69203 v1 The World Bank Least Cost Electricity Master Plan, Djibouti Final Report Volume 1 of 2 Main Report Submitted by November 2009 The World Bank Least Cost Electricity Master Plan, Djibouti Final Report Volume 1 of 2 Main Report Submitted by November 2009 THE WORLD BANK LEAST COST ELECTRICITY MASTER PLAN, DJIBOUTI FINAL REPORT VOLUME 1 of 2 MAIN REPORT November 2009 Prepared by Prepared for Parsons Brinckerhoff Ltd The World Bank (UK) Report Title : Least Cost Electricity Master Plan, Djibouti (Volume 1 of 2: Main Report) Report Status : Final Report Job No : 63579 A Date : November 2009 Prepared by : Mo Deif, Mark Fraser, Richard Gadsden...... Checked by : Richard Butlin .................................................. Approved by : Bruce Stedall .................................................... CONTENTS Page VOLUME 1: MAIN REPORT 1 INTRODUCTION ........................................................................................................... 1.1 1.1 CONTEXT OF THE ASSIGNMENT.................................................................................. 1.1 1.2 OBJECTIVES ............................................................................................................. 1.2 1.3 TECHNICAL MEMORANDUM ....................................................................................... 1.3 1.4 STRUCTURE OF THE REPORT ..................................................................................... 1.3 2 COUNTRY BACKGROUND ......................................................................................... 2.1 2.1 COUNTRY INFORMATION ........................................................................................... 2.1 2.2 THE ELECTRICITY SYSTEM IN DJIBOUTI ...................................................................... 2.3 3 DEMAND FORECAST .................................................................................................. 3.1 3.1 THE GENERAL PRINCIPLES OF DEMAND FORECASTING ................................................ 3.1 3.2 METHODOLOGY ........................................................................................................ 3.2 3.3 INPUT DATA .............................................................................................................. 3.6 3.4 SALES FORECAST ................................................................................................... 3.21 3.5 LOSSES .................................................................................................................. 3.23 3.6 CADLF’S ............................................................................................................... 3.27 3.7 DEMAND FORECAST ................................................................................................ 3.28 3.8 DEMAND FORECAST SCENARIOS ............................................................................. 3.30 3.9 DEMAND SIDE MANAGEMENT .................................................................................. 3.35 3.10 SUMMARY .............................................................................................................. 3.37 4 EXISTING & COMMITTED POWER SYSTEM ............................................................. 4.1 4.1 INTRODUCTION ......................................................................................................... 4.1 4.2 EXISTING AND COMMITTED PLANT .............................................................................. 4.1 4.3 INTERCONNECTOR .................................................................................................... 4.4 4.4 SUPPLY DEMAND BALANCE ....................................................................................... 4.8 4.5 TRANSMISSION ......................................................................................................... 4.9 4.6 DISTRIBUTION ......................................................................................................... 4.12 4.7 SUMMARY .............................................................................................................. 4.16 Least Cost Electricity Master Plan (Volume 1) Final Report November 2009 5 FUTURE DEVELOPMENT OPTIONS .......................................................................... 5.1 5.1 RENEWABLE RESOURCES.......................................................................................... 5.2 5.2 THERMAL RESOURCES .............................................................................................. 5.6 5.3 SCREENING CURVE ANALYSIS .................................................................................. 5.11 5.4 SUMMARY .............................................................................................................. 5.15 6 GENERATION PLANNING ........................................................................................... 6.1 6.1 INTRODUCTION ......................................................................................................... 6.1 6.2 GENERAL METHODOLOGY ......................................................................................... 6.1 6.3 GENERATION PLANNING ASSUMPTIONS ...................................................................... 6.7 6.4 GENERATION PLANNING SCENARIOS ........................................................................ 6.14 6.5 LEAST COST GENERATION EXPANSION PLAN (REFERENCE CASE) .............................. 6.15 6.6 SENSITIVITIES TO THE LEAST COST GENERATION EXPANSION PLAN ........................... 6.24 6.7 ADDITIONAL SENSITIVITIES ...................................................................................... 6.41 6.8 ENVIRONMENTAL CONSIDERATIONS ......................................................................... 6.42 6.9 SUMMARY .............................................................................................................. 6.43 7 TRANSMISSION PLANNING ....................................................................................... 7.1 7.1 GENERAL ................................................................................................................. 7.1 7.2 TRANSMISSION PLANNING CRITERIA ........................................................................... 7.1 7.3 DISTRIBUTION OF DEMAND ........................................................................................ 7.6 7.4 GENERATION DISPATCH ............................................................................................ 7.7 7.5 NETWORK MODEL ..................................................................................................... 7.8 7.6 NETWORK EXPANSION .............................................................................................. 7.9 7.7 SHORT-CIRCUIT STUDIES ........................................................................................ 7.13 7.8 TRANSIENT STABILITY STUDIES ................................................................................ 7.14 7.9 TRANSMISSION LOSSES........................................................................................... 7.16 7.10 TRANSMISSION EXPANSION COSTS .......................................................................... 7.16 7.11 SUMMARY .............................................................................................................. 7.18 8 DISTRIBUTION PLANNING ......................................................................................... 8.1 8.1 GENERAL ................................................................................................................. 8.1 8.2 PERFORMANCE OF EXISTING DISTRIBUTION NETWORK ................................................ 8.1 8.3 LOSSES .................................................................................................................... 8.2 8.4 DESIGN CONSIDERATIONS ......................................................................................... 8.4 Least Cost Electricity Master Plan (Volume 1) Final Report November 2009 8.5 NETWORK CONFIGURATION ....................................................................................... 8.4 8.6 OPTIMUM CONDUCTOR SIZE ...................................................................................... 8.4 8.7 AVERAGE INCREMENTAL COST OF DISTRIBUTION ........................................................ 8.5 8.8 ANNUAL LOAD RELATED AND NON-LOAD RELATED EXPENDITURE................................. 8.6 8.9 SUMMARY ................................................................................................................ 8.7 9 ISOLATED SYSTEMS .................................................................................................. 9.1 9.1 TADJOURA ................................................................................................................ 9.2 9.2 OBOCK ..................................................................................................................... 9.8 9.3 DISTRIBUTION ......................................................................................................... 9.14 9.4 SUMMARY .............................................................................................................. 9.15 10 CONCLUSIONS ....................................................................................................... 10.1 10.1 FUTURE DEVELOPMENT OPTIONS ............................................................................ 10.1 10.2 GENERATION PLAN ................................................................................................. 10.2 10.3 TRANSMISSION PLAN .............................................................................................. 10.3 10.4 DISTRIBUTION PLAN ................................................................................................ 10.4 10.5 ISOLATED SYSTEMS ................................................................................................ 10.5 10.6 INVESTMENT PLAN .................................................................................................. 10.5 Least Cost Electricity Master Plan (Volume 1) Final Report November 2009 VOLUME 2: APPENDICIES APPENDIX A: STUDY TERMS OF REFERENCE APPENIDIX B SALES FORECAST APPENDIX C: REVIEW OF GEOTHERMAL RESOURCES IN DJIBOUTI APPENDIX D: REVIEW OF WIND RESOURCES IN DJIBOUTI APPENDIX E: DESCRIPTION OF ASPLAN LEAST COST GENERATION PLANNING SOFTWARE APPENDIX F: GENERATION PLANNING RESULTS APPENDIX G: LOAD FLOW PLOTS APPENDIX H: TRANSIENT STABILITY RESULTS APPENDIX I: DISTRIBUTION NETWORK Least Cost Electricity Master Plan (Volume 1) Final Report November 2009 LIST OF ABBREVIATIONS AEO: Annual Economic Outlook AfDB: African Development Bank CADLF: Coincident After Diversity Load Factor, B.16 CAPEX: Capital Expenditure, C.32 COUE: Cost of Unserved Energy, F.3 DJF: Djibouti Franc DPA: Dubai Port Authority DSM: Demand Side Management EIA: Energy Information Administration EdD: Electricite de Djibouti, B.3 EEPCo: Ethiopian Electric Power Compnay ELDC: Equivalent Load Duration Curve, E.3 ENS: Energy Not Served, E.4 EUR: Euro, D.7 g/l: grams per litre, C.7 GDP: Gross Domestic Product, B.4 GoD: Government of Djibouti GoE: Government of Ethiopia GWh: Gigawatt hour HFO: Heavy Fuel Oil Hz: Hertz, C.21 kg/s: kg per second, C.13 km: kilometre, C.33 kV: kilovolt, C.33 kWh: kilowatt hour, C.26 LOLE: Loss of Load Expectation, F.3 LOLP: Loss of Load Probability, E.4 LRMC: Long Run Marginal Cost, F.3 LV: Low Voltage, B.6 mm: Millimetre MoU: Memorandum of Understanding, C.4 MV: Medium Voltage, B.12 MW: Megawatt, C.21 MWh: Megawatt hour MWso: Megawatt sent out NPV: Net Present Value Least Cost Electricity Master Plan (Volume 1) Final Report November 2009 O&M: Operation & Maintenance, C.4 OCGT: Open cycle gas turbine OPEX: Operational Expenditure, D.7 PB: Parsons Brinckerhoff, C.15 PPA: Power Purchase Agreement, C.4 RALF: Regression Analysis Load Forecast, B.14 SIL: Surge Impedence Loading t/h: tonnes per hour, C.10 UE: Unserved Energy, F.3 UN: United Nations UNDP: The United Nations Development Programme, C.3 USc: US cents, C.26 USD: US Dollar, D.7 VA: Value Added, F.3 Least Cost Electricity Master Plan (Volume 1) Final Report November 2009 SECTION 1 INTRODUCTION SECTION 1 INTRODUCTION 1 INTRODUCTION 1.1 Context of the assignment Djibouti is characterized by a large urban population. About 70 per cent of the population lives in the main town of Djibouti-Ville, 11 per cent live in secondary towns and the remainder in a rural setting, including a substantial nomadic population. The country’s electrification rate is about 50 per cent. Electricité de Djibouti (EdD), the national state-owned utility, report that there are approximately 38,000 electricity connections for the Djibouti-Ville metropolitan area. There is a total reliance on imported oil products as the fuels for electricity generation and the country has no hydroelectric potential. This has implied very high costs of production and of electricity generation in particular. Due to the high cost of electricity and high connection fees, the electrification rate remains relatively low and mostly available to the privileged, while performance of critical social and commercial sectors are hampered. Within the context of electricity, Djibouti’s main (Boulaos) and secondary (Marabout) power plants, running on costly heavy fuel oil (HFO) and diesel respectively, currently offer a total installed generation capacity of 95 megawatt (MW) to the main interconnected system that serves Djibouti-Ville and surrounding areas. There are a further four small diesel-fuelled generating stations serving the southern region, consisting of Dikhil and Ali-Sabieh, which are interconnected and between them have a nominal installed capacity of 3,328 kilowatt (kW); and the northern region, serving Tadjoura (2,240 kW) and Obock (1,760 kW), amounting together to a total installed capacity of 7.3 MW. The average available capacity in 2006, when taking into account the de-rating for ambient operating conditions, age of plant, the operating rules applied, and the increased number of technical problems; dropped to about half its installed potential (55 MW). That same year, peak demand reached 53 MW. Today’s situation has worsened and available capacity is now regularly insufficient to cover demand. Given the current rate of economic growth, including various large-scale ongoing infrastructure developments, new capacity is needed urgently. The transmission system in Djibouti is currently limited to a 63 kV inter-connector between the main substations of Marabout and Boulaos and two 20 kV transmission circuits from Djibouti town to the village of Arta, some 40 km away. There is also a 20 kV circuit between Dikhil and Ali Sabieh in the south of the country. The distribution system comprises 20 kV radial circuits emanating from the main substations. Most customers are supplied at low voltage (LV) via distribution substations. Least Cost Electricity Master Plan (Volume 1) Final Report November 2009 Page 1.1 SECTION 1 INTRODUCTION Despite very high electricity tariffs (over 30 cent/kWh on average), the financial performance of the sector, and of EdD in particular, has been deteriorating over the past few years due to various inefficiencies, staffing costs, and, by far the major contributor, high fuel costs. These reasons combined are severely impacting the utility’s cash-flows and capacity to ensure proper maintenance and new investments. Over recent months, Djibouti’s economy has been faced with an increasing inflow of foreign capital, primarily from Gulf Countries investors, coming to support numerous new projects, some expected to consume large quantities of electricity. Such projects range from new developments in the very active free zone, at the fast-expanding port terminal facilities, or at the new regional livestock triage centre, to the various real estate, industrial and hotel projects. Djibouti also possesses good potential for geothermal. A lot of preliminary exploration work has been carried out, but further testing up-front is needed, which is will be expensive. Until a complete investigation of that resource has been undertaken, it is not possible to know the exact size of a plant that could be supported, nor what the final investment costs will be. However, a Memorandum between the Government of Djibouti and the Icelandic Reykjavik Energy Invest Ltd (REI) has recently been signed, which paves the way for further drilling and lays out the tentative terms for an IPP arrangement, should it be proved that exploitation is possible and profitable. Wind offers promising short-and-medium-term opportunities, as comprehensive wind data and feasibility work for one site in Arta is readily available while partial data and pre-feasibility work for about 10 other sites exist, some of these showing very good wind conditions. Ongoing investments in the electric interconnection with Ethiopia (AfDB project) should fulfil a significant proportion of the demand for energy in the mid-term (mid 2010) and renewed momentum on the development of geothermal (with the support of Argeo, IGA, Proparco and IFC among others) should do so in the longer term. The short-term remains highly uncertain. 1.2 Objectives The World Bank appointed Parsons Brinckerhoff (PB) to undertake engineering consultancy services for the preparation of an electricity sector least cost master plan for Djibouti. The objectives of the assignment are: • To define the least-cost investment program for the development of Djibouti’s electric generation, transmission and distribution system for the next 25 years, particularly taking into consideration the country’s resources and recent economic and sector developments. Particular attention and detail should be given for the short-term forecast period (first 5 years) of the plan. Least Cost Electricity Master Plan (Volume 1) Final Report November 2009 Page 1.2 SECTION 1 INTRODUCTION • To provide EdD and the Government of Djibouti (GoD) a comprehensive report, model and database for the further development of its systems and updates of the plan as needed. • To provide EdD and the Ministry of Energy & Natural Resources (MENR) with some basic planning capacity and tools to update some key components of the master plan as needed. 1.3 Technical Memorandum The Terms of Reference for this master plan study requires the consultant to submit a Technical Memorandum (TM). The purpose of this document is to set out the main assumptions forming the foundations underpinning the analysis presented in this Report. The TM presented: • The derivation of the demand forecast, identifying the key assumptions employed; • The entire EdD electricity production is based on thermal powered plants using heavy fuel and gasoil, resulting in high electricity costs. Confronted with an ever increasing petroleum demand and a rapid growth in urbanisation; the GoD decided to direct the country’s energy policy towards the development of renewable energy resources. We highlight the main energy resources that potentially can be used for generating electricity with a view to evaluating these resources within the context of the least cost electricity master plan for Djibouti; • An identification of the economic inputs and parameters, together with the generation planning criteria to be utilised in deriving the least-cost generation expansion plan; • A review of the short-term supply/demand balance for EdD; and • Set out the proposed transmission planning criteria and the circuit parameters to be adopted for identifying the future transmission and distribution requirements. The TM was submitted in May 2009 for discussion. Following the receipt of the World Bank approval, we have used the information and assumptions published as the basis for the analysis presented in this report. 