000 PV Sola r-th erm a l 50000 Geo-th erm a l W in d 40000 Rep ow er CT 30000 CC 20000 10000 0 2010 2015 2020 2025 2030 Ye ar origin and shift in NOx emissions between these scenarios. In contrast, there is a slight drop in carbon emissions. Both policies result in reduced thermal dependence and gas consumption. However, Policy A does appear to result in more new renewable construction and lower carbon emissions, thermal dependence, and gas consumption than Policy D. This is not surprising given that the “bandless,” least-cost MRPR approach of Policy A results in greater renewable energy generation. Policy A produces 33 TWh of new renewable electricity in 2030, all wind, whereas Policy B produces 28 TWh because of the higher overall cost of the mixed renewable generation compared to wind. In other words, while both policies result in higher renewable generation, Policy A is more effective than D from the point of view of pure renewable development because gas competes more effectively against the mix of technologies required by Policy D than against wind only in Policy A. While both policies are more costly than the base case, the purchase requirement with technology bands is decidedly more expensive than the straight purchase requirement. Again, this is not surprising given that one of the goals of Policy D is continued purchases from more expensive renewable technologies. However, in both of these cases, sustained orderly development of renewable technology is assured, and this should be recognized as a benefit because this will avoid a “boom and bust” cycle. 32 CHAPTER 4 4.3.3 Policy B: Surcharge Policy In Policy B, a $620 million per year surcharge-based policy was simulated. While there are many ways to distribute surcharge monies collected to fund renewables programs (Wiser and Pickle 1997), we assume that monies are simply distributed to the least-cost renewable developers via an auction mechanism. It should be noted that the level of surcharge collection in this policy is quite large compared to many current surcharge policies and proposals. In California, for example, a total of $540 million is to be collected for renewables over a period of four years. Again, however, our goal here is to merely assess the viability of our approach and test some limiting cases. We did not actually choose the value of $620 million per year, as will be clear from the discussion below. However, we did seek to simulate an overall subsidy level in this general range because conventional wisdom during California’s MRPR debate was that the cost of a 15 percent MRPR could well be in this range. We therefore hoped to produce a case that would be roughly comparable to our MRPR simulations. Obviously, both A and B cases are somewhat extreme as we were seeking to exercise the model and identify boundaries We hope to assess smaller surcharge levels in future work. Our results are shown in Table 4-5 and Figure 4-5. Table 4-5. Summary of Elfin Pool Results with Surcharge Policy and a Neutral Environment Neutral (ON) $620M$/a Surcharge (BN) 2030 Cumulative New Renewable (GW) 0 29 2030 NOx Emissions (kt) 224 224 (100%) 2030 Carbon Emissions (Mt) 40 34 (85%) 2030 % Thermal 81% 64% 2030 Gas Consumption (EJ) 1.6 1.1 (73%) NPV System Costs (10 9 $)) 132 132 (100%) Compared with the base, or neutral, case, a $620M per year surcharge policy results in 29 GW of new renewable construction, 15 percent lower carbon emissions, thermal dependence of 64 percent and 27 percent less gas consumption. As in previous policy scenarios, however, there is no change in NOx emissions. 33 CHAPTER 4 Figure 4-5. Cumulative Capacity Construction Under Scenario BN MW 70000 Wind Repower 60000 CT 50000 CC 40000 30000 20000 10000 0 2010 2015 2020 2025 2030 Year 4.4 Comparing Scenario Results In this section, the results of all policy cases are compared directly. Table 4-6 reports the total generation for each case in each of five benchmark years. Elfin does not have hard generation or capacity constraints. That is, the level of construction of new capacity and the overall output are chosen economically, in this case, such that the returns to investors in new capacity is maximized. No unprofitable investments are made. Dispatch is organized on traditional cost-minimizing principles and failing to serve load under the assumed cost structure, notably the cost of unserved energy itself, is a less costly dispatch outcome, then it is chosen. In other words, adequate capacity to meet reserve requirements is not necessarily built, and load is not necessarily met. In comparing results, therefore, an important first question to consider is how much demand went unserved. The one notable result in Table 4-6 is that generation is lower in all years for the HN carbon tax case that is fully reported in Appendix F. This result emerges because costs are significantly higher in the HN case, and, therefore, unserved energy appears as a more attractive option to Elfin. 34 CHAPTER 4 Table 4-6. Total Generation (GWh) for N Cases 2010 2015 2020 2025 2030 ON 255,921 276,039 298,532 321,070 346,167 AN 255,921 276,037 298,532 321,069 346,153 BN 255,920 276,034 298,524 321,047 346,183 DN 255,921 276,038 298,532 321,070 346,170 HN 255,911 276,016 297,986 320,012 343,450 Table 4-7 shows the overall costs for the base and four policy cases, and Figure 4-6 shows how costs deviate from the ON case. The first column shows the internal net present cost of the scenario. This value represents the total cost of running the generation system through the end of the forecast period, including the cost of unserved energy. Table 4-7. Costs for N Cases NPC ANPC TSC per Generation Subsidies to M$ (1995) M$ (1995) per year B Cases (¢/kWh) (¢/kWh) 2006-2055 2006-2055 2010 2015 2020 2025 2030 2010 2015 2020 2025 2030 ON 132,002 132,002 1.6 2 2.2 2.5 2.7 AN 135,370 135,370 1.7 2.1 2.4 2.6 2.8 BN 132,001 139,506 1.6 2.1 2.5 2.7 2.8 2.7 2.0 1.8 1.5 th DN 148,382 148,382 1.8 2.2 2.6 2.8 3 HN 237,312 236,093 2.1 3.2 3.9 4.3 5.3 NPC (Net Present Cost) ANPC (for Benefits and Emission Taxes Adjusted Net Present Cost) TSC (Total Social Cost) 35 CHAPTER 4 Figure 4-6. Total Social Cost of Generation in Comparison to ON Case difference in cent/kWh 3 AN 2.5 BN DN HN 2 HN (excl. ENS) 1.5 1 0.5 0 year 2010 2015 2020 2025 2030 The second column reports an adjusted net present cost. The purpose of this value is to represent a more realistic picture of scenario costs by adjusting the NPC for the costs of subsidies and taxes faced by the electricity sector, which are transfer payments and not true economic costs. The AN and DN cases involve no tax or subsidy, so the adjustment is zero. However, in the BN case, because wind generators are receiving a subsidy, the true cost is higher than calculated internally by Elfin. On the other hand, true costs are lower in the carbon tax case because the tax revenue does not represent a true cost because the tax revenue is available for other purposes. Figure 4-8 summarizes these results by reporting deviations for each policy scenario from the 0N case. Again, the HN case stands out, and, as discussed above, has to be regarded with suspicion. Here however, it serves a useful purpose. Because costs are high in this simulation the amount on unserved energy is high, and this raises overall costs. In the figure the effect of this phenomenon is shown by reporting HN exclusive of the unserved energy cost. This just reinforces the notion that in Elfin, decisions are economic. In this simulation, a significant share of demand went unmet, and a cost was incurred as a result. The following columns show the per-kWh generation costs after the adjustment. These are average production costs and do not reflect the costs of transmission, distribution, billing, etc. For reference, Table 4-8 shows the comparative emission and thermal dependency results, which have been discussed above. 36 Table 4-8. Benefits for N Cases in Terms of Reduction of Emissions and Thermal Dependency C (Mt) NOx (kt) Thermal Dependency (%) 2010 2015 2020 2025 2030 2010 2015 2020 2025 2030 2010 2015 2020 2025 2030 ON 26.3 30.9 34.7 37 39.7 223 226 227.8 226.5 224.3 60 71 77 79 81 AN 24.4 28.7 32 33.5 36.1 222.2 225.8 227.2 225.3 223.9 53 64 69 69 71 AN %red. -7.5 -7.2 -7.8 -9.4 -9.1 -0.4 -0.1 -0.3 -0.5 -0.2 BN 25.8 27.8 28.2 30.7 33.6 225.7 225.4 223.5 223.8 223.9 57 61 58 61 64 BN %red. -1.9 -10.1 -18.7 -17.1 -15.2 1.2 -0.2 -1.9 -1.2 -0.2 DN 25.1 29.3 32.6 34.7 36.9 221.0 224.7 226.8 227.0 223.7 55 65 70 72 73 DN %red. -4.9 -5.3 -5.9 -6.2 -6.9 -0.9 -0.6 -0.5 0.2 -0.3 HN 16.4 12.2 16.8 18.9 21.7 177.3 138.5 184.0 180.1 184.9 53 60 63 66 68 HN %red. -37.7 -60.4 -51.4 -48.9 -45.4 -20.5 -38.7 -19.2 -20.5 -17.6 CHAPTER 4 Before presenting these benefits and costs in graphical form, it would be interesting to focus briefly on market electricity prices. As has been emphasized above, these prices are the Holy Grail of competitive market simulation. With a reasonable estimate of prices in hand, other aspects of simple project analysis are relatively trivial, but without them no accurate revenue forecast is possible. Figures 4-7 and 4-8 show some simplified price results. In this instance, the simulation has been run using only three subperiods for each of 12 typical weeks. The peak subperiod is each afternoon from 12:00 to 18:00 and the weekend is 18:00 Friday to 8:00 Monday. All remaining hours are off peak. Remembering that in these simulations, all dispatch is assumed centralized, we will refer to these prices as pool prices in this section. In actual fact, of course most prices will be set by bilateral contracting, and PX prices will serve primarily as a basis for comparison. Figure 4-7. ON Pool Price in 2025 250 200 ($1995/MWh) 150 100 50 0 1 2 3 4 5 6 7 8 9 10 Month 11 12 Weekend Off-Peak Peak One immediately interesting aspect of the results is how flat pool prices are overall. Outside of the later summer peak periods, pool prices stay within the range of $26 to 36 per MWh, the highest non-peak prices coming in the early winter period. The flatness of prices is somewhat of a surprise given the nature of the California electrical system. Load varies considerably, both diurnally and seasonally in the state, and the hydro resource, both in-state and imported, is also highly seasonal. However, the pool price appears to be dominated by the fact that traditional gas-fired resources almost always end up as the price-setting bidders, and, consequently, prices never collapse. 38 CHAPTER 4 Figure 4-8. AN Pool Price in 2025 250 200 ($1995/MWh) 150 100 50 0 1 2 3 4 5 6 7 8 9 10 11 12 Month Weekend Off-Peak Peak The second interesting aspect of the pool price results, quite obviously, is the extreme peak prices seen in the June to September period. The shortage of resources in this period is real, and quite clearly, none of the investment options we make available to Elfin can be profitably undertaken, despite these high prices. In other words, none of the peaking technologies available in these simulations, primarily solar and gas turbines, can be profitable despite the high peak pool prices, most likely because they appear in too few hours. It should also be noted that our simulations are incomplete in several ways that will tend to diminish the accuracy of on-peak pool price estimation. First, the peak pool price is heavily determined by the cost of unserved energy, as in all load duration curve based simulation. Second, imports together with other high cost options are treated quite crudely, and high cost bidders are likely to appear to peak periods that are not included in our simulation. Third, no demand response exists in the current version of Elfin. In a restructured industry, we expect more customers to face a real-time rate related to the pool price, and therefore, we can be sure that pool prices as high as these would result in load shifting. In ongoing work, a demand response module is being incorporated into Elfin that will allow demand in subsequent years to respond to high prices in any year. Another obvious question that comes to mind is how well do these prices reflect actual retail prices as the ratepayer will see them. A reasonable rule-of-thumb for this time period, that is well beyond the transition, might be that transmission and distribution costs will remain as they are now, and these plus the pool price might represent a reasonable estimate of retail energy rates, although marketing costs could be a significant factor. Currently transmission and distribution cost about 4 ¢/kWh. 39 CHAPTER 4 Figure 4-8 reports equivalent data for the AN scenario. The general pattern of results remains the same, but pool prices overall appear lower. This result actually must be true because a large amount of wind capacity appears in this scenario, bidding into the pool at a zero price. While the pool price is still being set by gas resources, the effect of the large block of wind generation is that pool prices are somewhat suppressed. This result raises a very interesting question, which is how much does this reduced pool price compensate for the cost of the various policies. In the AN case, the 2025 year-round average revenue going to generators is $37.4 per MWh compared to $44.1 per MWh in the ON case. In other words, the consumer comes out clearly ahead as a result of the AN policy. Although production costs rise, as seen in Table 4-8, pool prices fall, meaning a transfer takes place from the generator to the consumer. In other policy cases, when a subsidy is being given to a zero marginal cost generator, the value of the subsidy is partially offset by the falling pool price resulting from increased renewable participation. While quite predictable, this is an interesting result that merits further study. It should also be noted that not all renewable generators operate at close to zero marginal cost. In the DN case, for example, pool prices rise slightly over the ON case. In the BN case, direct evaluation of the benefit of the falling pool price relative to the increase in production cost is make significantly more complex by the fluctuations in the effect across the forecast period, and a much more detailed analysis would be needed in this case. Figure 4-9 summarizes the benefits and costs for the policy cases. All data are presented as percent deviations from the ON case. The positive values are the increases in costs. These represent the cost side of the policies. These are the total societal costs that represent true economic costs, free of transfers. The three subsequent indicators are all benefits, and are negative. In other words, falling thermal dependence and/or emissions is a benefit. As expected, the AN and BN cases have similar costs but BN delivers bigger benefits. This result is predictable because the total renewable construction is higher in the BN case, and because the additional renewable is wind in each case. This reinforces the earlier observation that the subsidy case is more effective than a purchase requirement in terms of total effect. However, development is less orderly. 