Greenhouse Gas Mitigation Options in the Sri Lanka Power Sector* February 2003 Joint UNDP/World Bank Energy Sector Management Assistance Programme (ESMAP) Contents Acknowledgement........................................................................................................... v Abbreviations and Acronymns ....................................................................................... vi 1. Introduction................................................................................................................ 1 1.1 Energy and GHG Emissions in Sri Lanka ....................................................................................................2 1.2 Analytical framework.....................................................................................................................................5 Power systems modeling.............................................................................................................................5 Resolving uncertainties using the options approach.............................................................................6 The SACC......................................................................................................................................................7 The dynamic abatement cost curve (DACC) ...........................................................................................8 Emission coefficients ...................................................................................................................................9 2. Business-as-Usual (Reference) Scenario................................................................. 11 2.1 Sri Lanka's power sector..............................................................................................................................11 2.2 The demand forecast...................................................................................................................................13 2.3 Transmission and distribution (T&D) assumptions...............................................................................14 2.4 Fuel costs ......................................................................................................................................................15 2.5 The capacity expansion plan......................................................................................................................17 2.6 Results: GHG emissions for BAU scenario ..............................................................................................19 3. The Reform Scenario................................................................................................ 21 3.1 T&D loss reduction and DSM ...................................................................................................................21 3.2 Generation capacity expansion plan..........................................................................................................24 3.3 Comparison of the impacts of the Sri Lanka reform scenario with Haryana........................................26 4. Mitigation Options.................................................................................................... 27 4.1 Alternative energy options.........................................................................................................................27 Small hydro..................................................................................................................................................27 Wind power.................................................................................................................................................29 Dendro-thermal power...............................................................................................................................32 Solar photovoltaic (PV) systems for remote rural applications...........................................................34 4.2 Fuel substitution options............................................................................................................................35 Oil steam-cycle plants................................................................................................................................35 Liquid natural gas (LNG)...........................................................................................................................35 Conventional hydro ...................................................................................................................................36 Biomass co-firing........................................................................................................................................37 4.3 DSM ...............................................................................................................................................................37 4.4 Options Not Considered .............................................................................................................................38 Nuclear power.............................................................................................................................................38 Rehabilitation of thermal power plants ...................................................................................................39 4.5 Results ...........................................................................................................................................................39 Impact on local air emissions....................................................................................................................43 Comparison to Indian GHG overlay studies...........................................................................................43 5. GHG Mitigation Scenarios ....................................................................................... 47 5.1 Combining mitigation options....................................................................................................................47 5.2 Developing the SACC .................................................................................................................................49 6. Resolving Uncertainty.............................................................................................. 51 6.1 Sources of uncertainty and the real options approach ..........................................................................51 6.2 Applying the options approach.................................................................................................................52 7. Summary and Conclusions ........................................................................................ 65 iii 7.1 Reform............................................................................................................................................................65 7.2 Residual oil....................................................................................................................................................65 7.3 Implementing the least-cost option...........................................................................................................65 7.4 Dendro-thermal power.................................................................................................................................65 7.5 Wind...............................................................................................................................................................66 7.6 Solar power....................................................................................................................................................66 7.7 Imported fuel oil............................................................................................................................................66 7.8 Transportation and liquid fuels ..................................................................................................................67 7.9 Dealing with uncertainty.............................................................................................................................67 Reference........................................................................................................................ 69 iv Acknowledgement This study was carried out in 1999-2000, with assistance from the Danish Trust Fund (Global Overlays Program), the ESMAP Energy-Environmental Review Program, and the World Bank Research Committee, by a team consisting of Mohan Munasinghe (ENVDR, Team Leader), Peter Meier (Consultant), Chitru Fernando (Consultant), Noreen Beg (ENVDR), and Sri Lankan counterparts including D. C. Wijeyratne (CEB) and Shavi Fernando (CEB). Valuable comments provided by the staff of the South Asia region and editorial assistance from Kate Sullivan and Nishanthi De Silva, are gratefully acknowledged. The document was prepared at the Munasinghe Institute for Development (MIND), Colombo -- a founding partner of the World Bank Global Village Energy Partnership (GVEP). v Abbreviations and Acronyms BAU business-as-usual CCGT combined cycle gas turbine CDM Clean Development Mechanism CFL compact fluorescent lightbulbs CGE computable general equilibrium DACC dynamic abatement cost curve DSM demand side management EEAC energy-efficient air conditioning EEAC energy-efficient air conditioning ENPEP Energy and Power Evaluation Program EPRI Electric Power Research Inst. ESDP Energy Services Delivery Project ESP electrostatic precipitator ICB initial contract bid IDA International Development Assoc IDC interest during construction IPCC Intergovernmental Panel on Climate Change IPPs independent power producers IRRs internal rates of returns JI joint implementation LECO Lanka Electricity Company LNG liquid natural gas LOLP loss of load probability LoS low sulphur LRMC long run marginal cost LT low tension MoU memorandum of understanding O&M operation & maintenance OCGT open-cycle combustion turbines PAD project appraisal document SACC standard abatement cost curve T&D transmission and distribution TJ terajoule UNFCC United Nations Framework on Climate Change WTP willingness-to-pay vi 1 Introduction Climate change induced by anthropogenic greenhouse gas (GHG) emissions will have severe consequences worldwide, especially in the developing countries.1 This volume contains an analysis ofGHG emission reduction options in Sri Lanka's power sector. The GHG Overlay Guidelines are taken as the general analytical framework for this study.2 Among the GHG abatement methodologies presented in the Guidelines,3 our approach falls into the "bottom-up" category. The guidelines suggest the following steps4: 1. Estimate GHG emissions for the reference (business-as-usual [BAU]) scenario, as presented in Section 2 of this volume. 2. Estimate GHG emissions for a reform scenario including pricing policy, and other interventions typically undertaken with World Bank assistance, as presented in Section 3. 3. Identify, evaluate, and screen a range of GHG mitigation options, as presented in Section 4. 4. Formulate alternative mitigation scenarios and estimate the GHG impacts of each scenario, and compare to the reform scenario to gauge the cost-effectiveness of the GHG emissions reduction options, as presented in Section 5. In addition, this study undertakes the following further step: 5. Discuss the results within a broader policy framework that seeks to deal with significant uncertainties facing Sri Lankan decisionmakers in responding to the opportunities offered by the Kyoto Protocol's flexibility instruments (i.e., Clean Development Mechanism, Joint Implementation, and eventually Emissions Trading). This is presented in Section 6. 1 See for example, C. Jepma and M. Munasinghe, 1998, and Munasinghe, 2001. 2 Global Environment Division, Guidelines for Climate Change Global Overlays, World Bank Environment Department Paper 47, Climate Change Series, Washington, D.C., February 1997 (hereinafter cited simply as "Guidelines"). 3 Guidelines, op.cit., p. 14. 4 Guidelines, op.cit., Exhibit I-3, p. 3. 1 2 Greenhouse Gas Mitigation Options in the Sri Lanka Power Sector 1.1 Energy and GHG Emissions in Sri Lanka Carbon emissions in Sri Lanka, both in absolute and per capita terms, are low even in comparison with the other countries of South Asia (see Table 1.1). For example, its CO2 emissions per capita (column 3) are 10 percent of the world average, and 40 percent of those of India. This is a consequence of the historical dominance of hydro in the electric sector, and of the relatively low energy intensity ofthe industrial sector. Table 1.1: Carbon emissions Country 1995 kg 1995 kg CO2/1990$ 1995 tons CO2 per CO2/1990$US of GDP GDP adjusted to capita PPP Sri Lanka 0.63 0.13 0.34 China 4.72 0.92 2.51 Germany 0.49 0.63 10.83 France 0.29 0.34 6.23 India 2.17 0.73 0.86 Indonesia 1.37 0.34 1.17 Japan 0.36 0.46 9.17 Pakistan 1.62 0.32 0.63 USA 0.85 0.85 19.88 Former USSR 5.22 2.61 8.48 World 0.94 0.75 3.92 GDP gross domestic product PPP purchasing power parity Source :International Energy Agency(IEA) CO2 database Indeed, until quite recently, Sri Lanka's power system was almost entirely dependent upon hydroelectricity (see Figure 1.1).5 Significant thermal generation occurred only during drought years. Consequently the power sector has contributed little to past GHG emissions.6 Moreover, even today, biomass accounts for the largest share of primary energy (Table 1.2). However, the trend toward a greater share of commercial energy is persistent and unmistakable. Since 1984, commercial energy has grown at an annual rate of 4.7 percent, while over the same period biomass energy has declined at an annual rate of 0.6 percent. 5 For a history of the development of the power sector in Sri Lanka, see P. Meier and M. Munasinghe, Incorporating Environmental Concerns into Power Sector Decisionmaking: A Case Study of Sri Lanka, World Bank Environment Department Paper 6, March 1994, Washington, D.C. 6 Carbon releases from hydro projects may safely be ignored. All of the remaining hydro projects under consideration by CEB are run-of-river with minimal pondages. Introduction 3 Figure 1.1: Historical electricity generation by type 6 5 9% 10% -11% 12% 4 5% 7% 10% GWh 3 2% 8%2% 3% 1000 9% 7% 10%2% 2 12% THERMAL 11%4%3%7% 5%7% 14%10%9% 1 HYDRO 0 1971 1973 1975 1977 1979 1981 1983 1985 1987 1989 1991 1993 1995 Table 1.2: Shares of total primary energy 1984 1996 %/year 1,000 TOE % 1,000 TOE % Biomass 4,203 71 3,925 57 -0.6 Commercial 1,700 29 2,952 43 4.71 Total 5,903 100 6,877 100 1.28 TOE = tons of oil equivalent. "Commercial" includes hydro and petroleum.7 Source: Energy Conservation Fund, Sri Lanka Energy Balance 1996. These circumstances are reflected in the results of a comprehensive GHG emissions inventory prepared by the Ministry of Forest and Environment8 (Table 1.3). However, the reported increases in power sector emissions from 1990 to 1992 are due solely to the variation of hydro generation (1991 and 1992 were dry years), rather than a change in generation capacity (since there were no capacity additions in 1990-1992). 7 The data need to be interpreted with care, because 1996 was a drought year, in which the share of hydro was lower than normal. However, 1996 is the latest year for which a complete energy balance is available. 8 Ministry of Forest and Environment, Environment Division, J. Ratnasiri, editor, Final Report of the Sri Lanka Climate Change Country Study, March 1998. 