1.4 Structure of the report This report comprises two volumes. Volume 1 contains the Main Report and Volume 2 the Appendices. Volume 1 comprises 10 Sections: Section 1 comprises this Introduction. Least Cost Electricity Master Plan (Volume 1) Final Report November 2009 Page 1.3 SECTION 1 INTRODUCTION Section 2 provides a country background and a review of the electricity system in Djibouti. Section 3 presents the derivation of the demand forecast and identifies the key underpinning assumptions. Section 4 discusses and reviews the existing and committed power system in Djibouti. Section 5 presents the future generation development options. Section 6 sets out and discusses the details of the generation planning analysis. Section 7 documents the work undertaken to produce a transmission expansion plan for Djibouti for the period from 2009 to 2035. Section 8 assesses the distribution system requirements to meet the load growth over the planning period. Section 9 presents the expansion plan for the isolated systems. And finally Section 10 presents our overall conclusions and recommendations together with the investment schedules derived for the least cost generation, transmission and distribution plans. Volume 2 comprises 9 appendices associated with the report: APPENDIX A: Study Terms of Reference APPENDIX B: Sales Forecast APPENDIX C: Review of Geothermal Resources in Djibouti APPENDIX D: Review of Wind Resources in Djibouti APPENDIX E: Description of ASPLAN Least Cost Generation Planning Software APPENDIX F: Generation Planning Results APPENDIX G: Load Flow Results APPENDIX H: Transient Stability Results APPENDIX I: Distribution Network Least Cost Electricity Master Plan (Volume 1) Final Report November 2009 Page 1.4 SECTION 2 COUNTRY BACKGROUND SECTION 2 COUNTRY BACKGROUND 2 COUNTRY BACKGROUND 2.1 Country information Djibouti is located in the horn of Africa and is bordered by Eritrea to the north, Ethiopia to the west and south and Somalia to the south east. Djibouti has approximately 315 kilometre (km) of coastline which borders, to the east, the Gulf of Aden, the Gulf of Tadjoura and the entrance to the Red Sea (the ‘Bab el Mandeb’). The total area of Djibouti is approximately 23,000 square kilometre (sq km), of which 22,980 sq km is land and 20 sq km is water. Djibouti is predominantly desert, with a coastal plain, plateau and central mountains. Djibouti has very little rainfall and temperatures are high throughout the year. During winter (November to April) daytime temperatures range from approximately 25 to 35 degrees Celsius whilst in the summer (May to October) temperatures are persistently above 45 degrees Celsius and the level of humidity is very high. Lac Assal, located to the west of the gulf of Tadjoura, is the lowest point in Africa at 155 metres below sea level. A map of Djibouti is presented below in Figure 2.1. Figure 2.1: Map of Djibouti Source: CIA World Fact Book Least Cost Electricity Master Plan (Volume 1) Final Report November 2009 Page 2.1 SECTION 2 COUNTRY BACKGROUND In 1977, the French Territory of the Afars and Issas became Djibouti. Towards the end of 1991 a civil war broke out between the Afars and the Issas, which devastated the country economically. Fighting continued for almost a decade, until on May 12 2001, a peace accord was finally signed. At present France maintains one of its largest military bases outside of France in Djibouti, whilst the United States of America (USA) have used Djibouti as a launching point for its war on terrorism since 2002. The war experienced between 1991 and 2001 had a significant detrimental impact on the economy of Djibouti, indicated by a fall in GDP per capita during this period. Although the situation has improved since the signing of the peace treaty and economic growth has resumed (albeit sporadically). Djibouti has experienced significant stability since the signing of the peace treaty, particularly in relation to the ongoing troubles experienced by its neighbours Somalia, Ethiopia and Eritrea. Despite the troubles within the region as a whole, Djibouti has forged close ties with Ethiopia. In 2004 a poverty reduction strategy was developed by the World Bank in coordination with the Millennium Development Goals (MDG) with the intent of reducing poverty and improving living standards. This strategy is currently being worked towards by the government of Djibouti and this interconnection study can be seen as a key indicator of this drive to improve the nations well being. The economy is driven by the service sector which has developed in relation to the country’s strategic location, controlling access to the Red Sea and situated on the key trade routes between Europe and the Middle East, India, Asia and Australasia. The service sector is, as such, dominated by the large port within Djibouti City (recently taken over by the Dubai Port Authority (DPA)). There is very little agricultural and industrial economic activity, although there is the potential to develop the fishing industry. The government is currently developing free-zones in and around the port area in an effort to attract industry and commerce to the country and encourage the expansion of the port. The location of several international army and air force bases in the country has helped stimulate the economy and also increase the international perception of safety and stability within Djibouti in recent years. Ethiopia’s ongoing conflict with Somalia and Eritrea has resulted in the landlocked country turning to Djibouti and its port to provide a hub for its import and export requirements. The current population of Djibouti is approximately 850,000 and about 80 per cent of this population is urbanised, living in the only major settlement, Djibouti City. The unemployment rate is currently very high and access to affordable electricity is low. Least Cost Electricity Master Plan (Volume 1) Final Report November 2009 Page 2.2 SECTION 2 COUNTRY BACKGROUND 2.2 The electricity system in Djibouti Electricité de Djibouti (EdD) was established in January 1960 to generate, distribute and sell electricity within the city of Djibouti and the towns of Arta, Oueah, Tadjoura, Obock, Dikhil, Ali Sabieh and Mouloud. Electricity access in Djibouti is estimated to be approximately 50 per cent, with approximately 80 per cent of the population living in urban areas. The per capita average annual electricity consumption in 2008 was 297 kWh, which is relatively high compared to Ethiopia. However, this level of consumption hides the disparities in the level of access between rural and urban areas. Energy access is very limited in rural areas and is found only in the vicinity of cities. Firewood is used in rural areas as a source of energy for cooking and heating. On the other hand, in urban areas, the majority of households use kerosene to satisfy their energy needs, but its high cost is an inhibiting factor particularly for poor households. The available generating capacity in the country is approximately 100 MW. This available generating capacity is divided between Djibouti City, Subdivision North (consisting of the towns Tadjoura and Obock) and Subdivision South (consisting of the towns Ali Sabiah and Dikhil). In the ‘Subdivision North’, there are 6 outlying generators in Tadjoura and a further 3 in Obock, providing a total available capacity of approximately 2.88 MW. In ‘Subdivision South’, there are outlying generators in Dikhil and Ali Sabieh providing a total available capacity of approximately 2 MW. The principal generating system consists of two power plants in the city of Djibouti. These are the Boulaos Power Station (82.8 MW) and the Marabout Power Station (14.4 MW). The power transmission system consists of 5 km of 63 kV underground cable connecting the Boulaos and Marabout transformer stations located in the central part of Djibouti. The distribution system within the city is at 20 kV. Dikhil and Ali Sabieh are supplied from a diesel plant located in Dikhil via 20 kV distribution lines. The 20 kV network is about 300 km long (240 km in Djibouti and 90 km in and between secondary cities). Energy is supplied to the customers through 279 distribution substations. In 2011 an interconnector between Ethiopia and Djibouti is due to begin operation. The interconnector will form the corner-stone for establishing a regional market whereby power (in the form of an energy contract) is traded between Ethiopia and Djibouti, increasing access to electricity at an affordable price. The introduction of the interconnector in 2011 will also see the incorporation of the Subdivision South (Ali Sabiah and Dikhil) load centres into the main network. Least Cost Electricity Master Plan (Volume 1) Final Report November 2009 Page 2.3 SECTION 2 COUNTRY BACKGROUND In 2008, the total electricity production was 326 GWh compared to 324 GWh in 2007 (an increase of 0.8 per cent) and 184 GWh in 1999 (an increase of 77 per cent during the ten year period). In 2008, energy sales amounted to 252 GWh and the peak demand was 56.9 MW. The services and domestic load components dominate the peak demand. The Djibouti system experiences two peaks between May to October, the first is at about 1500 hours when offices resume work after the mid-day break. The second occurs between midnight and morning due to air conditioning. The diversity in the occurrence of peak demands of the two countries is ideal for power transfer from Ethiopia to Djibouti. During cooler season, the minimum daily demand is as low as 10 MW and the evening peak is only about 15 MW. Least Cost Electricity Master Plan (Volume 1) Final Report November 2009 Page 2.4 SECTION 3 DEMAND FORECAST SECTION 3 DEMAND FORECAST 3 DEMAND FORECAST In this Section of the report we present the derivation of the demand forecast and identify its key underpinning assumptions. The determination of the demand forecast plays an important role for the electricity consumers of Djibouti, allowing for the development of the least cost generation expansion plan. The load forecast developed within this report is derived using econometric regression based techniques and utilises historical production/sales data provided by Electricité de Djibouti (EdD), historical economic data and data relating to the forecast level of development in the economy. To summarise, in this Section of the Report we: • Set out the general principles for robust demand forecasting techniques (Section 3.1). • Provide an overview of the methodology of the chosen demand forecasting technique (Section 3.2). • Present our assumptions relating to the historical economic performance of the country (Section 3.3) • Discuss the growth prospects of the economy, outlining the economic and demographic forecasts adopted for this study (Section 3.3). • Present and discuss the historical energy data provided by EdD, with particular emphasis on electricity sales for each defined consumer category (Section 3.3). • Present the forecast of electricity sales by consumer category, developed using our in-house regression model (Section 3.4). • Present the conversion of the sales forecast into the demand forecast in (Sections 3.5, 3.6, 3.7 and 3.8). • Discuss applicable Demand Side Management (DSM) techniques (Section 3.9). 3.1 The general principles of demand forecasting A demand forecast is the prediction of demand for power (MW) and energy (GWh) into the future. A demand forecast is a primary requirement for electricity planning. Electricity forecasts are needed for: • Generation planning, • Transmission planning, • Distribution planning, • Financial planning, • Pricing and tariff setting, and, • Operational planning (short-term). Least Cost Electricity Master Plan (Volume 1) Final Report November 2009 Page 3.1 SECTION 3 DEMAND FORECAST Different forecasts are needed to cover the short, medium or long term and for different levels of the system (e.g. generation, transmission substations, distribution substations and at consumer terminals). They may also be divided into one or more of the following: • Consumer group, • Voltage level, and, • Geographical area. The only certainty about a demand forecast is that it is unlikely to match or equal the actual outcome of events. To cover this eventuality it is essential to develop a demand forecasting technique which is suitable and appropriate for the work to which it relates. As a result, rigorous demand forecasting is necessary for a number of reasons. These reasons are: • It is very important for outside parties, e.g. bankers, IPP (Independent Power Producers), project shareholders and donor agencies, to be convinced of the reasonableness of future load growth before making a financial commitment. • Large consumers are often more optimistic about future growth than is justified by the prevailing economic climate. This may result in an over-estimate of load with a consequent over-investment of plant, especially at the LV (low voltage) level. • In markets where demand is approaching saturation, judgements formed from buoyant market growth in the past may not be a good guide to growth in the future. • Utilities will frequently over-estimate demand taking into account the time to get approval for projects and for construction. The demand forecast is particularly important because it is the principal input into expansion planning. Three characteristics of utilities heighten the negative consequences of inaccurate expansion plans: • They commit relatively large amounts of capital for long periods. • Comparatively long lead times are needed to add to their production capacity. • They provide a critical input to the production processes of many other industries. 3.2 Methodology The demand forecast developed for this study is derived using PB Power’s own in-house macroeconomic multi-variable regression analysis load forecast (RALF) model. This model investigates the likely presence of relationships between historical electricity sales for each customer category and economic drivers for consumption, such as population, per capita Gross Domestic Product (GDP) or sector specific GDP. Least Cost Electricity Master Plan (Volume 1) Final Report November 2009 Page 3.2 SECTION 3 DEMAND FORECAST Using our extensive experience and judgement of potential demand growth and where such statistical relationships are found from the historical data, we use these together with forecasts of the economic driving parameters to forecast sector sales of electricity in the future. The general form of the analysis uses the following standard econometric equation: Y = A + X1B + X2C + X3D …etc Where: Log (Y) = log (A) + B log (X1) + C log (X2) + D log (X3) …etc Y = Sector sales A = Constant B, C, D = Elasticities of demand (coefficients) X1, X2, X3 = Independent driving parameters. A strong historical relationship can be determined by analysing the statistical output that the model produces for any combination of consumer category and economic driver. There are three main statistical results that the analyst can use to determine the validity of any historical relationship. These are described below: • R2 (“R squared�) - The R2 statistic describes the correlation between historical sales (the dependent variable) and the economic drivers (the independent variables) selected in the regression. An R2 value of 1 would indicate ‘perfect correlation’. Whilst a value of zero would indicate ‘no correlation’. • Coefficients - The coefficients determine the impact of the economic drivers on historical sales. Regressions should only be accepted if the coefficients appear to be of the correct sign (positive or negative). For example, if price were included in a regression for domestic sales, the analyst should expect the price coefficient to be negative. A negative price coefficient indicates that a fall in price would result in an increase in sales (and vice-versa). The analyst should also expect the coefficients to be of the correct size. For example, the coefficient of any economic driver should not be too large, such that they indicate that a small change in the economic driver would result in an enormous change in the level of sales. • P Values - The P-values identify whether the economic drivers are adding anything to the regression. A P-value for any specific economic driver that is close to zero indicates that it is very likely that the particular economic driver in question plays a significant role in the determination of sales. Least Cost Electricity Master Plan (Volume 1) Final Report November 2009 Page 3.3 SECTION 3 DEMAND FORECAST Only those regressions which have fairly high R2 values (indicating good correlation), coefficients which are of reasonable magnitude and sign and low p-values should be accepted. Even if all these criteria are satisfied, the results obtained from the future projections should always be analysed to ensure that the growth rate is sensible within the context of overall expected energy sales and show limited volatility. The main premise with such a model is that it assumes that historical relationships will remain valid in the future. For this reason we interrogate the results of the analysis not only to ensure they are statistically robust, but to identify relationships which would yield forecasts that are not credible. In cases where no statistically robust relationships can be found, or an inappropriate relationship is identified, we can revert to alternative forecasting techniques, such as time series analysis. Through the adoption of the above guidelines, sales forecasts for each consumer category can be developed. Once a sales forecast has been developed, the model allows for a series of conversions to turn the sales forecasts (GWh) into a energy based generation level forecast (GWh) and a peak power demand forecast (MW) at the generation level. A step by step methodology to the derivation of the load forecast is presented in Figure 3.1. Figure 3.1: Load forecasting methodology Regression Analysis Consumer Level Sales Forecast Peak demand (GWh) CADLF's (MW) Energy Losses Power Losses Generation Level Peak Demand Forecast Forecast (GWh) (MW) Least Cost Electricity Master Plan (Volume 1) Final Report November 2009 Page 3.4 SECTION 3 DEMAND FORECAST To derive the generation level energy forecast, energy losses are added to the sales forecast. The level of losses is defined as “the difference between energy sales and sent out energy from the power station�, such that a forecast for sent out generation can be determined by adding system losses to sales. The user can make an assumption relating to system losses over the forecast period based on analysis of actual historical system losses and any loss reduction programmes that may be known. To derive the peak demand forecast at the generation level, the model uses Coincident After Diversity 1 Load Factors (CADLF ’s) and power losses. Load factors for each consumer category are estimated using consumer CADLF’s at the sales level. We have estimated these demands at sales level (i.e. before losses). This is because power losses are higher than energy losses, and if CADLFs were estimated inclusive of losses a change in losses would impact the CADLFs, even though the underlying consumer behaviour may not have changed. The CADLFs are estimated ensuring that the calculated demand at consumer level and the estimated demand at the same level are as close as possible over the historical time period under consideration. The above process provides a consumer level peak demand forecast. Power losses are then added to the consumer level demand forecast to produce the final peak demand forecast at the generation level. Relating the peak demand forecast to the generation level energy forecast, the system load factor is derived. The RALF model is a data intensive and data sensitive forecasting tool. The model requires the following raw data inputs: • Historical economic and demographic data, comprising of: - Population - GDP broken down by economic activity sectors • Forecasts of economic and demographic data, comprising of: - Population forecast - GDP forecasts for each economic sector 1 The CADLF is the load factor that relates energy sales to demand at time of system peak. Therefore, it includes the effects of both diversity (the maximum demand of a group of consumers is less than the sum of the individual demands) and coincidence (the peak demand of a group of consumers may not be at the time of system peak). Least Cost Electricity Master Plan (Volume 1) Final Report November 2009 Page 3.5 SECTION 3 DEMAND FORECAST • Historical consumer data, comprising of: - Historical sales data by consumer category - Historical consumer number data by consumer category - Historical tariff data by consumer category • Historical system data, comprising of: - Historical generation (sent out) - Historical system peaks We discuss our assumptions relating to each of the inputs above, in Sections 3.3 below. 3.3 Input data 3.3.1 Historical economic and demographic data Historical economic data Following independence from France in 1977 the economy of Djibouti has been unstable due to a civil war. A lack of investment and an influx of refugees from Somalia have additionally strained the country’s limited resources. However, the end of the civil war in Djibouti and the subsequent macroeconomic restructuring by the government has helped to bring positive economic growth back to the country. Unfortunately, this economic recovery has not been as fast as was hoped. A lack of jobs and an inability to create them has left the unemployment rate relatively high. The nation is not endowed with natural resources and relies heavily on imports, resulting in high prices. Reliable historic GDP data for the agriculture, manufacturing (industry) and service sectors is available for the years 1990 to 2008 from the World Bank and is shown in Table 3.1 below2. 2 GDP data for the years 1990 to 2007 extracted from the World Bank Data Query Tool website ( http://ddp- ext.worldbank.org/ext/DDPQQ/member.do?method=getMembers&userid=1&queryId=135 ). To complete the dataset for the year 2008 we have assumed the World Bank forecast of GDP growth as discussed in Section 3.3.2 of this report. Least Cost Electricity Master Plan (Volume 1) Final Report November 2009 Page 3.6 SECTION 3 DEMAND FORECAST Table 3.1: Historical GDP Data (DJF Millions, 1990 Prices) Year Agriculture Industry Services Total 1990 2,461 17,724 60,202 80,388 1991 2,449 16,666 57,834 76,949 1992 2,543 13,172 61,220 76,935 1993 2,532 10,910 58,400 71,841 1994 2,558 10,068 58,543 71,168 1995 2,222 10,606 55,857 68,686 1996 2,353 10,359 53,147 65,859 1997 2,310 10,021 53,036 65,368 1998 2,278 9,618 53,537 65,433 1999 2,333 9,955 54,570 66,858 2000 2,379 10,307 54,450 67,137 2001 2,424 10,723 55,365 68,511 2002 2,518 11,241 56,546 70,305 2003 2,621 11,725 58,208 72,555 2004 2,726 12,522 60,084 75,331 2005 2,746 12,906 62,067 77,719 2006 2,856 13,343 65,250 81,450 2007 3,266 14,303 67,138 84,708 2008 3,309 14,706 69,743 87,757 Figure 3.2 highlights the fall in economic activity as a result of the civil war conflict throughout the 1990s and the economic recovery which followed. Least Cost Electricity Master Plan (Volume 1) Final Report November 2009 Page 3.7 SECTION 3 DEMAND FORECAST Figure 3.2: Total GDP (DJF Millions, 2000 Prices) Agriculture Industry Services 120,000 100,000 80,000 GDP (DJF millions, 1990) 60,000 40,000 20,000 0 1990 1991 1992 1993 1994 1995 1996 1997 1998 1999 2000 2001 2002 2003 2004 2005 2006 2007 2008 In the early 1990’s total GDP fell at an average rate of - 1.8 per cent, with industry falling some 6.5 per cent per annum over the same period. GDP continued to decline during the second half of the 1990’s (the total GDP growth rate still fell by 0.2 per cent per annum), although a relatively strong growth in agriculture and a move away from negative growth in services was experienced. However, it can be seen that following the end of the conflict, recovery continued as total GDP grew by approximately 3 per cent per annum between 2000 and 2008, as all three sectors of the economy experienced positive growth. The annual total GDP growth rate is highlighted in Figure 3.3 overleaf. At present, the economy of Djibouti is a rent based economy with the majority of its income accrued from the large sea port, its military bases and grants. The service sector accounts for approximately 79.5 per cent of total GDP and completely overshadows the small industrial and agricultural base (16.8 per cent and 3.8 per cent respectively). Whilst the service sector has always dominated the economic demography of Djibouti, it is worth noting the impact that the civil war had on reducing the industrial capability of the country, with manufacturing only accounting for approximately 2.5 per cent of total GDP. The deficiency of natural resources and inhibiting weather has and will continue to restrict the agricultural sector’s role in the economy. Least Cost Electricity Master Plan (Volume 1) Final Report November 2009 Page 3.8 SECTION 3 DEMAND FORECAST Figure 3.3: Total GDP Annual Growth Rates 6% 4% 2% GDP (DJF millions, 1990) 0% 1991 1992 1993 1994 1995 1996 1997 1998 1999 2000 2001 2002 2003 2004 2005 2006 2007 2008 -2% -4% -6% -8% Historical population data There are many varying and conflicting population statistics for Djibouti and it is understood that a population census has not been carried out recently. It is however, commonly thought that approximately 80 per cent of the national population live in Djibouti City, whilst the remaining 20 per cent live in the other small towns (namely Obock, Tadjoura, Dikhil and Ali Sabieh) or alternatively live rurally. Reliable national population statistics are, however, available from the United Nations (UN) Department of Economic and Social Affairs for the years 1995 to 2008. The UN’s population data is 3 shown in Table 3.2 . We have adopted this data for use in this study. 