40 CHAPTER 4 Figure 4-9. Percentage Deviation in Costs and Benefits from ON Case in Year 2025 percentage deviation 80 % Change per kWh Social Cost % Change in annual CO2 Emissions 60 % Change in annual NOx Emissions % Change in Thermal Dependency 40 20 0 -20 -40 -60 ON AN BN DN HN cases 41 42 CHAPTER 5 Conclusions In this study, considerable progress has been made towards the goal of developing algorithms for simulating the operation of competitive electricity markets, and, specifically, of building a model capable of simulating the future operation of the competitive California market. Such models are useful both as planning and policymaking tools, and as a means of evaluating possible investments. Such models are particularly important to intermittent renewable generators because their revenue streams and the benefits they provide in terms of environmental benefits and reduced pool prices are hard to forecast. The expansion planning logic of Elfin, based on the ITRE algorithm, has been enhanced so that it more clearly represents conditions in a competitive market. A new algorithm has been developed that permits ITRE to choose only the most profitable investments and then find combinations of new market entry that result in a sustainable equilibrium with no new profitable entry possible. The likely equilibrium is the one that results in the maximum return to investors. When entry has been determined, by incorporating non-energy payments into dispatch decisions, market prices can be derived using traditional algorithms. The data sets of the incumbent utilities have been merged into one that contains all the key assets in California and those owned by California utilities but situated in neighboring states. Potential imports are represented by two single resources, one for the Northwest and one for the Southwest. The new algorithms and the California data set have been used to forecast investments in generation in the future California market in the post-restructuring quarter century, 2006- 2030. Under best-guess assumptions, no new renewable capacity is added and the state’s thermal generating dependency increases to 81 percent by 2030 with natural gas consumption 3.5 times today’s levels, and the overall carbon emissions of generation increasing from 113 to 126 g/kWh. A 15 percent renewable purchase requirement lowers the 2030 thermal dependency to 71 percent and lowers carbon emissions by nine percent. Such a policy raises production costs but lowers market prices, so consumers benefit but generators lose by the policy intervention. The same level of purchase requirement with technology bands delivers much less benefits and incurs higher costs. A direct subsidy to wind generators that averages $620M per year over the forecast period, but is fixed only in net present value terms, results in more benefits for similar costs. However, deployment of renewables is prone to a boom-and-bust cycle as developers tend to only invest during the time periods when it is clearly the most profitable. Therefore, a policy trade off exists between maximizing the benefits and achieving a sustained orderly development. 43 CHAPTER 5 This work represents a bold step towards simulation of competitive markets, but many technical problems remain to be solved. In our opinion, the key areas for future work fall into the two following areas: 1. Algorithm Development Subsidy simulation algorithms need to be developed that can estimate the effects of subsidies in a stable fashion. Currently, the search for sustainable equilibriums with subsidies in place is unstable because of the interplay of various effects of a change in the subsidy, especially across time periods. The search algorithm for MEPs performs satisfactorily for searches limited to small clusters of profit maxima, but needs to be enhanced such that it more effectively finds solutions that are remote from other local profit maxima. In this work, transmission constraints have been assumed away, but a more realistic analysis would bring these into the analysis. Current production cost modeling still functions without realistic account of demand response, a glaring deficiency given the effect that new load-shifting technologies and greater customer exposure to real-time prices might have. Repowering of existing stations is one of the most likely forms of entry into the California market. Unfortunately, the economics of investments in repowering are quite different under a competitive market in which alternative uses of the site are not only possible but potentially more profitable than in power generation. Furthermore, repowering potential is particularly hard to incorporate in models because the possibilities are numerous at any one site and data local to the site is needed. New algorithms for retirement and repowering decision making are needed as is better data on repowering options. 2. Data Enhancement The current data set focuses on in-state resources, whereas competition from out-of-state generators will be a key determinant of market conditions. The data set should be extended to incorporate these competitors. The renewable technology options used in this study are quite simplistic. Now that more realistic simulation of market operations is possible, better data on the characteristics of the renewable resources in California and nearby is needed. 44 References Cadogan, John B., Brian Parsons, Joseph M. Cohen, and Bertrand L. Johnson, “Characterization of Wind Technology Progress.” California Energy Commission (CEC) 1995. “1994 Electricity Report.” CEC Nov. 1995. CEC 1998a. “Renewable Energy Technology Program Guidebook.” CEC Jan 1998, 5 vols. CEC 1998b. “California Natural Gas End Use Price Forecast.” Commissioners Jananne Sharpless and Michael C. More, 25 February 1998. Deb, Rajat K., Richard S. Albert and Lie-Long Hsue. “Modeling Competitive Energy Market In California: Analysis of Restructuring. Report prepared for the CEC by LCG Consulting, 1997. Electric Power Research Institute (EPRI) 1993. “TAGTM Technical Assessment Guide Volume 1: Electricity Supply—1993 (Revision 7).” TR-102276-V1R7. June. EPRI 1997. “Renewable Energy Technology Characterizations.” Joint project of EPRI and the U.S. Department of Energy, Office of Utility Technologies, TR-109496, December 1997. Environmental Defense Fund (EDF). Electrical Financial Production Cost and Optimization: User’s Manual, v. 3.0, August 1997. Hadley, Stanton W., Lawrence J. Hill, and Robert D. Perlack 1993. “Report on the Study of the Tax and Rate Treatment of Renewable Energy Projects.” Oak Ridge National Laboratory. ORNL-6772. Kito, M. S. 1992. CPUC Report, October. Klein, Joel. Interim Staff Market Clearing Price Forecast for the California Energy Market: Forecast Method and Analytic Issues. CEC, December 1977. Marnay, Chris, and Steve Pickle. “Power Supply Expansion and the Nuclear Option in Poland.” Contemporary Economic Policy, January 1998, vol. XVI(1). Marnay, Chris, Suzie Kito, and Osman Sezgen. “Entry Into a Competitive Electricity Market: The Fate of Renewable Generators in the California Pool.” Paper presented at the USAEE- IAEE conference, International Energy Markets, Competition and Policy, Fairmont Hotel, San Francisco, CA, 7-10 September 1997. 45 Pickle, S. and R. Wiser. 1997. "Green Power Marketing: Boosting Demand for Renewables" Public Utilities Fortnightly, December. Sezgen, Osman, Chris Marnay, and Sarah Bretz. “Wind Generation in the Future Competitive California Power Market. Lawrence Berkeley National Laboratory. LBNL-41134. March 1998. U.S. Department of Energy (DOE) 1994. “Technology Characterizations.” Draft. May. Andersen Consulting 1995. The Role of Broadband Communications in the Utility of the Future. Wiser, R., S. Pickle, and C. Goldman 1996. “California Renewable Energy Policy and Implementation Issues—An Overview of Recent Regulatory and Legislative Action.” Lawrence Berkeley National Laboratory. LBNL-39247. September. Wiser, R. and S. Pickle. 1998. Selling Green Power in California: Product, Industry and Market Trends, LBNL-41807 May. Wiser, R. and S. Pickle. 1997. Green Marketing, Renewables, and Free Riders: Increasing Customer Demand for a Public Good, LBNL-40632 September. 46 APPENDIX A Detailed Results In this appendix we provide a more detailed set of figures and tables for each scenario modeled. Again, three environments for renewables were considered (neutral, good, and bad) and the policies modeled were: • Policy 0: No policy (i.e., no new or existing policies present) • Policy A: Non-hydro minimum renewable purchase requirement (set at 15% of total generation). • Policy B: Surcharge policy ($620M/a to lowest cost renewable) • Policy D: Policy A’s minimum renewable purchase requirement with technology bands (set at current market shares) • Policy H: Carbon tax (3 $/tC) Labels for each policy and environment combination are then: Environment Policy Neutral Good Bad None 0N 0G 0B A AN AG AB B BN BG D DN DG DB H HN HG HB 47 APPENDIX A Figure A.1 Resource Mix Under Scenario 0N 400000 350000 newr_epowers new_CT 300000 new_CC SW_coal_imports 250000 NW_coal_imports NW_hydro_mports GWh 200000 existing_solar existing_wind 150000 existing_biomass existing_hydro 100000 existing_geothermal existing_coal 50000 existing_oil/gas existing_nuclear 0 2010 2015 2020 2025 2030 Year Table A.1 Resource Mix Under Scenario 0N (GWh/a) 2010 2015 2020 2025 2030 existing_nuclear 33936 10960 0 0 0 existing_oil/gas 66411 56143 49044 45238 43900 existing_coal 14121 14286 14444 14097 13995 existing_geotherm a l 10463 9937 9414 8827 7820 existing_hydro 31500 31508 31600 31509 31508 existing_biom ass 7245 7245 7264 7245 7245 existing_wind 3069 3069 3076 3069 3069 existing_solar 911 911 913 911 911 N W _ h ydro_mports 16574 16575 16634 16575 16575 NW_coal_im ports 16881 18158 18487 18314 17636 SW_coal_im ports 21713 22559 22815 22684 22173 new_CC 16533 37485 42566 71928 111409 new_CT 1022 961 1092 1666 1520 new_repowers 19868 50414 85348 83049 72598 new_coal/nuke 0 0 0 0 0 new_biomass 0 0 0 0 0 new_wind 0 0 0 0 0 new_geotherm al 0 0 0 0 0 new_solar 0 0 0 0 0 O ther 4 61 66 187 0 Pumped storage 781 414 366 393 508 48 APPENDIX A Tables A-2 to A-6. Key Indicators for Scenario 0N Construction (units) 2010 2015 2020 2025 2030 CC 10 22 25 42 65 CT 43 46 58 58 58 Wind 0 0 0 0 0 Geo-thermal 0 0 0 0 0 Solar-thermal 0 0 0 0 0 Photo-voltaics 0 0 0 0 0 Repower 11 29 44 44 44 Emissions (t) 2010 2015 2020 2025 2030 NO 222986 225962 227828 226539 224277 SU 82708 85247 86427 86016 84775 PM 9701 11930 13715 14819 15906 RG 33524 34899 36100 36684 37204 CO 68793 68741 69212 69399 69719 CX 26344706 30883172 34651154 37006603 39683706 NG 233432 179637 134195 126077 126077 Thermal Usage 2010 2015 2020 2025 2030 % Thermal 60% 71% 77% 79% 81% EJ Gas 0.640 0.933 1.187 1.362 1.563 Billion m3 17 24 31 36 41 Cumulative Present Values 2006-2055 (1995B.O.Y Million $) Production Emission Shortage Fixed O&M Capital Net Present Cost 91429 0 0 12945 27628 132002 Market Revenues and Costs 2006-2055 (1995 B.O.Y. Million $) Energy Commit Spin Total Revenue Fuel Variable O&M Fixed O&M Capital Total Cost Ind. Profit 218944 9203 1333 229480 85214 4862 12945 27628 130649 98832 49 Figure A-2. Resource Mix Under Scenario 0G GWh new_wind 400000 newr_epowers 350000 new_CT SW_coal_imports 300000 NW_coal_imports NW_hydro_mports 250000 existing_solar 200000 existing_wind existing_biomass 150000 existing_hydro 100000 existing_geothermal existing_coal 50000 existing_oil/gas 0 existing_nuclear 2010 2015 2020 2025 2030 Year 50 APPENDIX A Table A-7. Resource Mix Under Scenario OG (GWh) 2010 2015 2020 2025 2030 existing_nuclear 33936 10960 0 0 0 existing_oil/gas 73200 49616 46947 44721 39791 existing_coal 15144 12970 13464 12900 11519 existing_geotherm 10366 9940 9049 8091 6500 al existing_hydro 31496 31511 31601 31471 31363 existing_biomass 7245 7245 7256 7125 6663 existing_wind 3069 3069 3075 2924 2534 existing_solar 911 911 913 883 796 NW_hydro_mports 16574 16575 16448 13779 10663 NW_coal_imports 18319 18465 18002 15779 13765 SW_coal_imports 22642 22776 21852 18298 15206 new_CC 0 0 0 0 0 new_CT 1474 1311 589 1590 3851 newr_epowers 25905 94891 69314 50645 44187 new_coal/nuke 0 0 0 0 0 new_biomass 0 0 0 0 0 new_wind 0 0 64345 117708 165422 new_geothermal 0 0 0 0 0 new_solar 0 0 0 0 0 Other 4 29 107 0 14 Pumped storage 866 423 1020 1961 2340 Tables A-8 through A-12. Key Indicators for Scenario OG A-8. Construction (units) 2010 2015 2020 2025 2030 CC 0 0 0 0 0 CT 58 58 58 58 58 Wind 0 0 131 241 341 Geo-thermal 0 0 0 0 0 Solar-thermal 0 0 0 0 0 PV 0 0 0 0 0 Repower 12 42 42 42 42 51 APPENDIX A A-9. Emissions (t) 2010 2015 2020 2025 2030 NO 228826 223672 222057 207683 187765 SU 87175 83368 82419 73463 63847 PM 9564 12203 10949 9774 8921 RG 33526 35012 34401 33193 31200 CO 69908 67683 66669 64955 60020 CX 26882477 30833779 27243183 23618085 21097815 NG 250777 149228 128664 124858 116044 A-10. Thermal Usage 2010 2015 2020 2025 2030 % Thermal 60% 71% 56% 44% 36% Gas (EJ) 0.634 0.947 0.704 0.548 0.495 Billion (m^3) 17 25 18 14 13 A-11. Cumulative Present Values 2006-2055 (1995 B.O.Y Million $) Production Emission Shortage Fixed O&M Capital Net Cost $67,705 $0 $0 $28,745 $56,871 $153,320 A-12. Market Revenues and Costs 2006-2055 (1995 B.O.Y. Million $) Energy Commit Spin Total Rev. Fuel Var. O&M $234,407 $7,745 $1,229 $243,381 $63,522 $3,739 Fixed O&M Capital Total Cost Ind. Profit $28,745 $56,871 $152,876 $90,505 52 APPENDIX A Figure A-3. Resource Mix Under Scenario 0B GWh newr_epowers 400000 new_CT 350000 new_CC SW_coal_imports 300000 NW_coal_imports NW_hydro_mports 250000 existing_solar 200000 existing_wind existing_biomass 150000 existing_hydro 100000 existing_geothermal existing_coal 50000 existing_oil/gas 0 existing_nuclear 2010 2015 2020 2025 2030 Year 53 APPENDIX A Table A-13. Resource Mix under Scenario OB (GWh) 2010 2015 2020 2025 2030 existing_nuclear 33936 10960 0 0 0 existing_oil/gas 66454 69451 59010 47257 43900 existing_coal 10909 