4 Greenhouse Gas Mitigation Options in the Sri Lanka Power Sector Table 1.3: Estimates of CO2 emissions by source, 1990-1992 In thousand tons 1990 1991 1992 Power (CEB) 8.5 226.6 643.2 Transport 2,214 2,317 2,334 Energy generation in industry 559 485 561 Commercial (fisheries, agriculture) 172 161 166 Domestic (LPG, kerosene) 550 578 651 Cement/lime production 249 244 436 Refinery own use 0.7 0.6 0.7 Total energy and industrial processes 3,899 4,120 4,894 (excluding biomass) Biomass 21,261 20,095 21,498 Emissions from agriculture, land use, 8,435 8,370 8,098 and waste CEB Ceylon Electricity Board LPG liquid petroleum gas Source: Ministry of Forest and Environment, Environment Division, J. Ratnasiri, editor, Final Report of the Sri Lanka Climate Change Country Study, March 1998 This situation will almost certainly change. For a number of reasons (to be discussed below), the share of hydro-electricity in Sri Lanka's energy mix will decline, as almost all new power plants presently anticipated as being necessary over the next two to three decades will be fossil-fueled. Wind power, mini hydro, and demand side management (DSM) will also increase in importance in the near future, but a substantial increase in the use of fossil fuel for power generation seems inescapable. Consequently, in the BAU case, we project that the share of GHG emissions accounted for by the power sector will increase from 18 percent of energy sector emissions in 1998 to 45 percent in 2018; in absolute terms the power sector will account for 20.1 million tons per year (mtpy) by 2018, compared with 1.7 mtpy in 1997, a more than 10-fold increase. The emissions trend--as expected on the basis of CEB's present expansion plan, and which is presented in detail in the business-as-usual scenario of Section 2--is depicted in Figure 1.2. It may be seen that transportation sector emissions have increased significantly over the 1992 estimates enumerated in Table 1.3: in 1998 these are estimated at 3.9 million tons compared with 2.3 million in 1992. The explanation lies in the extraordinary increase in transportation diesel consumption, whose annual growth rate averaged about 6.6 percent from 1986 to 1996.9 9 D. S. Jayaweera, Energy and Environment-Transportation Sector, Transport Policy Centre, Colombo, December 1998. Introduction 5 Figure 1.2: Projections of carbon emissions by energy sector 50 40 power sector 30 tons/year million 20 bunkers & aviation emissions, CO2 industry transportation 10 0 1998 2000 2002 2004 2006 2008 2010 2012 2014 2016 2018 domestic[LPG+Kerosene] 1.0 1.0 1.1 1.1 1.2 1.2 1.2 1.3 1.4 1.4 1.5 1.5 1.6 1.7 1.8 1.9 1.9 2.0 2.1 2.3 2.4 transportation 3.9 4.3 4.7 5.0 5.4 5.8 6.2 6.7 7.2 7.7 8.2 8.7 9.3 9.9 10.6 11.3 12.0 12.9 13.7 14.7 15.7 industry 1.0 1.1 1.1 1.2 1.3 1.4 1.4 1.5 1.6 1.7 1.8 1.9 2.0 2.2 2.3 2.4 2.6 2.7 2.9 3.1 3.2 bunkers & aviation 1.9 2.0 2.0 2.1 2.1 2.2 2.3 2.3 2.4 2.4 2.5 2.6 2.7 2.7 2.8 2.9 3.0 3.0 3.1 3.2 3.3 power sector 1.7 2.1 2.3 2.3 2.5 3.0 4.0 4.5 5.6 6.2 7.4 8.6 9.3 10.5 11.4 12.7 14.1 15.2 16.7 18.4 20.1 1.2 Analytical framework As noted, the study uses a typical bottom-up approach. It starts by estimating GHG emissions from the power sector associated with a reference (or BAU) scenario. The least-cost capacity expansion program is developed to satisfy the growth of electricity demand for the next 20 years, while meeting acceptable levels of reliability and environmental protection standards. Next, the GHG emissions associated with a reform scenario (including various policy options) are estimated. Several mitigation options are screened and then combined within alternative GHG mitigation scenarios, which are compared to the reference and reform scenarios. Finally, the effects of uncertainty and possible responses to the Kyoto Protocol flexibility mechanisms are analyzed. Power systems modeling Two standard least-cost power capacity expansion models are used in this study: ENVIROPLAN and WASP. ENVIROPLAN is used for mitigation option screening, benefit- cost analysis, sensitivity studies, and risk assessment. This model was developed in the early 1990s for use in a study of environmental issues in the power sector in Sri Lanka,10 but has since seen a variety of other applications, including by BCHydro for strategic planning;11 by the United States Agency for International Development (USAID) for the 1995 Integrated 10 See the annex in Meier and Munasinghe, op.cit., for a more detailed description. 11 BCHydro, 1995 Electricity Plan, Vancouver, B.C, September 1995. Volume C of this plan presents the economic and trade-off analysis. 6 Greenhouse Gas Mitigation Options in the Sri Lanka Power Sector Resource Plan of the Indian State of Andhra Pradesh,12 by the World Bank in 1997 and 1998 for the economic appraisal of power sector reform programs in the Indian states of Haryana and Andhra Pradesh,13 and by the World Bank in an ongoing study of environmental issues in the Indian power sector.14 WASP is the model used by CEB for generation expansion planning and is the basis for its official generation plan, which is issued annually.15 The version currently used by CEB is WASP-III plus, part of the Energy and Power Evaluation Program (ENPEP) software package developed by the International Atomic Energy Agency.16 This is a detailed model providing a sophisticated dynamic programming analysis for capacity expansion and probabilistic production costing subject to reliability constraints. In the study, we use WASP to confirm the ENVIROPLAN results for the three main scenarios (BAU, reform, and GHG mitigation), as well as for selected mitigation options (such as the scenario exploiting the full wind potential) where operational constraints and difficulties may only be revealed in a more detailed analysis.17 Resolving uncertainties using the options approach An important first step in the bottom-up approach is to identify the options available for reducing emissions. The standard abatement cost curve for GHG mitigation options is obtained by ranking specific GHG mitigation projects in the order of cost (or benefit in the case of win- win projects) of reducing a unit of carbon emissions (see Figure 1.3). The cost and benefit estimates of GHG mitigation are obtained by discounting future streams of costs and benefits back to the present time. This ranking implies that projects with a lower cost per unit of carbon reduction should be implemented prior to higher-cost projects. However, this static approach can be somewhat misleading in the recommendations that it provides, especially in the case of GHG mitigation projects, which can vary widely with regard to their time frame, the cost of reversal, and the uncertainty associated with future costs and benefits. This is because the standard abatement cost curve (SACC), which is based on the assessment of future costs and benefits at a particular instant in time, disregards the value of the options embedded in specific projects associated with changing the timing of their implementation. Taking proper account of such options could very well lead to a reversal in the ordering of projects along the SACC or control cost staircase. For example, higher cost (but 12 USAID, Integrated Resource Plan for Andhra Pradesh, Report to APSEB, Hyderabad, 1995. 13 P. Meier, Economic Analysis of the Haryana Power Sector Restructuring and Reform Program, World Bank, New Delhi, 1997; P. Meier, Economic Analysis of the Andhra Pradesh Power Sector Restructuring Program, World Bank, New Delhi, January 1999. 14 P. Meier, India: Environmental Issues in the Power Sector: A Case Study of Haryana, World Bank, 1999. 15 The most recently published plan of June 1998 is for the years 1998-2012. 16 A detailed description of the model and its application to CEB's generation planning studies may be found in Chapter 7 of CEB's 1998 Generation Planning Report. 17 ENVIROPLAN solutions are typically within 5 percent of those yielded by WASP, with essentially identical capacity expansion plans--differences in assumptions are much more important than choice of power systems planning model. Introduction 7 lower option-valued) projects might be undertaken prior to their lower-cost (but higher option- valued) counterparts. This option value will arise, in part, from the fact that the control cost curve is itself dynamic, with changes brought about in the course of time due to the evolution of science and technology, as well as socioeconomic conditions. Developing a dynamic abatement cost curve would require taking account of the significance of each of the above drivers of option value, for each of the GHG mitigation options that have been identified in the standard abatement cost curve. In the following discussion, we first explain the development of the SACC, thereafter turning to how it might change in a dynamic context. 18 The SACC An SACC for carbon emitted by a given country (like Sri Lanka) may be derived using conventional methods (e.g., power system modeling). The procedure is summarized in Figure 1.3, where the focus is on the energy sector, starting with the electric power subsector. First, using a systems-based, long-term planning model (e.g., WASP), the national least-cost generation expansion program is determined for at least the next 20 years, assuming a BAU or base case scenario. Second, various new electric subsector options (both technologies and policies) are introduced into the plan (e.g., renewable energy sources, loss reduction, price increases) to identify steps such as A, B, C, and so forth in the SACC staircase. The estimates of carbon emission reductions and the costs per unit of abatement are deterministic and based on the usual expected values of input data as the various incremental carbon abatement options are added to the base case. Finally, other sectors and subsectors such as oil, gas, coal, transport, and forestry could be analyzed in a similar fashion, to identify and insert further "steps" in the same SACC staircase. This next phase of analysis is planned in a follow-up study. Figure 1.3: Developing the SACC Cost/Ton SACC of Carbon G F Carbon Emissions Trading Price E D C B Carbon Abatement A BA 18 This discussion is based on a more detailed paper: C. Fernando and M. Munasinghe, A Real Options Framework to Assess Environmental Policy, Programs, and Strategy for GHG Mitigation, World Bank, December 1998. 8 Greenhouse Gas Mitigation Options in the Sri Lanka Power Sector In Figure 1.3, the options A and B are win-win outcomes that provide both economic benefits (in the form of cost savings) and carbon emission reductions. Options C, D, and E are profitable at the carbon emissions trading price shown in the figure and should be pursued under the deterministic framework of the SACC, if decisionmakers seek to maximize the net benefits of abatement to the country. However, as outlined below, such a simple analysis is not adequate to deal with the high levels of uncertainty and risk inherent in climate change decisionmaking. The dynamic abatement cost curve (DACC) Uncertainties of various kinds will affect the choice of policy options significantly. First, scientific uncertainties arise from the unpredictability of future climate change and its physical, ecological, and socioeconomic impacts. Second, technological uncertainties exist with regard to the evolution of future energy-producing and -using technology costs and fuel prices. Finally, sociopolitical uncertainties also abound, including the degree of consumer backlash against rising energy prices, the unpredictability of energy demand, the progress in United Nations Framework on Climate Change (UNFCC) negotiations, the evolution of the carbon market, and so on. We consider one simple example, to show how risk and uncertainty might affect decisionmaking. In the step curve above, let step A represent an increase in energy prices (e.g., removal of subsidies to bring prices closer to economically efficient levels), while step B indicates another option, such as energy conservation. In a riskless analysis, the decisionmaker would first carry out step A and then proceed to step B. However, in the real world, public reaction against energy price increases may involve too great a political risk, so that it would be more feasible to undertake step B first and defer step A to a later date. Thus, the introduction of political uncertainty and risk will change the decision-making process. A second example involves the use of option values. Step D might require commitment to a technology having high and unrecoverable sunk costs. Assuming that step E involves a smaller investment, together with the possibility of switching to another more cost-effective technology (which may become available in the future). In this case, step E has an additional option value over step D, because the former preserves the flexibility to change to a potentially cheaper and superior technology (at little cost) in the future. Once again, risk-based option value analysis could well favor step E over step D. Risk associated with the unpredictability of the global trading price for carbon abatement provides a third example. The future emissions trading price might rise rapidly in the future if climate change effects manifested themselves earlier and more strongly than currently anticipated (or vice versa). An options-based analysis would help to determine the profitability of exercising any specific carbon abatement option on the SACC staircase today, versus retaining the flexibility to use it at a later time. Introduction 9 Emission coefficients Wherever possible, emission coefficients are calculated by stochiometry, since this is the best way to ensure consistency between assumptions about fuel characteristics, generating plant efficiency, and assumptions about combustion technology and the pollution control options as may be applied.19 Where such stochiometric calculations are not possible, we use the emission coefficients provided by Intergovernmental Panel on Climate Change (IPCC). Table 1.4 describes the assumptions and coefficients used for each air emission. Table 1.4: Emission calculations C? ? Stochiometric calculation based on the carbon content of the fuel. If C is the weight percent of carbon in the fuel and f is the fraction of carbon oxidized in combustion, then the kg of C emitted per kg of fuel consumed = C f. SO2 Stochiometric calculation based on sulphur content in fuel. If S is the weight percent of sulphur in the fuel; ??is the fraction of SO2 in the flue gas removed by FGD; and f is the fraction of sulphur oxidized in combustion, then the kg of SO2 emitted per kg of fuel consumed = S f (1-?) 1.998. Particulates Stochiometric calculation based on the ash content of the fuel. If A is the weight percent of ash in the fuel; f is the fraction of ash to fly ash (and 1-f the bottom ash fraction), and ??is the fraction of particulates removed by the electrostatic precipitator, then the particulate emissions per kg of fuel consumed = A f (1-?). NOx Emission factor as provided by the equipment manufacturer (and/or) as per the project-specific Environmental Impact Assessment. CH4 Emission factor as per IPCC guidelines. Note: The fraction of carbon oxidized is as given in the Guidelines (p. 21): 0.98 for coal, 0.995 for gas, 0.99 for oil combustion. For each plant, the fuel consumption is calculated on the basis of the heat rate (as KCal/kWh) and the calorific value of the fuel (KCal/kg). Table 1.5 presents the characteristics of the fuels used in Sri Lanka. 19 A good example of why such an approach is desirable (and necessary) is flue gas desulphurization (FGD), the need for which has been debated in Sri Lanka every time a new proposal for a coal-fired power project has been advanced. Plants fitted with FGD systems will have lower overall efficiency (due to higher auxiliary consumption and impact on boiler efficiency) and may therefore increase CO2 emissions per net kWh by up to 5 percent. We might note parenthetically that in addition to the efficiency effect, SO2 removal, whether by lime or limestone FGD systems, or by fluidized bed combustion systems where calcination occurs in situ, involves the production of CO2 as indicated by the (greatly simplified) summary reactions CaCO3 --> CaO+CO2; CaO+SO2 +0.5O2 --> CaSO4. These chemical reactions imply that for every mole of SO2 reduced, one mole of CO2 is produced. However, in the case of a coal containing 1 percent sulphur by weight (which would be at the upper limit of sulphur content of typical export coals), the contribution of CO2 fromthese FGD reactions is about 1.5 percent of the CO2 produced by combustion. 10 Greenhouse Gas Mitigation Options in the Sri Lanka Power Sector Aggregations to carbon equivalent are made on the basis of the IPCC guidelines shown on Table 1.5. In this study we use the standard 100-year equivalents.20 Table 1.5: Direct global warming potentials (GWPs) of gases Time horizon 20 years 100 years 500 years Carbon dioxide 1.0 1.0 1.0 Methane 62.0 24.5 7.5 Nitrous oxide 290.0 320.0 180.0 Source: Guidelines, op.cit., p. 10 (Exhibit 3-2) 20 Since the power sector does not emit chloroflurocarbons (CFCs) and hydrochloroflurocarbons (HCFCs), these are omitted. Carbon monoxide, nonmethane hydrocarbons, and nitrogen oxides (other than nitrous oxide) are shown in the GHG Overlay Guidelines as having zero (or negligible) GWP and hence are also omitted. 2 Business-As-Usual (Reference) Scenario 2.1 Sri Lanka's power sector21 Hydro is the main indigenous source of primary commercial energy in Sri Lanka, with an estimated potential of 2,000 MW, of which some 1,205 MW have been harnessed to date.22 Due to the lack of any domestic resources, all fossil fuel is imported.23 Because of increasing difficulties in implementing large hydro projects, further exploitation of the remaining potential seems unlikely, and most of the capacity additions over the past few years have already been thermal. Table 2.1 lists the major generating stations of Sri Lanka. Table 2.1: Installed capacity, MW (as of December 1998) MW Fuel Commissioning Laxapana Hydro Complex Old Laxapana 3 ? 8.33 = 25 December 1950 2 ? 12.5=25 December 1958 Wimalasurendra 2 ? 25=50 January 1965 Polpitiya 2 ? 37.5=75 April 1969 New Laxapana 2 ? 50=100 February-March 1974 Canyon 2 ? 30=60 March 1983, March 1988 Mahaweli Hydro Complex Victoria 3 ? 70=210 1984-1986 Kotmale 3 ? 67=201 1985-1988 Randenigala 2 ? 61=122 July 1986 Ukuwela 2 ? 19=38 1976 Botawenna 1 ? 40=40 1981 Rantembe 2 ? 24.5=49 1990 21 For a more detailed presentation of Sri Lanka's power sector and its historical development, see, e.g., Meier and Munasinghe, op.cit. (Chapter 3). A detailed presentation of Sri Lanka's energy resources is found in M. Baldwin, editor, Natural Resources of Sri Lanka: Conditions and Trends, Natural Resources, Energy and Science Authority of Sri Lanka, 1991. 22 Including the 70 MW Kukule plant, under construction. 23 There are some small peat resources in marshy lands to the north of Colombo, but a 1989 feasibility study concluded that the quality and extent of the reserve was not commercially viable. 11 12 Greenhouse Gas Mitigation Options in the Sri Lanka Power Sector MW Fuel Commissioning Other hydro Small hydros 19 Samanalawewa 120 October 1992 Kukule (under construction) 70 2003 Thermal Kelanitissa Steam 2 ? 22=44 Fuel oil June 1962-September 1963 Kelanitissa CT 6 ? 20=120 Auto-diesel November 1980-March 1982 Sapugaskanda 4 ? 18=72 Residual oil May-October 1984 Sapugaskanda: Extension(ADB) 4 ? 