3 The population data used in this report (and presented in Table 3.2) is taken from the UN document entitled ‘World Population Prospects: The 2008 Revision’ ( http://esa.un.org/unpp/ ). The data is presented for the period 1995 through to 2050 in increments of 5 years and has therefore been used to determine both historical and forecasted population statistics for Djibouti. In order to derive annual population data for Djibouti a simple linear interpolation technique has been used. Least Cost Electricity Master Plan (Volume 1) Final Report November 2009 Page 3.9 SECTION 3 DEMAND FORECAST Table 3.2: Historical Population Statistics (1995 – 2008) Year Population Growth Rate 1995 624 1996 645 3.4% 1997 666 3.3% 1998 688 3.2% 1999 709 3.1% 2000 730 3.0% 2001 745 2.1% 2002 760 2.0% 2003 775 2.0% 2004 790 1.9% 2005 805 1.9% 2006 820 1.8% 2007 835 1.8% 2008 849 1.8% Forecasts of economic and demographic data 3.3.2 Economic and demographic data forecasts Forecast of economic activity A common method of developing a GDP forecast is through the use of growth trend analysis. This process involves analysing the historical growth rates experienced on a sector by sector basis, (i.e. agricultural GDP, industrial GDP and services GDP) in order to determine how these sectors of the economy are likely to grow in the future. The derivation of GDP forecasts using this method involves the identification of a ‘best fit curve’ for the historical annual growth rate. The mathematical formula for the ‘best fit curve’ would provide the basis for the projection of GDP into the future. Unfortunately, historical growth rates for Djibouti show a high degree of variability and do not provide a clear indication as to the likely medium and long-term growth trend. Hence, applying growth trend analysis to a small sample of data will inevitably provide inaccurate projections of future growth patterns. Djibouti’s short term historical GDP growth rates following the end of civil war are relatively high and, in the long-term, would be considered unsustainable inevitably leading to over estimating future GDP, particularly for the Services sector (which makes up a large proportion of the total GDP). This is because the country is recovering from an economic slump and as such has been growing at an Least Cost Electricity Master Plan (Volume 1) Final Report November 2009 Page 3.10 SECTION 3 DEMAND FORECAST unsustainable ‘recovery’ rate. It is for this reason that it would be inaccurate to assume that these short term growth rate trends would continue into the future. The World Bank published a 10 year total GDP forecast for Djibouti in August 2006 for the period 4 2006 to 2015 . This forecast is presented in Table 3.3 below. Table 3.3: World Bank GDP Projections for Djibouti Base Case High Case 2006-2009 3.6% 4.2% 2010-2015 3.5% 4.3% The GDP forecast developed for this study is based on the World Bank base case projection identified in Table 3.3 above. This GDP forecast is based on the assumption that without sustained structural policy reforms, Djibouti would not reap the potential benefits of investment and growth, and would continue with high unemployment and low growth. Hence, the World Bank advised that the base case should be based on the ‘status quo’ whereby some policy reforms will take place, but not enough to significantly improve the current condition, leading to approximately 3.5 per cent growth per annum. However, if Djibouti was to make sustained efforts to implement the key policy reforms it could lead to growth of approximately 4.5 per cent par annum, achieving the Millennium Development Goals and stimulating further investor confidence and further growth. As this forecast encapsulates only the years up to 2015 for the base and high case scenarios only, it is appropriate to extend this forecast for the remainder of the study period as well as to encapsulate a low case scenario. 4 Source – Country economic memorandum Least Cost Electricity Master Plan (Volume 1) Final Report November 2009 Page 3.11 SECTION 3 DEMAND FORECAST Table 3.4: GDP Projections Adopted for this Study Year Base Case High Case Low Case 2008 3.6% 4.2% 3.0% 2009 3.6% 4.2% 3.0% 2010 3.5% 4.3% 2.7% 2011 3.5% 4.3% 2.7% 2012 3.5% 4.3% 2.7% 2013 3.5% 4.3% 2.7% 2014 3.5% 4.3% 2.7% 2015 3.5% 4.3% 2.7% 2016 3.5% 4.3% 2.7% 2017 3.5% 4.3% 2.7% 2018 3.5% 4.3% 2.7% 2019 3.5% 4.3% 2.7% 2020 3.5% 4.3% 2.7% 2021 3.4% 4.2% 2.6% 2022 3.4% 4.2% 2.6% 2023 3.4% 4.2% 2.6% 2024 3.4% 4.2% 2.6% 2025 3.4% 4.2% 2.6% 2026 3.2% 4.0% 2.4% 2027 3.2% 4.0% 2.4% 2028 3.2% 4.0% 2.4% 2029 3.2% 4.0% 2.4% 2030 3.2% 4.0% 2.4% 2031 3.0% 3.8% 2.2% 2032 3.0% 3.8% 2.2% 2033 3.0% 3.8% 2.2% 2034 3.0% 3.8% 2.2% 2035 3.0% 3.8% 2.2% For the base case scenario, it has been assumed that the annual 3.5 per cent growth rate assumed by the World Bank for the years up to 2015 will continue up to and including the year 2020, after which the annual growth rate will fall to 3.4 per cent for the years 2021 to 2025, then to 3.2 per cent per annum for the years 2026 to 2030 and then to 3 per cent per annum for the years 2031 to 2035. For the high case scenario, it has been assumed that the annual 4.3 per cent growth rate assumed by the World Bank for the years up to 2015 will continue up to and including the year 2020, after which the annual growth rate will fall to 4.2 per cent for the years 2021 to 2025, then to 4 per cent per annum for the years 2026 to 2030 and then to 3.8 per cent per annum for the years 2031 to 2035. Least Cost Electricity Master Plan (Volume 1) Final Report November 2009 Page 3.12 SECTION 3 DEMAND FORECAST As the World Bank did not provide a low case scenario, it has been assumed that the growth in this scenario will be symmetrical to that of the base case and high case. Figure 3.4 presents a graphical comparison of the GDP growth rates assumed for the base, the high and the low case GDP forecast scenarios. Figure 3.4: Forecasted annual GDP growth rates Base Case High Case Low Case 5.0% 4.5% 4.0% 3.5% Real GDP Growth (%) 3.0% 2.5% 2.0% 1.5% 1.0% 0.5% 0.0% 2008 2009 2010 2011 2012 2013 2014 2015 2016 2017 2018 2019 2020 2021 2022 2023 2024 2025 2026 2027 2028 2029 2030 2031 2032 2033 2034 2035 Applying the assumed growth rates in Table 3.4 to total GDP provides a total GDP forecast. It is then possible to re-allocate the total GDP forecast to its sub-sectors based on the market share of each sub-sector. However, these sub sectors are unlikely to retain the same market share for the whole forecast period. Hence account should be taken of likely growth in each of the sub-sectors. Djibouti’s economy is predominantly a rent based economy centred on the port and associated industries, whilst there is very little scope for the development of the agricultural sector due to its lack of natural resources and workable land. As the service sector (the port) is expected to drive economic growth in Djibouti, we have assumed that the market share of this economic sub-sector will increase at the expense of industry and agriculture. It is assumed that by 2025 the service sector will account for 81 per cent of the economy (rising from 79.5 per cent in 2008), whilst industry will account for 16.5 per cent (falling from 16.8 per cent in 2008) and agriculture will account for the remaining 2.5 per cent (falling from 3.8 per cent in 2008). Least Cost Electricity Master Plan (Volume 1) Final Report November 2009 Page 3.13 SECTION 3 DEMAND FORECAST On the basis of the above assumptions, the detailed base case GDP forecast for the period 2008 to 2035 is shown in Table 3.5 and Figure 3.5. The continued dominance of the service sector as the key driver for growth is highlighted in Figure 3.5 where it overshadows any growth in the other sub sectors. Low and high GDP forecasts have also been developed using the same technique. Figure 3.6 is a graphical presentation of proposed GDP forecasts for this study. Table 3.5: Base case GDP forecast (DJF Millions, 2000 Prices) Year Agriculture Industry Services Total 2008 3,309 14,706 69,743 87,757 2009 3,350 15,118 72,448 90,916 2010 3,388 15,526 75,185 94,098 2011 3,448 16,070 77,875 97,392 2012 3,508 16,632 80,661 100,801 2013 3,568 17,214 83,546 104,329 2014 3,628 17,817 86,535 107,980 2015 3,688 18,440 89,631 111,759 2016 3,748 19,086 92,838 115,671 2017 3,807 19,754 96,159 119,719 2018 3,866 20,445 99,599 123,910 2019 3,924 21,161 103,161 128,246 2020 3,982 21,901 106,852 132,735 2021 4,049 22,646 110,553 137,248 2022 4,116 23,416 114,383 141,915 2023 4,182 24,212 118,346 146,740 2024 4,248 25,035 122,445 151,729 2025 4,314 25,886 126,687 156,888 2026 4,372 26,715 130,822 161,908 2027 4,428 27,570 135,091 167,089 2028 4,483 28,452 139,501 172,436 2029 4,538 29,362 144,054 177,954 2030 4,591 30,302 148,755 183,648 2031 4,729 31,211 153,218 189,158 2032 4,871 32,147 157,814 194,833 2033 5,017 33,112 162,549 200,678 2034 5,167 34,105 167,425 206,698 2035 5,322 35,128 172,448 212,899 Least Cost Electricity Master Plan (Volume 1) Final Report November 2009 Page 3.14 SECTION 3 DEMAND FORECAST Figure 3.5: Base case GDP forecast Agriculture Industry Services 250,000 200,000 GDP (DJF millions, 1990) 150,000 100,000 50,000 0 2008 2009 2010 2011 2012 2013 2014 2015 2016 2017 2018 2019 2020 2021 2022 2023 2024 2025 2026 2027 2028 2029 2030 2031 2032 2033 2034 2035 Figure 3.6: Base, High and Low GDP Forecasts (DJF Millions, 2000 Prices) Base Case High Case Low Case 300,000 250,000 GDP (DJF millions, 1990) 200,000 150,000 100,000 50,000 0 2008 2009 2010 2011 2012 2013 2014 2015 2016 2017 2018 2019 2020 2021 2022 2023 2024 2025 2026 2027 2028 2029 2030 2031 2032 2033 2034 2035 Least Cost Electricity Master Plan (Volume 1) Final Report November 2009 Page 3.15 SECTION 3 DEMAND FORECAST Forecast of population It is difficult to make an accurate estimate for the population in Djibouti due to the deficiency of information and the currently adopted data collection procedures. The UN Department of Economic and Social Affairs have managed however, to develop a reasonable population forecast for Djibouti up to the period 2035. The UN forecast, adopted for this study, is presented in Table 3.6 below. Table 3.6: Population Forecast (1000s) Year Base Case High case Low Case 2008 849 849 849 2009 864 864 864 2010 879 879 879 2011 894 895 892 2012 909 912 905 2013 923 928 919 2014 938 945 932 2015 953 961 945 2016 968 979 957 2017 983 997 969 2018 997 1014 981 2019 1012 1032 993 2020 1027 1050 1005 2021 1044 1071 1018 2022 1061 1091 1031 2023 1077 1112 1043 2024 1094 1132 1056 2025 1111 1153 1069 2026 1127 1174 1081 2027 1143 1194 1093 2028 1160 1215 1104 2029 1176 1235 1116 2030 1192 1256 1128 2031 1207 1276 1139 2032 1223 1296 1150 2033 1238 1316 1160 2034 1254 1336 1171 2035 1269 1356 1182 3.3.3 Historical consumer data The historic records for energy data made available from EdD covers the period 1996 to 2008. Unfortunately, the classification of consumer categories has changed over this period and as a result, we have made some simplifying assumptions to the consumer category classifications. As a result Least Cost Electricity Master Plan (Volume 1) Final Report November 2009 Page 3.16 SECTION 3 DEMAND FORECAST we have identified 8 key consumer categories. The consumer categories assumed for this study are described below: Social. The Social consumer category is comprised of the ‘Tarif Social’ tariff category. This low voltage consumer category is understood to be for those consumers who subscribe to a 1 kVA connection to the grid. This is typically defined as the ‘life line tariff’ for very small consumers who have a small house or run a small shop. Their electricity usage is basic, generally utilising only a single light and fan. Domestic. The domestic consumer category is comprised of the ‘Tarif Domestique’ and the ‘Agents EdD’ tariff categories. This low voltage consumer category is for domestic consumers who subscribe to a 3 kVA and up to 9 kVA connection to the grid. This type of connection is typically taken for consumers with bigger houses and villas, and those consumers who run larger shops. Their electricity usage is greater than the social consumers’, generally utilising lights, fans, air conditioners, a fridge and other household electrical equipment. LV Djibouti. The LV Djibouti consumer category is comprised of the ‘Autres CC < 36’ and ‘Autres CS > 36’ tariff categories5. The LV Djibouti is a commercialised consumer category for those consumers with a low voltage connection within the city of Djibouti City. The consumers defined by this consumer category are typically large shops (with multiple air-conditioning units and fridges), public offices and government buildings (typically two storeys). Public Lighting. The Public Lighting consumer category is comprised of the ‘Eclairage Public’ tariff category. All electricity consumption used in the lighting of public places (for example street lighting) is categorised within the public lighting consumer category. This is a low voltage consumer category. All public lighting operates between the hours of 6pm and 5.15am on a nightly basis. Chantier. The Chantier consumer category is comprised of the ‘Tarif Chantiers’ tariff category. The Chantier consumer category is a low voltage category and represents all consumers who require electricity during temporary construction operations. As this category is only for temporary construction, once construction is complete, the electricity consumption from the new development will fall into a different consumer category. MV Djibouti. The MV Djibouti consumer category is comprised of the ‘MT Djibouti’ tariff category. The MV Djibouti is a consumer category for those consumers with a medium voltage connection 5 These are amongst the consumer categories that have historically been subject to a number of reclassifications and combining these two categories together meant that we can obtain a consistent set of data. Least Cost Electricity Master Plan (Volume 1) Final Report November 2009 Page 3.17 SECTION 3 DEMAND FORECAST within the city of Djibouti City. The consumers defined by this consumer category are typically industries such as the sea port, the airport, the free zone and military camps. LV Interior Districts. The LV Interior Districts consumer category defines all those low voltage consumers located outside of Djibouti City, taking an electricity supply from a system that is not connected to Djibouti city (i.e.; Dikhil, Ali Sabieh, Tadjoura and Obock). MV Interior Districts. The MV Interior Districts consumer category defines all those medium voltage consumers located outside of Djibouti City, taking an electricity supply from a system that is not connected to Djibouti city (i.e.; Dikhil, Ali Sabieh, Tadjoura and Obock). Analysis of historical consumer data For each of the consumer categories defined above, it has been possible to determine statistics to cover: • Energy sales (GWh) • Tariff (cost per kWh) • Number of consumers; and • Specific consumption (kWh per consumer) Table 3.7 below presents the energy data by consumer category and the system data used in this study. Analysis of the data provided in Table 3.7 illustrates the problems experienced in Djibouti following the breakout of civil war in 1991. While the energy statistics are complete and believed to be accurate, they show the considerable impact of the civil war on electricity consumption. Electricity sales between 1996 and 2008 show the economic recovery post the civil war conflict where Djibouti’s external aid flows increased and inflation declined sharply. The economy’s real growth performance was relatively flat in 1996 and 1997, but a recovery took hold in 1998 owing to Ethiopia’s decision at midyear to channel its transit trade through the port of Djibouti. In 1998, a fire broke out in the Boulaos power station, causing the closure of several units. This fire had a significant impact on 6 the availability and reliability of electricity supply, and in turn on the level of consumption and sales . Nevertheless, even negating the impact of the fire, does not hide the unstable sales growth experienced during the 1990’s due to the civil war. This instability will restrain the historical correlation statistics; however, the accuracy of the relationships identified will not be compromised. A 6 For modelling purposes we have devised a dummy variable to be applied in the regression analysis model. The aim of this ‘fire dummy’ is to be an independent variable driver (i.e. like population or total GDP) that will negate the impact of the fire within the sales data of the consumer categories which were adversely effected. This will allow for improved regression results, which are not influenced by the adverse sales slump and subsequent sales recovery. Least Cost Electricity Master Plan (Volume 1) Final Report November 2009 Page 3.18 SECTION 3 DEMAND FORECAST sharp upturn in port operations more than offsets the negative impact of the reduction in France’s military presence and led to a positive real GDP growth. Since the late 1990s, EdD’s stronger growth performance reflected continued political stability, strong port performance, increased donor support; and fiscal reforms carried out by the Djibouti Government that led to substantial reduction in the budget deficit. The recent upward growth trend also reflects a surge in foreign direct investment (FDI) flows owing to investments in the new port, military aid from the US and France and continued donor support. It should be noted for this study however, that electricity consumption within the interior districts of Djibouti will not be included in the regression analysis and a forecast of electricity consumption for these areas will be conducted separately (see Section 9). Least Cost Electricity Master Plan (Volume 1) Final Report November 2009 Page 3.19 SECTION 3 DEMAND FORECAST Table 3.7: Energy data (1996 - 2008) 1996 1997 1998 1999 2000 2001 2002 2003 2004 2005 2006 2007 2008 Social Sales (GWh) 20.05 20.07 12.70 17.21 18.60 15.80 18.79 19.03 20.16 20.38 21.30 22.36 24.99 Tariff in FDJ per kWh 29.83 29.98 32.26 30.64 39.51 36.19 36.57 35.33 34.79 36.67 40.33 43.24 48.13 No. of Consumers 11,562 12,580 12,347 12,997 12,083 12,462 13,254 13,964 14,876 15,115 16,034 16,582 17,222 Specific Consumption (kWh) 1,734 1,596 1,029 1,324 1,540 1,268 1,417 1,363 1,355 1,348 1,328 1,349 1,451 Domestic Sales (GWh) 46.30 52.41 32.54 40.26 51.62 62.02 62.74 63.66 63.89 63.20 64.09 68.41 69.12 Tariff in FDJ per kWh 37.29 37.11 37.89 37.37 37.02 40.44 42.39 41.72 42.18 44.47 42.54 47.00 49.78 No. of Consumers 8,335 9,875 8,377 9,501 11,628 11,933 11,950 12,004 11,928 12,206 13,049 13,773 14,191 Specific Consumption (kWh) 5,555 5,308 3,884 4,237 4,439 5,197 5,250 5,303 5,356 5,177 4,911 4,967 4,871 LV Djibouti Sales (GWh) 31.49 30.24 24.21 32.84 40.87 40.08 42.99 46.92 44.44 50.26 50.30 51.33 63.02 Tariff in FDJ per kWh 145.60 144.90 147.57 144.89 167.79 189.64 196.34 196.21 234.83 209.68 65.27 71.39 67.48 No. of Consumers 3,049 3,113 3,089 3,183 3,234 3,247 3,316 3,413 3,276 3,477 3,668 3,740 3,957 Specific Consumption (kWh) 10,329 9,712 7,838 10,318 12,637 12,345 12,963 13,746 13,567 14,455 13,713 13,725 15,927 Public Lighting Sales (GWh) 2.64 2.39 1.17 1.76 1.80 1.42 2.01 1.97 2.35 2.28 2.20 2.72 2.71 Tariff in FDJ per kWh 36.00 36.00 36.00 36.00 36.00 39.07 40.90 41.84 45.12 45.72 49.11 53.98 57.26 No. of Consumers 168 168 168 169 169 169 169 173 174 175 177 177 187 Specific Consumption (kWh) 15,714 14,220 6,940 10,396 10,675 8,385 11,917 11,393 13,477 13,040 12,424 15,384 14,476 Chantier Sales (GWh) 1.05 0.91 1.33 0.58 0.86 0.74 0.69 0.90 1.07 1.34 1.15 3.69 2.27 Tariff in FDJ per kWh 53.94 54.53 50.77 56.25 52.54 55.38 59.29 57.32 62.69 55.43 70.31 73.36 78.75 No. of Consumers 86 87 81 76 82 71 74 63 64 76 94 105 111 Specific Consumption (kWh) 12,261 10,483 16,444 7,605 10,427 10,451 9,311 14,302 16,781 17,671 12,255 35,181 20,477 MV Djibouti Sales (GWh) 57.37 56.64 44.98 47.51 58.83 70.02 66.09 74.31 82.95 83.34 92.54 96.91 80.54 Tariff in FDJ per kWh 37.00 37.46 30.75 39.57 36.93 33.79 41.50 40.39 40.33 42.14 46.57 49.48 62.76 No. of Consumers 665 114 109 107 108 110 110 109 123 114 111 132 136 Specific Consumption (kWh) 86,263 496,881 412,634 444,055 544,742 636,546 600,802 681,765 674,370 731,092 833,712 734,197 592,169 LV Interioir Districts Sales (GWh) 5.33 4.91 4.98 4.44 4.84 4.87 4.84 6.28 5.77 6.58 6.71 7.17 7.32 Tariff in FDJ per kWh 41.02 41.05 41.03 40.91 41.06 43.97 48.20 40.79 35.57 43.95 48.76 53.54 57.22 No. of Consumers 2,215 2,366 2,371 2,362 2,315 2,344 2,424 2,542 2,507 2,869 3,043 3,247 3,430 Specific Consumption (kWh) 2,408 2,075 2,102 1,878 2,090 2,076 1,998 2,470 2,303 2,295 2,204 2,209 2,135 MV Interior Districts Sales (GWh) 1.34 1.30 1.13 1.16 1.08 1.16 1.52 1.73 4.49 0.83 0.93 2.11 1.79 Tariff in FDJ per kWh 51.50 51.27 51.88 57.30 53.21 55.72 56.60 51.72 35.86 70.73 69.82 64.13 72.11 No. of Consumers 58 10 10 10 9 9 11 11 12 10 10 10 12 Specific Consumption (kWh) 23,030 129,691 113,382 116,264 120,295 129,428 137,763 157,173 373,980 83,312 93,100 210,500 149,083 TOTALS Sales (GWh) 166 169 123 146 178 196 200 215 225 228 239 255 252 Tariff in FDJ per kWh No. of Consumers 26137 28313 26552 28405 29628 30345 31308 32279 32960 34042 36186 37766 39246 Specific Consumption (kWh) 6,335 5,965 4,634 5,131 6,025 6,463 6,377 6,655 6,830 6,704 6,611 6,745 6,415 1996 1997 1998 1999 2000 2001 2002 2003 2004 2005 2006 2007 2008 Generation Sent Out (GWh) 202.49 196.14 146.47 184.39 226.32 234.31 247.64 260.66 274.36 297.98 307.48 322.97 325.62 Energy Losses (%) 18.2% 13.9% 16.0% 20.9% 21.1% 16.3% 19.4% 17.6% 17.9% 23.4% 22.2% 21.1% 22.7% Maximum Demand (MW) 41.90 40.80 32.80 38.10 39.70 41.34 42.70 45.98 48.15 53.49 52.90 55.70 56.92 Load Factor (%) 55.2% 54.9% 51.0% 55.2% 65.1% 64.7% 66.2% 64.7% 65.0% 63.6% 66.4% 66.2% 65.3% Least Cost Electricity Master Plan (Volume 1) Final Report November 2009 Page 3.20 SECTION 3 DEMAND FORECAST 3.4 Sales forecast Using the RALF model we have developed regression based sales forecasts for each of the six defined