11320 11822 11435 9716 existing_geotherm 10439 9935 9414 8827 7820 al existing_hydro 31507 31505 31596 31507 31512 existing_biomass 7245 7245 7264 7245 7245 existing_wind 3069 3069 3076 3069 3069 existing_solar 911 911 913 911 911 NW_hydro_mports 16575 16575 16634 16575 16575 NW_coal_imports 5877 4567 3945 1452 72 SW_coal_imports 8236 9589 7376 2962 245 new_CC 28930 41806 42129 41600 42265 new_CT 27 820 6136 9635 2443 newr_epowers 36039 62447 103416 142759 184638 new_coal/nuke 0 0 0 0 0 new_biomass 0 0 0 0 0 new_wind 0 0 0 0 0 new_geothermal 0 0 0 0 0 new_solar 0 0 0 0 0 Other 0 56 9 24 4 Pumped storage 443 371 362 361 358 Tables A-14 through A-18. Key Indicators for Scenario OB A-14. Construction (units) 2010 2015 2020 2025 2030 CC 17 24 24 24 25 CT 9 43 71 71 71 Wind 0 0 0 0 0 Geo-thermal 0 0 0 0 0 Solar-thermal 0 0 0 0 0 PV 0 0 0 0 0 Repower 22 26 38 53 77 54 APPENDIX A A-15. Emissions (t) 2010 2015 2020 2025 2030 NO 177499 181858 177836 164306 151726 SU 48942 50413 48892 39318 27849 PM 9660 11667 13733 15197 16115 RG 33877 35356 36746 37629 38000 CO 69278 71406 71931 71419 70822 CX 24261221 29463608 33252907 35256617 37253430 NG 244801 224099 158584 126077 126077 A-16. Thermal Usage 2010 2015 2020 2025 2030 % Thermal 60% 71% 77% 79% 81% Gas (EJ) 0.847 1.190 1.488 1.728 1.958 Billion (m^3) 22 31 39 45 51 A-17. Cumulative Present Values 2006-2055 (1995 B.O.Y Million $) Production Emission Shortage Fixed O&M Capital Net Cost $66,551 $0 $0 $12,118 $23,831 $102,500 A-18. Market Revenues and Costs 2006-2055 (1995 B.O.Y. Million $) Energy Commit Spin Total Rev. Fuel Var. O&M $185,520 $13,907 $980 $200,406 $58,037 $8,287 Fixed O&M Capital Total Cost Ind. Profit $12,118 $23,831 $102,273 $98,133 55 APPENDIX A Figure A-4. Resource Mix Under Scenario AN GWh new_wind 400000 newr_epowers 350000 new_CT new_CC 300000 SW_coal_imports NW_coal_imports 250000 NW_hydro_mports existing_solar 200000 existing_wind 150000 existing_biomass existing_hydro 100000 existing_geothermal existing_coal 50000 existing_oil/gas existing_nuclear 0 2010 2015 2020 2025 2030 Year 56 APPENDIX A Table A-19. Resource Mix Under Scenario AN (GWh) 2010 2015 2020 2025 2030 existing_nuclear 33936 10960 0 0 0 existing_oil/gas 64988 56062 49854 45804 43900 existing_coal 14289 14257 14207 13955 14102 existing_geotherm 10445 9937 9414 8822 7820 al existing_hydro 31501 31509 31601 31510 31509 existing_biomass 7245 7245 7264 7245 7245 existing_wind 3069 3069 3076 3069 3069 existing_solar 911 911 913 911 911 NW_hydro_mports 16521 16575 16634 16575 16575 NW_coal_imports 16872 18308 18482 18194 17911 SW_coal_imports 21498 22580 22806 22388 22102 new_CC 0 6611 12072 42757 66395 new_CT 849 672 549 575 1031 newr_epowers 20925 60846 91191 82405 84293 new_coal/nuke 0 0 0 0 0 new_biomass 0 0 0 0 0 new_wind 17231 20672 24637 31008 33469 new_geothermal 0 0 0 0 0 new_solar 0 0 0 0 0 Other 3 43 62 77 0 Pumped storage 880 366 361 389 442 Tables A-20 through A-24. Key Indicators for Scenario AN A-20. Construction 2010 2015 2020 2025 2030 CC 0 4 7 25 39 CT 39 41 41 41 41 Wind 35 42 50 63 68 Geo-thermal 0 0 0 0 0 Solar-thermal 0 0 0 0 0 PV 0 0 0 0 0 Repower 12 34 49 50 50 57 APPENDIX A A-21. Emissions (t) 2010 2015 2020 2025 2030 NO 222178 225815 227198 225297 223901 SU 82134 85301 86025 85135 84823 PM 8984 11067 12647 13465 14476 RG 33074 34381 35487 35934 36323 CO 68000 68069 68575 68582 68629 CX 24372114 28666873 31962740 33522160 36067691 NG 232244 176894 137484 126077 126077 A-22. Thermal Usage 2010 2015 2020 2025 2030 % Thermal 53% 64% 69% 69% 71% Gas (EJ) 0.501 0.777 1.002 1.125 1.309 Billion (m^3) 13 20 26 29 34 A-23. Cumulative Present Values 2006-2055 (1995 B.O.Y Million $) Production Emission Shortage Fixed O&M Capital Net Cost $79,066 $0 $0 $16,836 $39,468 $135,370 A-24. Market Revenues and Costs 2006-2055 (1995 B.O.Y. Million $) Energy Commit Spin Total Rev. Fuel Var. O&M $214,934 $9,079 $1,066 $225,079 $72,712 $4,958 Fixed O&M Capital Total Cost Ind. Profit $16,836 $39,468 $133,974 $91,105 58 APPENDIX A Figure A-5. Resource Mix Under Scenario AG GWh 400000 new_solar new_wind 350000 newr_epowers new_CT 300000 SW_coal_imports NW_coal_imports 250000 NW_hydro_mports 200000 existing_solar existing_wind 150000 existing_biomass existing_hydro 100000 existing_geothermal existing_coal 50000 existing_oil/gas 0 existing_nuclear 2010 2015 2020 2025 2030 Year 59 APPENDIX A Table A-25. Resource Mix Under Scenario AG (GWh) 2010 2015 2020 2025 2030 existing_nuclear 33936 10960 0 0 0 existing_oil/gas 61607 52773 49720 45588 40344 existing_coal 14817 14317 14050 13599 11965 existing_geotherm 10186 9618 9076 8346 6823 al existing_hydro 31497 31509 31599 31493 31388 existing_biomass 7245 7245 7264 7211 6710 existing_wind 3069 3069 3076 3020 2602 existing_solar 911 911 913 906 869 NW_hydro_mports 16524 16575 16612 14888 11094 NW_coal_imports 17578 18453 18345 16675 14214 SW_coal_imports 21961 22742 22405 19825 15761 new_CC 0 0 0 0 0 new_CT 869 604 963 2515 4485 newr_epowers 23007 70802 75174 57731 45504 new_coal/nuke 0 0 0 0 0 new_biomass 0 0 0 0 0 new_wind 17232 20672 53101 103592 159790 new_geothermal 0 0 0 0 0 new_solar 0 0 342 340 338 Other 1 30 232 15 27 Pumped storage 1329 439 792 1874 2530 Tables A-26 through A-30. Key Indicators for Scenario AG A-26. Construction (units) 2010 2015 2020 2025 2030 CC 0 0 0 0 0 CT 41 41 41 41 41 Wind 35 42 72 212 329 Geo-thermal 0 0 0 0 0 Solar-thermal 0 0 0 0 0 PV 0 0 4 4 4 Repower 12 38 40 40 40 60 APPENDIX A A-27. Emissions (t) 2010 2015 2020 2025 2030 NO 224373 226219 225491 215382 191962 SU 84734 85960 84461 78073 66226 PM 9105 11209 11391 10383 9162 RG 33110 34458 34757 33889 31639 CO 67582 67700 67715 66193 61025 CX 24475132 28720813 28716385 25396181 21789267 NG 213427 159996 135355 125876 117116 A-28. Thermal Usage 2010 2015 2020 2025 2030 % Thermal 53% 64% 59% 48% 37% Gas (EJ) 0.488 0.775 0.785 0.621 0.514 Billion (m^3) 13 20 20 16 13 A-29. Cumulative Present Values 2006-2055 (1995 B.O.Y Million $) Production Emission Shortage Fixed O&M Capital Net Cost $64,094 $0 $0 $29,429 $58,380 $151,904 A-30. Market Revenues and Costs 2006-2055 (1995 B.O.Y. Million $) Energy Commit Spin Total Rev. Fuel Var. O&M $251,315 $14,303 $1,858 $267,477 $59,908 $3,389 Fixed O&M Capital Total Cost Ind. Profit $29,429 $58,380 $151,107 $116,369 61 APPENDIX A Figure A-6. Resource Mix Under Scenario AB GWh new_wind 400000 newr_epowers 350000 new_CT new_CC 300000 SW_coal_imports NW_coal_imports 250000 NW_hydro_mports existing_solar 200000 existing_wind 150000 existing_biomass existing_hydro 100000 existing_geothermal existing_coal 50000 existing_oil/gas existing_nuclear 0 2010 2015 2020 2025 2030 Year 62 APPENDIX A Table A-31. Resource Mix Under Scenario AB (GWh) 2010 2015 2020 2025 2030 existing_nuclear 33936 10960 0 0 0 existing_oil/gas 74048 66934 55471 45937 43900 existing_coal 11741 11537 11397 10739 9782 existing_geotherm 10423 9927 9413 8826 7820 al existing_hydro 31502 31505 31596 31508 31509 existing_biomass 7245 7245 7264 7245 7245 existing_wind 3069 3069 3076 3069 3069 existing_solar 911 911 913 911 911 NW_hydro_mports 16516 16575 16634 16575 16575 NW_coal_imports 7804 3822 1265 390 228 SW_coal_imports 12360 7303 4398 2069 834 new_CC 8617 20799 22817 34768 72327 new_CT 521 401 1405 1640 2086 newr_epowers 24265 68582 112452 130550 120674 new_coal/nuke 0 0 0 0 0 new_biomass 0 0 0 0 0 new_wind 17232 20672 24637 31008 33469 new_geothermal 0 0 0 0 0 new_solar 0 0 0 0 0 Other 1 18 14 32 23 Pumped storage 525 384 364 364 410 Tables A-32 through A-36. Key Indicators for Scenario AB A-32. Construction (units) 2010 2015 2020 2025 2030 CC 5 12 13 20 43 CT 33 36 44 44 44 Wind 35 42 50 63 68 Geo-thermal 0 0 0 0 0 Solar-thermal 0 0 0 0 0 PV 0 0 0 0 0 Repower 12 30 45 57 57 63 APPENDIX A A-33. Emissions (t) 2010 2015 2020 2025 2030 NO 190555 175291 167033 158537 152858 SU 57508 47226 40602 34365 28971 PM 8761 10843 12642 13599 14498 RG 33308 34813 36139 36629 36996 CO 69593 70418 70548 69656 69558 CX 23193838 26873604 29817279 31067270 33233060 NG 268848 220397 150687 126077 126077 A-34. Thermal Usage 2010 2015 2020 2025 2030 % Thermal 53% 64% 69% 69% 71% Gas (EJ) 0.677 1.051 1.323 1.469 1.662 Billion (m^3) 18 27 35 38 43 A-35. Cumulative Present Values 2006-2055 (1995 B.O.Y Million $) Production Emission Shortage Fixed O&M Capital Net Cost $56,998 $0 $0 $17,457 $39,819 $114,275 A-36. Market Revenues and Costs 2006-2055 (1995 B.O.Y. Million $) Energy Commit Spin Total Rev. Fuel Var. O&M $156,846 $14,824 $884 $172,555 $50,216 $6,425 Fixed O&M Capital Total Cost Ind. Profit $17,457 $39,819 $113,919 $58,635 64 APPENDIX A Figure A-7. Resource Mix Under Scenario BN GWh new_wind 400000 newr_epowers 350000 new_CT new_CC 300000 SW_coal_imports NW_coal_imports 250000 NW_hydro_mports existing_solar 200000 existing_wind 150000 existing_biomass existing_hydro 100000 existing_geothermal existing_coal 50000 existing_oil/gas 0 existing_nuclear 2010 2015 2020 2025 2030 Year 65 APPENDIX A Table A-37. Resource Mix Under Scenario BN (GWh) 2010 2015 2020 2025 2030 existing_nuclear 33936 10960 0 0 0 existing_oil/gas 70785 58830 49985 45971 43900 existing_coal 14543 14267 13751 13876 14024 existing_geotherm 10477 9801 9195 8815 7820 al existing_hydro 31499 31507 31599 31508 31510 existing_biomass 7245 7245 7264 7245 7245 existing_wind 3069 3069 3076 3069 3069 existing_solar 911 911 913 911 911 NW_hydro_mports 16573 16574 16588 16563 16573 NW_coal_imports 17657 18183 18061 18040 18219 SW_coal_imports 22149 22414 22014 22007 22232 new_CC 0 0 0 14811 26545 new_CT 1175 1082 787 1370 1759 newr_epowers 23299 55865 72505 84115 99596 new_coal/nuke 0 0 0 0 0 new_biomass 0 0 0 0 0 new_wind 6904 29532 57029 56978 56994 new_geothermal 0 0 0 0 0 new_solar 0 0 0 0 0 Other 5 40 24 72 0 Pumped storage 702 478 460 603 605 Tables A-38 through A-42. Key Indicators for Scenario BN A-38. Construction (units) 2010 2015 2020 2025 2030 CC 0 0 0 9 16 CT 47 49 49 49 49 Wind 14 60 116 116 116 Geo-thermal 0 0 0 0 0 Solar-thermal 0 0 0 0 0 PV 0 0 0 0 0 Repower 12 30 42 47 53 66 APPENDIX A A-39. Emissions (t) 2010 2015 2020 2025 2030 NO 225727 225427 223521 223805 223894 SU 84254 84720 83565 84126 84455 PM 9294 10610 11199 12377 13502 RG 33327 34143 34589 35281 35730 CO 69243 68195 67424 67827 67872 CX 25842860 27774071 28185354 30696316 33640414 NG 249285 186661 137874 126070 126077 A-40. Thermal Usage 2010 2015 2020 2025 2030 % Thermal 57% 61% 58% 61% 64% Gas (EJ) 0.585 0.717 0.762 0.937 1.137 Billion (m^3) 15 19 20 24 30 A-41. Cumulative Present Values 2006-2055 (1995 B.O.Y Million $) Production Emission Shortage Fixed O&M Capital Net Cost $67,004 $0 $0 $19,177 $42,820 $132,001 A-42. Market Revenues and Costs 2006-2055 (1995 B.O.Y. Million $) Energy Commit Spin Total Rev. Fuel Var. O&M $219,672 $9,202 $1,054 $229,928 $67,588 -$2,048 Fixed O&M Capital Total Cost Ind. Profit $19,177 $45,820 $130,536 $99,392 67 APPENDIX A Figure A-8. Resource Mix Under Scenario BG GWh 400000 new_solar new_wind 350000 newr_epowers new_CT 300000 SW_coal_imports NW_coal_imports 250000 NW_hydro_mports 200000 existing_solar existing_wind 150000 existing_biomass existing_hydro 100000 existing_geothermal existing_coal 50000 existing_oil/gas existing_nuclear 0 2010 2015 2020 2025 2030 Year 68 APPENDIX A Table A-43. Resource Mix Under Scenario BG (GWh) 2010 2015 2020 2025 2030 existing_nuclear 33936 10960 0 0 0 existing_oil/gas 61077 54313 49213 44562 40319 existing_coal 15177 14323 13701 12632 11793 existing_geotherm 10394 9744 8930 7969 6655 al existing_hydro 31499 31509 31597 31452 31382 existing_biomass 7245 7245 7264 7072 6725 existing_wind 3069 3069 3072 2890 2600 existing_solar 911 911 913 897 817 NW_hydro_mports 16575 16575 16023 13028 10918 NW_coal_imports 18395 18463 17534 15304 14187 SW_coal_imports 22675 22761 21215 17736 15543 new_CC 0 0 0 0 0 new_CT 728 698 671 2827 4532 newr_epowers 38564 73372 57872 45293 41965 new_coal/nuke 0 0 0 0 0 new_biomass 0 0 0 0 0 new_wind 0 16243 74967 124268 158702 new_geothermal 0 0 0 0 0 new_solar 0 0 0 0 5229 Other 3 81 82 13 21 Pumped storage 761 394 1300 1976 2542 Tables A-44 through A-48. Key Indicators for Scenario BG A-44. Construction (units) 2010 2015 2020 2025 2030 CC 0 0 0 0 0 CT 36 36 36 36 36 Wind 33 153 255 327 Geo-thermal 0 0 0 0 0 Solar-thermal 0 0 0 0 0 PV 0 0 0 0 62 Repower 19 37 37 37 37 69 APPENDIX A A-45. Emissions (t) 2010 2015 2020 2025 2030 NO 227852 226449 220817 204607 190926 SU 87345 85409 81125 71564 65406 PM 9897 11344 10452 9523 8977 RG 33622 34562 34170 32893 31523 CO 68177 68081 66879 64408 60856 CX 26511775 29272982 26130936 22904290 21264936 NG 211359 164898 134337 124211 117082 A-46. Thermal Usage 2010 2015 2020 2025 2030 % Thermal 60% 66% 53% 42% 36% Gas (EJ) 0.606 0.814 0.636 0.519 0.486 Billion (m^3) 16 21 17 14 13 A-47. Cumulative Present Values 2006-2055 (1995 B.O.Y Million $) Production Emission Shortage Fixed O&M Capital Net Cost 64953 0 0 28601 56894 150449 A-48. Market Revenues and Costs 2006-2055 (1995 B.O.Y. Million $) Energy Commit Spin Total Rev. Fuel Var. O&M 287549 15550 2576 305676 62223 1377 Fixed O&M Capital Total Cost Ind. Profit 28,601 56,894 149,096 156,579 70 APPENDIX A Figure A-9. Resource Mix Under Scenario DN GWh new_solar 400000 new_geothermal new_wind 350000 newr_epowers new_CT 300000 new_CC SW_coal_imports 250000 NW_coal_imports NW_hydro_mports 200000 existing_solar existing_wind 150000 existing_biomass existing_hydro 100000 existing_geothermal existing_coal 50000 existing_oil/gas existing_nuclear 0 2010 2015 2020 2025 2030 Year 71 APPENDIX A Table A-49. Resource Mix Under Scenario DN (GWh) 2010 2015 2020 2025 2030 existing_nuclear 33936 10960 0 0 0 existing_oil/gas 69866 57962 48560 46456 43900 existing_coal 14196 14083 14021 14086 14054 existing_geotherm 10438 9931 9413 8825 7820 al existing_hydro 31497 31507 31600 31508 31508 existing_biomass 7245 7245 7264 7245 7245 existing_wind 3069 3069 3076 3069 3069 existing_solar 911 911 913 911 911 NW_hydro_mports 16561 16575 16634 16575 16575 NW_coal_imports 15957 17716 18398 18153 17214 SW_coal_imports 21062 22237 22761 22551 21780 new_CC 11665 32404 37604 58859 98161 new_CT 1173 988 1004 1497 1028 newr_epowers 8806 37977 71286 71392 59040 new_coal/nuke 0 0 0 0 0 new_biomass 56 64 114 410 956 new_wind 2466 2958 3454 3942 4435 new_geothermal 8822 10379 12870 15147 17741 new_solar 2571 3225 3731 4458 5061 Other 26 97 63 213 0 Pumped storage 1043 478 378 411 737 Tables A-50 through A-54. Key Indicators for Scenario DN A-50. Construction (units) 2010 2015 2020 2025 2030 CC 7 19 22 34 57 CT 35 38 50 50 50 Wind 5 