10=40 Residual oil September 1997 Kelanitissa CT 115 Auto-diesel August 1997 Sapugaskanda: Extension(KfW) 4 ? 10=40 Residual oil Early 1999 Source: CEB Table 2.2: Demand structure, 1997 Number of Sales (GWh) kWh sold/ customers customer Domestic 1,611,102 1,191 739 Religious 13,155 22 1,672 General: small 202,059 389 1,925 General: medium 1,346 214 159 MWh General: large 26 89 3,423 MWh Industrial 23,008 1,430 62 MWh Bulk supply: LECO and municipal authorities: LT 2 7 11kV/33kV 57 650 Source: CEB Statistical Digest 1997. Electricity demand has been growing at a compound annual growth rate of around 7 percent since the early 1970s (Figure 2.1). Table 2.2 shows the structure of demand in 1997. IntroductionBusiness-as-Usual (Reference) Scenario 13 Figure 2.1: Historical electricity demand 5 4 3 Local Authorities GWh 2 Commercial Hotels Industry 1 Households 0 1980 1981 1982 1983 1984 1985 1986 1987 1988 1989 1990 1991 1992 1993 1994 1995 1996 1997 Households 187 212 252 297 309 337 358 369 392 408 496 623 681 803 909 1014 1026 1191 Industry 625 677 739 752 790 850 925 865 905 849 910 958 1057 1223 1406 1527 1361 1430 Commercial 223 220 235 244 301 280 304 341 358 357 424 466 499 551 575 631 592 689 Hotels 27 48 59 70 77 78 85 79 85 81 83 90 7 Local Authorities 357 394 433 451 419 502 543 571 601 631 657 572 545 536 609 683 542 657 Street Lighting 12 13 15 16 17 18 21 29 43 40 40 47 50 CEB conducts an annual long-term generation expansion planning study that assesses investment requirements in the power sector as a function of a range of input assumptions on load growth, fuel prices, and technology assumptions. The 1998 planning study, which covers the years 1999-2013, is presently under way (and is due to be released as the official plan of CEB in early 1999).24 The base case as presented in the CEB study is adopted for the GHG overlay as the BAU case. 2.2 The demand forecast As shown in Figure 2.2, electricity demand growth and GDP growth are highly correlated. Econometric studies conducted by CEB for the 1998 CEB demand forecast show GDP growth to be the most significant determinant of demand.25 Table 2.3 shows the GDP assumptions for the forecast.26 In the BAU demand forecast, the tariff is assumed to increase with inflation (no increase in real terms). By 2018 the busbar peak demand is estimated to grow to 4,292 MW (compared with 1,136 MW in 1998); the corresponding 2018 busbar energy demand is 20,678 GWh, up from 5,683 GWh in 1998. 24 CEB, Report on Long Term Generation Expansion Planning Studies, 1999-2013, Generation Planning Branch, Colombo, 1999. 25 CEB, Load Forecast for 1998 Generation Expansion Planning Studies, Colombo, 1998. 26 The econometrically determined forecasting equations for domestic, industrial, and commercial sectors are of the autoregressive form D(t)=a+bD(t-1) + cGDP(t-1) + dGDP(t) (where a,b,c,d are coefficients estimated by least squares, with R2>0.96 in all cases). 14 Greenhouse Gas Mitigation Options in the Sri Lanka Power Sector Figure 2.2: Electricity demand growth and GDP growth 15 8 10 6 growth 5 <> GDP 0 real electricity 2 -5 -10 0 1970 1975 1980 1985 1990 1995 2000 Table 2.3: GDP growth assumptions for the CEB forecast Forecast rate of annual GDP growth, % 1995 (actual) 5.5 1996 (actual) 3.8 1997 (estimated) 6.0 1998 6.5 1999 7.0 2000 7.5 2001 8.0 Source: Public Investment Programme 1997-2001: Department of National Planning, Colombo 2.3 Transmission and distribution (T&D) assumptions Total T&D loss rates and the system load factor (SLF) are taken as 17.6 percent (of generation) and 55 percent, respectively, which are taken at their 10-year average values. As indicated in Table 2.4, these have stayed remarkably constant over the past decade. Of the total T&D loss, 15.3 percent is assumed to be technical loss, and the balance of 2.3 percent is assumed to be nontechnical (commercial) loss.27 Box 2.1 presents an analysis of historical T&D expenditures that is used as a basis for projecting future T&D costs associated with load growth. 27 Studies are currently underway at CEB to confirm these estimates. IntroductionBusiness-as-Usual (Reference) Scenario 15 Table 2.4: T&D loss rates and SLF Year T&D loss rates SLF 1986 15.9 0.56 1987 16.8 0.54 1988 15.3 0.54 1989 17.7 0.53 1990 17.2 0.56 1991 18.8 0.56 1992 19.0 0.55 1993 17.8 0.56 1994 18.3 0.55 1995 18.1 0.56 1996 18.0 0.52 1997 17.8 0.54 Source: CEB 2.4 Fuel costs Fuel characteristics and costs are as given in Table 2.5. For the base case the fuel costs are taken as constant throughout the planning period: This is consistent with the most recent World Bank price forecasts for internationally traded fuels (see Figure 2.3).28 Table 2.5: Fuel characteristics Heat value Cost KCal/kg U.S. cents/Gcal Lanka auto-diesel 10,550 1,654 Lanka furnace oil (Bunker C) 10,100 1,180 Lanka residual 10,050 943 Australian coal (imported) 6,500 746 Source: Coal price for West Coast Coal Project Box 2.1: T&D expenditure The table below shows historical T&D capital expenditures, converted to $/kW of incremental peak load. Since there has been essentially no change in T&D loss rates over the period covered (1991-1997), this expenditure may be taken to be that required for new load growth, exclusive of any additional expenditure as may be necessary for T&D system rehabilitation to reduce T&D loss rates. The average computes to $850/kW, a figure that is used to derive probable T&D capital expenditures in the future (absent additional outlays for loss rate reduction, which are part only of the reform scenario, discussed in the next section). 28 The World Bank price forecast is for coal quality with less than 1 percent S, 12,000Btu/lb, 12 percent ash, fob Hampton Roads, Virginia. However, Australian export coals, the most probable source for Sri Lanka, are similar in specification, and would likely follow the broad market trends indicated by the World Bank forecast. 16 Greenhouse Gas Mitigation Options in the Sri Lanka Power Sector T&D expenditure 100 1400 $/incr .MW----> 1200 80 $/k 1000 Wi nc re 60 $ 800 m mil en lio tal n pe 40 transmission 600 ak loa distribution d 400 20 200 0 0 1991 1992 1993 1994 1995 1996 1997 Per CEB data 1990 1991 1992 1993 1994 1995 1996 1997 Distribution Rs million 2,321 3,442 3,158 3,204 2,817 3,350 4,399 Transmission Rs million 194 162 305 877 671 1,476 3,169 Economic costsa Distribution Rs million 1,625 2,409 2,211 2,243 1,972 2,345 3,079 Transmission Rs million 136 113 214 614 470 1,033 2,218 Exchange rate: end of Rs/$ 40 42.5 46 49.56 49.9 54.04 56.7 60 period Exchange rate: average Rs/$ 41.3 44.3 47.8 49.7 52 55.4 58.4 Distribution US$ million 38.2 52.4 44.6 44.9 36.5 41.4 51.3 Transmission US$ million 3.2 2.5 4.3 12.3 8.7 18.2 37 US$ deflator 1.08 1.06 1.04 1.03 1.02 1.01 1 Distribution US$97 million 41.4 55.6 46.4 46.3 37.2 41.8 51.3 Transmission US$97 million 3.5 2.6 4.5 12.7 8.9 18.4 37 Distribution expenditure Peak loadb MW 640 685 742 812 910 980 968 1,037 Incremental load MW 45 57 70 98 70 68.6 73.4 Distribution $/kW 919.7 975.4 663.1 472.5 531.8 608.9 699.2 Transmission expenditure Peak load MW 685 742 812 910 980 1,048.6 1,122 Incremental load MW 45 57 70 98 70 68.6 73.4 Transmission $/kW 76.9 45.9 64 129.3 126.7 268.3 503.7 Total transmission & $/kW 921 962 699 584 645 869 1,203 distribution a. Adjusted by rate of taxes and duties at 30 percent. b. In 1998, the recorded peak was 1,137 MW. The 1996 value of 968 MW (as opposed to the trend value of 1,049 MW) was due to the severe power shortages of that year. IntroductionBusiness-as-Usual (Reference) Scenario 17 Figure 2.3: World Bank crude oil and coal price forecasts 50 coal: fob HamptonRoads [12000Btu/lb,<1%S] 40 coal, constant 98 $ $/ton(coal) 30 high Arab Light 34API 20 $/bbl(oil); oil, constant 98 $ 10 low 0 1998 1999 2000 2001 2002 2003 2004 2005 2006 2007 2008 2009 2010 Arab Light 34API 15 17 17.5 17.8 18.1 18.4 18.7 19 19.4 19.8 20.2 20.6 21 low 12 12.0 12.0 12.0 11.8 11.5 11.3 11.0 10.9 10.8 10.7 10.6 10.5 high 18 22.0 23.0 23.6 24.5 25.3 26.2 27.0 27.9 28.8 29.7 30.6 31.5 coal 35.5 35.0 36.0 37.0 38.0 38.5 39.0 39.5 40.0 40.5 41.0 41.5 42.0 coal, 1998 $ 35.5 34.5 34.6 34.6 34.7 34.3 33.9 33.5 33.1 32.7 32.3 31.9 31.5 oil, 1998 $ 15.0 16.8 16.8 16.7 16.5 16.4 16.2 16.1 16.0 16.0 15.9 15.8 15.7 Source: World Bank 1998 fuel price forecasts 2.5 The capacity expansion plan Figure 2.4 summarizes the choice set for thermal plant in the form of a standard screening curve, based on the data in Table 2.6. Until such time as the large scale coal plant could be considered,29 diesel plant using residual fuel was least-cost for baseload, which explains the addition of significant diesel capacity over the past few years. Combined cycle plants are least-cost for the 20-60 percent plant factor range (which explains the recent decision to proceed with the first 150 MW combined cycle gas turbine or CCGT), while for peaking, open-cycle combustion turbines (OCGT) are least-cost for plant factors below 15 percent. Now that the system has grown sufficiently for coal plants to be warranted, the advantage of coal over diesel for baseload is evident. 29 The scale economies are substantial, particularly in regard to port facilitates. The thermal generation options study indicates the following unit costs as a function of total plant size: 150 MW: 961$/kW; 300 MW: 815$/kW; 500 MW: 691$/kW; 700 MW: 633$/kW. The proposal for a 3 x 300 MW plant on the west coast at Puttalam takes advantage of these scale economies. 18 Greenhouse Gas Mitigation Options in the Sri Lanka Power Sector Figure 2.4: Basic options screening curve 10.5 8.5 hydro 6.5 OCCT UScents/kWh CCCT 4.5 diesel pulv.coal 2.5 annual PLF 0.10 0.15 0.20 0.25 0.30 0.35 0.40 0.45 0.50 0.55 0.60 0.65 0.70 0.75 0.80 0.85 0.90 pulv.coal 6.10 5.64 5.27 4.97 4.72 4.51 4.33 4.17 4.03 3.91 OCCT 11.18 9.29 8.35 7.78 7.40 7.13 6.93 6.77 6.65 6.55 6.46 6.39 6.32 6.27 6.22 CCCT 12.65 9.60 8.08 7.16 6.55 6.12 5.79 5.54 5.33 5.17 5.03 4.91 4.81 4.72 4.65 diesel 6.50 6.00 5.60 5.27 5.00 4.77 4.57 4.40 4.25 hydro 7.36 Table 2.6: Assumptions for screening curves Pulv. coal OCCT CCCT Diesel Hydro - Gin (resid. oil) Ganga Total capacity MW 300 105 300 150 49 Capital cost $/kW 1,303 386 725 1,332 2,876 life years 30 20 30 25 50 Fixed O&M $/kW/month 0.56 0.36 0.27 0.92 Variable O&M million/kWh 2.79 2.85 2.72 7.36 Scheduled maintenance days 40 30 30 30 Forced outage rate 0.03 0.08 0.08 0.15 Heat rate, net KCal/kWh 2,293 3,060 1,890 1,954 0 PLFmin 0.40 0.00 0.00 0.40 0.45 Primary fuel Coal LoS Auto-diesel Auto-diesel Resid Hydro Heat content KCal/kg 6,300 10,550 10,550 10,300 Fuel price, cif plant gate $/bbl 0.0 24.0 24.0 9.9 $/ton 46 180 180 67 $/mmBTU 1.85 4.31 4.31 1.63 Fixed charge factor 0.106 0.117 0.106 0.110 0.101 Annual capital cost $/kW/year 138.2 45.3 76.9 146.7 290.1 Fixed Operation & $/kW/year 6.7 4.3 3.3 11.0 0.0 Mainten.(O&M))/year Total fixed cost $/kW/year 145.0 49.6 80.2 157.8 290.1 Fuel cost US cents/kWh 1.68 5.23 3.23 1.26 Variable O&M US cents/kWh 0.28 0.29 0.27 0.74 0.00 Source: SLEPTA, op.cit., Table 3.6. Fuel prices are based on a crude oil price of 16$/bbl for Saudi Light 34 (equivalent to $17.3$/bbl for Brent). IntroductionBusiness-as-Usual (Reference) Scenario 19 Figure 2.5: Capacity expansion plan, BAU 800 600 400 MW WestCoast>>> Trincomalee>>> 200 UpperKotmale Kukule 0 -200 1999 2001 2003 2005 2007 2009 2011 2013 2015 2017 hydro 70 150 coal 300 300 300 300 300 300 liquidfuels 106 300 105 105 105 105 105 300 gas diesel 100 -36 -36 renewables 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 CEB's base case generation capacity expansion plan, illustrated in Figure 2.5, is entirely consistent with these results. The first coal plant is required in 2004, with a total of six 300 MW units required over the planning horizon to 2018. Both the Kukule and Upper Kotmale hydro plants are taken as fixed30 (since the former is already under construction, the latter has been approved for construction, though presently suspended because of a court case).31 2.6 Results: GHG emissions for BAU scenario GHG emissions for the BAU case are shown in Figure 2.6. These are averaged over the five hydro conditions used in the capacity expansion planning models. Emissions increase from 1.3 mtpy in 1998 to 12.6 mtpy in 2018, an almost 10-fold increase over 20 years. Most of the increase is associated with the new coal plants. 30 However, as noted below, even in a free run these two plants are selected. 31 Environmental Foundation Ltd v. CEB & Ministry of Environment, 2001. 20 Greenhouse Gas Mitigation Options in the Sri Lanka Power Sector Figure 2.6: GHG emissions, BAU 15 0.8 0.6 10 tons/year] consumed [million BAU[#] 0.4 kg/kWh [right scale---->] emissions 5 GHG coal KgGHGemissions/kWh 0.2 liquid fuels 0 0 1998 2000 2002 2004 2006 2008 2010 2012 2014 2016 2018 Moreover, as the share of fossil-fueled plants in the total generation mix increases (Figure 2.7), emissions per kWh consumed also increase. Figure 2.7: The generation mix (1000 GWh) 20 15 GWh 10 diesel liquid fuels 1000 coal 5 hydro 0 1998 2000 2002 2004 2006 2008 2010 2012 2014 HYDRO 3.7 3.8 3.9 3.9 3.9 4.2 4.2 4.2 4.3 4.3 4.3 4.3 4.3 4.3 4.3 4.3 4.3 4.3 COAL 0.0 0.0 0.0 0.0 0.0 0.0 2.1 2.1 2.1 2.1 4.2 4.2 4.2 6.1 6.1 6.2 8.0 8.0 LQ 0.7 0.9 0.8 1.3 1.6 2.2 1.0 1.6 2.0 2.7 1.7 2.3 2.9 2.1 2.8 3.5 2.9 3.7 GAS 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 DIESEL 1.0 1.1 1.7 1.7 1.8 1.5 1.2 1.2 1.3 1.4 0.9 1.0 1.2 0.8 1.0 1.1 1.0 1.1 RENEW 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 3 The Reform Scenario In Sri Lanka, the specific components of a comprehensive reform initiative have yet to be articulated--unlike the Indian states of Haryana, Orissa, and Andhra Pradesh, where the packages of policy reform measures under way with Bank assistance have already been identified. Consequently, the individual programs that would be part of a reform and restructuring effort in Sri Lanka will themselves need some screening. In many cases, we cannot say a priori whether a particular measure (for example, the development of grid-connected mini hydro) would be part of a reform program based on purely the economic returns, or whether its economic returns are below the hurdle rate and would be justified only as part of GHG mitigation efforts. Nevertheless, it seems reasonable to suppose that the following elements might be part of a reform scenario: ?? a more aggressive set of assumptions for T&D loss reduction (as might be the consequence of distribution privatization), ?? implementation of the remaining cost-effective DSM options, ?? improvements to time-of-day tariff,32 and ?? an appropriate set of tariff reforms that would set the tariff based on a return-on- equity criterion higher than the present criterion for the government-owned CEB.33 3.1 T&D loss reduction and DSM Table 3.1 shows the assumptions for the reduction in technical and nontechnical losses in the reform scenario. In light of LECO's reduction in T&D loss rates achieved over the past 32 For a discussion of the anomalies of the present time-of-day tariff, see T. Siyambalapitiya, Electricity Pricing Policy in Sri Lanka, Institute of Policy Studies, Energy and Environmental Economics Series 5, July 1997. 33 Care will need to be exercised when evaluating pricing options in any bottom-up, partial equilibrium analysis to correct for welfare effects. Elimination of subsidies, reduction of nontechnical losses, or imposition of long run marginal cost (LRMC)-based tariffs implies loss of consumer surplus to some consumer groups (and possibly gains in producer surplus). Clearly, if system cost is used as the numeraire, then this must be adjusted for any changes in benefits associated with any particular option. 21 22 Greenhouse Gas Mitigation Options in the Sri Lanka Power Sector decade--its present loss rate at 11kV and below is 8 percent--the 12.5 percent target seems achievable.34 Table 3.1: Assumptions for T&D losses: reform scenario Technical Nontechnical Total losses losses 1999 15.5 2.3 17.8 2000 15.3 2.3 17.6 2001 15.3 2.3 17.6 2002 15.0 2.1 17.1 2003 14.0 2.1 16.1 2004 13.0 1.9 14.9 2005 12.0 1.7 13.7 2006 11.5 1.5 13.0 2007 11.0 1.5 12.5 2008 11.0 1.5 12.5 2009 and beyond 11.0 1.5 12.5 With regard to DSM, we assume that in the first stage of reform, DSM would be limited to lighting. The most recent (and comprehensive) analysis of DSM in Sri Lanka is that of Shresta, Fernando, and Shreshta (1998).35 As indicated in Table 3.2, all the appliances examined are economic (as measured by the cost of conserved energy being less than the LRMC), but only residential lighting has a net annual benefit to the utility. This is because the tariff in commercial and industrial sectors is higher than the LRMC (and therefore results in disproportionate revenue losses in consequence of energy savings).36 For the reform case DSM program, all of the lighting measures are included; energy efficiency motors and efficient air conditioning systems are considered later as potential mitigation options. 34 LECO makes purchases at 11kV, and therefore its loss rate does not include losses in transmission at the higher voltages. 35 Ram Shresta, W. J. L. S. Fernando, and Rabin Shrestha, "Environmental Emission Mitigation Potential of Efficient Electrical Appliances in Sri Lanka," International Journal of Energy Research, 22: 923-933, 1998. 