consumer categories (Social, Domestic, LV Djibouti, Public Lighting, Chantier and MV Djibouti). A summary of the regression based analysis for each consumer category is presented in Appendix A reporting Volume 2 of this report. Table 3.8 below presents the regression based sales forecasts for each consumer category. Table 3.8: Regression based sales forecast Public Year Social Domestic LV Djibouti Chantier MV Djibouti Total Lighting 2008 25.0 69.1 63.0 2.7 2.3 80.5 242.6 2009 24.4 76.3 58.9 2.7 1.8 99.0 262.9 2010 24.8 79.6 61.3 2.8 1.9 103.2 273.6 2011 25.2 83.1 63.9 2.9 2.0 107.6 284.6 2012 25.7 86.7 66.5 3.0 2.1 112.0 296.0 2013 26.1 90.5 69.2 3.1 2.2 116.6 307.7 2014 26.5 94.5 71.9 3.2 2.4 121.3 319.7 2015 26.9 98.7 74.7 3.3 2.5 126.0 332.1 2016 27.3 103.0 77.6 3.4 2.6 130.9 344.8 2017 27.7 107.5 80.5 3.5 2.8 135.9 358.0 2018 28.2 112.2 83.5 3.6 3.0 141.0 371.5 2019 28.6 117.1 86.5 3.8 3.1 146.2 385.4 2020 29.0 122.3 89.7 3.9 3.3 151.6 399.7 2021 29.5 127.5 93.3 4.0 3.5 157.8 415.5 2022 29.9 132.9 97.0 4.2 3.7 164.1 431.8 2023 30.4 138.6 100.8 4.3 3.9 170.6 448.6 2024 30.9 144.5 104.6 4.4 4.1 177.2 465.8 2025 31.4 150.6 108.6 4.6 4.4 184.0 483.6 2026 31.8 156.7 112.5 4.7 4.6 190.7 501.1 2027 32.3 163.0 116.5 4.9 4.8 197.6 519.0 2028 32.7 169.5 120.6 5.1 5.1 204.5 537.5 2029 33.2 176.3 124.7 5.2 5.3 211.7 556.4 2030 33.7 183.3 128.9 5.4 5.6 218.9 575.9 2031 34.1 190.2 133.0 5.5 5.9 226.0 594.8 2032 34.5 197.4 137.2 5.7 6.2 233.2 614.1 2033 35.0 204.8 141.5 5.9 6.4 240.5 634.0 2034 35.4 212.5 145.8 6.1 6.7 247.9 654.4 2035 35.8 220.5 150.2 6.2 7.0 255.5 675.3 We have been made aware however, that there are a number of relatively large committed projects which will come ‘on line’ within the next ten years. These include: Least Cost Electricity Master Plan (Volume 1) Final Report November 2009 Page 3.21 SECTION 3 DEMAND FORECAST • Heron City development, • Haramous City development, • A water desalination plant, • American Embassy, • The free zone, • Universities at Balbala and Malayslenne, • Business centres A and B • The ‘Port de Doraleh’, container and industrial zone, and • A large cement factory. We believe that the projects listed above have not been considered within the GDP forecast and therefore, following the initial regression analysis, we have modified the level of sales in the appropriate consumer categories to account for the extra demand for electricity created by these projects based on our evaluation and professional judgement of the specific projects. A summary of the “additional load� modifications assumed in this study are presented in Appendix A in Volume 2 of this report. A comment on desalination plant It should be noted that the additional load incurred by the development of the desalination plant has been modelled as a new consumer category with its own CADLF (assumed to be 75 per cent). The desalination plant is a significant load to be added to such a small system and by including it within the MV Djibouti consumer category (which has a CADLF of 100 per cent) would result in the over- estimation of its load factor and thus incorrectly increase the system load factor. For this reason, the additional load incurred by the development of a desalination plant is modelled as a new, separate The base case modified sales forecast derived for this study is shown in Table 3.9. The total consumer sales forecast is seen to grow from 243 GWh in 2008 to 1,145 GWh in 2035, an average annual increase of 5.9 per cent. A comparative analysis of the regression based sales forecast (Table 3.8) and the modified sales forecast (Table 3.9) highlights that the size and timing of the committed ‘additional loads’ have a significant impact on the sales forecast assumed for this study, particularly in the years 2009 to 2016. Least Cost Electricity Master Plan (Volume 1) Final Report November 2009 Page 3.22 SECTION 3 DEMAND FORECAST Table 3.9: Modified sales forecast LV Public MV Desalination Year Social Domestic Chantier Total Djibouti Lighting Djibouti Plant 2008 25.0 69.1 63.0 2.7 2.3 80.5 242.6 2009 24.4 82.0 58.9 2.7 1.8 113.0 282.7 2010 24.8 85.3 61.3 2.8 1.9 138.3 314.4 2011 25.2 100.2 63.9 2.9 2.0 163.6 10.5 368.3 2012 25.7 115.2 71.8 3.0 2.1 217.1 21.0 455.9 2013 26.1 130.4 90.2 3.1 2.2 242.7 47.3 542.0 2014 26.5 140.1 103.5 3.2 2.4 261.4 73.6 610.5 2015 26.9 155.6 106.3 3.3 2.5 266.2 99.9 660.6 2016 27.3 171.3 109.1 3.4 2.6 271.1 126.1 711.0 2017 27.7 187.2 112.0 3.5 2.8 276.1 126.1 735.5 2018 28.2 203.3 115.0 3.6 3.0 281.2 126.1 760.4 2019 28.6 213.9 118.1 3.8 3.1 286.4 126.1 780.0 2020 29.0 224.8 121.2 3.9 3.3 291.7 126.1 800.0 2021 29.5 230.0 124.8 4.0 3.5 297.9 126.1 815.8 2022 29.9 235.4 128.5 4.2 3.7 304.3 126.1 832.1 2023 30.4 241.1 132.3 4.3 3.9 310.8 126.1 848.9 2024 30.9 247.0 136.2 4.4 4.1 317.4 126.1 866.2 2025 31.4 253.1 140.1 4.6 4.4 324.2 126.1 883.9 2026 31.8 259.2 144.0 4.7 4.6 330.9 126.1 901.4 2027 32.3 265.4 148.0 4.9 4.8 337.7 126.1 919.3 2028 32.7 272.0 152.1 5.1 5.1 344.7 126.1 937.8 2029 33.2 278.8 156.2 5.2 5.3 351.8 126.1 956.7 2030 33.7 285.8 160.5 5.4 5.6 359.1 126.1 976.2 2031 34.1 292.7 164.6 5.5 5.9 366.1 126.1 995.1 2032 34.5 299.9 168.8 5.7 6.2 373.3 126.1 1014.5 2033 35.0 307.3 173.0 5.9 6.4 380.6 126.1 1034.3 2034 35.4 315.0 177.3 6.1 6.7 388.1 126.1 1054.7 2035 35.8 323.0 181.7 6.2 7.0 395.7 126.1 1075.6 3.5 Losses Energy losses are defined as the difference between historical energy sales and historical sent out energy from the power stations. In the RALF model the addition of system energy losses to the forecast of sales provides a forecast of sent out generation. To derive the sent out generation forecast it is therefore necessary to make an assumption with regard to the development of system losses into the future (i.e. a loss reduction programme). To facilitate the selection of a reasonable loss reduction programme assumption it is important to review the historical level of losses and to understand the cause of the losses. Figure 3.7 below presents the historical energy losses provided by EdD for the period 1996 to 2008. Least Cost Electricity Master Plan (Volume 1) Final Report November 2009 Page 3.23 SECTION 3 DEMAND FORECAST Figure 3.7: Historical Energy Losses (1996 – 2008) 30% 25% 20% Energy Losses (%) 15% 10% 5% 0% 1996 1997 1998 1999 2000 2001 2002 2003 2004 2005 2006 2007 2008 Unfortunately the historical dataset provided by EdD does not differentiate between technical and non-technical losses. To derive a reasonable loss reduction programme it is preferential to know how the losses are broken down (i.e. how much energy is lost to technical deficiencies and how much energy is lost to non-technical issues). In this respect, CabIRA and BCP have carried out a study (2008) on the reduction of technical and non-technical losses of energy for EdD. This study analyses data collated for the years 2005 to 2008. It is understood that this estimate is based on data analysis techniques and that detailed readings have not been carried out. The key results of the CabIRA/BCP study are summarised as follows: • The study concluded that technical losses could be reduced to around 7 per cent by balancing the loads across the three phases and by improving the power factor. • The report stated that measurements conducted between August and September 2008 identified; - Phase imbalance on approximately 1/3 of distribution transformers. - Poor load power factor of most MV customers. • There is no penalty for poor power factor and MV customers in particular are therefore being allowed to operate with poor power factors. This is resulting in Least Cost Electricity Master Plan (Volume 1) Final Report November 2009 Page 3.24 SECTION 3 DEMAND FORECAST increased system losses and is also limiting the amount of active power that may be delivered by the generators. • Some long lines contribute to relatively high losses but most 20 kV lines are loaded to less than 60 per cent of their rating and therefore require no new investment. • It is understood that EdD are addressing voltage issues at Arta by installing 20/0.41 kV or 19/0.4 kV distribution transformers. • The report also notes that a 6 MVAr capacitor bank could be used to improve the voltage at Arta. • The report concluded that re-configuration of the MV network would not result in significant loss reduction because most of the network is composed of adequately sized 150 mm2 feeders. • Reduction of non-technical losses could be through: - Improvement in billing systems - Installation of pre-payment meters. Table 3.10 below presents the breakdown of losses on the Djibouti system as estimated by CabIRA/BCP. It is apparent from this table that the total level of total losses (technical losses plus non-technical losses) are in line with those indicated by the data provided by EdD (and presented in Figure 3.7 above). Table 3.10: Breakdown of losses on the Djibouti system Unit 2005 2006 2007 2008 Energy Delivered GWh 286,364 294,927 308,795 313,381 Asymmetrical GWh 32,159 34,536 37,241 37,762 Technical Losses % 11.23 11.71 12.06 12.05 Non-technical GWh 42,296 42,145 33,782 46,850 Losses % 14.77 14.29 10.94 14.95 Total Losses GWh 74,455 76,681 71,023 84,613 % 26.00 26.00 23.00 27.00 Whilst the total losses indicated by the CabIRA/BCP report line up with the data provided by EdD, our experience and understanding of the Djibouti system suggests that technical losses contribute a larger proportion to total losses than non-technical losses. Least Cost Electricity Master Plan (Volume 1) Final Report November 2009 Page 3.25 SECTION 3 DEMAND FORECAST As such, for the year 2008 we estimate total losses to be 25.5 per cent. We believe that technical losses account for 15 per cent, whilst non-technical losses account for the remaining 10.5 per cent. For this study the following loss reduction programme has been assumed: • Whilst technical losses at 15 per cent can be assumed to be appropriate for the EdD system and any reduction could be a challenge, we believe that with the ongoing generation, transmission and distribution improvements, there may be room for a small improvement. We have therefore assumed technical losses to reduce from 2010 onwards at a rate of 0.2 per cent per annum until reaching 12 per cent in 2024 and remaining at that level thereafter.. • Non-technical losses are assumed to reduce from 2011 onwards, falling to 8 per cent in 2011, 6 per cent in 2012, 4 per cent in 2013, 2 per cent in 2014 and to 1 per cent in 2015. Non-technical losses are assumed to remain at 1 per cent for the remainder of the study period. Total losses are therefore assumed to fall below 20 per cent by 2013, below 15 per cent by 2015 and reach 13 per cent by 2024. The assumed loss reduction programme detailed above is ambitious; however, it is highly recommended that EdD strive to reduce losses as quickly as possible (see box below). A comment on the ambitious loss reduction programme assumed as part of this study It is recommended that EdD should take action to reduce the non-technical losses experienced on their system. This should be one of the key planning priorities for EdD. The investment required to reduce losses from their current high level of around 25 per cent to a more reasonable level of say 10 - 15 per cent* will be less than the amount of investment required in generation if losses were to remain at their current levels: if losses are high then additional generation would be required to satisfy the same amount of electricity sales. It can therefore be assumed that there will be savings to be gained from reducing losses, particularly in light of the high oil price currently experienced (and which is likely to remain for the foreseeable future). In other words, it will be cheaper for EdD to implement an active loss reduction programme than it will be for it to continue operating a system with a high level of losses. The demand forecast developed for this study assumes an active loss reduction programme as this process will be beneficial to both the electricity consumers of Djibouti and to EdD themselves. * The reduction of losses to a level below 10 per cent would incur significant additional costs and for the Djibouti system, this objective is not likely to be an economic solution at present. Least Cost Electricity Master Plan (Volume 1) Final Report November 2009 Page 3.26 SECTION 3 DEMAND FORECAST 3.6 CADLF’s We have estimated the load factors for each consumer category using consumer Coincident After Diversity Load Factors (CADLF’s) at the sales level. The CADLF is the load factor that relates energy sales to demand at time of system peak. Therefore, it includes the effects of both diversity (the maximum demand of a group of consumers is less than the sum of the individual demands) and coincidence (the peak demand of a group of consumers may not be at the time of system peak). We have estimated these demands before losses are taken into account. This is because power losses are higher than energy losses. If CADLFs are measured after losses then any change in losses will change the CADLFs, even though the underlying consumer behaviour may not have changed. The consumer characteristics are estimated by ensuring that the calculated demand at consumer level (i.e. removing power losses from the generation sent out maximum demand) and the estimated demand at the same level are as equal as possible over the time period under consideration. On this 7 basis, Table 3.11 sets out the CADLFs for each consumer group . The figures outlined in Table 3.11 are based on the technical understanding of the system and discussions with EdD staff. The correlation between actual system power sales and CADLF derived power sales is presented in Figure 3.8. The CADLF’s should be set at a level which results in the two lines shown in Figure 3.8 overlapping, however, in this case the difference between the two lines has been allowed to be greater than would normally be preferred. This is due to the specifics of the Djibouti system, in particular, the significant amount of additional loads belonging to the MV consumer category. Table 3.11: CADLFs by Consumer Category CADLF Consumer Category (%) Social 60.0% Domestic 65.0% LV Djibouti 60.0% Public Lighting 50.0% Chantier 80.0% MV Djibouti 80.0% Desalination Plant 60.0% 7 It should be noted that we have additional assumed a CADLF of 60 per cent for the ‘additional’ desalination plant. As there is no historical data for this temporary consumer category, the assumption will have no impact on the analysis presented in Figure 3.8. Least Cost Electricity Master Plan (Volume 1) Final Report November 2009 Page 3.27 SECTION 3 DEMAND FORECAST Figure 3.8: System power sales v CADLF based power sales System Power Sales (MW) CADLF Power Sales 45 40 35 30 Power Sales (MW) 25 20 15 10 5 0 1996 1997 1998 1999 2000 2001 2002 2003 2004 2005 2006 2007 2008 3.7 Demand forecast Using the CADLF and energy loss reduction programme inputs, we have derived the sent out base case MW demand scenario as shown in Table 3.12 and Figure 3.9. Figure 3.9 presents a comparison of the final base case demand forecast presented in Table 3.12 against the demand forecast if additional loads were not included. This figure clearly highlights the significant impact of adding committed additional loads to a purely regression based forecast. Least Cost Electricity Master Plan (Volume 1) Final Report November 2009 Page 3.28 SECTION 3 DEMAND FORECAST Table 3.12: Demand forecast - base case scenario Least Cost Electricity Master Plan (Volume 1) Final Report November 2009 Page 3.29 SECTION 3 DEMAND FORECAST Figure 3.9: Demand forecast – base case scenario Regression Based Demand Forecast Modified Demand Forecast 250 200 Maximum Demand (MWso) 150 100 50 0 1996 1998 2000 2002 2004 2006 2008 2010 2012 2014 2016 2018 2020 2022 2024 2026 2028 2030 2032 2034 3.8 Demand forecast Scenarios 3.8.1 High case demand forecast In addition to the base case demand forecast, we have developed a high case demand forecast. This forecast has been developing using the same methodology as described previously and the same regression relationships, with the only difference being the inputs, namely, we adopt a high case GDP forecast and a high case population forecast. For this scenario we assume that the size and timing of the additional loads remains the same as the base case scenario. The assumptions made in relation to the GDP and population high case forecasts are presented earlier in this Section of the report. The high demand forecast developed for this study is presented in Table 3.13 below. Least Cost Electricity Master Plan (Volume 1) Final Report November 2009 Page 3.30 SECTION 3 DEMAND FORECAST Table 3.13: Demand forecast - high case scenario Least Cost Electricity Master Plan (Volume 1) Final Report November 2009 Page 3.31 SECTION 3 DEMAND FORECAST 3.8.2 Low case demand forecast In addition to the base case demand forecast, we have developed a low case demand forecast. This forecast has been developing using the same methodology as described previously and the same 8 regression relationships , with the only difference being the inputs, namely, we adopt a low case GDP forecast, a low case population forecast and a low case assumption relating to the size and timing of the additional loads. The assumptions made in relation to the GDP and population low case forecasts are presented earlier in this Section of the report.. Table 3.14 below presents our assumptions relating to the size and timing of additional loads for the low case demand forecast scenario. The low demand forecast assumed for this study is presented in Table 3.15 below. 8 We have made a new assumption for the derivation of Social sales – see Appendix B. Least Cost Electricity Master Plan (Volume 1) Final Report November 2009 Page 3.32 SECTION 3 DEMAND FORECAST Table 3.14: Additional loads - low case scenario Least Cost Electricity Master Plan (Volume 1) Final Report November 2009 Page 3.33 SECTION 3 DEMAND FORECAST Table 3.15: Demand forecast - low case scenario Least Cost Electricity Master Plan (Volume 1) Final Report November 2009 Page 3.34 SECTION 3 DEMAND FORECAST 3.8.3 Demand forecast comparison Figure 3.10 below presents a comparison between the base, high and low demand forecast scenarios derived for this study. It is apparent from this diagram that the forecast is asymmetric and weighted towards a pessimistic viewpoint of future demand for electricity in Djibouti. In light of the current global recession and difficulties facing even the strongest economies over the coming years, it is reasonable to assume that future development in Djibouti is more likely to be lower than the base case than it is higher. It should also be noted that the size and timing of the assumed committed additional loads play a significant role in the demand forecasts developed in this study. As such, it is essential that EdD continue to monitor the development of these committed projects and update the demand forecast on a regular basis. Figure 3.10: Demand forecast comparison Base Case High Case Low Case 300 250 200 Maximum Demand (MWso) 150 100 50 0 2008 2010 2012 2014 2016 2018 2020 2022 2024 2026 2028 2030 2032 2034 3.9 Demand Side Management Demand-side management (DSM) usually refers to the actions of a utility to alter the pattern of end- use of electricity. These actions may be to increase or decrease demand or to shift demand away from peak demand periods to low load demand periods, or manage demand when there is intermittent load. These actions are carried out with the overall objective of reducing utility costs. In other words Least Cost Electricity Master Plan (Volume 1) Final Report November 2009 Page 3.35 SECTION 3 DEMAND FORECAST DSM is the implementation of any measures that help the customers to use electricity more efficiency and in doing so, reduce the cost of providing electricity by the utility to the consumer. Ideally, energy use would be optimised by supply and demand interactions in the market. For electricity use in particular, the price paid on the market is often regulated or fixed, and in many cases does not reflect the full cost of production. Electricity use can vary dramatically on short and medium time frames, and the pricing system may not reflect the instantaneous cost as additional higher-cost ("peaking") resources are brought on-line. In addition, the capacity or willingness of electricity consumers to adjust to prices by altering demand (elasticity of demand) may be low, particularly over short time frames. In many markets, consumers do not face real-time pricing at all, but pay rates based on average annual costs or other constructed prices. DSM activities should bring the demand and supply closer to a perceived optimum. DSM can be carried out by: • Developing and promoting energy efficient technologies. • Improving the efficiency of various end-uses, for example by correcting energy leakages, system conversion losses etc. • Adopting a tariff which reflects the costs of production (for example, higher prices during peak hours, concessional rates during off-peak hours, seasonal tariffs, interruptible tariffs, etc). • Incorporating development options, such as renewable energy systems, combined heat and power systems, independent PPAs etc that utility can implement to help meet the customer's demand at the lowest possible cost. The terms energy efficiency and DSM are often used interchangeably, but they should be used with caution. DSM refers explicitly to all activities that involve deliberate intervention by the utility in the marketplace so as to alter a consumer's load profile. Energy efficiency refers to any activity that would directly or indirectly lead to an increase in energy efficiency. For example, a program that encourages customers to install energy efficient lighting systems through a rebate program would fall under DSM. On the other hand, customer purchases of energy efficient lighting as a reaction to the perceived need for conservation is not DSM but energy efficiency gains. A loss reduction study has recently been completed for EdD and measures are in hand to implement the recommendations made. Furthermore, EdD is currently investigating the viability of introducing SMART metering into EdD’s distribution network and customer premises with a view to enhance metering, billing and collection as well as reduce the level of non-technical losses. Our demand forecast scenarios have taken account of the possible impact of such implementations. Least Cost Electricity Master Plan (Volume 1) Final Report November 2009 Page 3.36 SECTION 3 DEMAND FORECAST 3.10 Summary The demand forecast developed as part of this study and presented in this Section of the report can be summarised as follows: • The demand forecast technique adopted for this study is based on econometric principles and regression analysis. Regression analysis is widely regarded by financing institutes as a transparent, justifiable and competent demand forecasting technique. • The demand forecasts are developed using forecasts of GDP, population and historic sales data. • The regression based demand forecasts are supplemented by estimates of ‘additional loads’. These additional loads refer to a number of relatively large committed projects which will come ‘on line’ within the next ten years. • The addition of the projected loads from these projects to the regression based forecasts is based on the notion that their development is not considered within the GDP forecasts used to derive the regression based forecasts. • The size and timing of these additional loads have been based on our evaluation and professional judgement of the specific projects. • EdD should continuously monitor the progress of ‘additional load’ projects and revise forecasts as deemed necessary. • Whilst technical losses at 15 per cent can be assumed to be appropriate for the EdD system and any reduction could be a challenge, we believe that with the ongoing generation, transmission and distribution improvements, there may be room for a small improvement. • The level of non-technical losses in the EdD system is high (currently 10 per cent). The demand forecasts developed as part of this study assume an ambitious loss reduction programme (reducing losses from 25 per cent in 2009 to 13 per cent by 2024. • It is recommended that a thorough loss survey is undertaken and that EdD take subsequent action to reduce non-technical losses (and to a lesser extent technical losses) and that this action should be one of their key planning priorities. The demand forecasts developed for this study are based on the assumption of continued investment by EdD to increase the rate of electrification and reliability. Least Cost Electricity Master Plan (Volume 1) Final Report November 2009 Page 3.37 SECTION 4 EXISTING AND COMMITTED POWER SYSTEM SECTION 4 EXISTING and Committed Power System 4 EXISTING & COMMITTED POWER SYSTEM 4.1 Introduction In this sub-section we discuss and review existing and committed power system in Djibouti. 