6 7 8 9 Geo-thermal 11 13 16 19 22 Solar-thermal 7 9 10 12 14 PV 9 10 12 14 15 Bio 6 6 8 9 10 Repower 5 23 38 38 38 72 APPENDIX A A-51. Emissions (t) 2010 2015 2020 2025 2030 NO 221040 224704 226766 227021 223665 SU 80959 84140 86072 85646 83984 PM 9070 11240 12989 14311 15893 RG 33118 34468 35551 36231 36495 CO 68923 68579 68504 69277 69138 CX 25066855 29260940 32596073 34721397 36949270 NG 245648 185091 134294 126077 126077 A-52. Thermal Usage 2010 2015 2020 2025 2030 % Thermal 55% 65% 70% 72% 73% Gas (EJ) 0.550 0.813 1.028 1.177 1.347 Billion (m^3) 14 21 27 31 35 A-53. Cumulative Present Values 2006-2055 (1995 B.O.Y Million $) Production Emission Shortage Fixed O&M Capital Net Cost $88,168 $0 $0 $17,494 $42,719 $148,382 A-54. Market Revenues and Costs 2006-2055 (1995 B.O.Y. Million $) Energy Commit Spin Total Rev. Fuel Var. O&M $225,047 $12,396 $1,539 $238,983 $82,258 $4,469 Fixed O&M Capital Total Cost Ind. Profit $17,494 $42,719 $146,941 $92,042 73 APPENDIX A Figure A-10. Resource Mix Under Scenario DG GWh new_solar 400000 new_geothermal new_wind 350000 new_biomass newr_epowers 300000 new_CT 250000 SW_coal_imports NW_coal_imports 200000 NW_hydro_mports existing_solar 150000 existing_wind existing_biomass 100000 existing_hydro existing_geothermal 50000 existing_coal existing_oil/gas 0 existing_nuclear 2010 2015 2020 2025 2030 Year 74 APPENDIX A Table A-55. Resource Mix Under Scenario DG (GWh) 2010 2015 2020 2025 2030 existing_nuclear 33936 10960 0 0 0 existing_oil/gas 62368 55356 49348 45654 40695 existing_coal 14979 14734 14510 13867 11855 existing_geotherm 10257 9932 9305 8448 6910 al existing_hydro 31498 31506 31597 31497 31398 existing_biomass 7245 7245 7264 7229 6781 existing_wind 3069 3069 3076 3035 2713 existing_solar 911 911 913 907 877 NW_hydro_mports 16573 16575 16567 15091 11458 NW_coal_imports 18027 18464 18110 16501 14124 SW_coal_imports 22425 22774 22096 19932 15693 new_CC 0 0 0 0 0 new_CT 632 1202 936 1894 3460 newr_epowers 24019 68169 56596 44114 33968 new_coal/nuke 0 0 0 0 0 new_biomass 651 1399 5254 4959 4403 new_wind 2466 4436 51144 97795 153679 new_geothermal 8663 10136 12240 10256 9094 new_solar 2680 3302 3833 4569 4670 Other 1 90 162 11 11 Pumped storage 1199 380 1007 1800 2384 Tables A-56 through A-60. Key Indicators for Scenario DG A-56. Construction (units) 2010 2015 2020 2025 2030 CC 0 0 0 0 0 CT 41 44 44 44 44 Wind 5 9 104 200 316 Geo-thermal 11 13 16 19 22 Solar-thermal 7 9 10 12 14 PV 9 10 12 14 15 Repower 12 31 33 33 33 75 APPENDIX A A-57. Emissions (t) 2010 2015 2020 2025 2030 NO 226815 228363 229180 220215 195893 SU 86232 86459 85440 79507 66336 PM 10030 12967 17226 16080 14212 RG 33175 34497 34431 33787 31659 CO 67958 68496 68283 67075 62160 CX 25101437 29310753 27044122 24325920 20775604 NG 217551 173320 133604 125987 117704 A-58. Thermal Usage 2010 2015 2020 2025 2030 % Thermal 54% 64% 53% 43% 34% Gas (EJ) 0.500 0.787 0.630 0.502 0.410 Billion (m^3) 13 21 16 13 11 A-59. Cumulative Present Values 2006-2055 (1995 B.O.Y Million $) Production Emission Shortage Fixed O&M Capital Net Cost $65,670 $0 $0 $31,942 $64,578 $162,191 A-60. Market Revenues and Costs 2006-2055 (1995 B.O.Y. Million $) Energy Commit Spin Total Rev. Fuel Var. O&M $250,817 $19,678 $1,872 $272,368 $61,281 $3,771 Fixed O&M Capital Total Cost Ind. Profit $31,942 $64,578 $161,576 $110,794 76 APPENDIX A Figure A-11. Resource Mix Under Scenario DB GWh new_solar 400000 new_geothermal 350000 new_wind newr_epowers 300000 new_CT new_CC 250000 SW_coal_imports NW_coal_imports 200000 NW_hydro_mports existing_solar 150000 existing_wind existing_biomass 100000 existing_hydro existing_geothermal 50000 existing_coal existing_oil/gas 0 existing_nuclear 2010 2015 2020 2025 2030 Year 77 APPENDIX A Table A-61. Resource Mix Under Scenario DB (GWh) 2010 2015 2020 2025 2030 existing_nuclear 33936 10960 0 0 0 existing_oil/gas 72706 68045 54786 46452 43900 existing_coal 11507 11423 10471 10671 9912 existing_geotherm 10439 9938 9414 8826 7820 al existing_hydro 31502 31504 31593 31507 31510 existing_biomass 7245 7245 7264 7245 7245 existing_wind 3069 3069 3076 3069 3069 existing_solar 911 911 913 911 911 NW_hydro_mports 16563 16575 16634 16575 16575 NW_coal_imports 7299 3777 1197 736 139 SW_coal_imports 12400 7395 3929 2755 377 new_CC 16421 20338 26331 55724 81146 new_CT 479 407 1558 4722 1473 newr_epowers 21790 72008 115547 112356 119305 new_coal/nuke 0 0 0 0 0 new_biomass 11 18 12 34 13 new_wind 2466 2958 3454 3942 4436 new_geothermal 8880 10494 12951 15337 17759 new_solar 2551 3148 3606 4354 4846 Other 1 37 14 59 11 Pumped storage 492 364 361 366 373 Tables A-62 through A-66. Key Indicators for Scenario DB A-62. Construction (units) 2010 2015 2020 2025 2030 CC 10 12 15 32 50 CT 27 30 42 42 42 Wind 5 6 7 8 9 Geo-thermal 11 13 16 19 22 Solar-thermal 7 9 10 12 14 PV 9 10 12 14 15 Repower 11 30 45 45 57 78 APPENDIX A A-63. Emissions (t) 2010 2015 2020 2025 2030 NO 189864 175807 164275 160946 152254 SU 56089 46659 36735 35766 28834 PM 8923 11030 12834 14009 14803 RG 33361 34972 36272 36834 37165 CO 69468 70851 70579 70233 69897 CX 23689766 27619192 30430917 32398537 34269989 NG 263281 219570 148020 126077 126077 A-64. Thermal Usage 2010 2015 2020 2025 2030 % Thermal 55% 65% 71% 72% 73% Gas (EJ) 0.704 1.085 1.370 1.524 1.711 Billion (m^3) 18 28 36 40 45 A-65. Cumulative Present Values 2006-2055 (1995 B.O.Y Million $) Production Emission Shortage Fixed O&M Capital Net Cost $62,600 $0 $0 $17,637 $44,212 $124,450 A-66. Market Revenues and Costs 2006-2055 (1995 B.O.Y. Million $) Energy Commit Spin Total Rev. Fuel Var. O&M $156,848 $15,364 $1,023 $173,237 $55,630 $6,659 Fixed O&M Capital Total Cost Ind. Profit $17,637 $44,212 $124,139 $49,097 79 APPENDIX A Figure A-12. Resource Mix Under Scenario HN GWh 400000 new_geothermal new_coal/nuke 350000 SW_coal_imports 300000 NW_coal_imports NW_hydro_mports 250000 existing_solar 200000 existing_wind existing_biomass 150000 existing_hydro existing_geothermal 100000 existing_coal 50000 existing_oil/gas existing_nuclear 0 2010 2015 2020 2025 2030 Year 80 APPENDIX A Table A-67. Resource Mix Under Scenario HN (GWh) 2010 2015 2020 2025 2030 existing_nuclear 33936 10960 0 0 0 existing_oil/gas 52360 45144 49455 45072 43466 existing_coal 10921 9347 11826 12003 12610 existing_geotherm 9961 8817 9062 8585 7813 al existing_hydro 31498 31494 31586 31489 31462 existing_biomass 6536 5684 6486 6762 7023 existing_wind 3069 3063 3075 3069 3069 existing_solar 911 909 912 910 911 NW_hydro_mports 16552 15778 16575 16558 16568 NW_coal_imports 12502 9258 12750 11447 11586 SW_coal_imports 10254 1404 12555 9810 10600 new_CC 0 0 0 29907 53072 new_CT 0 0 0 0 0 newr_epowers 0 0 0 0 0 new_coal/nuke 53290 105530 106740 106914 107358 new_biomass 0 0 0 0 0 new_wind 0 0 0 0 0 new_geothermal 18848 33244 39663 39931 40107 new_solar 0 0 0 0 0 Other 81 462 2202 2174 2353 Pumped storage 1884 2527 2502 2342 2473 Tables A-68 through A-72. Key Indicators for Scenario HN A-68. Construction (units) 2010 2015 2020 2025 2030 CC 0 0 0 22 38 CT 0 0 0 0 0 Wind 0 0 0 0 0 Geo-thermal 25 50 50 50 50 Solar-thermal 0 0 0 0 0 PV 0 0 0 0 0 Nuclear 12 24 24 24 24 Repower 0 0 0 0 0 81 APPENDIX A A-69. Emissions (t) 2010 2015 2020 2025 2030 NO 177294 138479 184038 180111 184892 SU 54490 35160 60033 56308 59132 PM 6486 5240 6637 7909 9034 RG 30232 27665 30389 31796 32777 CO 59002 50883 58308 60912 63005 CX 16421002 12220729 16840799 18903321 21679018 NG 174045 140593 140219 125327 125830 A-70. Thermal Usage 2010 2015 2020 2025 2030 % Thermal 53% 60% 63% 66% 68% Gas (EJ) 0.230 0.165 0.176 0.360 0.519 Billion (m^3) 6 4 5 9 14 A-71. Cumulative Present Values 2006-2055 (1995 B.O.Y Million $) Production Emission Shortage Fixed O&M Capital Net Cost $138,631 $0 $0 $31,207 $67,351 $237,190 A-72. Market Revenues and Costs 2006-2055 (1995 B.O.Y. Million $) Energy Commit Spin Total Rev. Fuel Var. O&M $1,481,163 $10,646 $8,291 $1,500,101 $79,304 $5,852 Fixed O&M Capital Total Cost Ind. Profit $31,207 $67,351 $183,716 $1,316,384 82 APPENDIX A Figure A-13. Resource Mix Under Scenario HG GWh 400000 new_solar new_wind 350000 new_biomass new_CC 300000 SW_coal_imports NW_coal_imports 250000 NW_hydro_mports 200000 existing_solar existing_wind 150000 existing_biomass existing_hydro 100000 existing_geothermal existing_coal 50000 existing_oil/gas 0 existing_nuclear 2010 2015 2020 2025 2030 Year 83 APPENDIX A Table A-73. Resource Mix Under Scenario HG (GWh) 2010 2015 2020 2025 2030 existing_nuclear 33936 10960 0 0 0 existing_oil/gas 73172 66415 55672 41930 32831 existing_coal 14869 14470 14055 11665 9109 existing_geotherm 10051 9661 9184 7763 5816 al existing_hydro 31496 31498 31444 31322 31081 existing_biomass 7244 7236 7151 6122 4797 existing_wind 3069 3069 3076 2972 2348 existing_solar 911 911 913 897 744 NW_hydro_mports 16575 16575 16633 14960 10590 NW_coal_imports 18456 18412 18261 15639 12296 SW_coal_imports 21161 21569 20690 14299 9889 new_CC 23352 23201 22252 15290 10156 new_CT 0 0 0 0 0 newr_epowers 0 0 0 0 0 new_coal/nuke 0 0 0 0 0 new_biomass 2187 12831 17639 11412 10064 new_wind 0 33437 66609 132283 194198 new_geothermal 0 0 0 0 0 new_solar 3929 8949 10642 8871 7378 Other 136 1293 1093 1004 493 Pumped storage 1756 1948 2612 2516 2763 Tables A-74 through A-78. Key Indicators for Scenario HG A-74. Construction (units) 2010 2015 2020 2025 2030 CC 15 15 15 15 15 CT 0 0 0 0 0 Wind 0 68 136 272 408 Geo-thermal 0 0 0 0 0 Solar-thermal 15 30 30 30 30 PV 0 0 0 0 0 Bio 36 36 36 36 36 Repower 0 0 0 0 0 84 APPENDIX A A-75. Emissions (t) 2010 2015 2020 2025 2030 NO 227143 235185 235545 194649 157271 SU 84625 85514 83694 65490 49612 PM 12019 25513 31403 21793 18457 RG 33367 33943 33969 30357 26375 CO 70398 71924 71027 58286 45590 CX 26127316 25570417 23991051 18147924 13770713 NG 246190 197899 148432 116535 96503 A-76. Thermal Usage 2010 2015 2020 2025 2030 % Thermal 58% 51% 44% 31% 22% Gas (EJ) 0.597 0.513 0.399 0.237 0.155 Billion (m^3) 16 13 10 6 4 A-77. Cumulative Present Values 2006-2055 (1995 B.O.Y Million $) Production Emission Shortage Fixed O&M Capital Net Cost $144,646 $0 $0 $37,222 $75,953 $257,823 A-78. Market Revenues and Costs 2006-2055 (1995 B.O.Y. Million $) Energy Commit Spin Total Rev. Fuel Var. O&M $1,306,863 $21,987 $7,084 $1,335,935 $61,939 $3,442 Fixed O&M Capital Total Cost Ind. Profit $37,222 $75,953 $178,558 $1,157,377 85 APPENDIX A Figure A-14. Resource Mix Under Scenario HB GWh 400000 newr_epowers new_CT 350000 new_CC SW_coal_imports 300000 NW_coal_imports 250000 NW_hydro_mports existing_solar 200000 existing_wind existing_biomass 150000 existing_hydro 100000 existing_geothermal existing_coal 50000 existing_oil/gas existing_nuclear 0 2010 2015 2020 2025 2030 Year 86 APPENDIX A Table A-79. Resource Mix Under Scenario HB (GWh) 2010 2015 2020 2025 2030 existing_nuclear 33936 10960 0 0 0 existing_oil/gas 71895 79705 62461 48414 43900 existing_coal 11898 12605 12433 11772 9666 existing_geotherm 10484 9940 9414 8827 7820 al existing_hydro 31506 31510 31601 31510 31511 existing_biomass 7245 7245 7264 7245 7244 existing_wind 3069 3069 3076 3069 3069 existing_solar 911 911 913 911 911 NW_hydro_mports 16575 16575 16634 16575 16575 NW_coal_imports 7349 11844 8906 4171 1089 SW_coal_imports 11619 16017 10664 5027 1338 new_CC 6996 7002 7022 7002 6740 new_CT 258 8623 27816 40026 20340 newr_epowers 46391 64208 104526 140646 200131 new_coal/nuke 0 0 0 0 0 new_biomass 0 0 0 0 0 new_wind 0 0 0 0 0 new_geothermal 0 0 0 0 0 new_solar 0 0 0 0 0 Other 1 31 29 90 34 Pumped storage 371 363 362 362 361 Tables A-80 through A-84. Key Indicators for Scenario HB A-80. Construction (units) 2010 2015 2020 2025 2030 CC 4 4 4 4 4 CT 27 77 102 102 102 Wind 0 0 0 0 0 Geo-thermal 0 0 0 0 0 Solar-thermal 0 0 0 0 0 PV 0 0 0 0 0 Bio 0 0 0 0 0 Repower 22 22 34 46 70 87 APPENDIX A A-81. Emissions (t) 2010 2015 2020 2025 2030 NO 188540 205647 191219 172346 155255 SU 56669 68563 58087 44900 29989 PM 9588 11591 14137 15875 16535 RG 33742 35140 36973 38063 38241 CO 69777 72809 72989 72329 71084 CX 25006726 31144145 35011739 37259209 38481713 NG 266286 234874 161701 126077 126077 A-82. Thermal Usage 2010 2015 2020 2025 2030 % Thermal 60% 71% 77% 79% 81% Gas (EJ) 0.818 1.132 1.514 1.811 2.018 Billion (m^3) 21 30 40 47 53 A-83. Cumulative Present Values 2006-2055 (1995 B.O.Y Million $) Production Emission Shortage Fixed O&M Capital Net Cost $71,130 $0 $0 $10,171 $21,110 $102,412 A-84. Market Revenues and Costs 2006-2055 (1995 B.O.Y. Million $) Energy Commit Spin Total Fuel Variable Revenue O&M $168,077 $14,713 $1,544 $184,335 $62,202 $8,561 Fixed O&M Capital Total Cost Profit $10,171 $21,110 $102,046 $82,288 88 APPENDIX B The Expansion Planning Logic of Elfin B.1 Traditional Cost-Minimizing Capacity Expansion Planning The tradition of electric utility expansion planning using production cost models is based on the paradigm of the centralized, vertically integrated company, and applies cost-minimizing assumptions. The objective function is a grand net present cost function, C, which is the discounted cost of all utility operations from the beginning to the end of the planning period at time T. This cost function is the sum of several components as follows: N j c n,t % c t % c t g u e T n'1 C ' j t t'1 1 % d Within the Elfin context, C can be considered total net present cost, and cgn,t as the costs of running various n generating assets available to system. The denominator is the familar discounting term at a discount rate of d. The social cost of leaving energy unserved, that is, of letting the lights go out, is cut. Within traditional dispatch logic, resources are dispatched to meet load irrespective of cost. That is, demand is seen as fixed, and the need to meet it as absolute. No demand response of any type exists, although an interruptible load might be considered a supply-side asset. In other words, cut does not appear in the dispatch cost function meaning service cannot be interrupted on economic grounds alone. Elfin, unlike most expansion planning models, takes a more social welfare oriented approach to expansion planning. New capacity is built only if and only if it lowers cost, including the cost of not serving customers. Unserved load is treated no differently than other costs. The external costs of power generation, such as uninternalized environmental damage is represented by cet. These costs can be included in Elfin simulations if, appropriate values are specified by the user. Each cgn,t term can be thought of as a sum of the various elements of operating cost for a generator. These costs are normally summarized by categories of costs as follows: cgn,t = fuel costs + variable O&M (including labor) + fixed O&M + capital costs + other The other category can be a negative, if, for example, there is some subsidy, such as a renewables production credit, for which the resource is eligible. 