36 However, this study did not examine the impact of DSM in a dynamic systems planning model, but used an assumed LRMC as the yardstick for evaluating the cost of conserved energy. The Reform Scenario 23 Table 3.2: Economic and utility costs of efficient appliances Economic cost, Utility net annual US cents/kWh benefit, US$ million Residential 13 W compact 4.21 1.3 fluorescent-CFL 18 W CFL 3.02 2.0 23 W CFL 1.99 3.7 Commercial 36 W slim tube 0.67 -0.1 18 W CFL 2.48 -0.7 Efficient AC 6.21 -2.9 Industrial 36 W slim tube 0.67 -0.3 18 W CFL 2.48 -3.3 Energy-efficient 0.89 -35.6 motors (EEMs) Source: Shresta, Fernando and Shrestha[please verify spelling] (1998), op.cit: Tables 3 and 4. The impact of the lighting DSM measures is depicted in Figure 3.1, which shows representative daily load duration curves (one in each season). By 2007, lighting DSM reduces the peak evening demand by 330 MW, about 15 percent of the total evening peak load. Figure 3.1: 2007 load duration curves with lighting DSM 2500 d(DSM):=330 MW d(DSM):=326 MW 2000 16.2% 15.0% 15.3% 15.0% 14.0% 14.3% 1500 MW d(DSM):=44 1000 9.3% 9.5% 10.0% MW 1000 4.5% 500 3.5% 3.6% feb-may2007 jun-nov,jan2007 dec2007 0 0 Figure 3.2: Capacity expansion plan: Reform 20 40 60 80 24 Greenhouse Gas Mitigation Options in the Sri Lanka Power Sector Figure 3.2: Capacity expansion plan: reform 600 400 200 MW 0 -200 1999 2001 2003 2005 2007 2009 2011 2013 2015 2017 hydro 70 150 coal 300 300 300 300 300 liquid fuels 106 300 105 105 gas diesel 100 -36 -36 renewables 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 3.2 Generation capacity expansion plan As shown in Figure 3.2, the combination of reduced T&D losses and the peak-shaving impact of the lighting DSM program have a significant impact on the capacity expansion plan. Over the planning period, there is a reduction of the number of required baseload coal units from six to five, and a significant reduction in the number of thermal peaking units required toward the end of the planning horizon (needing two rather than six 105 MW OCCTs from 2010 to 2018). This reduction in thermal peaking would be expected, given the effect of lighting DSM on the evening lighting peak load. These adjustments result in a significant reduction in GHG emissions, from 118 million tons over the planning horizon in the base case to 98 million tons under reform. The year-to- year changes are shown in Figure 3.3. The Reform Scenario 25 Figure 3.3: Reform GHG emissions compared with base case 15 BAU[#] Reform 10 tonnnes) (10..6 5 emissions GHG 0 changerelativeto:BAU[#] -5 1998 2000 2002 2004 2006 2008 2010 2012 2014 2016 2018 The impact of reforms on system costs and GHG emissions can be displayed as a trade-off plot as indicated in Figure 3.4. The impact of T&D loss reduction and DSM is shown separately (each as perturbations of the BAU case). Reform is clearly win-win. Indeed, reform is also win-win with respect to other environmental attributes for local air emissions.37 Figure 3.4: Impact of reform 2800 BAU[#] 2700 million T&D loss Red $US 2600 (costs) DSM PV2500 Reform 2400 2300 95 100 105 110 115 120 total GHG emissions 37 However, when these benefits are monetized, they account for only a small fraction of the carbon offset benefits. 26 Greenhouse Gas Mitigation Options in the Sri Lanka Power Sector 3.3 Comparison of the impacts of the Sri Lanka reform scenario with Haryana Comparisons with Haryana, the second Indian state (after Orissa) to embark on comprehensive power sector reform, are instructive, particularly since Haryana is among the smaller Indian states and comparable in size to Sri Lanka.38 As indicated in Table 3.3, while in Sri Lanka the consequences of reform are economic benefits as well as reductions (about 15 percent) in GHG emissions, in Haryana the economic benefits are some six times larger than in Sri Lanka, but reform also results in a 25 percent increase in GHG emissions. This is a consequence of the massive inefficiencies in the Haryana system (such as T&D losses of 35 percent, almost twice the Sri Lanka level) and the significant present curtailments (25 percent of demand), which would be eliminated by reform (by restoring commercial creditworthiness, allowing presently stalled independent power producers (IPPs) to come to financial closure). Table 3.3: Impact of reform: comparison with Haryana Units Sri Lanka Haryana Present installed capacity (MW) MW 1,610 2,392 of which hydro MW 1,205 1,058 GHG emissions BAU Million tons 118 277 Reform Million tons 98 377 GHG reduction due to reform Million tons 20 -100 (increase) Economic impact of reform US$ million 227 1,418 Kg CO2/kWh BAU 0.48 0.85 Reform 0.41 0.79 Impact of reform 0.07 0.06 (15% reduction) (7.1% reduction) 38 World Bank, Economic Analysis of the Haryana Power Sector Reform and Restructuring Program, Washington, D.C., November 1997. 4 Mitigation Options In this section, we analyze the impact of individual mitigation options. Each option is assessed one at a time as a perturbation of the reference (reform) case. The goal is to derive a cost of avoided carbon for individual options, and to compare the cost-effectiveness of each. Based on this screening process, GHG mitigation scenarios are constructed in Section 5. In these scenarios, individual options are combined to generate a GHG mitigation option supply curve that takes into account the interactions among individual programs. The focus in this section is on establishing the economic costs39 and on the magnitude of the physical GHG emission reduction benefits, which we express as the economic cost of avoided carbon. Many of the options have upward-sloping supply curves, which in the case of generic alternative energy options is due to two main reasons. The first applies to situations where capital costs are likely to be relatively constant across individual projects, but where energy output (as quantified by annual plant factors) between the best and worst sites shows large differences (e.g., as in the case of wind power). The second applies to the converse of this, where plant factors might show relatively small differences, but where construction costs may show high variation (e.g., as in the case of mini hydro, where civil works costs across individual projects show larger variations than run-off coefficients that determine annual load factors). 4.1 Alternative energy options Small hydro Small hydro is an excellent starting point for the assessment of mitigation options, for it is already a component of the International Development Assoc.(IDA}/Global Environment Facility (GEF) Energy Services Delivery Project (ESDP).40 Indeed, grid-connected mini hydro is its largest single component ($14.4 million of $24.2 million). Four such small hydro plants were connected to the CEB grid in 1996 and 1997: Hydrotech Lanka Dickoya (Pvt) (960 kW), Talawakelle Tea Factory (112 kW), Ritigala Tea Factory (112 kW), and Seethaeliya Tea Factory (108 kW). 39 All references to cost in this section therefore mean economic cost, without taxes and duties, and without interest during construction. 40 World Bank Project Appraisal Document, Energy Services Delivery Project, Report 16063-CE, Energy and Project Finance Division, Country Department 1, South Asia Region, February 27, 1997. 27 28 Greenhouse Gas Mitigation Options in the Sri Lanka Power Sector The mini hydro potential of Sri Lanka was recently assessed by Fernando (1998).41 Small hydro is defined by this assessment as plants of less than 5 MW. A total of 257 sites were identified, broken down as indicated in Table 4.1. Fernando assumes an average capital cost figure of $1,500/kW of installed capacity at 1998 price levels.42 He concludes that 45 MW of the estimated potential could run at annual plant factors of 50 percent, thereby generating around 200 GWh/year.43 Table 4.1: Classification of small hydro sites Site Status No. of Used Exploitable sites capacity, potential, kW kW Old estate sites Abandoned 49 3,343 10,367 Not in operation 14 544 2,555 In operation 74 2,228 10,746 New estate 71 20,733 sites Nonestate sites 47 53,016 Total 257 97,407 Source: Fernando, op.cit., pp. 21-24. A number of more detailed feasibility studies were undertaken in 1996 as part of the background work for the ESDP.44 Table 4.2 shows the main results of these studies.45 41 Sunith Fernando, An Assessment of the Small Hydro Potential in Sri Lanka, Intermediate Technology Development Group, Colombo, March 1998. 42 This is substantially higher than the $1,030/kW figure used by the ESDP. 43 Fernando, op.cit, p. 30. 44 These were prepared by Consultancy and Professional Services (Pvt) Ltd., for the following projects: Peakfield Estate at Maskeliya, Delmar Estate, Carolina Estate, Delta Estate at Pupuressa, and Ellapita Ella. The latter project was used as a basis for the financial analysis in the ESDP project appraisal document (PAD). 45 A number of issues arise from these studies. The price levels and exchange rates are not stated, but we have assumed the average 1996 exchange rates for the calculation of the $/kW indicated in Table 4.2. Assuming 100 percent foreign import for the electromechanical and power transmission equipment, we can estimate the foreign content also as shown (though again, this is our assumption). There is no distinction between economic and financial analysis, and the internal rates of returns (IRRs) calculated are financial, based on current price levels (so the real IRR would be lower by roughly the assumed inflation rate of 5 percent). Mitigation Options 29 Table 4.2: Results of five mini hydro feasibility studies (at Rs 55/$) kW GWh/ Annual IRR $/kW Foreign year PLF content Delta Estate: Pupuressa 292 1.45 0.56 20.53 1,069 0.51 Carolina Estate 2,500 14.04 0.64 20.48 934 0.62 Delmar Estate 339 1.75 0.59 24.05 915 0.47 Peakfield Estate: Maskeliya 2,700 12.3 0.52 19.85 908 0.48 Ellapita Ella: Maliboda Estate 580 2.26 0.44 16.3 950 0.65 A second study for the ESDP examined 11 projects, whose capital costs and plant factors are shown in Figure 4.1.46 Figure 4.1: Mini hydro costs 1 Hope:60kW Mooloya:150kW Hapugastenne:1500kW 0.8 Lelopitiya:100kW AVERAGE Cecilton:140kW factor Leobecondera:150kW plant Balangoda:350kW Wedemulle1:330kW 0.6 Diyagama:440kW Delta:270kW Wedemulle2:210kW 0.4 0 50 100 150 200 250 $/kW Based on these data, we divide the mini hydro potential into two tranches: a first tranche of 10 MW of the larger projects (such as Carolina and Maskeliya), with capital costs of less than $1,100/kW, implemented in 2000-2002, and a second tranche of 35 MW of smaller projects in the range of $1,100-$1,500/kW, implemented between 2001 and 2007. Wind power Sri Lanka's wind climate is characterized by two monsoon systems, the southwest monsoon, lasting from May to early October, and the northeast monsoon, from December to late February. Southwesterly winds are stronger and also display a greater directional 46 Posch and Partners, Sri Lanka Micro Hydro Feasibility Study, Report to the Asia Alternative Energy Unit, World Bank, January 1994. 30 Greenhouse Gas Mitigation Options in the Sri Lanka Power Sector persistence than the northeasterly winds. With strong to moderate winds prevailing for about eight months of the year in some parts of the country, Sri Lanka can be considered a country with a modest potential for wind power generation. A 3 MW wind power demonstration is under way with the support of the ESDP, with an estimated capital cost of 1,175$/kW. Sri Lanka lacks historical wind data of good quality to make a realistic assessment of the wind energy potential. From the viewpoint of wind energy assessment, existing wind data at meteorological stations suffer from several shortcomings: ?? inadequate height of measurement, typically at three to six meters; ?? inappropriate frequency of measurement, typically three hourly during the day and none in the nighttime; ?? poor exposure of the anemometer; and ?? lack of historical information on instrument calibration, replacement, change of measuring locations, landscape, and so forth. Nevertheless, some tentative conclusions regarding the classification of windy regions in Sri Lanka could be drawn from existing wind data and also from natural indicators. The situation is much better in the southern and southeastern lowland region, where CEB conducted a three- year wind monitoring program covering about 1,500 km2. Data gathered in this program consist of hourly average wind speed and maximum wind speed for each hour at 10, 15, and 20 minute intervals and wind direction at 20 minute intervals. Using both types of information, the wind potential in Sri Lanka can be summarized as indicated in Table 4.3, and a corresponding tentative estimate of the exploitable wind power potential is presented in Table 4.4. This analysis does not take into account the Central and Knuckles mountain regions because of the vast uncertainties associated with wind power prediction in complex terrain without firm data and modeling work. Table 4.3: Wind potential classification Typical terrain characteristics and Estimated range of mean annual wind Region landscape features speed at 40m (m/s) Southern and Largely flat terrain with large tracts of Inland areas: 4.5-5.5 southeastern open landscapes. On the coast: 6.0-6.5 lowlands Puttalam-Kalpitiya Largely flat terrain but extensively covered In open tracts: 5.5-6.0 up to Dutch Bay with coconut plantations; but quite a few On the coast: 6.0-6.5 open tracts of land are available. Mannar-Jaffna Insufficient information to describe In open tracts: 5.5-6.0 properly, but map information suggests On the coast and over the lagoon: that large open tracts are available all along 6.0-7.0 the coastal belt extending from Mannar- Pooneryne to the Jaffna peninsula. The Jaffna lagoon, particularly the Elephant Pass region, offers a lot of potential for shallow offshore wind development. Mitigation Options 31 Typical terrain characteristics and Estimated range of mean annual wind Region landscape features speed at 40m (m/s) Eastern coastal A vast open expanse of sandy, desert-like Inland and coast: 6.5-7.0 region of the Jaffna region with hardly any vegetation. peninsula Jaffna to TrincomaleeInsufficient information to describe Inland areas: 4.5-5.5 properly, but map information suggests Coastal: 6.0-6.5 that large open tracts are available all along the coastal belt, extending from the Jaffna peninsula-Mulativu to Trincomalee. Central mountains Rugged mountainous terrain, with the Very strong SW winds can be found in landscape characterized by tea plantations. mountain gaps and valleys, but unable to quantify because of the site-specific nature of winds. Ambewela plains close to Nuwara Eliya are a large open region with a slightly undulating landscape that offers some potential for wind power generation. NE monsoon winds rarely cross over to the central mountain range, which makes this region windy only for about 4-5 months each year. Knuckles range Rugged mountainous terrain, with the Very strong SW winds can be found in landscape characterized by tea plantations. mountain gaps and valleys, but unable to quantify because of the site-specific nature of winds. Table 4.4: Wind power potential Location Approx. size of the Fraction Area Potential MW region Length Width km2 developed km2 MW/km2 MW km km Mirijjawila-Hambantota 12 1.0 12 50% 6 6 36 Karagam Lewaya 3 1.5 5 50% 2 6 14 Kirinda-Palatupana 12 1.5 18 50% 9 6 54 Weligatta-Weerawila 6 6.0 36 50% 18 6 108 Sevanagala sugar plantation 6 6.0 36 50% 18 6 108 Suriyawewa-Mirijjawila 12 1.5 18 50% 9 6 54 Puttalam-Kalpitiya-Dutch Bay 40 1.0 40 50% 20 6 120 Region around Mannar, lat. 8o 45' to 9o 5' 40 1.5 60 50% 30 6 180 Ponnaweli­Pooneryn, lat. 9o 15' to 9o 35' 1.5 75 50% 38 6 225 Jaffna lagoon, lat. 9o 25' to 9o 35' 25 6.0 150 50% 75 6 450 Manalkadu­Chundikulam, lat. 9o 30' to 9o 50 3.0 150 50% 75 6 450 45' Total potential 600 300 1,799 Detailed studies using the WASP model were undertaken to examine the maximum about of wind power that could be absorbed by the system while still meeting the planning 32 Greenhouse Gas Mitigation Options in the Sri Lanka Power Sector criteria established by CEB using loss of load probability (LOLP of less than 0.1 percent in the years after 2001). These studies reveal that as much as 2,000 MW--that is, slightly more than the 1,800 MW indicated in Table 4.4 (which reflects 50 percent development of the respective areas)--could be absorbed into the system over the planning horizon. This maximum wind power scenario is shown in Figure 4.2. We assume a decrease in capital costs from the present $1,175/kW to $1,000 over the planning horizon. Figure 4.2: Capacity expansion plan, maximum wind development case 800 600 400 MW 200 Upper Kotmale Kukule 0 -200 1999 2001 2003 2005 2007 2009 2011 2013 2015 2017 hydro 70 150 coal 300 300 300 liquid fuels 106 300 105 105 105 105 105 300 gas diesel 100 -36 -36 renewables 3 30 30 30 60 60 75 75 75 75 150 150 150 150 150 150 150 150 150 150 Dendro-thermal power There are many proposals for dendro-thermal power plants for Sri Lanka.47 The main rationale is their renewable nature, avoiding costly fossil fuel imports. Equally important is employment generation in areas presently characterized by high unemployment. We assess the potential of dendro-thermal power plants for GHG mitigation by postulating a scenario in which baseload coal plants are replaced by dendro-thermal units in 50 MW increments.48 This results in the capacity expansion plan indicated in Figure 4.3. It is assumed that gasifier rather than combustion (steam cycle) technology is employed: the assumptions are given in Table 4.5. 