4.2 Existing and committed plant Demand for electricity in Djibouti is served by two main power stations, namely Boulaos and Marabout. The Boulaos power station is located to the east of the city. The Marabout power station is located next to the main port and container terminal on the Djibouti peninsula. The locations of the power plants are shown in Figure 4.1. Figure 4.1: Location of power plants in Djibouti Marabout Power Station Djibouti Port & Container Terminal Boulaos Power Station Source: http://wikimapia.org/ Least Cost Electricity Master Plan (Volume 1) Final Report November 2009 Page 4.1 SECTION 4 EXISTING and Committed Power System 4.2.1 Boulaos power station At present the Boulaos power station comprises thirteen medium speed diesel generating units. Of the thirteen installed units, only twelve are currently in operation. At present the Boulaos power station has a total sent out capacity of 84.4 MWso. The thirteenth unit is due to return to operation at the beginning of 2010 with a sent out capacity of 6.5 MWso. The latest investment programme obtained from EdD (dated 14 March 2009), indicates that there are two new medium speed diesel generating units to be added to the site in 2010. Each of these units has a sent out capacity of 4.7 MWso, thus raising the total installed capacity at the Boulaos power station to 100.3 MWso in 2010. EdD have additionally committed themselves to a programme of refurbishment of some of the units at the Boulaos power station. The refurbishment proposals will require certain units to be out of operation for a period of time before returning back onto the system. For this study we have assumed that the refurbished units will be “off-line� for a period of one year. Table 4.1 below presents the existing and committed plant capacity schedule for the Boulaos power station. Following a review of the maintenance schedule, the running hours and the refurbishment programme, Table 4.1 sets out our estimated retirement dates based on the currently adopted operation and maintenance policy. These retirement dates do not take into account the impact of the Ethiopia-Djibouti interconnector whereby the EdD units are likely to operate less hours and hence there would be a possibility of delaying the postulated retirement dates. The fuel cost for each unit of this power plant is to be derived using the specific fuel consumption of each unit and the fuel price forecast and thus working out the fuel price for each individual unit based on the relative fuel mix. All units at the Boulaos power station are HFO-fired units and use gas oil for starting up and shutting down. The average annual fuel mix approximately consists of 97.4 per cent HFO and 2.6 per cent gas oil. As such, the operating cost per kWh will reflect the relative efficiency of the units and therefore a unit with a higher level of efficiency will consume less fuel to produce the same output compared to a less efficient plant. The generation planning model takes account of operational cost changes on an annual basis, in line with the fuel price forecast and any assumed changes in efficiency of plant following refurbishment. Least Cost Electricity Master Plan (Volume 1) Final Report November 2009 Page 4.2 SECTION 4 EXISTING and Committed Power System Table 4.1: Boulaos capacity schedule Plant Type Capacity in Date of Post Retirement Date 2009 Refurbishment Refurbished or Installation / Installation Capacity (MWso) (MWso) G11 Medium Speed Diesel 4.0 2012 6.5 Beyond study horizon G12 Medium Speed Diesel 6.5 N/A 6.5 2034 G13 Medium Speed Diesel 4.5 N/A 4.5 2030 G14 Medium Speed Diesel 4.5 N/A 4.5 2030 G15 Medium Speed Diesel 4.5 N/A 4.5 2031 G16 Medium Speed Diesel 4.5 N/A 4.5 2031 G17 Medium Speed Diesel 6.5 N/A 6.5 2033 G18 Medium Speed Diesel 6.5 N/A 6.5 2034 G21 Medium Speed Diesel 10.0 2013 13.4 Beyond study horizon G22 Medium Speed Diesel 15.0 N/A 15 Beyond study horizon G23 Medium Speed Diesel 0.0 2009 6.5 Beyond study horizon G24 Medium Speed Diesel 4.5 2013 6.5 Beyond study horizon G25 Medium Speed Diesel 13.4 N/A 13.4 2030 G31 Medium Speed Diesel 0.0 2010 4.7 Beyond study horizon G32 Medium Speed Diesel 0.0 2010 4.7 Beyond study horizon 4.2.2 Marabout power station The Marabout power station comprises six medium speed diesel generating units of 2.4 MWso each, providing a total sent out capacity of 14.4 MWso. In spite of their age and the number of running hour 9 accumulated, all six units are assumed to be available for operation throughout the study period . All units at the Marabout power station are operated on 100 per cent gas oil. The operational and performance parameters for the Marabout plant are set out in Table 4.2. 9 Generally these units are required during the summer months only and hence the utilisation is relatively low. Utilisation will further reduce with the commissioning of the Ethiopia-Djibouti interconnector. Least Cost Electricity Master Plan (Volume 1) Final Report November 2009 Page 4.3 SECTION 4 EXISTING and Committed Power System Table 4.2: Marabout capacity schedule Plant Type Capacity Availability (MWso) (%) M1 Medium Speed Diesel 2.4 95 M2 Medium Speed Diesel 2.4 95 M3 Medium Speed Diesel 2.4 95 M4 Medium Speed Diesel 2.4 95 M5 Medium Speed Diesel 2.4 95 M6 Medium Speed Diesel 2.4 95 The operating costs for the units at this plant will be derived in the same way as described for the Boulaos plant (see section 4.2.1). As these units are gas oil-fired, it stands to reason that the operating cost of the Marabout plant is higher than the operating costs of the Boulaos plant. Again, it is important to note that the generation planning model takes account of the operational cost changes on an annual basis, in line with the fuel price forecast. 4.3 Interconnector While Djibouti faces an increasing reliance upon fossil fuel fired thermal generation to meet its growing demands, Ethiopia possesses abundant exploitable hydroelectric potential. The Ethiopian system is expected to have surplus energy available when on-going hydroelectric power projects are commissioned. The amount of excess energy available will vary periodically, declining gradually as demand for electricity grows and increasing markedly as each new hydroelectric project is commissioned. Against this backdrop, the two countries prepared a feasibility study of power interconnection with a view to cooperation in the development of their energy resources for mutual benefits. The Government of Ethiopia (GoE) and the GoD have agreed to implement the interconnection between the two systems supported by finance from the African Development Bank (AfDB). 4.3.1 Availability of supply At present there is a Power Purchase Agreement (PPA) between EEPCo and EdD governing the conditions of use for the soon to be built interconnector. The fundamental basis of this PPA is that the supply will, in general, be from Ethiopia to Djibouti and that the supply will be in the form of energy only and have no capacity element. To this end the supply is based on four periods. These periods are shown below in Table 4.3. Least Cost Electricity Master Plan (Volume 1) Final Report November 2009 Page 4.4 SECTION 4 EXISTING and Committed Power System Table 4.3 : Interconnector supply periods Peak Period Off-Peak Period Season / Time of Day (18:30 to 21:30) (21:30 to 18:30) st th Wet (July 1 to November 7 ) Supply Available Supply Available No Contractual Dry (November 8th to June 30th) Supply Available Supply All Times GMT +3 The definition of Peak and Off-Peak are based on Ethiopian daily load curves are set out in Table 4.3 above, the only period where there will be no contractual supply available to Djibouti will be during the Ethiopian Dry Peak period. The EdD system is characterised by two seasons: a. The summer high load season (May to October) when the EdD peak demand occurs around midnight (later than the EEPCo peak period); and b. The winter low load season (November to April) when the EdD peak period coincides with the EEPCo peak period. The EdD maximum demand during these months does not exceed 75 per cent of the annual system peak demand and can be as low as 55 per cent of system peak demand in January. This is illustrated in Table 4.4 below. Table 4.4 : EdD monthly system maximum demand (kW) September November December Maximum February January October August March June April May July 2005 27,800 29,760 35,500 38,970 46,850 51,650 50,500 49,460 53,490 48,020 37,900 32,500 53,490 2004 27,350 27,540 31,770 36,370 45,130 46,540 46,400 45,830 48,150 43,150 32,510 30,400 48,150 % of system 54.3% 56.4% 66.2% 74.1% 90.5% 96.6% 95.3% 93.8% 100.0% 89.7% 69.3% 61.9% maximum demand Therefore, the months of May and June are the only months when the EdD high load season overlaps with the EEPCo dry season with the consequence that EdD will supply the Djibouti system peak demand at those times when there is no contractual supply from EEPCo across the interconnector. 4.3.2 Available energy The power purchase agreement (PPA) currently states that the amount of energy that can be supplied by the interconnector on an annual basis is limited. The PPA estimates that energy available from the interconnector could range from between 180 GWh per annum up to a maximum of approximately 300 GWh per annum. Recent discussions between EdD and EEPCo have, however, indicated that Least Cost Electricity Master Plan (Volume 1) Final Report November 2009 Page 4.5 SECTION 4 EXISTING and Committed Power System the level of energy transfer available via the interconnector may potentially exceed 300 GWh per annum. Amongst numerous clauses within the PPA, it requires EdD to continue maintaining its existing power generation capacity and will carry on investing in new generating capacity such that EdD is able to cope with outages, unforeseen or planned, or with planned contract interruptions. 4.3.3 Operating philosophy The underpinning hypothesis of the operating philosophy revolves around the notion that the introduction of the interconnector between Ethiopia and Djibouti would be beneficial to both countries, and in particular, lower the electricity tariff in Djibouti. It is therefore vital that any potential benefits from introducing the interconnector are maximised. In this sense, it is essential that the operation of the interconnector and EdD’s power plants are optimised in the generation planning process. The starting point for this analysis is dictated by the wording of the PPA. The PPA is expressed in energy terms, not capacity terms. This means that Djibouti can take the agreed energy during any period except for the Peak Dry period. Given that the energy will not be imported during the peak dry period, it can be assumed that the cost of sale of energy over the interconnector will be significantly lower than the despatch costs of plant owned and operated by EdD. As previously mentioned, the PPA states that EdD must maintain sufficient capacity to supply its own electricity requirements during outages and therefore there can be no consideration of capacity savings on the Djibouti side of the PPA. It is also important to note that the level of MW import over the interconnector will only be limited by the thermal loading limit of the interconnector. Based on the arguments outlined above, the ‘Objective’ for EdD is to minimise the amount it spends on fuel, variable O&M and capacity, subject to maintaining full capacity to meet its own demand at all times. In other words, the objective is to maximise the sum of the fuel cost savings from using the interconnector whilst keeping the capacity costs as low as possible. The difference in cost between the fossil fuel fired generation in Djibouti and the hydroelectric generation in Ethiopia is so large that Djibouti is likely to import most, if not all, of the energy made available. This situation would continue until Djibouti installs some form of low cost generation utilising indigenous resources (most probably geothermal or wind). In order to maximise fuel cost savings from the introduction of the interconnector, import energy should displace energy from the most expensive EdD plant first. Thereafter, energy should be displaced from the plant with the next highest cost and so on. Least Cost Electricity Master Plan (Volume 1) Final Report November 2009 Page 4.6 SECTION 4 EXISTING and Committed Power System To maximise fuel cost savings, within the context of a load duration curve, energy imports should be located on the load duration curve from the top-down until the area under the curve is equal to the available import energy. Figure 4.2 below shows the load duration curve (for the period when energy imports from EEPCO are available) for the Djibouti system with 300 GWh of energy imports displacing energy generated by peaking and mid-merit order generating units. It can be seen that the interconnector displaces some 70 MW of generating capacity at peak load time, effectively displacing all EdD’s peaking units as well as a considerable amount of energy that would have been generated by the small medium speed diesel units at Boulaos, thus maximising fuel savings. Figure 4.2: Example of load duration curve with the interconnector stacked from the top 120 Small low speed diesel units at Boulaos and Marabout units displaced 100 Loading 80 Load in 2010 (MW) 70 MW approx 60 Energy Imports over the interconnector (300 GWh) 40 20 Generation from Large Low Speed Diesel Units @ Boulaos 0 0 1000 2000 3000 4000 5000 6000 7000 8000 Time (Exc Peak Dry Period) The interconnector with Ethiopia (and the energy that will flow from Ethiopia to Djibouti) is a committed future energy resource and it will play a significant role in shaping the development of the electricity sector in Djibouti into the future. The size of the potential energy flows available from Ethiopia, the relatively small demand in Djibouti and the high level availability of the interconnector means that any additional generating plant to be built in Djibouti is unlikely to be required to run for significant periods of time. In this sense, the least cost generation plan is likely to be skewed towards thermal plant with low capital costs and small, base load renewable generation. Least Cost Electricity Master Plan (Volume 1) Final Report November 2009 Page 4.7 SECTION 4 EXISTING and Committed Power System 4.3.4 Project progression and status The Governments of Ethiopia and Djibouti made a bilateral agreement on the implementation strategy for the interconnector. This bilateral agreement was signed between The Governments of Ethiopia and Djibouti in March 1998. A PPA between EEPCo and EdD was made on 21 April 2006. This agreement remains in force for the duration of construction plus a further 25 years. The construction, operation and maintenance agreement for the interconnector was also made on 21 April 2006 and this agreement also remains in full force and effect for the duration of construction plus 25 years. With regard to the physical construction of the line, it is understood that the project is approximately 40 per cent complete. The target date for completion is April 2010 and we understand that it is highly probably that the line will be available for operation around that date. For this study therefore, we have assumed that the full commercial operation of the line will commence at the beginning of 2011, allowing sufficient time for any minor delays, training of staff supporting documentation to be finalised. 4.4 Supply Demand Balance As part of the interconnector energy transfer agreement, EdD must maintain sufficient capacity to supply its own electricity requirements should the interconnector not be available. As such, Figure 4.3 below presents the balance between peak demand in Djibouti (as described in Section 3) and the existing and committed capacity available to supply this demand (as described in Section 4.2). This supply demand balance identifies the timing for new generating plant. It can be seen that the balance between demand and supply is positive for the years 2008 to 2012. When allowance is made for the need to carry reserve of capacity, however, (say at 20 per cent) to cover maintenance outages, the balance is positive for the years 2009 to 2011 only. This analysis suggests that even though the interconnector is estimated to begin operations in 2011, EdD will still have to invest in generating capacity around this time. Least Cost Electricity Master Plan (Volume 1) Final Report November 2009 Page 4.8 SECTION 4 EXISTING and Committed Power System Figure 4.3: Supply demand balance Total Capacity (MWso) Demand (MW) Reserve Margin (%) 300 120% 250 80% Capacity / Demand (MWso) 200 40% Reserve Margin (%) 150 0% 100 -40% 50 -80% 0 -120% 2008 2009 2010 2011 2012 2013 2014 2015 2016 2017 2018 2019 2020 2021 2022 2023 2024 2025 2026 2027 2028 2029 2030 2031 2032 2033 2034 2035 4.5 Transmission 4.5.1 General The EdD power system is split into three regions to supply Djibouti town, the south of the country and the north of the country respectively. The three regions are not electrically interconnected at present although the south of the country will be interconnected with Djibouti town as part of the Ethiopia- Djibouti Inter-connector project. 4.5.2 Djibouti town The Djibouti town region includes the two main power plants at Boulaos and Marabout which are electrically interconnected at 63 kV. Figure 4.4 shows a single line diagram for the main electrical equipment and configuration of the Djibouti town power system. The maximum demand for the Djibouti town region was 57 MW in 2008. Least Cost Electricity Master Plan (Volume 1) Final Report November 2009 Page 4.9 SECTION 4 EXISTING and Committed Power System Figure 4.4: Existing power system (Djibouti town) DIgSILENT 1 X 400 MM2 AL (4.8 KM) MARABOUT 63 BOULAOS 63 2 x 36 MVA 36 MVA G G G ~ ~ ~ G21 G22 G25 BOULAOS 20 MARABOUT 20 MARABOUT 15 2 x 1.5 MVAr G G G G G G ~ ~ ~ ~ ~ ~ BOULAOS LOAD MARABOUT LOAD M1 M2 M3 M4 M5 M6 G G G G G G G G G G ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ G1 G12 G13 G14 G15 G16 G17 G18 G23 G24 63 kV <= 20 kV The smaller generating units at Boulaos are connected via generator transformers to the 20 kV busbar, while the larger units are connected via generator transformers to the 63 kV busbar at Boulaos. The Marabout generators are connected via a single generator transformer to the Marabout 63 kV busbar. The 63 kV interconnector between Boulaos and Marabout comprises a 400 mm2 Al cable with a thermal rating in the summer of 36 MVA. The 63/20 kV transformers at Boulaos and Marabout are also rated for 36 MVA. There are 2 x 1.5 MVAr capacitor banks at Marabout to provide voltage support when required. 4.5.3 Division South Division South includes the towns of Dikhil and Ali Sabieh. There are two diesel power plants, one at Dikhil (2.5 MW installed capacity) and one at Ali Sabieh (800 kW installed capacity). The two power plants are interconnected at 20 kV and supply a maximum demand of around 1.4 MW. Least Cost Electricity Master Plan (Volume 1) Final Report November 2009 Page 4.10 SECTION 4 EXISTING and Committed Power System 4.5.4 Division North Division North includes two diesel power plants at Tadjoura and Obock which are not currently interconnected. Tadjoura power plant has an installed capacity of 2.16 MW and supplies a maximum demand of around 960 kW for the town. Obock power plant has an installed capacity of around 720 kW and supplies a peak of around 440 kW. 4.5.5 Inter-connector The Ethiopia-Djibouti inter-connector project is currently ongoing and due for completion in April 2010. Following completion of the inter-connector and associated infrastructure, there will be a significantly larger transmission system in the country. Initially, a single circuit 230 kV transmission line will be strung on one side of a double circuit tower construction. The 230 kV transmission line will comprise Twin Ash conductors (2x180mm2 AAAC). The 230 kV inter-connector will terminate at a 230/63/20 kV substation at PK-12, comprising two 230/63 kV, 63 MVA transformers. Two 63 kV, 65 MVA transmission circuits will connect PK-12 with Djibouti town via Palmeraie on the outskirts of the town. One circuit will terminate at Marabout and the other at Boulaos. A further 63 kV double circuit line will be installed from PK-12 to Ali Sabieh. These developments are shown geographically in Figure 4.5 and schematically in Figure 4.6. Figure 4.5: Existing and committed transmission system Marabout PK12 Boulaos Palmeraie Note: Palmeraie-Boulaos-Marabout-Palmeraie Existing Under construction 63 kV circuits are underground. All other transmission circuits 230 kV Are overhead. 