89 APPENDIX B B.2 The MC-ITRE Algorithm In expansion planning, keeping the search area within the limits of computational tractability is accomplished by representing potential new additions as a small number of generic alternatives. Elfin uses multiple algorithms for solving the expansion planning problem but here we focus on just one alternative, MC-ITRE. The MC-ITRE algorithm searches on the C cost surface as follows: 1. A table is built of the per MW net present value (NPV) of adding or deleting each generic resource in each year of the planning period. 2. Elfin adds new units up to a user-specified limit of the available expansion options. 3. Elfin then recalculates the table with the chosen additions in place. This operation completes one iteration. 4. The next iteration is commenced and Elfin again searches for cost reducing additions and reductions. On this and all subsequent iterations, Elfin also tests the benefits of deleting prior additions from the plan. 5. When no further cost reducing additions or deletions can be found, searching ceases. However, the final lowest cost plan is further tested by swapping in and out construction choices to verify that the plan is truly is lowest cost. This search algorithm has proven to be quite stable and efficient. Some tricks are used to avoid getting trapped in a local cost minimum, but, in general, costs fall quickly as the iterations progress and a minimum is found that can be verified to be a reasonable minimum by the simple swapping of resource options in search of lower costs. B.3 Towards a Competitive Expansion Logic While traditional dispatch logic may persist in competitive market systems, clearly, expansion decision making will be performed in quite a different way from what the current centralized utility paradigm encourages. Investment decision making will be decentralized and based on individual investor returns rather than net present system operating cost. The goal here is to move Elfin's expansion planning logic incrementally towards a credible model of a competitive market system, of the kind proposed for California. The key change made to Elfin's expansion planning logic for the purposes of this study is a move away from the omnipotent centralized cost minimizing view of the old logic and towards a competitive paradigm driven by the decentralized entry decisions of new generating technologies. Remembering that the intent of expansion planning models is not the accurate simulation of actual operations, but rather 90 APPENDIX B the approximation of outcomes at a level sufficient only for mid to long run forecasting (beyond 5 years), and that computational burdens must be kept to a minimum to enable lengthy search procedures to complete, the basic logic of the approach is three-pronged. 1. It is assumed that either an ISO will continue to run system unit commitment and dispatch in a similar way as territorial utilities operate today, or competitive pressure will lead towards similar minimum cost solutions. Therefore, simulation of actual operations need be only modestly revised. (Section 4, below.) 2. The most important determinant of capacity construction under a competitive regime is free entry as far as is profitable. (Section 5, below.) 3. The search algorithm must be similar to the current one so that changes to Elfin are manageable and understandable. (Section 7, below.) B.4 Market Dispatch Logic A key initial assumption made here is that overall unit commitment and dispatch will tend towards the same sort of result current models would achieve for the same system and demand; that is, the cost minimum solution subject to constraints imposed by limits on various operations will be the outcome of both traditional and ISO dispatch unit commitment. The significant difference is the manner in which investments in new capacity are made. Elfin does not currently have good multi-area modeling capability that might be used to simulate the effect of local transmission constraints, and, therefore, strategic bidding is assumed non- existent. The modifications required to Elfin are manageable and need only address the fact that payments from the market will diverge from the simple minimum cost in the following minor ways. 1. An energy payment accrues to each generator that produces during a period. The payment is equal to the generators output times a weighted market price. The weighted market price is the sum of bid prices of generators that emerged as the marginal one dispatched during the period weighted by the share of the time each was marginal. 2. A commit payment is assumed to exist. This payment is made to the last generator committed during a period if it fails to break even from its market revenues. The payment simply makes this last generator whole and is given to all generators who are committed during the period. 3. A spin payment is assumed to exist. This payment is made to any generator whose output was curtailed to meet the spinning requirement even though it bid below the 91 APPENDIX B market price. The payment is specific to the generator ramped down and is equal to the lost revenue that it would have collected if it had been free to generate. Consequently, the revenue stream obtained by any generator is the sum of three payment types, although the energy payment is by far the largest of the three. B.5 How Much Entry Will Occur A net present profit function can be written for the industry as follows: Bi = Bx + Be where, Bi = profits of the industry as a whole Bx = profits of existing generating assets, and Be = profits earned by entering generating assets The paradigm adopted is one of competing technologies. Consider first the profits accruing to exiting capacity, Bi. Given that dispatch in this study almost follows the traditional rules and no strategic bidding exists, existing generating capacity is essentially passive. It has no control over its profit function and passively accepts its lot. If its net present market revenues exceed its net present costs, then it generates profits, otherwise not. However, since these generating assets are typically largely depreciated and bid into the market at their marginal cost including variable O&M, only failure to cover fixed O&M results in losses. In a sense, existing assets have no entry decisions to make. Their profits will most likely be highest if no entry occurs, thereby pushing up market prices, and vice-versa. The one complication is that in some cases, the retirement of existing units is linked to repowers at the same site. From this perspective, this amounts to a unit being removed from the existing term of the profit function as its repower appears in the entering term. For the repower to be profitable, the overall profitability of the site must exceed the profit stream at the site were the existing plant to remain in place. The focus here, of course, is on profits accruing to entering capacity, Be. Note that, from a modeling point of view, entering capacity never becomes existing capacity. Entering capacity covers all capacity built throughout the study period. There are two fundamental assumptions governing entry. First, entry by at least one technology is unrestricted, and second, investors as a group will try to establish the pattern of new entry that will result in their own maximum profit. Consider a breakdown of the entering capacity net present profit function by technology and year of construction. That is, capacity net present profit function by technology and year of construction. That is, 92 APPENDIX B H T B ' j j Bh ,t e h ' 1 t' 1 where, h = entering technologies, i.e. nuclear, gas combined cycle, wind, etc. and t = the years of the study period The net present profit function for any technology built in any year, Bh, t, depends on how much total capacity is built which will determine the revenue stream from market payments, and, of course, on the costs of the technology. This perspective essentially treats the construction of units of one given technology in one given year as a separate competitor. Since all units of a given technology are identical, and clearly more capacity lowers the market price, we can picture this profit function as follows. In Figure B-1, the first two units of technology h built in year t generate profits and the net present profit function stays positive. If the third unit is built, however, the profit function turns negative. By the rule that all profitable entry occurs, this industry will build two units in this year. This rule is equivalent to saying that while we are looking at one technology as represented by one industry, it is, nonetheless, a competitive industry and it cannot increase Figure B-1. Technology Profit Function Last Profitable Unit πh, t (x) 0 1 2 Units of Technology, h 93 APPENDIX B profits by restricting entry. Further, all producers in this industry are homogeneous; that is, all plants built are identical. However, for at least two reasons, positive profits can exist. First, the lumpiness of this and all other technologies precludes entry to the point of zero profits. This phenomenon makes careful specification of generic resources imperative. The realities of limited computation time require the use of as small a number of generic options as possible, and the specification of large generic resources. However, if resources are specified as too large, artificial lumpiness is being introduced. This is a particularly big problem with renewable technologies, such as wind, which, obviously, could be built on quite small scales. And second, since the Elfin algorithm looks at all years in deciding which additions to choose, not absolutely all profitable entry is made in every year. Elfin chooses the most profitable entry throughout the forecast period, which means that in the short-run, entry will not continue to the zero profit point. Note that the value of the net present profit function is much more complex than it seems because it depends not only on the build of this technology but also on the decisions made by all other technologies regarding their builds. Bh, t Xh, t , P(Xh, t , Yh, t , Zt), Ch, t That is, the net present profit for technology h in year t depends not only on Xh, t, the number of units of technology h built in year t, but also on the stream of expected market revenues, P, which depends of the capacity of technology h built in all years, Xh,t, and the capacity build decisions of all other entrant technologies in all years, Yh,t, and on the existing capacity stock in all years, Zt, and the net present costs of construction and operation, Ch, t. The profit function of entering technologies, Be, can be further broken down into those technologies that have limited entry and those which are unlimited. The later category are the true generic resources. How many units of these technologies are built is entirely at the discretion of the model. And, each simulation must contain at least one unlimited technology if entry is to be truly free. The limited entry technologies are more troubling. For example, consider a specific type of geothermal site on which no more than two generating units can be constructed. This technology will benefit if the two units are built and yield profits, but, assuming that they are built, they will benefit the most if further entry is limited. Therefore, it seems at first blush that these technologies must be excluded from consideration in the same way that existing technologies are excluded. However, this is not so. The key to this paradox is that entry can still occur, even though not of this specific technology. Investors will seek the entry combination that results in maximum profit including the limited entry technologies. Even though a plan in which both of the limited entry technology units are built may seem inadmissable, in fact it is. A combination of new construction under which no further new capacity can be profitably built; that is, no additional entry is possible, is called a market equilibrium plan (MEP). Obviously, multiple MEPs exist for any combination of expansion alternatives. 94 APPENDIX B B.6 Finding the Best Plan Once as many of the MEPs as possible have been identified, choosing a winner from them is trivial. In as much as those investing in the industry will choose the plan that maximizes their private profit, the plan with the highest Be must be selected. B.7 Revised Algorithm Given the goal of approximating a market system with free entry rather than traditional cost minimization, the following adjusted MC-ITRE algorithm has been developed and implemented in Elfin: 1. Because there is good reason to believe that MEPs lie close to the minimum cost point, and because minimum cost searches are efficient and stable, the minimum cost point is found. 2. Beginning at the minimum cost point, the first step in the algorithm, then is to build a table akin to the MC-ITRE table that shows whether any entry by a given technology in a given year can be profitable, given all other entry (and exit) decisions. The basic format of MC-ITRE is retained. For example, all decisions are made in discrete one-year time steps. If there is potentially profitable entry, then it assumed to take place. 3. This process continues until all the entrant profit functions are positive, but if a unit of any technology anywhere is built, then the profit function of its industry turns negative; that is, given the response of all other technologies, the last unit built loses money, which, because by definition, all units of a given technology are homogeneous, means they all lose money. 4. Unfortunately, because the profit surface is craggy but fairly level, numerous combinations of construction may meet this basic criterion. Therefore, the search algorithm must make subsequent searches in such a way that as many candidate plans as possible are identified in an unbiased manner. 5. When Elfin finds itself searching in a place it has visited before, searching ceases. 6. A swap step attempts to find new productive areas for searching. 7. When as many MEPs as possible have been found, the one with the highest entrant profit is selected as the winner. 