47 See, for example, R. Wijewardene and P. G. Joseph, Growing Our Own Energy: Complementing Hydro- power for Sustainable Thermal Energy and Rural Unemployment in Sri Lanka, Govt. of Sri Lanka, Colombo, 1999. 48 This is purely for modeling convenience. In actuality, the increments could be in much smaller units, more closely matching the year-by-year incremental capacity requirements. Mitigation Options 33 Table 4.5: Assumptions for dendro-thermal power plants Dendro-thermal Comparable (gasifier) values for coal plant Unit size 10 2 x 300 Capital cost $/kW 1,200 899 Fixed O&M $/kW/month 1.8 0.64 Variable O&M (nonfuel) $/MWh 2.5 3.54 Heat rate KCal/kWh 4,560 2,162 Sulphur content % 0.02 0.6 Heat value as receiveda 3,800 6,300 Plant gate cost US$/ton 22 48 Land area for fuelwood Ha/MW 360 plantation a. Assuming 20 percent moisture for fuelwood Source: Based on information provided by P. G. Joseph of the Sri Lanka Energy Conservation Fund, Colombo. Figure 4.3: Capacity expansion plan for the dendro-thermal scenario 500 400 300 200 MW Upper Kotmale 100 Kukule 0 -100 1999 2001 2003 2005 2007 2009 2011 2013 2015 2017 hydro 70 150 coal liquid fuels 106 300 105 105 105 gas diesel 100 -36 -36 renewables 0 0 0 0 10 100 0 100 100 150 0 150 0 200 200 0 200 0 200 0 Dendro-thermal power plants require substantial land for the necessary fuelwood plantations. However, the advocates note that 1,000 MW of dendro-thermal would require only some 20 percent of present scrub and chena lands (Table 4.6). Table 4.6: Land requirements for dendro-thermal plants 34 Greenhouse Gas Mitigation Options in the Sri Lanka Power Sector Total land area of Sri Lanka (ha) 6,500,000 Scrub lands 600,000 Chena lands 1,000,000 Land requirement for: 10 MW 3,600 ha plantation 1,000 MW 360,000 ha = 20% of scrub and chena lands A more potent question concerns the fuel cost (assumed at $22/ton, delivered to the plant gate). This is based on the present price of fuelwood as delivered in Colombo. Fuelwood produced in properly managed plantations, with power plants sited at the center of such plantations, may reduce this cost: To test the sensitivity of the resulting cost of avoided carbon to fuel cost assumptions, we therefore consider two dendro-thermal fuel cost scenarios, one at $22/ton, the second at $14/ton.49 Solar photovoltaic (PV) systems for remote rural applications Solar PV systems for remote rural homes ("solar homes") represent another component of the ESDP. Such systems have been marketed for some time by commercial vendors, but GEF/ESDP support is warranted by the many institutional and financing constraints that have impeded greater market penetration to date.50 The solar homes scenario is illustrated in Figure 4.4. We assume that by 2018, about 85 percent of the then remaining unelectrified homes would be served by solar systems, representing about 500,000 systems in place. This is significantly greater than the present estimate of between 100,000 to 215,000 homes considered by the private vendors as the present potential market, taking into account affordability criteria.51 However, over the 20-year planning horizon considered here, and given the GDP growth projection also used as a basis for electricity demand forecasting, a doubling of real income (and, say, a consequent doubling of the potential market) ought to be achievable. Moreover, we assume that the present cost of $6,000/kW will decrease over time, reaching $1,000/kW by 2018, and the unit output per system would increase from the present 30-50 W52 to 200 W by 2018. These postulated improvements (as well as the market penetration assumptions) may well be optimistic, but seem 49 The analysis of fuelwood plantations using eucalyptus grandis in a seven-year growing cycle by N. Walpita (Biomass Development and Environmental Consequences, Govt. of Sri Lanka, August 1997) shows a financial cost of $21/ton of fuelwood. However, the short-rotation coppice system shows considerable promise for lowering these costs. Fast-growing coppicing species, such as gliricidia, leucaena, casuarina, and felcipium, would be planted at about one-meter spacing and are initially cropped at about one meter of height. Subsequent harvesting can be twice a year, with an overall biomass yield per hectare about double that of traditional fuelwood trees. 50 The ESDP will provide a $100 grant per solar home module of 30 W or greater. 51 See, for example, P. Jayewardene and V. Perera, Sri Lanka: The Rural Electrification Problem, Report by Power and Sun (Pvt) Ltd., Colombo March 1991. 52 The ESDP that targets 2,200 medium-income (income greater than Rs 3,000/month) rural households in the Galle district assumes that 30 percent of the system sold would be 30 W units, 40 percent would be 40 W units, and 30 percent would be 50 W units. Mitigation Options 35 warranted to determine the bounds of potential cost-effectiveness for GHG mitigation. Figure 4.4: The solar homes program 3 2 unelectricifed households million 1 cumulative (% market share) 5 0 6 4 8 3 % % % 1 5 1 9 2 3 3 0 4 6 8 1 0 1 3 % % % % 0 . 2 0 . 4 0.8 1 2 2 3 % % % % % units % % % % % % % sold 0 1998 1 9 9 9 2 0 0 0 2 0 0 1 2 0 0 2 2 0 0 3 2 0 0 4 2 0 0 5 2 0 0 6 2 0 0 7 2 0 0 8 2 0 0 9 2 0 1 0 2 0 1 1 2 0 1 2 2 0 1 3 2 0 1 4 2 0 1 5 2 0 1 6 2 0 1 7 4.2 Fuel substitution options Oil steam-cycle plants The economics of oil steam-cycle plants are studied in some detail in another study.53 For this GHG assessment, we assume that 3.5 percent high-sulphur fuel oil would be imported from Singapore, and that FGD would be required (as uncontrolled emissions would exceed standards and are higher than from plants burning Australian low-sulphur coal). The plant characteristics are taken from the Electrowatt Thermal Options Study. Liquid natural gas (LNG) There is no question that from an environmental point of view, LNG is an attractive option when compared with coal- or oil-based plants.54 Moreover, LNG has the advantage that it is readily burnt in combustion turbines (with or without a combined cycle) that are characterized by high efficiency and that have seen significant recent decreases in capital costs.55 53 International Development and Energy Associates, Ltd., Sri Lanka Electric Power Technology Assessment (SLEPTA), Draft Report to the World Bank, November 1999. 54 For example, among fossil fuels, natural gas has the lowest emissions of GHG emissions per unit of generation (15.2 tons C per terajoule (TJ) as compared with 25.8 for coal and 21.1 for heavy fuel oil) 55 A comparison of the 1996 and 1998 generation plan assumptions is instructive. In 1996, the capital costs of open-cycle gas turbines were estimated at 514$/kW (CEB 1996); in 1998, this had dropped to 372$/kW (CEB 1998), an assumption documented by the actual price of a 105 MW unit commissioned in 1997. Nevertheless, it is worth noting that in the 1988 Thermal Options Study (by Black and Veatch), the capital costs of a 300 MW coal plant were estimated at about $1,600/kW, which a decade later had fallen to $1,451 (for the first unit of the Puttalam project), including port development in both cases. In 1988 prices, this is about $1,000/kW, implying a 40 percent decrease in capital cost. 36 Greenhouse Gas Mitigation Options in the Sri Lanka Power Sector But whether in the case of Sri Lanka these advantages offset the disadvantages--notably the very high cost of LNG--is unclear. A detailed assessment of the economics of LNG for Sri Lanka has again been undertaken elsewhere.56 For the overlay study, we take the LNG price, delivered to the plant gate, at $5/mmBtu (based on a Saudi Light crude oil price of 16$/bbl). This includes the cost recovery of the upfront infrastructure costs (terminal, gas storage, regasification facility, etc.) Capital and operating costs of the power generation plant, heat rate, and so forth may be taken as the same as for diesel-fueled combined-cycle plants.57 Conventional hydro Among larger conventional hydro schemes, only the Kukule and Upper Kotmale (Talawakelle) projects are in the least-cost generation expansion plan, and consequently, to examine the effect of additional hydro as a GHG mitigation option, the remaining hydro projects have to be forced into the solution, with exogenously specified commissioning dates.58 This maximum hydro development scenario is shown in Table 4.7.59 Table 4.7: Additional hydro projects for the maximum hydro scenario Assumed start- MW Capital cost $/MWa up Broadlands 2005 40 2,548 Gin Ganga 2005 49 2,127 Moragolla 2007 27 3,002 Uma Oya 2010 150 2,152 a. Pure (overnight) project cost, without interest during construction (IDC), taxes, duties. 56 SLEPTA, op.cit. 57 In SLEPTA, other LNG development scenarios are also considered. For example, one might hypothesize that a much larger LNG project would be set up, of the size being proposed in India (at around 2.0-2.5 mtpy). In this case, other gas users (transportation, industry) would account for the balance of the gas off-take. However, even in this case, where one may be able to benefit from economies of scale, LNG is still very costly. 58 One of the uncertainties of the GHG emission reductions that may actually be achievable by hydro plants over the sort of time scales appropriate to analysis of climate change is the impact of siltation. Recent surveys of the impoundments in the Upper Mahaweli scheme show high rates of physical soil loss in the catchment areas, and correspondingly high rates of reservoir siltation. The siltation level of the Polgolla reservoir in 1994, some 14 years after commissioning, has been estimated at 44 percent of its total capacity (IDA 1997). The importance of watershed management has been recognized in Sri Lanka's National Environmental Action Plan (1990), and is one of the main components of a new World Bank-supported project to strengthen institutional capacity in watershed management and environmental regulation (IDA 1997). 59 However, as noted in SLEPTA, under the assumption of tied concessionary financing (i.e., financing that would not be available to Sri Lanka for other sectors), these hydro plants are also in the (economic) least-cost plan. These hydro plants would also enter the least-cost plan at lower discount rates. (For example, in one sensitivity analysis conducted by CEB for a zero discount rate, all these hydro candidates are selected.) Mitigation Options 37 Biomass co-firing Given the high cost and uncertainty surrounding dendro-thermal gasifiers, biomass co- firing at the proposed coal plants may be possible. Supplying 15 percent of the energy input from fuelwood should be feasible at minimal modification of the coal plant (about $75/kW). Table 4.8 summarizes the assumptions used for this case, which are based on the Electric Power Research Inst. (EPRI) Technical Assessment Guide (TAG). The plantation fuelwood that might be used in Sri Lanka would have a higher heat value and lower moisture content than the wood fuel blend (of urban wood, mill, forest, and agricultural crop residues) assumed in the EPRI study. This option will be studied in detail in the planned follow-up study. Table 4.8: Assumptions for wood co-firing EPRI TAG assumptions As used in this volume Power consumption in wood 60 kWh/dry ton pulverizers Reduction in boiler efficiency 1.5% Wood characteristics 33% moisture 20% moisture 5,495 Btu/lb=3,022KCal/kg 3,800 KCal/kg Hence: increase in net plant 166 Btu/kWh 45 KCal/kWh heat rate Wood pulverizers $19/kW (at 1993 prices) $25/kW Fuel-handling facilities $31/kW (at 1993 prices) $40/kW Modifications of boilers and $6/kW (at 1993 prices) $10/kW other facilities 4.3 DSM A program of lighting DSM is assumed to be included in the reform case, as described above in Section 3. For the further GHG mitigation cases, we separately examine EEMs and energy-efficient air conditioning (EEAC). Table 4.9 and Figure 4.5 show the results of the screening analysis of individual options. Table 4.9: Screening of DSM programs $/ton CO2 Tons of avoided carbon Avoided Measure Cumulative IND: slim tube -51.4 353 353 COMM: slim tube -48.3 693 1,046 RES: 23 Wcfl -46 481 1,527 IND: EEM -41.9 403 1,930 RES: 18 Wcfl -31.3 525 2,455 IND: incand>18 Wcfl -23.8 412 2,867 RES: 13 Wcfl -14.2 1,181 4,048 COMM: EEAC -10.2 417 4,465 COMM: incand>Sl comfort 2.2 779 5,244 Note: Negative values are win-win 38 Greenhouse Gas Mitigation Options in the Sri Lanka Power Sector Figure 4.5: Cost of avoided carbon, $/ton CO2 20 COMM:incand>SLcomfort 0 tonCeq] [$/ COMM:EEAC RES:13Wcfl carbon -20 IND:incan>18Wcfl avoided RES:18Wcfl of cost-40 IND:EEM RES:23Wcfl COMM:slimTube IND:slimTube -60 0 2 4 6 avoided carbon [undiscounted, million tons] 4.4 Options Not Considered Finally we may briefly explain why some mitigation options considered in other countries (particularly in India) have not been considered in Sri Lanka. Nuclear power However one may assess the broader sociopolitical issues of nuclear power, advocates argue that nuclear power could make a significant contribution to GHG emission reduction efforts.60 Nuclear power was included as an option in a recent environmental assessment of the Indian state of Andhra Pradesh, which indicated significant GHG emission reductions would be possible at a cost of avoided carbon of about $9.50/ton C.61 The GHG overlay study for Bihar shows the avoided costs of nuclear power to be only $2.5/ton C.62 However, in the case of Sri Lanka, nuclear power does not arise as a technically feasible option for at least the next 20 years. Reliability and system stability considerations require that the largest single unit in a system ought not to account for more than 10 percent of the peak load, particularly in a system that is not interconnected.63 Presently, commercial units of around 500 60 See, for example, K. Matsui, "Global Demand Growth of Power Generation, Input Choices and Supply Security," Energy Journal 19(2): 93, 1998. 61 Administrative Staff College of India, India: Global Overlay for Andhra Pradesh, Report to the World Bank, March 1999, Table 4.3. This estimate is based on undiscounted GHG emissions over a 20-year planning horizon. 62 Sone Command Area Development Authority (SCADA), India, Global Overlay for Bihar, Report to the World Bank, March 1999. Again, this estimate is based on undiscounted GHG emissions. 63 Interconnecting Sri Lanka with southern India by a transmission line across the Palk Straits was actively discussed in the mid-1970s, but interconnection to South India does not now seem very likely for some time to come. Mitigation Options 39 MW unit size could therefore be accommodated only once the peak demand reaches 5,000 MW, which the load forecasts indicate as being unlikely before about 2015.64 Rehabilitation of thermal power plants All of Sri Lanka's older thermal plants have recently been rehabilitated: the Kelanitissa combustion turbines in 1998, the Kelanitissa steam plant in 1991, and the Sapugaskanda diesel plant in 1991. In each case, life extension of 10 to 15 years was achieved. However, CEB studies show that beyond 2000, the fixed costs of maintaining these plants would be uneconomic, and they should therefore be retired. In view of the advanced age and condition of these plants, further life extension is not viewed as practicable. In any event, even if their heat rates consequent to further rehabilitation could be improved, they would still be relatively low in the merit order and would be used very infrequently. Consequently, any incremental GHG emissions attributable to such use would be negligible, and the question of thermal plant rehabilitation as a GHG emissions reduction measure does not therefore arise in Sri Lanka.65 4.5 Results The results of the screening analysis are summarized in Table 4.10. Note that the reductions in emissions in columns 4 and 5 cannot be added, since the individual data points represent perturbations of the reform scenario, one at a time (and some are mutually exclusive, such as LNG, oil steam-cycle with FGD, or the dendro-thermal options at different delivered wood costs). Negative costs of avoided carbon indicate win-win options. 64 New technology--modular, fail-safe units in the 150-350 MW size range--that could be accommodated into smaller systems may change this assessment. However, commercial availability of such units seems at least 10 to 15 years distant. 65 This is in sharp contrast to India (see Table 4.11). 40 Greenhouse Gas Mitigation Options in the Sri Lanka Power Sector Table 4.10: Cost of avoided carbon System ??(cost) CO2 ? (CO2) ? (C) Cost of cost, $ (relative emissions (relative avoided million to reform) (million tons)to reform) carbon, $/ton C Reform 2,449 98.4 +DSM: ind: EEM 2,430 -18.5 96.6 -1.8 -0.5 -37.6 Oil steam (resid. oil) 2,325 -123.7 83.1 -15.3 -4.2 -29.7 Oil steam (resid.