63 kV 20 kV Least Cost Electricity Master Plan (Volume 1) Final Report November 2009 Page 4.11 SECTION 4 EXISTING and Committed Power System Figure 4.6: Single line diagram following completion of Inter-connector DIgSILENT INTER-CONNECTOR: 230 kV DOUBLE CIRCUIT TWIN ASH CONDUCTOR (283 KM) ETHIOPIAN NETWORK 2 x 63 MVA 120 KM 127 KM 40 KM DIRE DAWA 230 ADIGALA 230 PK-52 230 PK12 230 PK12 63 ETHIOPIA DJIBOUTI 40 MVA PK12 20 63 kV ASTER (8 KM) 63 kV DOUBLE CIRCUIT ASH CONDUCTOR (72 KM) PK12 LOAD 2 X 400 MM2 AL (4 KM) 2 X 400MM2 AL (5 KM) 1 X 400 MM2 AL (4.8 KM) MARABOUT 63 BOULAOS 63 ALI SABIEH 63 2 x 36 MVA 36 MVA MARABOUT 20 MARABOUT 15 BOULAOS 20 12 MVA G G G ~ ~ ~ G21 G22 G25 2 x 1.5 MVAr ALI SABIEH 20 G ~ G ~ G ~ G ~ G ~ G ~ BOULAOS LOAD MARABOUT LOAD M1 M2 M3 M4 M5 M6 ALI SABIEH LOAD G G G G G G G G G G ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ G1 G12 G13 G14 G15 G16 G17 G18 G23 G24 230 kV 63 kV <= 20 kV 4.6 Distribution 4.6.1 General In Djibouti, 20 kV is used for medium voltage (MV) and 400/230V is used for low voltage (LV) distribution. The distribution network in Djibouti comprises MV feeders supplying 20/0.4 kV distribution substations. Several LV feeders emanate from each distribution substation which in turn provides supplies to customers. Figure 4.7 shows the basic distribution system arrangement. In the urban area, the MV feeders are either connected across two primary substations or they are connected across separate busbar sections of a single primary substation. Each MV feeder has a normally open point which may be moved as required for operational reasons. The rural areas may be served by a single radial MV feeder as shown. Least Cost Electricity Master Plan (Volume 1) Final Report November 2009 Page 4.12 SECTION 4 EXISTING and Committed Power System Figure 4.7: Distribution System Arrangement Primary 20 kV Substation Distribution LV feeders substation Open point Typical rural feeder Typical urban feeders Open point Primary 20 kV Substation 4.6.2 MV Feeders The MV distribution system for Djibouti town and the village of Arta comprises 20 kV feeders originating from the primary substations of Marabout and Boulaos as shown in Appendix G. Some of the feeders are connected across Boulaos and Marabout while for others, both ends of the feeder are connected to either Marabout or Boulaos substation. In this case, the two ends of the feeder are connected to separate sections of the Marabout or Boulaos Primary switchboard. These feeders are meshed in the urban areas and may be supplied from either end which, (provided the circuits are well maintained) should result in a relatively high level of reliability at the medium voltage level. The feeders are operated with a normally open point, which through manual switching may be moved as required. Least Cost Electricity Master Plan (Volume 1) Final Report November 2009 Page 4.13 SECTION 4 EXISTING and Committed Power System Table 4.5 provides a list of the MV feeders showing the primary substation from which they originate and the length of the feeder in each case. The feeders are seen to range from 2.6 to 70 km, with the longer feeders supplying the outlying villages including Arta, Oueah and Damerjog. Table 4.5: MV Feeder Lengths Main Substation MV Feeder Length (m) Telex 5,315 Esperey 6,972 Artois 2,854 Pharmacie 2,585 Marabout Batignolles 6,125 Conteneur 4,850 Total 70,108 Z. Franche 2,930 Rocade 2,940 Sogik 8,288 Cooperation 11,072 Sirage 7,799 190LGTS 36,596 Boulaos LYS 5,460 Lettelier 5,583 O. Kamil 10,650 G. Batal 53,528 Bowling 8,605 Total length 252,260 The majority of medium voltage distribution feeders within the built-up areas of Djibouti town are underground cables comprising 150mm2 and 185 mm2 Al conductors. There are medium voltage overhead lines for distribution within the outlying areas of Djibouti town and also within the smaller towns of Ali Sabieh, Dikhil and Arta. Three sizes of overhead conductor are used for the medium voltage circuits; 34 mm2, 117 mm2 and 148 mm2 AAAC conductors. These are strung on a combination of wood and steel poles. 4.6.3 Distribution Substations Distribution substations are spaced at intervals as required along the length of each MV feeder. The majority of distribution substations are the ‘cabine’ type, where the equipment is housed within a building. There are however a number of pole mounted transformers (‘H61’ type), particularly in the outlying villages such as Arta. Least Cost Electricity Master Plan (Volume 1) Final Report November 2009 Page 4.14 SECTION 4 EXISTING and Committed Power System For the ‘cabine’ type substations, the equipment comprises; MV switchgear, one or two 20/0.4 kV transformers and an LV fuse-panel. The MV switchgear is in a typical ring-main arrangement comprising (in the majority of substations) two feeder panels and a transformer panel. The transformer panel takes the form of a fuse-switch, while the feeder panels are simply switches without protection. In some cases the substation includes both an EdD transformer and a customer transformer. The transformers may be rated between 250 kVA and 1000 kVA although the majority on the system are 400 kVA and 630 kVA. The pole mounted substations comprise a transformer, hung on the pole and directly connected to the MV feeder. Connected to the LV side of the transformer is a set of fuses and a single LV feeder. The rating of the pole mounted transformers ranges from 40 – 160 kVA. 4.6.4 LV Feeders Originating from each distribution substation are two-four LV feeders each around 400 m long. The majority of LV feeders comprise overhead aerial bundled conductors (ABC 3x70mm2 + 54mm2). There are however LV underground cable feeders in central Djibouti town (3x150mm2 + 70mm2). Table 4.6 shows the quantities and unit costs of distribution system equipment in Djibouti town. Table 4.6: Distribution System Unit Costs and Quantities (Djibouti town) 2009 Unit 2009 Unit 2009 Cost Cost Quantity (DJF 000's) (US$ 000's) 20 kV Feeder Circuit Breakers Boulaos 16,019 91 10 Marabout 16,019 91 10 20 kV Circuits Underground Cable (150mm2 + 185mm2) 9,200 52 140 OHL (148 mm2) 8,000 45 51 OHL (34 mm2) 4,500 25 77 Pole Mounted 50 kVA 835 5 1 100 kVA 1,051 6 6 160 kVA 1,241 7 7 Ground Mounted 250 kVA 13,184 74 22 400 kVA 13,435 76 113 630 kVA 13,894 78 98 800 kVA 14,161 80 12 1000 kVA 14,552 82 10 LV Underground Cable (3*150mm2+70mm2) 7,000 40 100 LV ABC (3*70mm2+54mm2) 4,350 25 248 Least Cost Electricity Master Plan (Volume 1) Final Report November 2009 Page 4.15 SECTION 4 EXISTING and Committed Power System 4.7 Summary The existing and committed plant detailed in this Section of the report can be summarised as follows: • Generation: - Demand is currently served from two main power stations. The Boulaos power station comprises 13 medium speed diesel generating units and has an installed capacity of 84.4 MW (sent out). The Marabout power station comprises of 6 medium speed diesel generating units and has an installed capacity of 14.4 MW (sent out). Therefore, total installed capacity in Djibouti is 98.8 MW (sent out). - Additional capacity will be added to Boulaos power station through a series of refurbishments and installation of new plant. By the start of 2014, it is assumed that there will be 108.2 MW of sent out capacity available from Boulaos. Total sent out system capacity would therefore reach 122.6 MW by 2014. - Our supply demand balance analysis indicates that the balance between demand and supply is positive for the years 2008 to 2012. When allowance is made for the need to carry reserve capacity, however, (say 20 per cent) to cover maintenance outages, the balance is positive for the years 2009 to 2011 only. This analysis suggests that even though the interconnector is assumed to begin operations in 2011, EdD will still have to invest in generating capacity around this time. - There are outlying power plants in the towns of Dikhil and Ali Sabiah in the South and in Tadjoura and Obock in the North of the country. The total installed capacity in the South is approximately 3 MW. The total installed capacity in the North is also 3 MW. • Transmission: - The EdD power system is split into three regions; Djibouti town, Davison South and Division North. - The three regions are not presently interconnected, although the south will be interconnected with Djibouti town as part of the interconnector project. Least Cost Electricity Master Plan (Volume 1) Final Report November 2009 Page 4.16 SECTION 4 EXISTING and Committed Power System - In Djibouti town, the two main power plants are electrically interconnected at 63 kV. The smaller generating units at Boulaos are connected via generator transformers to the 20 kV busbar, while the larger units are connected via generator transformers to the 63 kV busbar at Boulaos. The Marabout generators are connected via a single generator transformer to the Marabout 63 kV busbar. The 63 kV interconnector between Boulaos and Marabout comprises a 400 mm2 AI cable and has a summer thermal rating of 36 MVA. The 63/20 kV transformers at Boulaos and Marabout are also rated at 36 MVA. - In Division South, the power stations at Dikhil and Ali Sabieh are connected at 20 kV. - In Division North, the power stations at Tadjoura and Obock are not connected. - A double circuit interconnector is currently under construction between Ethiopia and Djibouti and is set to be complete by April 2010. The interconnector will provide Djibouti with a low cost supply of energy at certain times of the day, with a restriction on the maximum amount of energy that will flow from Ethiopia and Djibouti. - The interconnector will play a significant role in shaping the development of the electricity sector in Djibouti into the future. The size of potential energy flows available from Ethiopia, the relatively small demand in Djibouti and the high level of availability of the interconnector means that some of the generating plant in Djibouti is unlikely to be required to run for significant periods of time and therefore the least cost plan is likely to be skewed towards thermal plant with low capital costs and small sized base load generation. - Following the completion of the interconnector and associated infrastructure, there will be a significantly larger transmission system in the country, connecting the towns of Dikhil and Ali Sabiah the Djibouti town network. • Distribution: - In Djibouti, 20 kV is used for medium voltage (MV) and 400/230 V is used for low voltage (LV). Least Cost Electricity Master Plan (Volume 1) Final Report November 2009 Page 4.17 SECTION 4 EXISTING and Committed Power System - The distribution network comprises MV feeders supplying 20/0.4 kV distribution substations. Several LV feeders emanate from each distribution substation which in turn provides supply to customers. Least Cost Electricity Master Plan (Volume 1) Final Report November 2009 Page 4.18 SECTION 5 FUTURE DEVELOPMENT OPTIONS SECTION 5 FUTURE DEVELOPMENT OPTIONS 5 FUTURE DEVELOPMENT OPTIONS There are no known oil or gas resources in Djibouti, hence, at present, the Republic of Djibouti imports 85 per cent of its energy needs as hydrocarbon products, and produces only 15 per cent through indigenous wood and charcoal. The entire EdD electricity production is based on thermal powered plants fired on imported HFO and gasoil, resulting in high electricity costs. In the second quarter of 2010 a new interconnector between Ethiopia and Djibouti is expected to be commissioned. This interconnector will go a long way to reducing Djibouti’s reliance on the imported fuel products and thus reducing production costs. The amount of energy provided by the interconnector however is limited, both in terms of total energy that can be supplied from Ethiopia to Djibouti as well as the time of day and season. As a consequence, there is a commitment by the Government of Djibouti to continue to invest in generating plant such that EdD will be able to meet demand as if the interconnector were not built. Confronted with an ever increasing petroleum demand and a rapid growth in urbanisation; the Government of Djibouti (GoD) have also decided to direct the country’s energy policy towards the development of renewable energy resources (geothermal and wind in particular). In order to determine the least-cost generation solution for Djibouti into the future, it is necessary to understand how various plant types and technologies (both from the renewable sector and the thermal sector) compare in relation to each other and how this fits with their intended use on the system. In this Section we highlight the main energy resources that could potentially be used for generating electricity in Djibouti with a view to evaluating these resources within the context of the least cost electricity master plan for Djibouti. Where applicable, and for the purposes of this study, a number of “candidate� new plant options have been considered. Within these broad category types there are different set sizes available. These 10 options are discussed below . It should be noted that in the last two or more years the worldwide power generation market has experienced an upsurge in demand and this, together with increases in commodity prices, has resulted in significant and ongoing increases in the prices of new generating plant. This makes it difficult to estimate with confidence the ‘overnight’ cost of such plant. 10 Unless otherwise indicated, the estimates of capital cost per kW (also known as ‘specific capital cost’) given in this Section are based on the ‘sent out’ rating of the plant at site, not the ISO rating. All capital cost estimates exclude interest during construction (IDC). Least Cost Electricity Master Plan (Volume 1) Final Report November 2009 Page 5.1 SECTION 5 FUTURE DEVELOPMENT OPTIONS 5.1 Renewable resources 5.1.1 Geothermal Geothermal power is defined as the power extracted from heat that is stored in the earth. This geothermal energy source originates from the original formation of the planet, from the solar energy absorbed from the sun at the earths’ surface and from the radioactive decay of minerals. The Republic of Djibouti is located on a triple junction of the Red Sea, Gulf of Aden and East African rifts giving rise to a geodynamical situation that gives the area a remarkable position for the development of the geothermal energy. In spite of the significant geothermal studies and the deep drilling explorations conducted since 1970 on several geothermal prospect zones, geothermal energy in Djibouti is still to be developed. Effectively, the underground heat sources are expressed on the surface by numerous hot springs and fumaroles mainly distributed on the Western part of the country and along the Gulf of Tadjoura ridge. Surface explorations have identified the geothermal provinces of the country. The Asal rift, Nord Goubhet, Hanle and Gaggade areas have been identified as priority sites for geothermal explorations. Appendix C in Volume 2 of this report provides a review of studies conducted on the potential use of geothermal energy for electricity production in Djibouti. Details of potential development alternatives (and the associated economics) are also discussed in detail in Appendix C. These studies suggest that there is potential for geothermal power plants in Djibouti, and sizes of up to 30 MW have been mooted. Although there is potential for geothermal power development in Djibouti, further studies, drilling and exploration are required in order to determine if this resource is suitable for large-scale power generation. As such the development of this resource can not be considered as part of the least cost planting programme and must be considered as a sensitivity to the reference case only. If the resource were available for large-scale power generation, any geothermal power plant would have to be operated as a base load plant with continuous operation as the output of geothermal cannot be controlled in the same way as, say, a conventional thermal power plant. The minimum load in Djibouti is low and therefore we believe that the introduction of a 30 MW geothermal units (following the commissioning of the first unit) would be constrained, requiring a higher growth rate for minimum load. For this reason, and considering the uncertainties of the resource itself, our sensitivity studies have considered new geothermal units sized at 20 MW. Table 5.1 and Table 5.2 below present the cost and operating parameters for two generic geothermal power plants (20 MW and 6 MW). Whilst generic data is provided for a 6 MW unit as well as a 20 MW Least Cost Electricity Master Plan (Volume 1) Final Report November 2009 Page 5.2 SECTION 5 FUTURE DEVELOPMENT OPTIONS unit, it should be noted that a 6 MW unit would only be feasible if the larger 20 MW geothermal plant already existed (see Appendix C in Volume 2 of this report). Table 5.1: Operational parameters for generic geothermal candidate plant (20 MW & 6 MW) ISO Available Build Life of Unit FOR POR Availability Rating Capacity Period Plant (MW) (MW so) (Years) (%) (Days/yr) (%) (Years) 20 MW Geothermal 22.0 20.0 3 3.0% 26 90.1% 25 6 MW Geothermal 6.5 6.0 2 3.0% 26 90.1% 25 Table 5.2: Cost and performance parameters for generic geothermal plant (20 MW & 6 MW) Capital Net Fixed Variable Unit Fuel Type Heat Rate Cost Efficiency O&M Cost O&M Cost ($/kW) (KJ/kWh) (%) ($/kW/yr) ($/MWh) 20 MW Geothermal Geothermal 4620 0 N/A 0.00 13.60 6 MW Geothermal Geothermal 4333 0 N/A 0.00 12.83 With regard to the implementation programme for a medium sized geothermal development, we have considered the following: • Lead time to define, finance and prepare for implementation of the exploration drilling programme (1 to 3 years) • Execution of the exploration drilling programme (1 ½ years) • Lead time to define, finance, and prepare for implementation of the production drillings and construction of plant (1 to 3 years) • Execution of the programme (3 years) Consequently, the earliest date of commercial operation of a geothermal development would be approximately 6 ½ years. The latest date of commercial operation of a geothermal development would be approximately 10 years. Based on the information available from the studies previously carried out, it is assumed for the sensitivity scenarios of this study that the first geothermal power plant would be available in 2016, and further units could be built at 3 year intervals. Least Cost Electricity Master Plan (Volume 1) Final Report November 2009 Page 5.3 SECTION 5 FUTURE DEVELOPMENT OPTIONS A comment on geothermal development Although there is undoubted potential for geothermal development in Djibouti it must be highlighted that over 30 years of research and analysis has been undertaken on this subject and, at present, there is still not a single MW of installed geothermal capacity. Recent studies conducted by ISOR have demonstrated that past investigations were not carried out according to rational and risk mitigated plans, and therefore led to poor results. To mitigate the risks in geothermal development, the steps that are of paramount importance (with go/no go decisions at the end of each) are: (i) geo-scientific reviews leading to (through integration of various complementary disciplines) the definition of a conceptual model, (ii) exploration drillings (small diameter, not at full depth, so much cheaper than production drillings), and finally, (iii) production drillings and construction of steam field, power station and inter-connection facilities. A further complication to geothermal development in Djibouti is the presence of large risks which are specific to small, developing countries (geographic isolation, low domestic technical capacity, and governance). These risks can affect the institutional arrangements necessary to mobilise capital and other resources. Although we acknowledge the potential for geothermal development in Djibouti, it should be noted that the development of this resource is not straight forward and significant time, effort and cooperation will be required in order to harness the potential of this natural resource. As such, the development of geothermal power is considered as a sensitivity to the base case only. It should also be noted that geothermal power could play a role in remote areas where electricity is required near potential geothermal developments and thus minimising the need for transmission. 5.1.2 Wind Since the turn of the century, the GoD have been interested in the potential of wind energy resources in Djibouti. Since garnering this interest, several reports and studies have been undertaken to identify this potential. Appendix C in Volume 2 of this report provides a review of the wind resources of Djibouti based on these previously published documents. The conclusion of this analysis is that there is potential for electricity production using wind in Djibouti. Although there is potential for wind power development in Djibouti, further evidence and exploration of the availability of this resource is required in order to determine if this resources is suitable for large- scale power generation. As such, the development of this resource can not be considered as part of Least Cost Electricity Master Plan (Volume 1) Final Report November 2009 Page 5.4 SECTION 5 FUTURE DEVELOPMENT OPTIONS the least cost planting programme and must be considered as a sensitivity to the reference case only. Five potential key locations have been identified as the optimal wind resource locations for electricity production. These locations are: • Ghoubet • Ali Sabiah • Djibouti Ville • Egralyta • Bada Wein Table 5.3 and Table 5.4 below present the cost and operating parameters for the various wind candidate options. It is assumed that a wind turbine would only be available from 2015 onwards. Table 5.3: Operating parameters for wind candidate plant ISO Available Build Life of Unit FOR POR Availability Rating Capacity Period Plant (MW) (MW so) (Years) (%) (Days/yr) (%) (Years) Ghoubet 10.2 10.2 1 3.0% 15 93.0% 20 Ali Sabieh 10.2 10.2 1 3.0% 15 93.0% 20 Djibouti Ville 10.2 10.2 1 3.0% 15 93.0% 20 Egralyta 10.2 10.2 1 3.0% 15 93.0% 20 Bada Wein 5.1 5.1 1 3.0% 15 93.0% 20 Table 5.4: Cost and performance parameters for wind candidate plant Capital Net Fixed Variable Unit Fuel Type Heat Rate Cost Efficiency O&M Cost O&M Cost ($/kW) (KJ/kWh) (%) ($/kW/yr) ($/MWh) Ghoubet Wind 1869 0 N/A 0.00 14.00 Ali Sabieh Wind 1528 0 N/A 0.00 14.00 Djibouti Ville Wind 1415 0 N/A 0.00 16.00 Egralyta Wind 1869 0 N/A 0.00 18.00 Bada Wein Wind 2426 0 N/A 0.00 27.00 It should be noted that it is hard to predict (particularly on a short term basis) when it is going to be windy, how strong the wind will be and how long it will last for. As a result, wind capacity cannot be treated as firm capacity and may only form part of the least cost development plan if its introduction can offer fuel cost savings which offset the costs of building firm capacity to carry out the same role. Least Cost Electricity Master Plan (Volume 1) Final Report November 2009 Page 5.5 SECTION 5 FUTURE DEVELOPMENT OPTIONS It should also be noted that wind power could play a role in remote areas where electricity is required near potential wind turbine sites and thus minimising any transmission requirement. 