95 APPENDIX B B.8 Conclusion A variation on the MC-ITRE algorithm has been developed to simulate entry into a competitive electricity market. Estimates are made of the profitability of construction of one new unit of generating capacity for each candidate technology. The most profitable capacity in the most profitable year is built first and the future operation of the system resimulated with the additions in place. Subsequent iterations add more profitable entry until no more is possible, combination of investments called an MEP. The choice between multiple MEPs is made so as to maximize overall profits to entrants. This algorithm has been implemented in Elfin together with a system of energy, commit, and spin payments. 96 APPENDIX C Resource Options C.1 Overview In this appendix, we summarize the ranges of costs and operational parameters that were found in the literature for the 12 potential new resource additions modeled in this project and the assumptions that we ultimately used to model these resources. C.2 Ranges of Costs and Other Operational Parameters In this section, we describe the range of cost and other parameters we found in the literature for the 12 resource options that we included in our data set. The primary sources for this information were EPRI (1993), U.S. DOE (1994), and the resource characteristics of California utilities found in the Elfin data sets created for the 1994 Electricity Report. In each section, we summarize costs and other parameters in a table to facilitate side-by-side comparison of the range of cost and other parameters that we found from these various sources. We have converted all of the cost figures to 1995 dollars using the Gross Domestic Product (GDP) implicit price deflator and assume that natural gas costs will be the same for new units as for existing ones. In most cases, we found very wide ranges of costs for these 12 technologies and, thus, we have represented the resource options with base case, high, and low capital costs. We have also chosen representative parameters from the ranges presented for plant size, plant life, variable O&M, fixed O&M, heat rates, fuel costs, forced outage rates, and maintenance rates. The costs and other parameters ultimately selected for inclusion in this analysis are presented in Section C.3.3. C.2.1 Gas Combined Cycle We found a wide range of costs for combined cycle (CC) technologies, with capital costs ranging from approximately $600/kW to $1,400/kW (see Table C-1). Siting differences explain at least some of these differences. Construction of CCs at existing sites with appropriate infrastructure tends to cost less than new sites, with potentially more stringent permit requirements and possible public opposition. For this study, we use a range of different capital costs: $600/kW for the base case, $500/kW for the low-cost case, and $800/kWh for the high-cost case. The base case of $600/kW is consistent with EPRI (1993), SDG&E ER94 data, and Hadley, Hill, and Perlack (1993). The low-cost assumption assumes technical progress by 2005, which is the year in which we consider resource additions. In $ addition, we use the other data elements specified by EPRI (e.g., 225 MW for plant size, 30 kW@a for fixed O&M, etc.). EPRI’s variable and fixed O&M costs differ from the utilities’ because EPRI assumes that more of the O&M costs are fixed and the utilities assume that more are 97 APPENDIX C variable. The fuel for this and all gas-fired technologies is ordinary natural gas, priced equally for all technologies. Table C-1. Gas Combined Cycle Costs and Other Parameters ER94, SCE Option ER94 SCE Option ER94 SDG&E ER SDG&E EPRI Tag (1993) #7, Existing Site #10, New Site Option #13, Option #10, Out- Data Elements CT/ CC 16.3 In-Basin Out-of-Basin In-County CC of-County CC Capital Costs 623 979 1384 702 (692 for 2) 794 (911 for 2) (1995$/kW) Fixed O&M Costs 27.8 10.2 10.2 8.84 8.85 (1995 $ ) kW@a Variable O&M 0.0004 0.0027 0.0027 0.0040 0.0040 Costs (0.87% real esc.) (0.87% real esc.) (0.66% real esc.) (0.66% real esc.) (1995$/kWh) Heat Rate (AHR 11,089 - 56 8,810 - 124 9,443 - 115 11,848 - 44 11,896 - 43 kJ/kWh) - Block 8,778 - 113 8,388 - 157 9,021 - 146 8,552 - 131 8,616 - 129 Size (MW) 7,934 - 169 7,702 - 210 8,229 - 195 7,808 - 218 7,840 - 216 7,702 - 225 8,156 - 292 8,189 - 289 7,934 - AA 7,770 - 366 7,801 - 362 7,723 - 436 7,755 - 428 Forced Outage 4.6% 3% 3% 4.2% 4.2% Rate Maintenance Rate 6.9% 5% 5% 4.2% 4.2% Unit Capacity 225 210 195 472 (NC) 464 (NC) (MW) 436 (DC) 428 (DC) Plant Life (a) 30 29 29 30 30 +We have converted all of the figures into 1995 dollars using the GDP Implicit Price Deflator. ++Natural gas fuel costs will be the same for existing and new units in the Elfin model. +++ Heat Rate and Block Sizes for SDG&E are summer values (June to October). AA = Average Annual NC = net capacity DC = dependable capacity for reliability calculations C.2.2 Repowers As discussed elsewhere in this report, repowers are one of the most important yet difficult to characterize capacity options. On reason repowers are inherently problematic resources in capacity expansion modeling is because repower projects are unique to specific sites and equipment, whereas the computational constraints of modeling dictate that expansion options be as small a set of generic options as possible. In other words, it is inherently difficult to represent repower resources as a generic option. In addition, possible repower options at any one site are numerous and, obviously, the choice of any one project will have a major impact on other projects. For the purposes of this study, the data used for repowers was a low-end 98 APPENDIX C estimate made on the basis of green field gas-fired combined cycle technology and ER94 data on potential repowers. The fuel is ordinary natural gas. C.2.3 Gas Combustion Turbine We also found a wide range of costs for gas combustion turbines (see Table C-2). We use $450/kW as the base-case option, $350/kW for the low-cost option, and $600/kW for the high-cost option. The base-case costs are consistent with EPRI (1993), the low cost are consistent with Hadley, Hill, and Perlack (1993), and the high costs are consistent with SCE’s ER94 data set. The low-cost assumption assumes technical progress by 2005, which is the year in which we consider resource additions. We use EPRI operational parameters (e.g., fixed and variable O&M, plant capacity, etc.), except for forced outage and maintenance rates, where we use SCE’s values. EPRI’s variable O&M costs are substantially lower than the utilities’. Table C-2. Gas Combustion Turbine Costs and Other Parameters ER94, SCE ER94, SCE EPRI Tag Option #11, Option #12, ER94, SDG&E ER94, SDG&E (1993) Existing Site In- Existing Site Option #16, Option #18, Data Elements CT 15.4 Basin Out-of-Basin GT-GE7F GT-LM 6000 Capital Costs 453 610 1330 621 1047 (1995$/kW) Fixed O&M Costs 10.7 8.2 8.2 3 4.3 (1995 $ ) kW@a Variable O&M 0.0001 0.0053 0.0053 0.0061 0.0087 Costs (0.87% real esc.) (0.87% real esc.) (0.66% real esc.) (0.66% real esc.) (1995$/kWh) Heat Rate (AHR 18,031 - 38 11,817 - 144 12,080 - 139 32,707 - 15 29,177 - 8 kJ/kWh) - Block 13,230 - 75 19,191 - 42 16,209 - 10 Size (MW) 11,827 - 113 14474 - 69 13,399 - 17 11,711 - 150 12,959 - 97 12,157 - 24 12,882 - AA 11,966 - 124 11,195 - 31 11,682 - 151 10,806 - 38 Forced Outage 10.4% 4% 4% 4.2% 4.2% Rate Maintenance Rate 6.9% 4% 4% 4.2% 4.2% Unit Capacity 150 144 139 151 (DC) 38 (DC) (MW) 163 (NC) 42 (NC) Plant Life (a) 30 29 29 24 24 +We have converted all of the figures into 1995 dollars using the GDP Implicit Price Deflator. ++Natural gas fuel costs will be the same for existing and new units in the Elfin model. +++ Heat Rate and Block Sizes for SDG&E are summer values (June to October). AA = Average Annual DC = dependable capacity NC = net capacity for reliability calculations 99 APPENDIX C C.2.4 Wind Wind capital costs range from a low of $620/kW in 2005 to a high of $1,600/kW for at least one SDG&E option. In addition, DOE (1996) estimates that capital costs will move from a current $825/kW to $625/kW in 2030. The ranges we use in our analysis differ slightly from those found in Table C-3. For the base-case capital costs, we assume that capital costs are $900/kW in 1995 and fall to $600/kW in 2030. These costs are generally consistent with DOE (1996), EPRI (1993), Wiser and Kahn (1996), Hadley, Hill, and Perlack (1993), Hamrin and Rader (1993), and Williams and Bateman (1995). For low capital costs, we assume that capital costs fall from $800/kW in 1995 to $500/kW in 2026. For the high cost, we assume that current prices of approximately $900/kW remain constant. We assume fixed O&M costs of 26 kW@a $ , no variable costs, maintenance rates of 2.5 percent, forced outage rates of zero percent, and a nameplate capacity of 250 MW. Table C-3. Wind Plant Costs and Other Parameters EPRI Tag (1993) EPRI Tag (1993) ER94 Wind 24.1 Wind 24.1 ER94 SCE ER94 SCE SDG&E ER94 SDG&E Data Elements (1995 Costs) (2005 Costs) Option #21 Option #20 Option #28 Option #42 Capital Costs 860 620 1159 969 1632 957 (1995$/kW) Fixed O&M Costs 26.4 26.4 15.5 15.5 71.8 2.6 (1995 $ ) kW@a Variable O&M 0 0 0.0082 0.0082 0.014 0.014 Costs (0.87% real (0.87% real (0.66% real (0.66% real (1995$/kWh) esc.) esc.) esc) esc) Load Shape see WIN1 see USWP Forced Outage 2.5% 2.5% 5.8% 5.8% Rate Maintenance Rate 2.5% 2.5% 4.8% 4.8% Plant Capacity 50 (NC) 50 (NC) 50 (DC) 50 (DC) 11 (DC) 12 (DC) (MW) 250 (NC) 250 (NC) 75 (NC) 80 (NC) Plant Life (a) 30 30 29 29 20 50 +We have converted all of the figures into 1995 dollars using the GDP Implicit Price Deflator. ++EPRI Tag O&M numbers are expected to decline in the future. We have used 1995 and 2005 capital costs DC = dependable capacity NC = net capacity for reliability calculation 100 APPENDIX C C.2.5 Wind with Combustion Turbine Backup There were no existing capital costs or operational parameters for a wind plant backed up by a combustion turbine (CT). We combined the costs of a wind and a CT plant for the capital cost options. We assume that costs fall from $1,350 to $1,050 in 2030 for the base case, that costs fall from $1,150 to $850 in 2030 for the low-cost case, and that costs remain constant to $1,500/kW for the high-cost case. We assume a 250 MW facility, with maintenance and forced outage rates of four percent, no variable O&M costs, and fixed costs O&M costs of 40 $ kW@a . We assume no fuel costs when the wind plants are generating energy and a heat rate of 12,000 units when the CT is operating and using gas. Emissions are the same as for the CT provided above. C.2.6 Geothermal Table C-4. Geothermal Costs and Other Parameters EPRI Tag EPRI Tag (1993) ER 94 SCE ER94 SCE ER94 SDG&E ER94 SDG&E (1993) Dual Flash Option #35, Option #22, Option #24, Option #26, Data Elements Binary 21.1 21.2 Binary Dual Flash Binary Dual Flash Capital Costs 2158 1275 4658 4244 4359 3891 Fixed O&M Costs 51.5 39.1 192.2 268.5 192.2 107.2 Variable O&M 0 0 0.0015 0.0068 0.015 0.0099 Costs (0.87% real (0.87% real (0.66% real (0.66% real esc) esc) esc) esc) Heat Rate (AHR 43,732 - 3 29,753 - 24 kJ/kWh) - Block 37,233 - 7 Size (MW) 36,446 - 10 36,273 - 13 35,144 - AA 30,649 - AA Forced Outage 1.5% 1.0% 5% 5% 7% 4% Rate Maintenance Rate 2.3% 2.7% 3% 3% 3.1% 3.8% Unit Capacity 2 x 13 MW 24 MW 100 MW 100 MW 30 (DC) 33 (DC) (MW) 70 (NC) 40 (NC) Plant Life 30 30 29 30 25 25 +We have converted all of the figures into 1995 dollars using the GDP Implicit Price Deflator. ++No fuel costs because steam field has been purchased for the utilities' options. AA = Average Annual In addition to the numbers presented above (Table C-4), DOE technology characterizations presents capital cost figures for geothermal binary, geothermal flashed steam, and geothermal hot dry rock. For a 30-MW geothermal binary plant, DOE estimates capital costs of $ $3,590/kW for 1995 falling to $1,870/kW in 2030 and O&M costs of 114 kW@a in 1995 falling to $58/kW in 2030. For 50-MW geothermal flashed steam plant, DOE estimates capital costs 101 APPENDIX C of $2,310 in 1995 falling to $1,560/kW in 2030 and O&M costs of 124 kW@a$ in 1995 falling to $ 60 kW@a in 2030. Finally, for a 10-15 MW geothermal hot dry rock plant, DOE estimates capital costs of $5,640/kW for a base system, $2,530/kW for a second generation system, and $ $ $ $1,880/kW for a goal system, with corresponding O&M costs of 181 kW@a , 78 kW@a , and 62 kW@a . These figures are in 1990 dollars, so 1995 values would be about 7.5 percent higher. For our analysis, we use capital costs of $2,300 falling to $1,600 in 2030 for the base case, with most of the decrease occurring by 2000. This assumption is taken from DOE (1996) and is consistent with Hadley, Hill, and Perlack (1993), Hamrin and Rader (1993), and Williams and Bateman (1995). We use $1,300/kW for the low cost case, which is consistent with EPRI (1993). Finally, we use $2,300/kW for the high cost case, essentially using DOE (1996) numbers but assuming no technological innovation or cost decreases over time. We also use $ low, medium, and high fixed O&M costs of 40 kW@a , 50 kW@a $ , and 190 $ . We assume a plant kW@a size of 100 MW, a forced outage rate of five percent, maintenance rate of three percent, no variable costs, and fuel costs similar to PG&E geothermal facilities (i.e., heat rate of 22,000 and steam price of $0.63/mbtu). 