+FGD 2,362 -87.2 83.1 -15.3 -4.2 -20.9 +DSM: Comm AC 2,441 -7.6 96.4 -2.0 -0.6 -13.6 Oil steam+FGD 2416 -33.2 88.5 -9.9 -2.7 -12.3 +T&D>10% 2,440 -9.2 95.2 -3.2 -0.9 -10.4 Mini hydro: I 2,448 -0.8 97.5 -1.0 -0.3 -3.2 Mini hydro: II 2,450 1.5 95.7 -2.7 -0.7 2.1 Dendro-thermal (14$/t) 2,492 42.7 39.2 -59.3 -16.2 2.6 Oil steam (0.3% S) 2,465 16.5 88.6 -9.8 -2.7 6.1 Dendro-thermal (22$/ton) 2,580 131.1 39.6 -58.8 -16.0 8.2 Diesel ($800/kW) 2,494 45.3 83.8 -14.6 -4.0 11.4 LNG 2,545 96.3 77.3 -21.1 -5.8 16.7 Diesel ($1,140/kW) 2,580 130.9 83.8 -14.6 -4.0 32.8 Max hydro 2,553 104.0 88.3 -10.1 -2.8 37.7 Wind (max) 2,870 421.2 74.8 -23.7 -6.5 65.3 Solar homes 2,490 41.6 97.6 -0.8 -0.2 184.5 Figure 4.6 displays these data in a different way, plotting system cost against GHG emissions (undiscounted), with the four quadrants defined by the reform case. The slope of the line connecting any point with the reform case defines the cost of avoided carbon (column 5 of Table 4.10). Mitigation Options 41 Figure 4.6: System cost v. emissions 3000 Wind[Max] 2800 BAU[#] T&DlossRed USmillion$ 2600 Dendrothermal[22$/ton]Diesel[$1140/kW] Wind[~] (costs) LNG MaxHy DSM PV Dendrothermal[14$/t] Diesel[$800/kW] SolarHome Oil Steam s MiniHydro:I +DSM:CommAC Reform +T&D>10%(F=1.5 OilSteam+FGD 2400 +T&D>10%(F=1.0) ) +DSM:ind:EE Minihydro:II M OilSteam[Resid]+FGD OilSteam[Resid] 2200 20 40 60 80 100 120 140 total GHG emissions There are few surprises in these results. The most expensive in terms of the cost of avoided carbon is the solar homes program at $184/ton C. However, this result is a good example of the hazards of conducting the analysis in terms of changes in system costs rather than changes in net benefits. When properly adjusted for the much higher economic value of the benefits provided (rural lighting), the cost of avoided carbon reduces to $ 132/ton C. Indeed, under some assumptions, however, this may fall to as little as $1/ton C (see Box 4.2). Box 4.2: Economic benefits of the solar homes program The solar homes program targets primarily lighting service, for which the economic benefit (willingness to pay) is much higher than for electric service as a whole. (Experience with solar systems sold thus far suggests that the next most valued use is for TVs.66) The rationale for using system cost in this study is that in all cases the demand forecast is satisfied, and hence the level of benefits to consumers stays unchanged. However, the assumption is that the solar homes would be implemented to those rural areas unconnected to the CEB grid, therefore by definition adding to the economic benefits. The incremental value of these benefits must therefore be subtracted from the system cost. This incremental benefit can be estimated by calculating the value of kerosene that is displaced (though this calculation would still underestimate the benefits, since it ignores the economic benefits consequent to the higher quality of light, improved convenience, safety, and better indoor 66 See L. Gunaratne, Making the Investment: Innovative Ideas in Alternative Energy Policy Financing and Services: Annex Market Study Commissioned by National Development Bank, Paper presented at the FINESSE Workshop, Kuala Lumpur, Malaysia, October 1991. 42 Greenhouse Gas Mitigation Options in the Sri Lanka Power Sector air quality). Such a calculation was the basis for the cost-benefit analysis of the ESDP solar homes subcomponent, which estimated a net present value (NPV) of benefits at almost exactly the NPV of the program costs (resulting in an IERR of 12 percent).67 In other words, the incremental cost of the program would be (at least) zero, but with some GHG emission reduction. Our own calculations suggest that the NPV of incremental costs of the solar homes program is $44.5 million, which is reduced to $41.6 million when CEB's offsetting fuel savings are taken into account. Using standard assumptions,68 we estimate the value of kerosene savings displaced at $40.5 million, thereby reducing the incremental cost from $41.6 million to $1.1 million, for which the revised cost of avoided carbon is $1.3/ton (making this one of the most cost-effective GHG mitigation programs). Unfortunately, use of the standard assumptions (in this as well as the ESDP calculation)--in which one starts with the lumen output of the electric light and converts it into kerosene equivalent--does not appear to be consistent with actual consumption of kerosene in Sri Lanka. According to Ceylon Petroleum Corporation (CPC), the present consumption of kerosene is some 238,000 tons/year. Making the conservative assumption that this is consumed only for lighting in the roughly 2.5 million nonelectrified households, the average kerosene consumption in such households is 95 liters/year, compared with 131 liters/year resulting from the standard assumptions. This is particularly a problem since the assumed output of the solar systems is assumed to increase over time, making the per household kerosene equivalent equal to 197 liters/year by 2005. If we assess willingness-to-pay (WTP) by the actual average purchase of kerosene per nonelectrified household, then the value of avoided kerosene purchase falls to $14.8 million, reducing the incremental cost from $44.5 million to $29.7 million, with a resultant cost of avoided carbon of $ 132/ton. The discrepancy arises because the level of lighting increases once solar electricity is provided. However, the ability to purchase the equivalent level of kerosene, especially by the rural poor, is constrained by income. Thus, WTP as revealed by actual purchase of kerosene for lighting services is much lower than that implied by the standard calculation. 67 ESDP PAD, op.cit., Annex 4B. 68 Luminous efficiency of 12 kilolumen-hours per kWh, and a kerosene consumption of 0.1 kg per kilo- lumen hour (in a kerosene mantle lamp). Mitigation Options 43 Figure 4.7: Correlation between GHG emissions and SOx emissions 0.26 BAU[#] 0.24 T&D loss Red DS 0.22 +DSM:ind:EEMM Mini hydro:II +T&D>10%(F=1.0) Solar Homes+DSM:CommAC Wind[~] SOx)emissions Reform +T&D>10%(F=1.5) 0.2 PV(: Max Hy Wind [Max] 0.18 IMPACT 0.16 Diesel [$1140/kW] LOCAL Dendro thermal [22$/ton] LN Oil Steam+FGD 0.14 G OilSteam [Resid]+FGD 0.12 20 40 60 80 100 120 140 GLOBAL IMPACTS: GHG emissions [to 2018], undiscounted Impact on local air emissions Implementation of GHG emissions reduction measures brings significant reduction in local air emissions (SOx, NOx, particulates). Figure 4.7 shows the relationship between GHG emissions and SOx emissions for the individual mitigation measures (expressed as perturbations of the reference scenario). However, the monetized value of these local environmental benefits is small in comparison with the cost of avoided carbon.69 Comparison to Indian GHG overlay studies The Sri Lankan results may be compared to those of the two GHG overlays recently prepared for the Indian states of Andhra Pradesh70 and Bihar.71 Our results are similar to the results in India for some technologies (lighting DSM, further reduction of T&D losses), but quite different for others (such as LNG), as indicated in Table 4.11. These results are all expressed in terms of undiscounted emissions. Table 4.11: Comparison with Indian GHG overlay study results: Cost of avoided carbon in $/ton C (measured against reform scenario) 69 For a detailed description of how environmental costs and benefits are measured, see Munasinghe 1992. 70 Administrative Staff College of India, op. cit., March 1999. 71 SCADA, op. cit., February 1999. 44 Greenhouse Gas Mitigation Options in the Sri Lanka Power Sector 5 Sri Lanka Andhra Pradesh Bihar DSM -13 to -38 (EEM, AC) -28.9 (lighting)a -18.0 DSM-agriculture -14.3 Biomass 8.2 (dendro-thermal) -18 bagasse -44 bagasse Refinery residue 3.3 Further T&D rehabilitation -10.4 -19.4 -36.0 LNG 17.0 0.7 24.1b Thermal plant rehabilitationc -71.5 Mini hydro -3.2 to 2.1 4.0 -21.1 Large hydro 38.0 -8.4d 34.0 Nuclear 9.5 2.5 Wind 65.0 22.0 Solar PV 184.0 100.0 108.0 Coal washing 167.0 Note: Indian results are assessed and reported as Rs/ton, converted to US$ at Rs 43/$. a. In Sri Lanka, lighting DSM is assumed to be part of the reform program. b. Pipeline gas imported from Bangladesh at $4/mmBTU. c. In Andhra Pradesh, rehabilitation of the Kothagudem thermal plant. In Bihar, part of reform. Not applicable to Sri Lanka. d. The hydro in Andhra Pradesh is the powerhouse-only plant at Jurala, at an existing irrigation project. The LNG results in the case of Andhra Pradesh, which suggest a cost of avoided carbon of only $0.70/ton rather than $17/ton as in Sri Lanka,72 are particularly interesting, since, as in Sri Lanka, LNG is assumed to replace imported Australian coal.73 However, as is evident from Table 4.12, the Andhra Pradesh results are based on a lower LNG price (the difference of 0.50$/mmBTU reflects the difference in scale, Indian LNG developments allowing 2,000 MW plant sizes) and a much higher capital cost for imported coal;74 consequently the cost of avoided carbon in Andhra Pradesh is much lower. Table 4.12: Comparison of assumptions, Andhra Pradesh v. Sri Lanka 72 However, this is the undiscounted value. When properly discounted (as is necessary for an analysis of carbon offset revenues), the corresponding cost of avoided carbon climbs from $16.70/ton to $79.20/ton (see Table 6.2). 73 According to the firm ASCI, Andhra Pradesh's own coal resources are limited, and they would be able to command only some small share of the production of the large coal fields in the Talcher region of Orissa (which requires shipment by rail to Paradeep, then coastal sea transportation to ports on the Andhra Pradesh coast). 74 The Sri Lanka estimate is that of the Electrowatt Thermal Options Study, as described previously, and includes the cost of the pier and unloading facilities. The Andhra Pradesh costs are described as economic costs (and therefore exclude escalation and IDC), and "include the cost of pier construction" (and are thus comparable to the Sri Lanka basis). The explanation of the difference is that the Andhra Pradesh costs are based on the actual proposals of IPPs for memorandum of understanding (MoU)- based projects, while the Electrowatt estimate is based on costs revealed by initial contract bid (ICB). However, costs for LNG combined-cycle plants in Andhra Pradesh, as in Sri Lanka, are both based on ICB (in Andhra Pradesh, based on the cost of the Vijeswaram II plant). Mitigation Options 45 Andhra Pradesh Sri Lanka Coal Capital cost $1,139/kW $899/kW Fuel cost $2.16/mmBTU $1.85/mmBTU LNG Capital cost $697/kW $634/kW Fuel cost $4.50/mmBTU $5.00/mmBTU 5 GHG Mitigation Scenarios 5.1 Combining mitigation options Starting with BAU, individual mitigation measures are brought into the solution cumulatively. The criterion for introducing the next measure could be the next lowest cost of avoided carbon or the measure most likely to be implemented. For example, starting with BAU, the first step could be either DSM (which has the best cost of avoided carbon) or T&D rehabilitation (which experience in India shows is more easily implemented as the first step of reform). In Figure 5.1, we show T&D loss reduction as the first step. Table 5.1 shows the order in which the measures are brought into solution, reaching the least-cost scenario after introduction of all of the DSM, mini hydro, and further T&D loss reduction.75 Any further measures to reduce GHG emissions would then begin to increase costs. Figure 5.1: Defining the GHG mitigation scenario 2800 [1] BAU [#] 2700 [2] T&D loss Red ++Wind cost] 2600 [system ++Max Hy DSM PV 2500 ++Diesel [10] ++Dendro ++LNG [3] Reform [4]+DSM:ind:EEM [5] ++DSM:CommAC [6]++mini Hy I 2400 [9] ++Oil St+FGD [7]++ T&D>10% [8]++Mini Hy II 2300 40 60 80 100 120 140 GHG undiscounted, million tons 75 The least-cost solution could be further improved if we assume additional quantities of residual oil available for diesel-powered generation from a future refinery expansion. However, given the uncertainties of such an expansion, we limit residual oil use to that presently available. 47 48 Greenhouse Gas Mitigation Options in the Sri Lanka Power Sector By the time we reach the economic least-cost scenario (all the DSM, mini hydro, and further reduction of T&D losses)--the lowest point on the curve--the number of coal plants over the planning horizon has reduced further from the five in the reform scenario to four (see Figure 5.2). Table 5.1: Definition of mitigation scenarios DSM Mutually exclusive options CEB T&D Light- Ind: Comm Mini T&D Mini Oil Dendro Diesel LNG Max Wind Solar base(>12.5 ing EEM AC hy: I >10 hy: IIsteam hy homes case %) % +FGD $/ton C -82 -35 -50 -20 -16 -14 -0.5 -21 8 11 16 38 66 150 1. BAU x x 2. +T&D>12.5% x x 3. Reform x x x 4. +DSM: Ind x x x x EEM 5. +DSM: Comm x x x x x AC 6. +Mini hydro: I x x x x x x 7. +T&D>10% x x x x x x x 8. +Mini hydro: x x x x x x x x II 9. Oil steam x x x x x x x x x (+FGD) 10.Dendro-thermal x x x x x x x x x 11.Diesel x x x x x x x x x 12.LNG x x x x x x x x x 13.+Max hydro x x x x x x x x x x 14.+Wind x x x x x x x x x x x 15.+Solar homes x x x x x x x x x x x x GHGMitigation Scenarios 49 Figure 5.2: Capacity expansion plan, least-cost scenario 400 300 200 MW Upper Kotmale 100 Kukule 0 -100 1999 2001 2003 2005 2007 2009 2011 2013 2015 2017 hydro 70 150 coal 300 300 300 300 liquid fuels 106 300 105 105 gas diesel 100 -36 -36 renewables 0 3 8 10 5 5 5 5 5 0 0 0 0 0 0 0 0 0 0 0 5.2 Developing the SACC Table 5.2 shows the avoided costs of carbon and the percentage reductions (relative to the BAU case) achievable in each scenario. Table 5.2: GHG mitigation scenarios Scenario System (cost) CO2 (CO2) $/ton C % cost ($ reduction million) (BAU) BAU(#) 2,775 382 118.0 30 -46.4 T&D loss reduction 2,653 260 107.1 19 -49.5 9 Reform 2,449 56 98.4 11 -19.4 17 +DSM: ind: EEM 2,430 38 96.6 9 -15.7 18 ++DSM: Comm AC 2,425 32 95.3 7 -15.7 19 ++Mini hy: I 2,423 30 94.6 7 -16.2 -20 ++T&D>10% 2,403 10 90.5 3 -13.8 23 ++Mini hy: II =least-cost 2,393 87.8 26 ++Oil steam+FGD 2,402 10 79.6 -8 4.4 33 ++Dendro 2,481 89 46.2 -42 7.8 61 ++Diesel 2,500 107 75.4 -12 31.6 36 ++LNG 2,465 73 70.2 -18 15.2 41 ++Max hydro 2,540 147 49.8 -38 14.2 58 ++Wind 2,639 246 50.1 -38 23.9 58 50 Greenhouse Gas Mitigation Options in the Sri Lanka Power Sector It may be noted that win-win options when considered as a stand-alone perturbation of the reform case--such as oil steam+FGD--may not necessarily still be win-win when considered together with other win-win options in a scenario. As shown in Table 5.2, introducing oil steam+FGD after the last tranche of mini hydro increases costs, and is no longer win-win. The GHG mitigation supply curve--where we introduce each measure in order of avoided cost--is shown in Figure 5.3. Figure 5.3: Standard abatement cost curve (GHG mitigation supply curve) 60 40 20 tonCeq] [$/ [8]+Mini HyII [10] Dendro 0 [7]+T&D>10% [6]+mini H y I carbon [5]+DSM:Comm AC [4]+DSM:Ind:EEM Reform avoided -20 of cost -40 [2]+DSM -60 0 20 40 60 80 avoided carbon [undiscounted, million tons] 6 Resolving Uncertainty The results outlined in the previous section represent only the starting point for identifying an appropriate GHG strategy for Sri Lanka. Foremost among the problems facing decisionmakers is the large uncertainty associated with such calculations. And among these many sources of uncertainty is the question of the value of the carbon offset--which could range from zero, if a market for carbon reductions does not develop, to values in excess of $20/ton if the physical impacts of increasing global warming begin to appear in the short term. 6.1 Sources of uncertainty and the real options approach This decision problem can be analyzed using areal options framework, which has been extensively developed in other fields and which has potential application to GHG mitigation strategy. Box 6.1 illustrates the basic concept using a simple numerical example. A detailed description of the analytical framework is given elsewhere.76 Uncertainty is a key driver that calls for the options-based approach to be used to enhance the traditional NPV rule, and it features prominently in the assessment of sustainable development projects. Munasinghe et al (1998) classifies climate change-related uncertainty into two broad categories: 1. Scientific uncertainty, which arises as a result of limited knowledge of the rates of natural emission and absorption, and the environmental consequences of greenhouse gases; and 2. Socioeconomic and technological uncertainty linked to human actions, including uncertainties that arise because of the difficulty of predicting the future rates of emissions arising from human activity, the inability to value in monetary terms the future impact of GHG accumulation and other forms of environmental degradation, lack of knowledge about human responses to environmental changes, and lack of knowledge about the development of technological options that would affect all aspects of the environmental equation". As noted earlier--for example, in the case of dendro-thermal plants--there is uncertainty about whether fuelwood can actually be grown at the scale and cost envisaged by its 76 Fernando and Munasinghe, op cit., December 1998. 51 52 Greenhouse Gas Mitigation Options in the Sri Lanka Power Sector advocates, about the cost of gasifier technology, and about the value of carbon emissions avoided from fossil fuels that are displaced. It is also quite evident that long time frames are very much a factor in sustainable development projects, with typical forecast periods starting in 2010 and ranging up to 100 years or more beyond this time.77 Even a relatively small amount of uncertainty can lead to a very wide range of outcomes over such long periods. Box 6.1: Calculation of option values78 Consider an environmental project that requires an investment of $100 million. Based on the information available today (time 0), this project will yield an environmental credit valued at $200 million (good state) or $50 million (bad state), with equal probability, a year from now (time 1). The project terminates at time 1. Assuming a discount rate of 10 percent, the expected value of the investment (NPV) is given by NPV = 0.5 {200/1.1} + 0.5 {50/1.1} ­ 100 = $13.6 million Return in year 1 Return in year 1 Cost of xPr{good state} xPr{bad state} investment The standard NPV rule would recommend that this project be undertaken to realize the expected positive NPVof $13.6 million. Now suppose the project can be postponed at a cost of $5 million. Suppose further that one year from now, it will be known whether the state is good or bad. If the good state is observed, the investment will be made, with a return of $200 million at time 2. If the bad state is observed, the investment will not be made. Therefore, the NPV of postponing the investment decision until t=1 is given by NPV = 0.5 {200/1.12} - 0.5 {100/1.1} + 0.5*zero ­ 5 = $32.2 million Return in year 2 Cost of investment Pr{bad state} Cost of xPr{good state} x[no investment] delay Thus, despite having a positive NPV if undertaken at time 0, it will never be optimal to invest in this project at time 0, since investors can always do better by waiting until time 1, when the uncertainty is resolved and the expected value increases to $32 million. The option value associated with delaying a decision by one year is therefore $32.2 million - $13.6 million = $18.6 million. 