5.1.3 Solar There have not been any specific studies carried out regarding the potential for solar powered electricity production in Djibouti. Needless to say however, there is certainly enough sunlight in Djibouti for this to be a viable, although perhaps not economic option for future electricity production. Table 5.5 and Table 5.6 present the cost and performance parameters for a generic 20 MW solar concentrating power plant. Table 5.5: Operating parameters for solar candidate plant ISO Available Build Life of Unit FOR POR Availability Rating Capacity Period Plant (MW) (MW so) (Years) (%) (Days/yr) (%) (Years) Solar 22.0 20.0 4 5.0% 42 84.1% 25 Table 5.6: Cost and performance parameters for solar candidate plant Capital Net Fixed Variable Unit Fuel Type Heat Rate Cost Efficiency O&M Cost O&M Cost ($/kW) (KJ/kWh) (%) ($/kW/yr) ($/MWh) Solar Solar 5000 0 N/A 36.00 7.00 Due to the nature of solar powered electricity production, solar capacity cannot be treated as firm capacity and may only form part of the least cost development plan if its introduction can offer fuel cost savings which offset the costs of building firm capacity to carry out the same role. It should also be noted that smaller-scale solar panel technology could play a role in remote areas where small amounts of electricity are required without significant transmission requirements. 5.2 Thermal resources 5.2.1 Oil-fired The entire EdD electricity production is based on thermal powered plants fired on imported HFO and gasoil. To consider oil-fired candidate plant a fuel price forecast for refined oil products must be developed. This forecast is presented below. Least Cost Electricity Master Plan (Volume 1) Final Report November 2009 Page 5.6 SECTION 5 FUTURE DEVELOPMENT OPTIONS Fuel price forecast The base commodity for evaluating the economic cost of refined fuels used in the power sector is crude oil. Oil prices have been very volatile over the last few years. After a period of relatively stable prices, 1998 saw crude prices fall below USD 10 per barrel (bbl) at the start of 1999. Following the depletion of surplus oil in the trading markets (which was primarily due to cuts in OPEC production quotas) prices rose during 1999 eventually to exceed USD 30 per barrel by the third quarter of 2000. Prices fluctuated between the long-term averages of USD 20-30 per barrel to the end of 2002. Continued high demand growth for oil across the world (particularly from the US, China and India) saw prices rise from about USD 30 per barrel at the start of 2003 to USD 140 per barrel by 2008. Oil prices have recently dropped to around USD 60 per barrel. This trend is illustrated in Figure 5.1. Figure 5.1: Historical Average Monthly Data (1988-2008) (real prices) Projections of long term oil prices are performed regularly by third party oil market analyst groups and can show significant variation of results. In this report we use the most recent oil price forecast (2009 to 2030) produced by the EIA (Energy Information Administration) in their Annual Energy Outlook 11 2009 report (AEO2009), updated to reflect 2008 prices. The prices in the AEO2009 takes account of the recent and sharp fall in the oil price at the end of 2008 and predicts that world prices for imported crude oil will rise sharply to approximately USD 100 per barrel by 2015, continuing to rise to USD 130 per barrel by 2030. It has been assumed by this consultant that the oil price will remain constant (in real terms) between 2030 and 2035. Table 5.7 and Figure 5.2 present the reference world imported crude oil price projections (corrected to 2006 prices) over the period 2009 to 2035. 11 Energy Information Administration, Annual Energy Outlook 2009, (http://www.eia.doe.gov/oiaf/aeo/index.html), (April 2009). Least Cost Electricity Master Plan (Volume 1) Final Report November 2009 Page 5.7 SECTION 5 FUTURE DEVELOPMENT OPTIONS Table 5.7: World oil price projections for imported crude oil (2008 – 2035) Imported Year Crude Oil (USD / bbl) 2008 100.2 2009 40.2 2010 50.9 2011 64.4 2012 74.9 2013 84.1 2014 92.0 2015 100.5 2016 105.8 2017 110.3 2018 114.8 2019 117.1 2020 118.9 2021 120.3 2022 121.0 2023 122.2 2024 122.8 2025 120.5 2026 122.0 2027 123.7 2028 125.9 2029 126.6 2030 129.1 2031 129.1 2032 129.1 2033 129.1 2034 129.1 2035 129.1 Figure 5.2 : World oil price projections for imported crude oil 140 120 Imported Crude Oil Price (USD / bbl) 100 80 60 40 20 0 2008 2009 2010 2011 2012 2013 2014 2015 2016 2017 2018 2019 2020 2021 2022 2023 2024 2025 2026 2027 2028 2029 2030 2031 2032 2033 2034 2035 Least Cost Electricity Master Plan (Volume 1) Final Report November 2009 Page 5.8 SECTION 5 FUTURE DEVELOPMENT OPTIONS The imported crude oil price forecast portrayed in the preceding sub-section has been adjusted upwards to take into account carriage, insurance and freight (CIF). The level of adjustment to take into account CIF is assumed to be 8 per cent on the basis of receipted documentation provided by EdD regarding payments to Shell Djibouti for the period between June and July 2006. The plants currently in operation in Djibouti burn HFO or Gasoil. On average, the cost ratio of HFO to crude oil is approximately 80 per cent and we have assumed that the ratio of gas oil to crude oil is 150 12 per cent and the ratio of lube oil to crude oil is 500 per cent . Fuel price forecasts for all crude oil derivatives to be used in the generation planning model are as shown in Table 5.8 below. Table 5.8 : Fuel price forecasts for crude oil products (2006 prices) Fuel Price (USD / bbl) Year Crude Oil HFO Gas Oil Lub Oil Diesel Oil 2008 108.2 86.5 162.3 540.9 173.1 2009 43.5 34.8 65.2 217.3 69.5 2010 54.9 44.0 82.4 274.7 87.9 2011 69.5 55.6 104.3 347.6 111.2 2012 80.9 64.7 121.4 404.6 129.5 2013 90.8 72.7 136.2 454.1 145.3 2014 99.4 79.5 149.1 496.9 159.0 2015 108.5 86.8 162.8 542.6 173.6 2016 114.3 91.4 171.5 571.5 182.9 2017 119.1 95.3 178.7 595.6 190.6 2018 124.0 99.2 185.9 619.8 198.3 2019 126.5 101.2 189.7 632.4 202.4 2020 128.4 102.7 192.6 642.0 205.5 2021 129.9 103.9 194.8 649.4 207.8 2022 130.6 104.5 195.9 653.1 209.0 2023 132.0 105.6 198.0 660.0 211.2 2024 132.7 106.1 199.0 663.3 212.2 2025 130.2 104.1 195.2 650.8 208.3 2026 131.7 105.4 197.6 658.7 210.8 2027 133.6 106.9 200.5 668.2 213.8 2028 136.0 108.8 203.9 679.8 217.5 2029 136.7 109.4 205.1 683.7 218.8 2030 139.5 111.6 209.2 697.3 223.1 2031 139.5 111.6 209.2 697.3 223.1 2032 139.5 111.6 209.2 697.3 223.1 2033 139.5 111.6 209.2 697.3 223.1 2034 139.5 111.6 209.2 697.3 223.1 2035 139.5 111.6 209.2 697.3 223.1 12 All ratios refer to the crude oil price inclusive of the CIF adjustment. Least Cost Electricity Master Plan (Volume 1) Final Report November 2009 Page 5.9 SECTION 5 FUTURE DEVELOPMENT OPTIONS Oil-fired candidate plant options There are three main oil-fired candidate plant options for Djibouti. These options are: • Diesel turbine fired on HFO • Gas turbines (OCGT) fired on gasoil • Oil-fired steam Whilst both the HFO-fired diesel plant and the gasoil-fired OCGT plant are viable for Djibouti, an oil- fired steam plant cannot be considered for this study. Oil-fired steam plants are typically large in size and would be too large for the Djibouti system. Table 5.9 and Table 5.10 present the cost and performance parameters for a generic HFO-fired diesel and generic gasoil-fired OCGT power plant considered in our studies for the derivation of the least cost generation expansion plan for Djibouti. Table 5.9: Operating parameters for oil-fired candidate plant ISO Available Build Life of Unit FOR POR Availability Rating Capacity Period Plant (MW) (MW so) (Years) (%) (Days/yr) (%) (Years) Diesel HFO Plant (750 RPM) 7.0 7.0 3 3.0% 30 89.0% 25 Diesel HFO Plant (600 RPM) 12.0 12.0 3 3.0% 30 89.0% 25 GT - MAN 1304-10 9.3 7.8 2 2.0% 25 91.3% 20 GT - KHI 180 D 18.0 15.0 2 2.0% 25 91.3% 20 Table 5.10: Cost and performance parameters for oil-fired candidate plant Capital Net Fixed Variable Unit Fuel Type Heat Rate Cost Efficiency O&M Cost O&M Cost ($/kW) (KJ/kWh) (%) ($/kW/yr) ($/MWh) Diesel HFO Plant (750 RPM) HFO/Gas Oil 1318 9500 37.9% 39.54 8.71 Diesel HFO Plant (600 RPM) HFO/Gas Oil 1445 9000 40.0% 43.35 8.71 GT - MAN 1304-10 Gas Oil 993 13300 27.1% 31.65 8.71 GT - KHI 180 D Gas Oil 1055 12500 28.8% 29.79 8.71 5.2.2 Gas-fired Djibouti is not known to have any natural reserves of gas and as such, no gas infrastructure exists in Djibouti. Neighbouring Yemen however, intend to start producing natural gas in the not so distant future and this has given rise to the hypothetical possibility of Djibouti tapping into the gas resource Least Cost Electricity Master Plan (Volume 1) Final Report November 2009 Page 5.10 SECTION 5 FUTURE DEVELOPMENT OPTIONS via a sub-sea pipeline which would allow imported gas from the Yemen to be used in power stations in Djibouti. It is understood however, that at present there is insufficient reserves of gas in Yemen to allow for a supply to any of its neighbouring countries within the 25 year horizon of this study. Even if Yemen were to possess sufficient gas reserves for a supply to Djibouti the project is certain not to be commercially viable. This presumption is made on the basis that the demand for gas in Djibouti would be very small, particularly given that the presence of the interconnector between Ethiopia and Djibouti is likely to result in reduced thermal power plant operations. A low demand for gas combined with the high costs of building a 30 km sub-marine pipeline across one of the world’s busiest shipping lanes in a location with strong currents and active seismic activity (plus a further 250km of over land pipeline) would deem any such proposal a non-starter. 5.2.3 Coal-fired The peak demand in Djibouti in 2008 was approximately 57 MW. Economically viable coal-fired power plants are typically large - in the region of 100 MW or greater. Given the disparity between the size of economically viable coal plants and the demand for electricity in Djibouti, a coal-fired power plant would not be a feasible development option in the foreseeable future. It should also be noted that Djibouti does have a large port which may be suitable for receiving coal, but at present there are no reception facilities to offload, treat or store the coal. If a new coal-fired power plant were to be developed in Djibouti, these facilities would have to be built. Whilst this option is technically viable, the cost of investing in these facilities would not be commercially viable given the relative small scale coal plant that would have to be developed to satisfy the relatively small demand. As a result, coal-fired candidate power plants have not been considered any further in this study. 5.2.4 Nuclear Nuclear power plants are not available in a size suited to Djibouti demand and therefore this option is not considered for this study. 5.3 Screening curve analysis Before commencing the least cost generation expansion planning analysis (in Section 6) we have employed a screening curve approach to form an idea of the likely competitiveness and role for the principal power plant options. The principal power plant options are: • HFO-fired diesel turbine Least Cost Electricity Master Plan (Volume 1) Final Report November 2009 Page 5.11 SECTION 5 FUTURE DEVELOPMENT OPTIONS • Gas oil-fired OCGT Geothermal (assuming the resource is confirmed as being available for large scale power generation)Although screening curves by themselves cannot determine the appropriate timing of plant, nor whether the system needs base load, mid-merit or peaking plant, they do suggest the relative competitiveness of different technologies for certain duties within the merit order. The plant 13 data used in these curves are presented in Section 5.1 and Section 5.2 of this report . The screening curves for the principal candidate plant for this study are shown in Figure 5.3 below. The intercept on the y-axis shows the annual expenditure (on a per kW net basis) of fixed costs (including capital cost and the fixed component of the operation and maintenance costs). These costs are incurred whether or not the plant is operated. The slope of each line represents the expenditure on variable costs (mainly fuel costs) against plant utilisation. The total cost (fixed and variable) depends on the capacity factor assumed for the plant. Figure 5.3: Screening curve (firm capacity options only) Diesel HFO Plant (750  RPM) Diesel HFO Plant (600  RPM) GT � MAN 1304�10 GT � KHI 180  D 20 MW Geothermal 6MW Geothermal 1800 1600 1400 1200 1000 800 600 400 200 0 0% 5% 10% 15% 20% 25% 30% 35% 40% 45% 50% 55% 60% 65% 70% 75% 80% 85% 90% 95% 100% Figure 5.3 indicates that (if the resources are developed correctly) a 20 MW geothermal power plant is the least cost option for capacity factors greater than 85 per cent (and greater than 75 per cent for a 13 It should be noted that the screening curves are based on capital cost data inclusive of IDC. Least Cost Electricity Master Plan (Volume 1) Final Report November 2009 Page 5.12 SECTION 5 FUTURE DEVELOPMENT OPTIONS 6 MW plant). The figure also shows that the diesel generators (fired on HFO) are the cheapest option at all other realistic operating capacity factors (5 per cent to 75-85 per cent). It should be noted however, that the size of the potential energy flows available from Ethiopia over the interconnector and the relatively small demand in Djibouti means that some of the additional generating plant to be built in Djibouti is unlikely to be utilised for significant periods of time (if at all). In this sense, the least cost generation plan is likely to be skewed towards thermal plant with the lowest capital costs. As such we would expect a few OCGTs to be selected. To summarise, if the geothermal resource is not available, we would expect the system to be a mix of import energy for peak lopping, diesel plant to be selected to operate at all operational capacity factors (including base load) and OCGTs to take the role of stand-by capacity. If, however, the geothermal resource is developed, we would expect the system to be a mix of geothermal (for base load operation), import energy for peak lopping, fewer diesel plant operating at lower capacity factors and OCGTs to take the role of stand-by capacity. A comment on OCGT selection The screening curve discussion detailed above strives to identify the best candidate plant options to add to the capacity of the system (and meet the LoLE) based on self sufficiency. Recognising that the presence of the interconnector combined with the seasonality of the load will result in units that are idle for long periods OCGT seems to be the ideal choice whereby it has a lower capital cost and will not be used long enough for expensive diesel units to offset the possible savings. That said, it is recognised in our analysis that OCGT units are at a disadvantage (compared to diesel units) in higher temperatures as well as EdD’s lack of experience with OCGT’s in the past.. As such, similar sized diesel units, whilst having a higher capital cost (and may be diesel- fired due to their relatively small size), could replace OCGT in our proceeding generation planning analysis (these are also flexible in the sense that in case of base unit unplanned outages, diesel can be used as base load generators without fuel cost being excessive). This option would not represent the least cost solution but the extra capital cost for diesel units could be mitigated by EdD’s familiarity with such units. The above mitigation costs can not be quantified and therefore the decision would lie purely with EdD. As stated in the previous sub-Section, neither wind nor solar power offers firm capacity to the system. These technologies may still form part of the least cost development plan if there introduction can offer fuel cost savings which offset the costs of building firm capacity to carry out the same role. Least Cost Electricity Master Plan (Volume 1) Final Report November 2009 Page 5.13 SECTION 5 FUTURE DEVELOPMENT OPTIONS Figure 5.4 shows the same screening curve as above but with the inclusion of both the wind and solar candidate plant options at their respective operational capacity factors. Figure 5.4: Screening Curve (including non-firm capacity) Diesel HFO Plant (750  RPM) Diesel HFO Plant (600  RPM) GT � MAN 1304�10 GT � KHI 180  D Ghoubet Ali Sabieh Djibouti Ville Egralyta Bada Wein 20 MW Geothermal 6MW Geothermal Generic Solar Concentrating Power Plant 1800 1600 1400 1200 1000 800 600 400 200 0 0% 5% 10% 15% 20% 25% 30% 35% 40% 45% 50% 55% 60% 65% 70% 75% 80% 85% 90% 95% 100% It is apparent from Figure 5.4 that a solar-concentrating power plant will not form part of the least cost plan. This is because the technology behind solar power is not mature and therefore the capital cost of a large solar-concentrating power plant still outweighs the capital and operating costs of other competing technologies. Figure 5.4 does show however, that there is potential for wind power to provide fuel cost savings as Ghoubet, Ali Sabiah, Djibouti-ville and Egralyta are all shown to be the least cost option at all reasonable operational capacity factors below 45 per cent. The Bada Wein wind plant is more expensive than the other wind options and the diesel plant up to a capacity factor of 35 per cent. To summarise, we will analyse a least cost development option consisting of diesel plant and OCGT together with varying levels of imports from the interconnector. We will also carry out additional analysis to determine the role that geothermal and wind power may be able to play into the future if these resources are (i) confirmed, and, (ii) developed. Least Cost Electricity Master Plan (Volume 1) Final Report November 2009 Page 5.14 SECTION 5 FUTURE DEVELOPMENT OPTIONS 5.4 Summary The future development options for Djibouti can be summarised as follows: • The Ethiopia - Djibouti interconnector is set to be in operation by the end of the first quarter of 2010 and will go along way to reducing Djibouti’s reliance on imported fuel products and thus reducing production costs. • The amount of energy that is likely to be available is in the range between 180 GWh and 700 GWh per annum. In our study we have assumed a commercial operation date of January 2011 and propose to analyse the impact of the interconnector assuming imports of 180 GWh and 700 GWh per year. • There is a commitment by the GoD to continue to invest in generating plant such that EdD will be able to meet demand as if the interconnector were not built. • Confronted with an ever increasing petroleum demand and a rapid growth in urbanisation; the GoD have also decided to direct the country’s energy policy towards the development of renewable energy resources (geothermal and wind in particular). • Although there is potential for geothermal power development in Djibouti, further studies, drilling and exploration are required in order to determine if this resource is suitable for large-scale power generation. As such the development of this resource can only be considered as a sensitivity to the reference case. • Our analysis also indicates that there is potential for wind power development in Djibouti. Again however, further evidence and exploration of the availability of this resource is required in order to determine if this resource is suitable for large-scale power generation. As such, the development of this resource is considered as a sensitivity to the reference case. • Solar power development is considered as a viable power generation option. • HFO fired diesel plant is expected to form part of the least cost generation expansion plan and we therefore consider this technology in our base case analysis. • OCGT fired on gasoil is also expected to form part of the least cost generation expansion plan and we therefore consider this technology in our base case analysis. • Oil-fired steam plant is not considered in our studies as these plants are typically large in size and therefore would be too large for the current system. • Gas-fired power plant is also discounted from this study. This is because Djibouti does not have any natural gas resources and a mooted pipeline connection between Yemen and Djibouti would not be commercially viable within the horizon of this study. Least Cost Electricity Master Plan (Volume 1) Final Report November 2009 Page 5.15 SECTION 5 FUTURE DEVELOPMENT OPTIONS • Coal plant is discounted from this study. This decision is based on the economic viability of small coal-fired power plants and the relatively small size of the Djibouti system. • Nuclear plant is also discounted from this study on the basis that nuclear power plants are not available in a size suited to the Djibouti demand. • The oil price forecast assumed for this study is based on the 2009 AEO forecast (developed by EIA). This forecast suggests oil will fall to US$ 43.5 per barrel in 2009, gradually rising to almost US$ 100 per barrel by 2014, with continued price rises through to 2030, where by the price is forecasted to reach US$ 140 per barrel. • Our principal screening curve analysis indicates that if geothermal resources are not available into the future, we would expect the system to be a mix of import energy for peak lopping, diesel plant to operate at all operational capacity factors (including base load) and OCGTs to take the role of stand-by capacity. If, however, geothermal resources are developed, we would expect the system to be a mix of geothermal (for base load