102 APPENDIX C C.2.7 Solar Thermal Table C-5. Solar Thermal Costs and Other Parameters ER94 EPRI Tag ER94 SCE ER94 SCE SDG&E (1993) Option #18, ST Option #19, ST Option #33, ER94 SDG&E Option Data Elements 23.1 Hybrid w/gas w/o gas ST Pond #32, ST w/gas Capital Costs 3399 5150 4862 6085 3575 (1995$/kW) Fixed O&M 35.7 61.3 53.6 82.9 40.9 Costs (1995 $ ) kW@a Variabe O&M 0.0049 0.0102 0.0102 0 0.0041 Costs does this (0.87% real esc) (0.87% real (0.66% real esc) (1995$/kWh) include esc) fuel? Heat Rate for 12,977 - 20 Gas (AHR 11,922 - 40 kJ/kWh) - Block 10,867 - 60 Size (MW) 9,835 - 80 Forced Outage 4% 7% 7% 2% 7% Rate Maintenance 3.8% 7% 5% 2% 3.8% Rate Unit Capacity 80 80 80 3 (DC) 80 (DC) (MW) 5 (NC) 91 (NC) Plant Life (a) 30 33 33 30 30 +We have converted all of the figures into 1995 dollars using the GDP Implicit Price Deflator. ++Backup fuel is gas. In addition to these figures (Table C-5), DOE (1994) provides capital cost figures for the following solar thermal technologies: power tower system, parabolic dish, 7.5-kW module, parabolic dish, 25-kW module, parabolic through power plant. The capital costs and O&M costs for these technologies are as follows: Table C-6. DOE Solar Thermal Plant Costs Solar Thermal Solar Thermal Parabolic Dish, 7.5 Parabolic Dish 25 kW Solar Thermal Parabolic Power Tower System kW Module Module Through Power Plant Capital Costs 2,310 in 2000 falling to 5,700 in 1995 falling to 2,000 in 2005 falling to 3,125 in 1995 falling to (1995$/kW) 2,240 in 2005 3,800 in 2000 1,400 in 2012 2,573 in 2000 Fixed O&M Costs 28 in 2000 falling 77 45 in 2005 falling to 34 52 in 1995 falling to 45 in $ to 25 in 2005 in 2010 and 23 in 2020 2000 and 33 in 2010 (1995 kW@a ) For the base case, we assume that costs fall from $3,100 in 1995 to $2,600 in 2000. This 103 APPENDIX C assumption is consistent with DOE (1996) estimates for solar thermal parabolic through technology and Hadley, Hill and Perlack (1993), Hamrin and Rader (1993), and Williams and Bateman (1995). For the low cost scenario, we assume that capital costs fall and remain constant at about $2,200/kW. This is consistent with DOE’s estimate for a power tower system. Finally, for high capital costs we assume that solar thermal costs remain constant at $3,100/kW. In addition, we generally use the operational parameters specified by EPRI, although we assume fixed costs of $55/kW, maintenance rates of five percent, and forced outage rates of seven percent (from SCE). C.2.8 Photovoltaic Table C-7. PV Plant Costs and Other Parameters EPRI Tag EPRI Tag EPRI Tag EPRI Tag SDG&E (1993) (1993) (1993) (1993) SCE PV Central Distributed Data Elements 22.1B 22.1C 22.3 22.3 Station Flat Plate PV Capital Costs 2986 2659 3463 2870 2592 5067 (1995$/kW) Fixed O&M Costs 9.2 6.6 23.4 6 5.9 11.3 (1995 $ ) kW@a Variable O&M 0 0 0 0 0 0 Costs (1995$/kWh) Load Shapes or Heat Rates for Gas? Forced Outage Rate 3% 3% 3.% 3% 1% 7% Maintenance Rate 3.8% 3.8% 3.8% 3.8% 1% 1% Plant Capacity 5 50 5 50 50 2.5 (MW) Plant Life(a) 30 30 30 30 32 20 +We have converted all of the figures into 1995 dollars using the GDP Implicit Price Deflator. In addition to the values given above, DOE (1994) estimates that costs for a concentrating photovoltaic plant would be about $5,000/kW in 1995 falling to $1,200 in 2028. We use this estimate as our base case. For our low-cost case, we assume that photovoltaic costs would fall from $4,000/kW to $1,000/kW in 2020. Finally, for our high-cost case, we assume that costs would fall from $5,000/kW to $3,000/kW in 2030. We are aware current photovoltaic costs are about $7,000/kW for small projects, but we are assuming that with larger scale projects, current costs could easily fall to $5,000/kW and to $4,000/kW in the most optimistic scenario. For other operational parameters, we rely primarily upon EPRI (1993). 104 APPENDIX C C.2.9 Nuclear Table C-8. Nuclear Plant Costs and Other Parameters Data Elements EPRI Tag EPRI Tag EPRI Tag (1993) (1993) (1993) 28.1 - 28.2 - Passive 28.3 - Evolutionary Safety ALWR ALMR ALWR Capital Costs 1562 1938 1818 Fixed O&M Costs 63.2 78.9 68.7 (1995$/kW) Variable O&M Costs 0.0003 0.0031 0.0003 (1995$/kWh) Heat Rate (AHR kJ/kWh) - 10,762 - 1350 10,973 - 600 10,271 - Block Size (MW) 11,089 - AA 11,300 - AA 1,488 10,582 - AA Forced Outage Rate 9.8% 7.7% 5.3% Maintenance Rate 8.2% 7.3% 3.7% Unit Capacity (MW) 1350 600 1488 Plant Life 30 30 30 +We have converted all of the figures into 1995 dollars using the GDP Implicit Price Deflator. AA = Average Annual For the base-case scenario, we assume capital costs of $2,500/kW, which is consistent with EPRI (1993). For low capital costs, we assume $1,800/kW and for high costs, we assume $5,000/kW, roughly the cost of building the Diablo Canyon plant in California. We assume variable O&M costs of 0.3 ¢/kWh, fixed O&M costs of 75 kW@a$ , and fuel costs of 0.5 ¢/kWh. 105 APPENDIX C C.2.10 Coal Gasification—Combined Cycle Table C-9. Coal Gasification Combined Cycle Plant Costs and Other Parameters EPRI Tag EPRI Tag 10.1A - 10.3A - EPRI Tag 10.4B Entrained Flow - Entrained - Moving Bed - Medium Flow - High Data Elements Integration Nonintegrated Integration SCE #30 Capital Costs 1776 2044 1784 2829 Fixed O&M Costs 53.6 61.9 52.5 20.4 (1995 $ ) kW@a Variable O&M 0.0004 0.0006 0.0015 0.0112 Costs (0.87% real (1995$/kWh) esc) Heat Rate (AHR 16,058 - 125 15,984 - 125 15,024 - 125 12,661 - 93 kJ/kWh) - Block 11,426 - 250 11,374 - 250 10,688 - 250 11,817 - 186 Size (MW) 9,833 - 375 9,791 - 375 9,205 - 375 10,129 - 278 9,211 - 500 9,168 - 500 8,620 - 500 9,707 - 371 9,485 - AA 9,443 - AA 8,884 - AA Forced Outage 10.1% 11.6% 10.1% 4.6% Rate Maintenance Rate 4.7% 4.7% 4.7% 3% Unit Capacity 500 500 500 371 (MW) Plant Life 30 30 30 29 +We have converted all of the figures into 1995 dollars using the GDP Implicit Price Deflator. AA = Average Annual For base case capital costs, we use $2,000/kW falling to $1,500 in 2030. For low costs, we assume $1,500/kW and for high costs, we use $2,800/kW, which is consistent with SCE’s ER94 data set. We use EPRI’s operational parameters for an entrained flow—medium integration unit. 106 APPENDIX C C.2.11 Advanced Coal Table C-10. Advanced Coal Plant Costs and Other Parameters EPRI 6.1 - EPRI 5.1A - EPRI 5.5A - Pressurized EPRI 6.2 - Pressurized Atmospheric Atmoshpric Fuidized-Bed Fuidized-Bed Fluidized-Bed Fluidized- Bed Combustion - Combustion - Bubbling SDG&E #22 - Combustion- Combustion - Bubbling - - Subcritical (non Atmospheric Data Elements Bubbling Circulating Subcritical (reheat) reheat) Fluidized Bed Capital Costs 1630 1956 2109 1448 3111 (1995$/kW) Fixed O&M Costs 37.9 39.4 7435 44.8 41.7 (1995 $ ) kW@a Variable O&M 0.0024 0.0013 0.0036 0.0035 0.0041 Costs (1995$/kWh) Heat Rate (AHR 13,777 - 50 14,215 - 50 - 20 - 20 15,332 - 48 kJ/kWh) - Block 11,321 - 100 11,679 - 100 9,976 - 40 10,196 - 40 15,012 - 77 Size (MW) 10,717 - 150 11,058 - 150 9,501 - 60 9,712 - 60 12,928 - 106 10,521 - 200 10,855 - 200 9,248 - 80 9,452 - 80 12,780 - 134 10,731 - AA 11,072 - AA 9,525 - AA 9,736 - AA 12,338 - 163 12,265 - 192 Forced Outage Rate 4.7% 4.1% 11.7% 12.2% 1.02% Maintenance Rate 5.7% 5.7% 8% 8% 9.4% Unit Capacity (MW) 200 200 80 80 192 (DC) 220 (NC) Plant Life (a) 30 30 30 30 30 +We have converted all of the figures into 1995 dollars using the GDP Implicit Price Deflator. ++ Heat Rate and Block Sizes for SDG&E are summer values (June to October). AA = Average Annual We use $1,500/kW for base-case capital costs, consistent with EPRI (1993), $3,100/kW for high capital costs, consistent with SDG&E’s ER94 data set, and $1200/kW for low capital costs. 107 APPENDIX C C.2.12 Biomass Table C-11. Biomass Plant Costs and Other Parameters Data Elements EPRI (1993) EPRI (1993) EPRI (1993) EPRI (1993) SDG&E SDG&E 26.1 Wood 26.2 Wood 26.3 Tree 26.4B Agricultural Biomass Waste Forest Waste Capital Costs 1973 2304 1474 2279 2773 2157 Fixed O&M Costs 91.9 97.5 61.3 107.2 31.7 26.5 (1995 $ ) kW@a Variable O&M 0.0085 0.0093 0.0074 0.0093 0.0049 0.0048 Costs (0.66% real (0.66% real (1995$/kWh) esc) esc) Fuel Costs 2/mmbtu 2.5/mmbtu Heat Rate (AHR 14,658 - 50 14,627 - 50 11,241 - 100 13,050 - 100 vary vary kJ/kWh) - Block 15,098 - AA 15,066 - AA 11,578 - AA 12,740 - AA Size (MW) Forced Outage 10.0% 10.0% na na 9.8% 9.8% Rate Maintenance Rate 5.6% 5.6% na na 5.7% 5.7% Plant Capacity 50 50 100 100 17.8 (DC) 21.2 (DC) 28 (NC) 26.3 (NC) Plant Life 30 50 30 30 20 20 AA = Average Annual In addition to the data provided above, DOE (1994) estimates costs for a number of biomass technologies, including biomass to electricity direct fired technology (na), biomass power gasification system ($1,200 falling to $1,000 for this near commercial technology), and biomass power-biocrude combustion turbine ($2700 to $1,500 in 2030 for this technology currently under development). We use $2,000/kW falling to $1,500/kW for the base case. This falls within the parameters provided by SDG&E’s ER94 data set, EPRI (1993) and Hadley, Hill, and Perlack (1993), Hamrin and Rader (1993), and Williams and Bateman (1995). For low cost, we use $1,600/kW falling to $1,200/kW in 2030, which is slightly higher than the DOE costs for the biomass power gasification system. C.3 Summary of Parameters Used in this Analysis For this analysis, we include 12 resource options. For each option, we specify low, medium, and high capital costs (see Table C-12). Although non-capital costs and other parameters vary among sources, for simplicity’s sake, we only vary capital costs. For each option, we also specified the size of the plant, plant life, variable operation and maintenance (O&M), fixed O&M costs, fuel costs, forced outage rates, and maintenance rates (see Table C-13). Emissions rates are found in Table C-14 and offset values are shown in Figure C-1. 108 APPENDIX C Table C-12. Summary of Capital Costs Assumptions High Cost Case Resource Options Base Case ($/kW) ($/kW) Low Cost Case ($/kW) Combined Cycle 600 800 500 Combustion Turbine 450 600 350 Repower Wind Power Plant 900 - falling to 600 900 800 - falling to 500 in in 2030 2030 Wind Power w/CT 1,350 - falling to 1500 1,150 - falling to 850 in 1,050 in 2030 2030 Geothermal 2,300 - falling to 2300 1300 1,600 in 2030 Solar Thermal 3,100 - falling to 3100 2200 2,600 in 2000 Photovoltaic 5,000 - falling to 5,000 - falling to 4,000 - falling to 1000 in 1,200 in 2030 3,000 in 2030 2020 Nuclear 2500 5000 1800 Coal Gasification 2,000 falling to 2800 1500 1,500 in 2030 Advanced Coal 1500 3100 1200 Biomass 2,000 falling to 2000 1,600 falling to 1,200 in 1,500 in 2030 2030 Repower 500 700 400 109 Table C-13. Other Characteristics of Generic Technologies Plant Plant Variable Fixed Heat Rate (Block Size) Size Life O&M O&M kJ/kWh MWh Forced Main- (MW) (a) (1994$/kWh) (1995 $ ) Fuel Costs Pmin to Phigh Outage tenance kW@a CC 225 30 0.0004 30 $2/MMBtu (1.5%) 11,089 (56); 8,778 (113); 7,934 (169); 7,702 4.6% 6.9% (0.75%) (225) CT 150 30 0.0001 10 $2/MMBtu (1.5%) 18,031 (38); 13,230 (75); 77,827 (113); 11,711 4% 4% (0.75%) (150) Wind 250 30 0 26 na na 0% 2.5% Wind w/CT 250 30 0 40 $2/MMBtu (1.5%) 12,661 (250) 4% 4% Geothermal 100 30 0 40/50/190 $0.63/MMBtu+ 23,211 (100) 5% 3% Solar 80 30 0.0049 55 $2/MMBtu (1.5%) 12,977 (20); 11,922 (40); 10867 (60); 9835 (80) 7% 5% Thermal (0.75%) Photovoltaic 50 30 0 7 na na 3% 4% Nuclear 600 30 0.0031 75 0.005 na 7.7% 7.3% Coal 500 30 0.0004 54 $1.5/MMBtu 16,058 (125); 11,426 (250); 9,833 (375); 9,211 10.1% 4.7% Gasification (0.75%) (1.5%) (500) Advanced 30 30 0.0024 38 $1.5/MMBtu 13,777 (7.5); 11,321 (15); 10,717 (22.5); 10,521 4.7% 5.7% Coal (0.75%) (1.5%) (30) Biomass 100 30 0.0074 62 $2.50/MMBtu 10551 (100) 10% 5.6% (0.75%) Repower 400 30 0.0027 10 $2/MMBtu (1.5%) 9,232 (236); 8,810 (300); 8,177 (400) 5.0% 5.0% (0.75%) + Escalated at same rate as gas price APPENDIX C Table C-14 Emissions Characteristics of Generic Technologies (lbs/mbtu or lbs/kWh) NOx SO2 PM ROG CO Carbon cc & repower 0.005/mbtu 0.001/mbtu 0.013/mbtu 0.008/mbtu 0.01/mbtu 33/mbtu ct 0.005/mbtu 0.001/mbtu 0.013/mbtu 0.008/mbtu 0.01/mbtu 33/mbtu repower 0.005/mbtu 0.001/mbtu 0.013/mbtu 0.008/mbtu 0.01/mbtu 33/mbtu wdct 0.005/mbtu 0.001/mbtu 0.013/mbtu 0.013/mbtu 0.01/mbtu 33/mbtu geothermal 0 0 0 0 0 0.041/kWh solar thermal 0.08/mbtu 0.001/mbtu 0.007/mbtu 0.002/mbtu 0.037/mbtu 33/mbtu photovoltaic 0 0 0 0 0 0 nuclear 0 0 0 0 0 0 coal gasification 0.0002/kWh 0.0004/kWh 0.0001/kWh 0.000001 0.000002 1.91/kWh /kWh /kWh advanced bed 0.038/mbtu 0.038/mbtu 0.013/mbtu 0.003/mbtu 0.083/mbtu 64.9/mbtu biomass 0.17/mbtu 0.03/mbtu 0.28/mbtu 0.01/mbtu 0.05/mbtu 0.0815/kWh Table C-14. Emissions Characteristics of Generic Technologies (g/kWh +) NOx SO2 PART ROG CO C cc & repower 0.017 0.003 0.043 0.027 0.033 109.369 ct 0.025 0.005 0.066 0.04 0.054 166.3 repower 0.018 0.004 0.046 0.028 0.035 116.111 wdct 0.027 0.005 0.071 0.044 0.054 179.784 geothermal 0 0 0 0 0 18.614 solar thermal 0.339 0.004 0.03 0.008 0.157 139.662 photovoltaic 0 0 0 0 0 0 nuclear 0 0 0 0 0 0 coal gasification 0.091 0.182 0.145 0.0005 0.001 867.14 advanced bed 0.172 0.172 0.059 0.014 0.376 298.746 biomass 0.772 0.136 1.271 0.045 0.227 37.001 + At Average Full Load Heat Rate 111 APPENDIX C Figure C-1. Forecasts of Offset Costs 500 Combined Cycle Combustion Turbine 400 Repower Integrated Coal Gasification/Advanced Fluidized Bed Solar Thermal Trough 300 200 100 0 1994 1996 1998 2000 2002 2004 2006 2008 2010 2012 2014 2016 2018 2020 2022 2024 2026 2028 2030 C.4 Offsets Offset costs are required for combined cycles, combustion turbines, wind plants backed by combustion turbines, solar thermal backed by gas, coal gasification, advanced coal, and biomass. The offset costs differ considerably depending upon which air quality basin the plant will be located. Combined cycles, combustion turbines, and repower show the most variation in offsets across basins. We use the lowest offset values for combined cycles, combustion turbines, and repowers. The solar thermal offset values remained constant across utilities as all were building these plants in Mohave Air Quality Management District (AQMD). We chose the highest value for integrated coal gasification (and assume that advanced coal is similar), and the only value for biomass. Biomass costs start at $475/kW in 1995 and increase well above $3,000/kW by 2030. Biomass presumably has such high offset costs because it has high emissions compared to the other technologies. No offsets are required for wind, geothermal, photovoltaic, or nuclear. 