6.2 Applying the options approach We begin the analysis of uncertainty with Table 6.1, in which we examine the use of oil steam (+FGD) as a mitigation measure, given the least-cost scenario as a starting point. As shown in panel A, this increases the system cost by $9.7 million and decreases GHG emissions 77 See, for example, Jepma and Munasinghe (1998). 78 Adapted from Fernando and Munasinghe, op.cit., Section 3.1. Resolving Uncertainty 53 by 8.1 million tons, with a discounted cost of avoided carbon of $21.20/ton79 (which falls to $4.40/ton when undiscounted). Panel B shows the probability distribution for the value of the carbon offset, expressed as $/ton carbon. The expected value of the carbon offset, E{V(c)}, is $22.00. The corresponding expected value of carbon offset revenue, E{R(c)}, is $10.1 million when discounted at 10 percent. E{R(c)} is given by E{R(c)}=? VjP{j}? C where ? C is the avoided carbon (8.1 million tons in this case), and P{j} is the probability of the carbon offset having value Vj. Panel C then derives the expected value of implementing this mitigation option today, which is the cost as given in Table 5.2 (NPV of $2,402.3 million), less the expected value of the revenue from carbon offsets ($10.1 million). This is $0.4 million less than the least-cost case ($2,392.6 million), whose expansion plan is depicted in Figure 5.2. This reduction in cost (though quite small--about 4 percent of the expected value of carbon offset revenue) would normally justify undertaking the mitigation measure (given the expected value of $22/tonne for carbon offsets).80 In Panel D, we now consider the implications of delaying a decision by five years. Since a commitment to baseload plant in the least-cost solution would otherwise need to be made now (to meet demands five years hence), this delay is not costless, for in its place one must build some other of interim capacity--say, diesels--to meet the demand. We therefore recalculate the capacity expansion plan such that the first coal unit is delayed by five years, with 300 MW of diesels (at $1,100/kW) built over the five-year period. Then five years from now, one would reconsider whether to then build a coal plant, or whether to implement the mitigation measure. The new costs and GHG emissions are noted in Panel D. Because the presence of the interim diesels has some impact on the merit order dispatch, GHG emissions are different from those shown in Panel A, with a slightly higher cost of avoided carbon ($21.90/ton, line 6 of panel D). 79 The figures in panel A for system cost, GHG emissions, and avoided cost match the data given in Table 5.2. 80 We would of course expect this mitigation measure to be undertaken, since the cost of avoided carbon ($21.20/ton) is below the expected value of the carbon offset revenue (at $22.00/ton). 54 Greenhouse Gas Mitigation Options in the Sri Lanka Power Sector Table 6.1: Oil steam (+FGD) compared with least-cost solution A. Cost and avoided carbon if implemented today Units Baseline: Mitigation measure ? Least-cost ++Oil steam + FGD 1. System cost US$ million 2,392.6 2,402.3 9.7 Undiscounted 2. GHG [CO2] Million tons 87.8 79.6 -8.1 3. Cost of avoided CO2 $/ton CO2 1.2 4. Cost of avoided carbon $/ton C 4.4 Discounted 5. GHG [CO2] Million tons 25.4 23.8 -1.7 6. Cost of avoided CO2 $/ton CO2 5.8 7. Cost of avoided carbon $/ton C 21.2 B. Expected value of revenue from carbon offsets (US$ million) Pr{ } R(c) E{R(c)} (US$ million) ($/ton C) 1. 0.025 2. 10.00 0.150 4.38 0.7 3. 20.00 0.500 9.15 4.6 4. 30.00 0.250 13.73 3.4 5. 40.00 0.075 18.31 1.4 6. Discounted value 10.1 C: Expected value of implementing measure today (US$ million) 1. Cost with mitigation measure ++Oil 2,402.3 (from A, line 1) steam+FGD 2. Less expected value of carbon -10.1 offsets (from B, line 6) 3. Expected cost of mitigation measure = line 2,392.3 1 + line 2 4. Cost without mitigation measure Least-cost 2,392.6 (from A, line 1) 5. Hence cost reduction 0.4 6. Benefit: proceed Resolving Uncertainty 55 D. Cost and avoided carbon if measure: ++oil steam+FGD is implemented at t+5 Units Baseline: Mitigation measure ? Least- (9) Oil cost+5 steam+FGD+[t+5] +Interim measure Diesel Diesel 1. System cost US$ million 2,434 2,441 7.1 Undiscounted 2. GHG emissions (CO2) million tons 84.7 78.5 -6.3 3. Cost of avoided CO2 $/ton CO2 1.1 Cost of avoided C $/ton C 4.1 Discounted 4. GHG emissions [CO2] 24.6 23.5 -1.2 5. Cost of avoided C $/ton CO2 6.0 6. Cost of avoided C $/ton C 21.9 E. Impact of delay Baseline Mitigation measure Least- (9) Oil cost+5 steam+FGD+[t+5] 1. System cost US$ million 41.6 39.0 2. GHG (CO2) undiscounted Million tons -3.1 -1.2 In panel E, we show the impacts of delay. The costs increase in both cases by a roughly similar amount.81 We know that the least-cost case builds a first coal plant in 2006: It necessarily follows that if one were to delay the coal plant and build something else to meet the demand in the interim, costs will increase by $41.6 million. GHG emissions, however, decrease by 3.1 million tons, because emissions from a diesel plant are lower than from the corresponding generation from coal. Similarly, because diesels have slightly greater efficiency than steam-cycle plants, replacing the first oil plant by diesels also reduces GHG emissions relative to no delay, and so GHG emissions decrease by 1.2 million tons per year. 81 Even though 300 MW of diesels are built in each case, the increase in cost is not identical because of the varying impact on subsequent merit order dispatch (and fuel costs). The calculations of system cost and GHG emissions are again done in ENVIROPLAN; the interim measure--in this case, diesels--must necessarily be forced in (since diesels are not part of the least-cost solution). 56 Greenhouse Gas Mitigation Options in the Sri Lanka Power Sector F. Decision if uncertainty resolved: calculation of option value Price Pr{ } PV[E{R(c)}] (9) Oil Least- ? cost Decision Benefit E{benefit} $/ton steam+FG cost+5 D+[t+5] 1. 0 0.025 0 2,441 2,434 7 Do not build 0.0 0.0 2. 10.00 0.150 3.2 2,438 2,434 4 Do not build 0.0 0.0 3. 20.00 0.500 6.5 2,435 2,434 1 Do not build 0.0 0.0 4. 30.00 0.250 9.7 2,432 2,434 -3 Build 17.9 0.65 5. 40.00 0.075 12.9 2,428 2,434 -6 Build 26.2 0.44 6. Total, expected value of carbon offsets 1.09 7. Benefit if implemented today (from C, line 5) 0.37 8. Hence option value of delay 0.72 In panel F, we now recalculate the expected values, given resolution of the uncertainty at t+5. If the actual value of the offset is less than the cost of avoided carbon ($21.90/ton), then the mitigation measure is not undertaken; but it is undertaken if the actual value is greater than $21.90/ton, which in the case of the hypothetical discrete probability distribution assumed here is the case for actual offset values of $30/ton and $40/ton. The expected value of benefits is calculated at $1.09 million, which is greater than the $0.37 million calculated if the decision is made today. In other words, the option value is $1.09 million - $0.37 million = $0.72 million. While the numbers are small, nevertheless we note that the option value of delay is almost twice the expected value of the benefit of making the commitment to the mitigation measure today, and it represents about 7 percent of the expected value of carbon offset revenue. However, perhaps more important is the illustration of the importance of proper discounting. The GEF suggests that carbon savings should be expressed in undiscounted terms. Since undiscounted carbon emissions over some given planning horizon will always be greater than the corresponding discounted sum, it necessarily follows that the apparent cost of avoided carbon will be lower than if discounted. However, the economic flows that result from carbon offset payments must be discounted for valid analysis, which effectively increases the cost of avoided carbon (from $4.40/ton to $21.20/ton in this example).82 The option value will vary with the extent of uncertainty: the higher the variance of the probability distribution for the future value of the carbon offset, the greater the option value. This is illustrated in Figure 6.1, which shows the option value, again for oil steam+FGD, for five different probability distributions, each with expected value of $22/ton but with different variances. The probability distribution used in Table 6.1 corresponds to probability distribution number 3 in this figure (with standard deviation of $7.8 million). We note than in the case of high uncertainty (probability distribution number 1), the option value increases to $2.85 million, 82 At the same time, one must be sure that end-effects are properly accounted for, since the life of some mitigation measures will extend beyond the planning horizon. ENVIROPLAN accounts for these end- effects by calculating a salvage value of assets whose economic life extends beyond the planning horizon, and whose value is subtracted from the system cost. Resolving Uncertainty 57 some seven times greater than the expected value of a decision to proceed today. Figure 6.1: Option value as a function of variance of the carbon offset payment 3 1 2 $million value, 2 option 1 3 4 5 0 0 5 10 15 20 25 standard deviation, $million Table 6.2 shows the results for LNG (measured against the reform option). This shows that a commitment to LNG today has an expected value of -$69.5 million: In other words, the costs of the carbon offsets are insufficient to compensate for the higher costs. This follows logically, given a cost of avoided carbon of $79.20/ton compared with an expected value of the carbon offset value of $22.00/ton. Table 6.2: LNG compared with reform A. Cost and avoided carbon if implemented today Units Baseline Mitigation ? measure Reform LNG 1. System cost US$ million 2,448.8 2,545.1 96.3 Undiscounted 2. GHG [CO2] Million tons 98.4 77.3 -21.1 3. Cost of avoided CO2 $/ton CO2 4.6 4. Cost of avoided carbon $/ton C 16.7 Discounted 5. GHG (CO2) Million tons 28.2 23.8 -4.5 6. Cost of avoided CO2 $/ton CO2 21.6 7. Cost of avoided carbon $/ton C 79.2 58 Greenhouse Gas Mitigation Options in the Sri Lanka Power Sector Waiting for five years does not change the attractiveness of LNG, and there is no option value. Obviously, even if the uncertainty is resolved after five years, if the avoided cost of carbon is $66/ton (panel D, line 6), then even if the actual value of the carbon offset is $40/ton, the LNG option would still not be implemented. B. Expected value of revenue from carbon offsets US$ million Price $/ton C Pr{ } R(c) E{R(c)} (US$ million) ($/ton C) 1. 0 0.025 0.00 0.0 2. 10.00 0.150 12.15 1.8 3. 20.00 0.500 24.30 12.1 4. 30.00 0.250 36.45 9.1 5. 40.00 0.075 48.60 3.6 6. Discounted value 26.7 C: Expected value of implementing measure today (US$ million) 1. Cost with mitigation measure (from A, line 1) LNG 2,545.1 2. Less expected value of carbon offsets (from -26.7 B, line 8) 3. Expected cost of mitigation measure = line 1 2,518.4 + line 2 4. Cost without mitigation measure (from A, line Reform 2,448.8 1) 5. Hence cost reduction -69.5 6. Cost: do not proceed D. Cost and avoided carbon if measure LNG is implemented at t+5 Units Baseline: Mitigation measure ? Reform+5 LNG+5 + interim measure Diesel Diesel 1. System cost US$ million 2,515 2,566 50.7 Undiscounted 2. GHG emissions (CO2) Million tons 94.3 79.5 -14.8 3. Cost of avoided CO2 $/ton CO2 3.4 Cost of avoided C $/ton C 12.6 Discounted 4. GHG emissions (CO2) 27.1 24.3 -2.8 5. Cost of avoided C $/ton CO2 18.2 6. Cost of avoided C $/ton C 66.6 Resolving Uncertainty 59 E. Impact of delay Units Baseline: Mitigation measure Reform+5 LNG+5 1. System cost US$ million 66.2 20.6 2. GHG (CO2) undiscounted Million tons -4.1 2.3 F. Decision if uncertainty resolved: calculation of option value Price $/ton Pr{ } PV[E{R(c)}] LNG+5 Reform+5 ? cost Decision Benefit E{benefit} 1. 0 0.025 2,566 2,515 51 Do not build 0.0 0.0 2. 10.00 0.150 7.6 2,558 2,515 43 Do not build 0.0 0.0 3. 20.00 0.500 15.2 2,551 2,515 35 Do not build 0.0 0.0 4. 30.00 0.250 22.8 2,543 2,515 28 Do not build 0.0 0.0 5. 40.00 0.075 30.4 2,535 2,515 20 Do not build 0.0 0.0 6. Total, expected value of carbon offsets 1.8 7. Benefit if implemented today (from C, line 5) Cost -53.5 8. Hence option value of delay Table 6.3 shows the option value calculations for the dendro-thermal option. The expected value of a commitment today is a benefit of $31.4 million, which again follows immediately from a comparison of the cost of avoided carbon ($10.90/ton) with the expected value of the carbon offset. However, the impact of delay is quite different from the two cases examined above. As before, forcing diesels into the least-cost solution for the interim period reduces GHG emissions. But in the case of the dendro-thermal option, diesels will significantly increase GHG emissions--by 9.1 million tons (panel E, line 2). This results in an aggregate cost of avoided carbon that jumps to $42.6/ton (panel D, line 6). Hence, given the maximum value of $40/ton in the discrete probability distribution assumed for carbon offset revenue, dendro-thermal would not be implemented at t+5. Clearly, there is no option value in this case. However, this counterintuitive result is readily explained: Since the diesels have an economic life that extends considerably beyond the assumed five-year interim period, new dendro-thermal plants at t+5 must now compete only against the variable cost of diesel plant operation, since the capital costs are sunk and no longer relevant at t+5. 60 Greenhouse Gas Mitigation Options in the Sri Lanka Power Sector Table 6.3: Dendro-thermal relative to least-cost solution A. Cost and avoided carbon if implemented today Units Baseline: Mitigation measure ? Reform ++Dendro System cost US$ million 2,448.8 2,482.4 33.6 Undiscounted GHG (CO2) Million tons 98.4 44.9 -53.5 Cost of avoided CO2 $/ton CO2 0.6 Cost of avoided C $/ton C 2.3 Discounted GHG (CO2) Million tons 28.2 16.9 -11.3 Cost of avoided CO2 $/ton CO2 3.0 Cost of avoided C $/ton C 10.9 B. Expected value of revenue from carbon offsets (US$ million) Price $/ton C Pr{ } R(c) E{R(c)} US$ million $/ton C 0 0.025 0.00 0.0 10.00 0.150 29.53 4.4 20.00 0.500 59.07 29.5 30.00 0.250 88.60 22.2 40.00 0.075 118.14 8.9 Discounted value 65.0 C: Expected value of implementing measure today (US$ million) Cost with mitigation measure ++Dendro 2,482.4 (from A, line 1) Less expected value of carbon -65.0 offsets (from B, line 8) Expected cost of mitigation measure = line 1 2,417.4 + line 2 Cost without mitigation measure Reform 2,448.8 (from A, line 1) Hence cost reduction 31.4 Benefit:proceed Resolving Uncertainty 61 D. Cost and avoided carbon if measure ++dendro is implemented at t+5 Baseline: Mitigation measure ? Reform+5 Dendro+5 +interim measure Diesel Diesel System cost US$ million 2,515 2,602 87.0 Undiscounted GHG emissions (CO2) Million tons 94.3 54.0 -40.4 Cost of avoided CO2 $/ton CO2 2.2 Cost of avoided C $/ton C 7.9 Discounted GHG emissions (CO2) 27.1 19.6 -7.5 Cost of avoided carbon $/ton CO2 11.6 Cost of avoided C $/ton C 42.6 E. Impact of delay Units Baseline Mitigation measure Reform+5 Dendro+5 System cost US$ million 66.2 119.7 GHG (CO2) undiscounted Million tons -4.1 9.1 F. Decision if uncertainty resolved: calculation of option value $/ton Pr{ } PV[E{R(c)}] Dendro+5 Reform+5 ? cost Decision Benefit E{benefit} 1. 0 0.025 0.0 2,602 2,515 87 Do not build 0.0 0.0 2. 10.00 0.150 20.4 2,582 2,515 67 Do not build 0.0 0.0 3. 20.00 0.500 40.8 2,561 2,515 46 Do not build 0.0 0.0 4. 30.00 0.250 61.2 2,541 2,515 26 Do not build 0.0 0.0 5. 