112 APPENDIX D Extreme Search Test For Market Equilibrium Plans The Elfin search for the best Market Equilibrium Plan (MEP) is not based on a global optimization procedure. Elfin starts with the minimum cost plan and searches for a profit maximizing plan which does not allow any further profitable entry. It is likely that this process is path independent. The following test was done to see if Elfin converges to the same plan when it starts at plans which are radically different. The intention of this exercise is to look more closely at Elfin’s behavior when searching for MEPs and how broad an area of possible technology combinations as MEP candidates Elfin considers. The search domain can very well be displayed in a picture of a volcanic crater. The starting point is a costly expansion plan, and could be considered a point high on the rim. The valley of the crater is a quite flat but craggy area and represents roughly the minimum cost level. Since MEPs must be close to the minimum cost solution, they can be pictured as small peaks scattered over the crater floor. The best MEP is a plan in which no further entry is possible and profits are maximized, so it can be thought of as the highest of these local peaks. In general in this work, MEPs are found by beginning with a minimum cost search, which is equivalent to finding a quick route the bottom of the crater. Then the crater is searched for MEPs. The highest peak found is declared the best MEP. In this exercise the search area for MEPs, and thus often the result (the best MEP), is shown to be path dependent, meaning dependent on the combination of technologies that serves as starting point for the MEP search. Different starting points are used and the progress of the exploration reported. While some cases cover big areas, some are local, leading to very different results for the generation plan chosen by Elfin. D.1 Procedure In our extreme search MEP test runs, the search is started from different extreme plans shown in Table D-1, rather than from the minimum cost plan. Each starting point depends on a large amount of one specific technology, in addition to combustion turbines. The start plan for the minimum cost (MINC) case is the usual minimum cost plan. 113 APPENDIX D Table D-1. Start Plans for the MEP Search Test Extreme Case NUCL SOLAR COAL WIND GEO MINC Start Plan nuke 67 solar 81 coal 67 wind 67 geo 67 ccs 10 (number of units cts 81 cts 46 cts 81 cts 81 cts 81 cts 67 of the technolo- rpcc 69 gies in place in wind 3 final year) We then ran these cases searching for market equilibrium in the neutral policy environment (ON) case. Since we did not pay attention to meeting a specific generating capacity expansion in creating our start plans, Elfin initially needed to add or delete many resources, in some cases, before starting to search for market equilibria, notably in the GEO case, where the capacity of a plant is only 100 MW . D.2 Findings Elfin found multiple potential market equilibrium plans for each case and chose the best market equilibrium plan among these, except for the COAL case which was stopped after 188 iterations. The results of the searches starting at different initial plans are shown in Table D-2 and Figure D-1. In all cases, the extreme technologies given in the start plans are deleted and only combustion turbines, repowers, and combined cycles remain in the MEPs. Combined cycles only exist in the GEO case. Two distinct clusters of MEPs can be seen in Figure D-1, ones with over 80 combustion turbines (cts) and about 65 repowers (rpcc) and ones with 45-58 cts and 79-83 rpcc. Note from Table D-1 that MEPs which appear close in the plot can exhibit wildly different profits, and that the same plan can apparently yield different profits because construction programs involving different years can reach the same end year construction totals. Looking more carefully at the cases: NUCL: The nuclear case searches in both clustered areas. The best market equilibrium plan is one with high rpcc and lower cts. MINC: The minimum cost case only searches in one of the two areas and never reaches the second cluster. Its best MEP is also one with high rpcc and lower cts. SOLAR: The solar case behaves like the minc case with lower profits. COAL: The coal case is special in that it, as the nuclear case, reaches both areas but it picks its best MEP in the opposite area, the one with high cts and lower rpcc. It is the only one which identifies this market equilibrium plan with the second highest profits of all cases of $895 M. 114 APPENDIX D GEO: The geothermal case is outstanding in it is the only three-dimensional case. It searches the area with high rpcc and lower cts as well as the one with lower rpcc and higher cts but differs in that it for the second area (high cts, low rpcc) includes a third technology (ccs). Since this case is undominated and gives highest profits, it represents the best MEP. WIND: We did not consider the wind case more closely since it gives lower profits and would not lead to more insights. Table D-2. MEPs Found by Each Search Combustion Turbines Repowers Combined Cycle Profit N1 81 66 123 N2 56 79 36 N3 53 80 88 N4 50 81 97 N5 49 81 155 N6 47 82 47 N7 47 82 76 N8 47 82 113 N9 45 83 203 N10* 45 83 265 MCP1 54 80 265 MCP2 51 81 245 MCP3 50 81 265 MCP4 48 82 72 MCP5 47 82 240 MCP6 45 83 237 MCP7* 50 81 484 MCP8 51 81 82 S1 53 80 159 S2* 53 80 188 C1* 81 65 895 C2 55 79 296 C3 56 79 206 C4 55 79 299 G1* 81 65 2 1250 G2 56 79 618 G3 58 78 405 G4 52 80 552 G5 52 80 532 G6 53 80 211 G7 50 81 230 G8 50 81 264 G9 49 81 518 G10 50 81 200 G11 49 81 566 G12 50 82 200 * best MEP found in case 115 APPENDIX D Figure D-1. MEPs Found by Each Search 85 N9, N10*, MCP6 N6, N7, N8, MCP5 G12 MCP2, MCP8 * or ** indicates best MEP found for case (N, MCP, S, C N3, S1, S2*, G6 MCP4 or G) 80 MCP1 ** This plan also includes N5, G9, combined cycle plant G4, G5 G3 G11 Repowers Note: Similar results on this 75 N4, MCP3, MCP7*, N2, C3, G2 figure generate different profit G7, G8, G10 levels due to the difference in C2, C4 timing of construction 70 N1 C1*, G1** 65 40 45 50 55 60 65 70 75 80 85 Combustion Turbines D.3 Conclusion Elfin does not necessarily reach all relevant areas in its search algorithm and does not lead to one consistent ‘best MEP’ for all different cases. This shows that the MEP search algorithm is path dependent. While for some starting plans the area may well be explored, in others it is not. This is particularly well displayed investigating the coal and solar cases. The coal case searches both clusters in Figure D-1 and finds its best MEP in the area with very high profits. The solar case, on the other hand, restricts the search to only one area and never reaches the cluster where the coal case MEP is located. This indicates that Elfin occasionally misses potentially better combinations of technologies. These results underscore the difficulty of finding the solution sought in this work. Plans that differ in small details can result in significantly divergent profits, and the path by which the results can be found is not at all clearly marked. On the other hand, qualitatively MEPs do appear to cluster and if all clusters could be found and searched, reasonable results are feasible. In future work, the search algorithm will be further refined. 116 APPENDIX E Simulation Difficulties in the B Policy Simulation We encountered a fundamental problem with the B case, unfortunately, one not amenable to ready solution. The problem is simply that meeting a target level of total subsidy payments by searching across various levels of subsidy results in a highly unstable search. Consider the per kWh and total subsidy results of the run that we report here, as shown in Figures E-1 and E-2. Figure E-1 shows the average per-kWh subsidy applied in the B policy, while Figure Figure E-1. Per kWh Subsidy to Wind - BN 3.2 2.8 2.4 2.0 cent/kWh 1.6 1.2 0.8 0.4 0.0 2006 2008 2010 2012 2014 2016 2018 2020 2022 2024 2026 2028 2030 year 117 APPENDIX E E-2 shows the total annual cost of the subsidy. The first obvious characteristic of these two graphs is that, although the level of subsidy declines in an apparently predictable manner, the overall cost of the subsidy is erratic, rising dramatically, then falling from 2013 to 2018. The second interesting feature of the subsidy level is that required subsidies are high, starting at over 3 ¢/kWh and never reaching 1 ¢/kWh. Figure E-2. Net Total Subsidy to Wind - BN 1050 900 750 1995M$ 600 450 300 150 0 2010 2012 2014 2016 2018 2020 2022 2024 2026 2028 2030 year The cause of these results is simply that the search for the correct level of subsidy required to meet a predetermined target total subsidy cost is highly unstable. This effect apparently 118 APPENDIX E arises from three sources. First, changes made in any year affect the construction choices made in all other years. Therefore, a minor change in the level of subsidy in any year will change not only the amount of subsidy collected by new resources built in that same year, but also the amount of subsidy required to meet the obligation to new generation built in all other years. Second, the level of the subsidy must not only compensate the investor in a renewable technology for the high cost of today’s technology but also for the lost opportunity to invest in future years’ improved technology. For example, if a wind generator is built this year, not only is it not viable compared to thermal generation options, but it is also not viable compared to wind technology in future years. Therefore, the subsidy must compensate for both of these effects, effectively raising the required subsidy level in early years beyond expectations. And third, increasing generation by zero marginal cost generators tends to dampen market prices, thereby diminishing the value of the subsidy to developers. These problems underscore one feature and one limitation of Elfin. First, a key feature of the ITRE logic is perfect foresight. New construction is chosen not only by technology but also by year. A new plant will never be built today if it is more profitable in present value terms to wait for a future year’s technology. An alternative search algorithm, such as ICEM, that lacks this foresight would result in more early construction of falling-cost technologies, but neither perfect nor absent foresight are credible assumptions. 5 Second, Elfin lacks the capability of imposing a fixed level of subsidy directly. While developing a suitable algorithm for this feature was beyond the scope of this project, it must be undertaken if useful analysis of fixed subsidies is to be possible. 5 The Iterative Cost Effectiveness Method (ICEM) is an alternative search algorithm to ITRE that considers investments year-by-year without foresight. 119 120 APPENDIX F Carbon Tax Policy: Policy H In Policy H, the introduction of a modest carbon tax of $3.00/tC is assumed. Our results are shown in Table F-1. Table F-1. Summary of Elfin Pool Results with a Low Carbon Tax and a Neutral Environment Neutral (ON) Carbon Tax (HN) 2030 Cumulative New Renewable (MW) 0 5000 2030 NOx Emissions (t) 224277 184,892 (82%) 2030 Carbon Emissions (t) 39683706 21,679,018 (55%) 2030 % Thermal 81% 68% 2030 Gas Consumption (EJ) 1.563 0.519 (33%) NPV System Costs ($M) $132,002 $237,190 (180%) Curiously enough, Policy H results in only 5 GW of new renewable construction (all geothermal). However, this case results in a staggering 14.4 GW of new nuclear construction, producing the most dramatic reductions in emissions, thermal dependence and gas consumption of any policy modeled relative to the neutral, or base, case. Under the policy’s carbon tax, NOx emissions are cut by 18 percent, carbon emissions by 45 percent, thermal dependence by 13 percent, and gas consumption by a stunning 67 percent. Naturally, these savings are not achieved on the back of just 5 GW of new renewable energy source capacity alone, and major source of these benefits is the nuclear construction. Total costs for Policy H are 180 percent of costs in the neutral case. Costs increase in all components of overall system costs. Production costs rise by over 50 percent, and because of the high cost of nuclear construction and O&M, these costs also rise significantly. As discussed below, Policy H represents the high end for both potential emissions reduction and total cost of any of the policies modeled here. However, the results of this carbon case should be viewed with some suspicion. The equivalent HB and HG cases result in quite different outcomes, neither of which involve nuclear. Interestingly, the HG case includes biomass, the only case in which that renewable proves competitive under our assumptions. The HB results in an outcome quite similar to the OB case, although output is lower. The results of the HN case, then, are quite disturbing. It seems that our search has not found a credible MEP in this instance. We report the HN case here primarily to demonstrate that the MEP search algorithm, described in Appendix B, exhibits quirky behavior and work is ongoing to improve its performance. However, it should be noted that the nuclear results can 121 APPENDIX F be highly sensitive to carbon tax scenarios. In other work conducted at Berkeley Lab, this effect has been investigated more rigorously. 6 6 See Marnay and Pickle (1998). 122 APPENDIX G Cost Duration Curves for the California Pool This appendix contains some cost duration curves for busbar cost. Remembering that in this work, marginal cost bidding is assumed, these curves show the basic pattern of competitive prices. Curves are shown for each fifth year from 2010 to 2030 in Figures G-1 to G-5. The qualitative pattern of prices is the same in each graph, and similar to the pattern shown in the body of this report. The numbers presented are for the best guess scenario (ON case). For most hours, the variation in prices is minimal. The peaks occur during summer afternoons. Gas is clearly the marginal fuel most of the time, and variation in prices is driven by heat rate variation. For a few hours, however, prices peak dramatically, and these peaks become more spectacular in later years. The marginal costs increase during the years 2010 to 2030. The peaking hours aside, marginal costs for the flat region in Figure G-1 range from 27 $/MWh to 19 $/MWh. This range moves up to 34 $/MWh to 26 $/MWh by the year 2030 and this escalation happens evenly during the period studied. The marginal cost during the peak hour escalates from 50 $/MWh in 2010 to about 275 $/MWh in 2025, and from there on declines to about 150$/MWh by the year 2030. One possible interpretation of these results is that since our expansion planning options contain a limited number of alternatives, once it is no longer profitable to built the technology most suited for peaking duty, no more capacity is built, resulting in a shortfall and the inevitable peak in prices. Together these graphs show more clearly the basic pattern of pool price results described in the body of this report. 123 Figure G-1. Cost Duration Curve for 2010 300 250 Marginal Busbar Cost ($/MWh) 200 150 100 50 0 0 1,000 2,000 3,000 4,000 5,000 6,000 7,000 8,000 9,000 10,000 Hours Figure G-2. Cost Duration Curve for 2015 300 250 Marginal Busbar Energy Cost ($/MWh) 200 150 100 50 0 0 1,000 2,000 3,000 4,000 5,000 6,000 7,000 8,000 9,000 10,000 Hours Figure G-3. Cost Duration Curve for 2020 300 250 Marginal Busbar Cost ($/MWh) 200 150 100 50 0 0 1,000 2,000 3,000 4,000 5,000 6,000 7,000 8,000 9,000 10,000 Hours Figure G-4. Cost Duration Curve for 2025 300 250 Marginal Busbar Energy Cost ($/MWh) 200 150 100 50 0 0 1,000 2,000 3,000 4,000 5,000 6,000 7,000 8,000 9,000 10,000 Hours . Figure G-5. Cost Duration Curve for 2030 300 250 Marginal Busbar Energy Cost ($/MWh) 200 150 100 50 0 0 1,000 2,000 3,000 4,000 5,000 6,000 7,000 8,000 9,000 10,000 Hours