40.00 0.075 81.6 2,520 2,515 5 Do not build 0.0 0.0 6. Total, expected value of carbon offsets 0.00 7. Benefit if implemented today (from C, line 5) 31.37 8. Hence option value of delay No option value However, one of the advantages of dendro-thermal is that it can be implemented in small capacity increments. Unlike a commitment to a coal plant, which would be for at least 600-900 MW to capture the necessary economies of scale, dendro-thermal can be built in increments of as little as 1-5 MW. Therefore, instead of the intermediate five-year commitment to diesels before choosing coal or dendro-thermal, we make a five-year commitment to a dendro-thermal demonstration program, and then decide whether to go to coal or continue with a full-scale dendro-thermal program. The corresponding option value calculation is presented in Table 6.4. The expected value for a present permanent commitment to dendro-thermal is unchanged at $31.4 million. However, with a dendro-thermal demonstration as the interim measure, the impact of delay is to 62 Greenhouse Gas Mitigation Options in the Sri Lanka Power Sector create a win-win situation, for the mitigation measure has costs of $2,482 million, that are below that of the least-cost solution (again with the interim dendro-thermal demonstration) of $2,515 million, though of course still higher than the least-cost solution alone (of $2,448 million). However, the calculated option value of $59.4 million indicates the significant value of being able to abandon the dendro-thermal option at t+5, regardless of the value of the carbon offset. Table 6.4: Dendro-thermal option relative to least-cost solution, with dendro- thermal demonstration as the interim option A. Cost and avoided carbon if implemented today Units Baseline Mitigation ? measure Reform ++Dendro 1. System cost US$ million 2,448.8 2,482.4 33.6 Undiscounted 2. GHG (CO2) Million tons 98.4 44.9 -53.5 3. Cost of avoided CO2 $/ton CO2 0.6 4. Cost of avoided carbon $/ton C 2.3 Discounted 5. GHG (CO2) Million tons 28.2 16.9 -11.3 6. Cost of avoided CO2 $/ton CO2 3.0 7. Cost of avoided carbon $/ton C 10.9 B. Expected value of revenue from carbon offsets (US$ million) $/ton C Pr{ } R(c) E{R(c)} (US$ million) ($/ ton C) [1] Zero 0.025 0.00 0.0 [2] 10.00 0.150 29.53 4.4 [3] 20.00 0.500 59.07 29.5 [4] 30.00 0.250 88.60 22.2 [5] 40.00 0.075 118.14 8.9 [6] discounted value 65.0 C. Expected value of implementing measure today (US$ million) 1. Cost with mitigation measure ++Dendro 2,482.4 (from A, line 1) 2. Less expected value of carbon offsets -65.0 (from B, line 8) 3. Expected cost of mitigation measure = line 1 2,417.4 + line 2 4. Cost without mitigation measure Reform 2,448.8 (from A, line 1) 5. Hence cost reduction 31.4 6. Benefit:proceed Resolving Uncertainty 63 D. Cost and avoided carbon if measure ++dendro is implemented at t+5 Baseline: Mitigation measure ? Reform+5/dendro ++Dendro +interim measure Dendro demo Dendro demo 1. System cost US$ million 2,515 2,482 -32.6 Undiscounted 2. GHG emissions (CO2) Million tons 94.3 44.9 -49.5 3. Cost of avoided CO2 $/ton CO2 -0.7 Cost of avoided C $/ton C -2.4 Discounted 4. GHG emissions (CO2) 27.1 16.9 -10.2 5. Cost of avoided carbon $/ton CO2 -3.2 6. Cost of avoided C $/ton C -11.7 E. Impact of delay Units Baseline Mitigation measure Reform+5/dendro ++Dendro 1. System cost US$ million 66.2 0.0 2. GHG (CO2) undiscounted Million tons -4.1 0.0 F. Decision if uncertainty resolved: calculation of option value $/ton Pr{ } PV[E{R(c)}] ++Dendro Reform ? cost Decision Benefit E{benefit} +5/dendr o 1. 0 0.025 0.0 2,482 2,15 -33 Build 32.6 0.82 2. 10.00 0.150 26.4 2,456 2,515 -59 Build 59.1 8.86 3. 20.00 0.500 52.9 2,430 2,515 -86 Build 85.5 42.76 4. 30.00 0.250 79.3 2,403 2,515 -112 Build 112 27.99 5. 40.00 0.075 105.8 2,377 2,515 -138 Build 138.4 10.38 6. Total, expected value of carbon offsets 90.82 7. Benefit if implemented today (from C, line 5) 31.37 8. Hence option value of delay 59.44 7 Summary and Conclusions 7.1 Reform Undoubtedly, the first priority for GHG emission reduction in Sri Lanka is power sector reform, which is assumed to permit a reduction in T&D losses, and the implementation of a comprehensive DSM lighting program. Unlike reform in India, reform is win-win with respect to both net economic benefits and GHG emission reduction. 7.2 Residual oil Once reform has been implemented, use of residual oil for diesel power generation represents the GHG mitigation option that would be easiest to implement from an institutional and managerial standpoint--and, indeed, it will likely bring economic benefits as well. However, the availability of this option (and its economic benefits) depends upon an expansion of Sri Lanka's domestic refinery capacity, the feasibility of which is still under study. 7.3 Implementing the least-cost option The least-cost option is determined to include additional DSM and T&D loss reduction (beyond levels as might be envisaged in a first step of reform), plus an aggressive implementation of some 45 MW of mini hydro schemes. However, although these are win-win in terms of economic and GHG emission reduction benefits, the financial and institutional barriers are significant. For this reason, efforts to implement mini hydro schemes should continue to be supported by GEF and the ESDP. 7.4 Dendro-thermal power Dendro-thermal power plants offer substantial potential for GHG emissions reductions at a reasonable incremental cost of avoided carbon. However, many of the assumptions can only be verified by field demonstration on a sufficiently representative scale to be able to answer critical questions about plantation management, land and labor availability, and plant gate cost. It is therefore recommended that a demonstration plantation be established at a scale to 65 66 Greenhouse Gas Mitigation Options in the Sri Lanka Power Sector provide fuelwood for at least a 1 MW unit (which requires about 400 ha). In its initial stages, the output of such a project could be marketed as fuelwood in Colombo. If a decision is made to proceed with the coal-fired unit on the west coast, we recommend that a study be made of the incremental costs of providing the second 300 MW with co-firing of fuelwood. This involves tested technology and is already practiced in many places worldwide. The timing of any such commitment to co-firing is such that it could be based on a successful demonstration of the plantation concept. Undoubtedly, the more important constraint to the dendro-thermal option is the capital cost of the gasifier. The experience of the GEF demonstration project in Brazil suggests that gasifier costs may be substantially greater than $1,200/kW, and reach costs in the range of $1,200-$1,500 only after 5-10 years of large-scale implementation. If this were indeed the case, then the best strategy for Sri Lanka would be to await the benefits of this experience in larger countries, and use co-firing at the coal plant in the interim. In any event, even if co-firing were to prove too expensive (or otherwise undesirable from the point of view of any IPP operator of the coal plant), the wood from the demonstration plantation could still be sold in the Colombo fuelwood market. Therefore, the risks of such a demonstration program would appear manageable. 7.5 Wind Sri Lanka has substantial wind power potential, but the costs are high. Even if capital costs were to fall to 1,000$/kW (from the present $1,175/kW), this would still be an expensive mitigation option, given the low plant factors achievable. For the maximum wind scenario examined in this study, the cost of (undiscounted) avoided carbon is 65 $/ton. In any event, the results of the ESDP-supported 3 MW demonstration program should be awaited and evaluated as a first step. 7.6 Solar power There may be many reasons for implementing the solar homes program, given the importance that may be attached to rural lighting (and the expense and difficulties of extending grid service to remote villages). But as a GHG mitigation measure, it would have low priority. Even under the highly optimistic assumptions about market penetration, the quantity of GHG emissions reduced is less than 1 percent of the reform scenario emissions. 7.7 Imported fuel oil High sulphur fuel oil can be imported at relatively low cost from Mideast gulf refineries, and over the past decade, there have been many periods when this would have been a cheaper fuel than coal. However, given the present volatility of oil prices (which shows no sign of moderation) and the experience of the 1970s oil shocks (whose effects lasted until January 1986), the government of Sri Lanka might be quite cautious about any commitment to a large steam-cycle oil plant. Indeed, while the historical dependence on hydro insulated the electric Summary and Conclusions 67 sector from the 1970s oil price shocks, the risks of over-dependence on hydro were amply demonstrated during the 1996 drought, which caused severe dislocations to the economy. Thermal generation reduces the potential impact of this hydro risk, but one may hardly be eager to replace it with a new dependence on oil. However, the hedging strategy that suggests itself is again dual-firing capability at the coal plant. Assessment of the incremental capital costs of boiler modification may be straightforward. The question of the amount of oil storage to be provided is more complicated. The larger the volume of storage, the greater the ability to take opportunistic advantage of periods of low oil prices, and the less is the need for futures-market hedging. However, the larger the volume of storage, the greater the upfront capital cost. 7.8 Transportation and liquid fuels This study has been limited to mitigation options in the power sector. However, even if Sri Lanka moves to coal for meeting its baseload power generation requirements, liquid and transportation fuels will continue to account for a substantial proportion of GHG emissions. There may well be options involving the transportation, industrial, and domestic sectors that have lower costs of avoided carbon than the power sector, and these should therefore be placed into the abatement cost curves before Sri Lanka makes any commitments to projects under the Joint Implementation (JI) or Clean Development Mechanism (CDM) mechanisms of the Kyoto Protocol. We recommend that a follow-up study examine such potential projects and integrate the results into those presented in this volume.83 7.9 Dealing with uncertainty In this study, a preliminary framework was explored to deal with high levels of uncertainty and irreversibility, as well as long time frames that characterize GHG mitigation projects. In brief, such characteristics may sometimes make it optimal for decisionmakers to defer making GHG mitigation investments or otherwise preserve their flexibility to change course midstream until some of the uncertainty is resolved. By doing so, they avoid getting locked into investments that may turn out to be suboptimal ex post. Some pilot calculations carried out in this volume show that using an options-based framework facilitates the analysis of GHG mitigation projects. Such an approach complements the conventional risk-adjusted cost-benefit analysis used to assess these projects, in order to more accurately value the costs and benefits associated with these projects and more precisely determine when these projects should be undertaken. Timing is especially critical, since it is not possible to rule out, prima facie, the prospect of either high costs or high benefits accruing from 83 There are many candidates for transportation sector projects with significant GHG emission implications, including railway electrification, the proposed dry container port, the proposed Colombo- Katunayake Expressway (as well as other new trunk highways that would relieve increasing congestion on the present road system), conversion of automobiles to LPG (which already accounts for the significant recent growth in LPG consumption), and the rationalization of tax and import duty policy on transport fuels (whose present anomalies cause significant distortions in transportation sector fuel consumption). 68 Greenhouse Gas Mitigation Options in the Sri Lanka Power Sector the delay of these projects. We argue that measuring the option value associated with delay often provides a useful and straightforward decision rule to determine the timing of environmental projects. The options framework suggests that GHG mitigation projects should be characterized along dimensions such as uncertainty, irreversibility, time frame, and downside risk to ascertain what is likely to be the best approach to be undertaken for various classes of projects. This would affect the ranking of GHG mitigation projects at the global or country level. It may be optimal in the face of uncertainty to defer long-duration, high sunk-cost GHG mitigation projects, even when these projects rank ahead of short-duration, low sunk-cost projects based on conventional cost-benefit analysis. Further work along these lines would be very helpful in helping to make decisions with respect to the valuation and timing of international transfers under the provisions of the Kyoto Protocol. Taking account of option value is especially important for developing countries that are considering giving up their emission rights to developed countries in exchange for investments by the latter in GHG mitigation projects. These emission rights could sometimes be significantly more valuable to the developed countries than their locally avoided costs, leaving aside the costs of actually implementing the projects in the developing countries. An options value-based framework will help to estimate the value of the premium that should be attached to the "avoided-cost" price of these emission rights and determine the actual timing of these transfers. It is possible that in some cases, developing countries may be better off cashing in their emission rights now, whereas in other cases, deferring this decision to a future time may be the optimal strategy. A more detailed follow-up study is recommended to further develop this approach and apply it to a typical country like Sri Lanka. References Administrative Staff College of India, 1999 India: Global Overlay for Andhra Pradesh, Report to the World Bank, Andhra Pradesh, India. Baldwin, M. (editor), 1991 Natural Resources of Sri Lanka: Conditions and Trends, Natural Resources, Energy and Science Authority of Sri Lanka, Colombo, Sri Lanka. BCHydro, 1995 Electricity Plan, Vancouver, B.C., September 1995. Volume C of this plan presents the economic and trade-off analysis. CEB, 1998 Load Forecast for 1998 Generation Expansion Planning Studies, Colombo, Sri Lanka. CEB, 1999 Report on Long Term Generation Expansion Planning Studies 1999-2013, Generation Planning Branch, Colombo, Sri Lanka. CEB, 1998 Generation Planning Report, Chapter 7, Colombo, Sri Lanka. Fernando, C. and M. Munasinghe, 1998 A Real Options Framework to Assess Environmental Policy, Programs, and Strategy for GHG Mitigation, World Bank,Wash. DC. Fernando, Sunith 1998 An Assessment of the Small Hydro Potential in Sri Lanka, Intermediate Technology Development Group, Colombo, Sri Lanka. Gunaratne, 1991 Making the Investment: Innovative Ideas in Alternative Energy Policy Financing and Services: Annex Market Study Commissioned by National Development Bank, Paper presented at the FINESSE Workshop, Kuala Lumpur, Malaysia. International Development and Energy Associates, Ltd., 1999 Sri Lanka Electric Power Technology Assessment (SLEPTA), Draft Report to the World Bank. Jayaweera, D.S. 1998 Energy and Environment-Transportation Sector, Transport Policy Centre, Colombo, Sri Lanka. Jayewardene P. and V. Perera, 1991 Sri Lanka: The Rural Electrification Problem, Report by Power and Sun (Pvt) Ltd., Colombo, Sri Lanka. Jepma, C.J. and M. Munasinghe, Climate Change Policy, Cambridge University Press, London, UK, 1998. Matsui, K. 1998 "Global Demand Growth of Power Generation, Input Choices and Supply Security," Energy Journal 19(2): 93. 69 70 Greenhouse Gas Mitigation Options in the Sri Lanka Power Sector Meier, P. and M. Munasinghe, 1994 Incorporating Environmental Concerns into Power Sector Decision-making: A Case Study of Sri Lanka, World Bank Environment Department Paper 6, Washington, D.C. Meier, P. 1999 Economic Analysis of the Andhra Pradesh Power Sector Restructuring Program, World Bank, New Delhi, India. Meier, P. Economic Analysis of the Haryana Power Sector Restructuring and Reform Program, World Bank, New Delhi, 1997. Meier, P. 1999 India: Environmental Issues in the Power Sector: A Case Study of Haryana, World Bank, Wash. DC. Munasinghe, M., 1993 Environmental Economics and Sustainable Development, World Bank, Wash. D.C. Munasinghe, M., 2001 "Sustainable Development and Climate Change: Applying the Sustainomics Trans-disciplinary Meta-framework", Int. Journal of Global Environmental Issues, 1(1): 13-55. Ministry of Forest and Environment, J. Ratnasiri, editor, 1998 Final Report of the Sri Lanka Climate Change Country Study, Environment Division, Minsitry of Environment, Colombo, Sri Lanka. Posch and Partners, 1994 Sri Lanka Micro Hydro Feasibility Study, Report to the Asia Alternative Energy Unit, World Bank, Wash. D.C. Shresta, Ram, Fernando, W. J. L. S. and Rabin Shrestha, 1998 "Environmental Emission Mitigation Potential of Efficient Electrical Appliances in Sri Lanka" International Journal of Energy Research, 22: 923-933. Siyambalapitiya, T. 1997 Electricity Pricing Policy in Sri Lanka, Institute of Policy Studies, Energy and Environmental Economics Series 5. Sone Command Area Development Authority (SCADA), 1999 India, Global Overlay for Bihar, Report to the World Bank, Wash. D.C. USAID, 1995 Integrated Resource Plan for Andhra Pradesh, Report to APSEB, Hyderabad. Walpita, N. 1997 Biomass Development and Environmental Consequences, Govt. of Sri Lanka. World Bank Project Appraisal Document, 1997 Energy Services Delivery Project, Report 16063-CE, Energy and Project Finance Division, Country Department 1, South Asia Region, World Bank, Wash. D.C. World Bank, 1997 Economic Analysis of the Haryana Power Sector Reform and Restructuring Program, Washington, D.C. World Bank, 1997 Guidelines for Climate Change Global Overlays, Global Environment Division, Environment Department Paper 47, Climate Change Series, Washington, D.C. Summary and Conclusions 71