operation), import energy for peak lopping, fewer diesel plant operating at lower capacity factors and OCGTs to take the role of stand-by capacity. • The inclusion of solar plant in our screening curves indicates that solar-concentrating power (at this present moment) is more expensive than other technologies with whom it would compete in the merit order and this technology is therefore discounted from this study. • Our screening curve analysis also indicates however, that there is potential for wind power (if developed) to provide fuel cost savings as the sites at Ghoubet, Ali Sabiah, Djibouti-ville and Egralyta are all shown to be the least cost option at all reasonable operational capacity factors below 45 per cent. Least Cost Electricity Master Plan (Volume 1) Final Report November 2009 Page 5.16 SECTION 6 GENERATION PLANNING SECTION 6 GENERATION PLANNING 6 GENERATION PLANNING 6.1 Introduction Generation expansion planning has been undertaken for this study within the context of determining the least cost electricity master plan for Djibouti. To this end, a considerable number of scenarios have been selected for analysis with a view to encompassing a wide range of eventualities that cover the uncertainties inherent in the demand growth projections as well as uncertainties relating to the level of imports available over the interconnector and the potential development of renewable resources such as geothermal and wind power. In this Section we set out and discuss the details of the generation planning analysis. We start however, by outlining the methodology adopted in determining the least cost generation expansion plan. 6.2 General methodology We are aware that EdD employs a reliability criterion equivalent to a reserve margin of 50 per cent of the peak demand plus the size of the largest unit on the system. We believe this is high by international standards, even for small systems. The implementation of probabilistic criteria such as LOLE may be more appropriate and our review presented in Appendix F has identified a loss of load expectation of 24 hours per year (equivalent to a loss of load probability of 0.742 per cent) as a suitable reliability criterion for the least cost plan. Traditionally we would use a probabilistic generation programme to determine the least cost development plan for Djibouti in preference to deterministic methods. WASP or A/Splan are highly sophisticated generation planning programs that are in wide spread use throughout the World. In particular they utilise a dynamic programming optimisation technique that ensures that the least cost programme is drawn from a consideration of all possible options. Recognising that the PPA for the interconnector represents an energy contract and that EdD is expected to continue investing in new generating plant such that EdD is able to meet their own system peak demand without the reliance on power transfers over the interconnector, the generation planning programmes WASP or A/Splan can be considered as ideal tools for the derivation of the long-term least cost generation plan for Djibouti. To this end, we have used our generation planning software A/Splan to identify the least cost plan for any scenario representing a particular set of assumptions. The derived expansion plan can then be simulated to obtain the various cost streams for NPV calculation purposes, and the results can be displayed in tabular and/or graphical format. Least Cost Electricity Master Plan (Volume 1) Final Report November 2009 Page 6.1 SECTION 6 GENERATION PLANNING Recognising the need to maintain chronology of the daily variation in the level of load demand for simulation purposes, such that the impact of seasonality on the availability of energy imports over the interconnector is accounted for, we will use our in-house probabilistic generation planning simulation model (GPSM) based on statistical premise. This model uses the highly sophisticated ‘@Risk’ add-in to excel which is defined as a risk analysis, problem-solving tool that uses Monte Carlo simulation techniques. Monte Carlo simulation is based on the representation of variable factors by statistical distributions. A random number generator is then used to draw a sample figure from a defined statistical population to simulate an actual event. By simulating a large number of events the underlying convolved distribution can be estimated with increasing accuracy. This probabilistic approach allows system simulation to be based on statistically derived distributive means, reflecting thousands of scenarios for a single time period, and would provide an accurate and in-depth analysis of the Djibouti generation sector. 6.2.1 Software A/S Plan Generation Planning Model A/Splan is a pc based generation expansion planning model licensed by Analytical Solutions of the USA. A/Splan has been developed from WASP, the best known and most widely used generation expansion planning model. It uses the same Booth-Baleriaux algorithm for probabilistic production simulation using equivalent load duration curves (ELDCs) to model energy despatch and system security (i.e. loss of load probability and energy not served), and the same optimality techniques using Bellman's Principle of Optimality to select the least cost plan from thousands of possible options. The model comprises of three main functions, the load forecast, cost production simulation, and generation plan optimisation. A detailed description of the model is set out in Appendix E. The GPSM The software programme @Risk is defined as a risk analysis, problem-solving tool that uses Monte Carlo simulation techniques. Monte Carlo simulation is based on the representation of variables by statistical distributions. A random number generator is then used to draw a sample figure from a defined statistical population to simulate an actual event (an ‘iteration’). By simulating a large number of events (a ‘simulation’) the underlying convolved distribution can be estimated with increasing accuracy. The GPSM uses @Risk to assign appropriate statistical distributions that describe these variables. This probabilistic approach allows least-cost generation planning and tariff analysis to be based on statistically derived distributive means which reflect thousands of scenarios for a single time period. This will provide an accurate and in-depth analysis of the Djibouti generation sector. Least Cost Electricity Master Plan (Volume 1) Final Report November 2009 Page 6.2 SECTION 6 GENERATION PLANNING The use of the Excel based @Risk add-in also provides a significant amount of adaptability to the modelling process. This is significant in relation to modelling the interconnector between Ethiopia and Djibouti. Other generation planning software programmes would not be able to handle the complexities involved in the operation and interaction of the interconnector within the Djibouti electrical system. However, this has been easily achieved using @Risk. The GPSM is initially developed from three main inputs. These are: • The load duration curve, • The peak load forecast, and; • The plant data for both existing and candidate plant. The GPSM is developed using a normalised load duration curve. In other words, daily load data is represented as a percentage of maximum demand in relation to the year from which the load data is sampled. The resulting normalised load duration curve will map time as a percentage on the x-axis and load as a percentage of maximum demand on the y-axis. An example of a normalised load duration curve is shown in Figure 6.1. The time element of the load duration curve is modelled by a uniform distribution varying between 0 per cent and 100 per cent. It can be seen in Figure 6.1 that by using a normalised load duration curve and by randomly sampling a point on the time axis, it is possible to determine the percentage of maximum load. Furthermore, the use of a normalised load duration curve make it possible to repetitively determine any load level for any year of the study period by combining the normalised load duration curve and the demand forecast. Hence for any given randomly sampled time, an ‘iterative demand’ for any year of the study can be estimated. It is this ‘iterative demand’ that the existing and candidate plant is then despatched to serve. Least Cost Electricity Master Plan (Volume 1) Final Report November 2009 Page 6.3 SECTION 6 GENERATION PLANNING Figure 6.1: A normalised load duration curve Maximum Demand (%) 100 60 Time (%) 100 The availability of plant dispatched to meet the iterative demand is modelled by a binomial distribution. The mathematical formula for the binomial distribution is provided in Equation 6.1 below. The GPSM determines the capacity available from each unit for every randomly sampled time according to this distribution. For each random time sample, the plant available is determined and placed in merit order (cheapest plant comes on stream first) to meet the ‘iterative demand’. If the cumulative available capacity of the plant is greater than the iterative demand then demand will be met and there will be no Loss of Load Expectation (LOLE) event. However, if the cumulative available capacity of the plant is less than the iterative demand then a LOLE event is recorded. When the iterative demand exceeds available capacity the difference between iterative demand and available capacity represents the level of un-served energy. Equation 6.1: Equation for the binomial distribution (P + Q) n Where = Q = (1 – P), Q = probability of being available P = probability of outage n = the number of units The process of randomly selecting a point in time to determine an iterative level of load and calculating the available capacity to meet that load before dispatching in merit order is repeated Least Cost Electricity Master Plan (Volume 1) Final Report November 2009 Page 6.4 SECTION 6 GENERATION PLANNING thousands of times (in other words, there are thousands of iterations) for each year of the forecast. This repetitive process is known as a ‘simulation’. This process is illustrated in Figure 6.2 and can also be summarised in the following steps: • A random point on the time axis of the normalised load duration curve is selected using a uniform distribution. • The ‘iterative demand’ is determined for the randomly sampled time point using the normalised load duration curve and peak load forecast. • The plants (existing and candidate) that are available to serve the iterative load are determined and sorted into merit order. • Using a binomial distribution, the available capacity from each unit is determined. Plant is then dispatched on a least-cost basis to meet the ‘iterative demand’. • System statistics are recorded. This includes whether demand exceeds total available capacity (LOLE criterion), plant despatch (generation), system operational costs, capital costs, un-served energy and other statistics such as the system marginal price and peak/off-peak energies. • Points one to five are repeated ‘X’ number of times. Repeating steps 1 to 5 over and over is known as a simulation and provides a statistically accurate representation of the system. • The mean results of the simulation are then analysed. Analysing the mean results from the simulation provides information regarding the system LOLE, un- served energy, lifetime costs, generation by plant and various others outputs for each year of the planning period. Critically when analysing the simulation results, if the mean LOLE rises above a predetermined planning criteria level in any given year there will be a need for new generating plant to be built in that year to meet the load. Any new plant added to the system is placed appropriately within the merit order for future dispatch. The system would then be ‘stable’ such that demand could be met without infringing the maximum LOLE criterion. It is imperative that the combination of candidate plant is selected at least-cost. Consequently a present value lifetime cost function provides a comparative indicator of least-cost attainability, and is referred to throughout the planning exercise, particularly when adding candidate plant to the system. Least Cost Electricity Master Plan (Volume 1) Final Report November 2009 Page 6.5 SECTION 6 GENERATION PLANNING Figure 6.2: The process of generation planning using the GPSM 1/ Randomly Sampled Peak Demand Point on Time Axis using Forecast Uniform Distribution Unitised Load Peak Demand Duration Curve for Year 3/ Plant Data 2/ Iterative Demand 4/ Determine Plant 6/ Repeat Availability using Process 'X' times Binomial Distribution 4/ Despatch Plant 5/ Measure System Outputs 7/ Analyse Mean Results from Simulation 6.2.2 Modelling issues Section 4.3 detailed the available energy and the operating philosophy for the interconnector. The interconnector can be modelled as an energy limited plant (e.g.; Hydro plant). This can be easily done within a software package such as WASP or A/Splan. However, for the GPSM model, the interconnector is treated as follows: • The load duration curve (LDC) which is essentially a relationship between MW and time, is used as a basis; • The LDC is used to derive a relationship between energy demand and time. This is generally referred to as the ‘integrated load duration curve’ (ILDC); • The ILDC is used to determine the ‘time elapsed’ (peak lopping basis- see Figure 4.2) to absorb the import energy made available; • A MW level is determined on the LDC that corresponds to the ‘time elapsed’; and • This MW level determines the level of EdD’s base load generation. Least Cost Electricity Master Plan (Volume 1) Final Report November 2009 Page 6.6 SECTION 6 GENERATION PLANNING Geothermal generating units cannot easily be controlled in an economical manner that follows load variation (i.e., increase and decrease the level of output). Geothermal plant is therefore represented as a base load ‘must take’ energy producer. In other words, all available energy and capacity from a geothermal plant has to be despatched. Similarly for wind farms, as wind cannot be stored and its pattern of availability cannot be predicted with any level of certainty, power and energy generated by a wind farm are assumed to be of a ‘must take’ nature. A maximum capacity factor is assumed and its equivalent MW capacity is assumed to be available throughout the year on the basis that any energy generated by the wind farm must be despatched by the operator. 6.3 Generation planning assumptions The input data for the analysis includes: • a national demand forecast (Section 3); • technical performance data for existing plant and committed new plant (Sections 4); • capital cost and technical performance data for candidate plant (Section 5); • fuel costs (Section 5.2.1); • load characteristics • data describing the required level of security of supply; • time horizon; and, • discount rate. 6.3.1 Load characteristics Daily load data for the year 2005 was provided by EdD and has been used to produce a normalised load duration curve. Figure 6.3 below portrays the normalised load duration curve used in this study, mapping time as a percentage on the x-axis and load as a percentage of maximum demand on the y- axis. This load duration curve has been produced using data for each day of the year in half hourly 14 increments . 14 Load data for a whole year was not provided. Where data was missing, we have determined the missing data using linear interpolation techniques in order to acquire load readings for each half hourly period for each day of the year. Least Cost Electricity Master Plan (Volume 1) Final Report November 2009 Page 6.7 SECTION 6 GENERATION PLANNING Figure 6.3 : The load duration curve for Djibouti 100% 80% Maximum Demand (%) 60% 40% 20% 0% 0% 10% 20% 30% 40% 50% 60% 70% 80% 90% 100% Time (%) 6.3.2 Planning criteria The objective of a generation plan is to develop the least-cost plan for that system whilst meeting an acceptable level of system security. This indicates that there is a need to balance the cost of achieving some measure of security against the cost, as a more secure system can be achieved with more plant, but at a higher cost. Loss of Load Expectation (LOLE) and Unserved Energy (UE) are the two most common probabilistic measures of system security. The value of LOLE in any one year is a measure of the expected duration that the combination of number, sizes and availability of generating sets will be unable to meet the total demand on the system. Normally a predetermined value of LOLE is assumed as a maximum tolerable level which can not be exceeded in any given period. As a system load grows and as plant is retired, it is the LOLE criterion which usually determines the timing of the introduction of new plant onto the system. Electrical and economic theory suggests that the determination of an acceptable maximum tolerable level of LOLE can be derived through the simple derivation of the Long Run Marginal Cost (LRMC) of generation and the Cost of Unserved Energy (COUE). In fact it is commonly accepted that: LOLE = LRMC / COUE The COUE and the LRMC can be estimated using economic, demographic and generic generating plant data. In this sub-section we make an estimate of an acceptable and justifiable LOLE planning criterion to be used in the @Risk least cost generation planning modelling for this study. Least Cost Electricity Master Plan (Volume 1) Final Report November 2009 Page 6.8 SECTION 6 GENERATION PLANNING Non Domestic COUE If the electricity supply to Commerce and Industry were to stop then the resulting outcome would be a stoppage of economic activity. Based on the assumption that a power failure will only result in the loss of output for the period that electricity is not available (i.e. it is assumed that in most cases raw materials can be utilised later on), then the loss from the power failure is the difference between output value and input costs for the foregone production. This difference is commonly referred to as Value Added (VA). It is understood that a rationally operated business would be willing to pay to avoid this loss of value added – the business would be willing to pay for this privilege until a value just less than the VA that has been lost due to the power failure. It can therefore be assumed that at this limit the cost of outage equates to the level of VA. The sum of all the VA in the economy is collected by Government Statisticians and presented as GDP. The historical GDP figures for Djibouti (in Constant 2008 USD) are presented in Table 6.1 below. Table 6.1: Djibouti GDP (USD millions, 2008 prices) (1996 – 2008) Year Agriculture Industry Services Total 1996 25.3 111.4 571.8 708.5 1997 24.9 107.8 570.6 703.2 1998 24.5 103.5 575.9 703.9 1999 25.1 107.1 587.1 719.3 2000 25.6 110.9 585.8 722.3 2001 26.1 115.4 595.6 737.0 2002 27.1 120.9 608.3 756.3 2003 28.2 126.1 626.2 780.5 2004 29.3 134.7 646.4 810.4 2005 29.5 138.8 667.7 836.1 2006 30.7 143.5 702.0 876.2 2007 35.1 153.9 722.3 911.3 2008 35.6 158.2 750.3 944.1 By dividing the VA (i.e. total GDP) by the total amount for energy (kWh) used by the productive sectors then an estimate of the cost of losing a kWh of supply can be derived (i.e. the COUE). The derivation of the non-domestic sector COUE is shown in Table 6.2 below. Least Cost Electricity Master Plan (Volume 1) Final Report November 2009 Page 6.9 SECTION 6 GENERATION PLANNING Table 6.2: Non Domestic COUE (2008 prices) (1996 – 2008) Sales (GWh) Total Total Non- Non Domestic LV Public MV Total GDP Year Social Domestic Chantier Domestic Domestic COUE Djibouti Lighting Djibouti (USDm) Sales Sales (USD / kWh) 1996 20.1 46.3 31.5 2.6 1.1 57.4 66.4 92.5 708.5 7.7 1997 20.1 52.4 30.2 2.4 0.9 56.6 72.5 90.2 703.2 7.8 1998 12.7 32.5 24.2 1.2 1.3 45.0 45.2 71.7 703.9 9.8 1999 17.2 40.3 32.8 1.8 0.6 47.5 57.5 82.7 719.3 8.7 2000 18.6 51.6 40.9 1.8 0.9 58.8 70.2 102.4 722.3 7.1 2001 15.8 62.0 40.1 1.4 0.7 70.0 77.8 112.3 737.0 6.6 2002 18.8 62.7 43.0 2.0 0.7 66.1 81.5 111.8 756.3 6.8 2003 19.0 63.7 46.9 2.0 0.9 74.3 82.7 124.1 780.5 6.3 2004 20.2 63.9 44.4 2.3 1.1 82.9 84.1 130.8 810.4 6.2 2005 20.4 63.2 50.3 2.3 1.3 83.3 83.6 137.2 836.1 6.1 2006 21.3 64.1 50.3 2.2 1.2 92.5 85.4 146.2 876.2 6.0 2007 22.4 68.4 51.3 2.7 3.7 96.9 90.8 154.7 911.3 5.9 2008 25.0 69.1 63.0 2.7 2.3 80.5 94.1 148.5 944.1 6.4 It should be noted that the cost of outages seem to be falling slightly over time. This may possible be due to a removal of suppressed demand. Domestic COUE For the domestic sector we adopt the argument that people make a choice between work and leisure. If the value of leisure is higher than the marginal wage rate then people will work fewer hours and if the value of leisure is lower than the marginal wage rate then people will work more hours. This assumption is not presumed to be a day by day decision but rather reflects a long term equilibrium state. If it is also assumed that a black out destroys all leisure (i.e. there would be no television, no reading, no cooking etc) at peak times then it is reasonable to assume that the value of lost leisure equates to the marginal wage rate and thus it is possible to derive an estimate of the cost of outage for the domestic sector (Domestic COUE). The derivation of the Domestic COUE is shown in Table 6.3 below. Least Cost Electricity Master Plan (Volume 1) Final Report November 2009 Page 6.10 SECTION 6 GENERATION PLANNING Table 6.3: Domestic COUE (2008 prices) National Population (2008) 849400 Average Household Size 12 People per House 2008 Total GDP 944.1 USD millions (2008 prices) 2008 GDP per Capita 1111 USD (2008 prices) 2008 GDP per Household 13338 USD (2008 prices) Number of Hours Worked per Year 2304 Hours (48 hours for 48 weeks) Average Household Earnings per Hour 5.79 USD (2008 prices) Domestic Specific Consumption 2995.9 kWh per Connection CADLF 60% Peak Usage 0.6 kWh per hour at peak Domestic COUE 10.16 USD per kWh Deriving an estimate to the LOLE In order to minimise the sum of the producers and consumers costs it is essential to make sure that the marginal cost of generation is equal to the marginal cost of an outage on a per kWh basis. If we consider the last kW of capacity available from the most expensive plant (assumed to be small diesel plant at Marabout), this unit will run for the numbers of hours per year equal to the LOLE (since the next kW of capacity can only come from disconnecting consumers). An estimate of the capital cost of this plant and thus the LRMC of generation is shown in Table 6.4 below. Table 6.4: LRMC of generation (2008 prices) Price base 2008 Price base of calculation Capital cost of plant 1000 US$ per kW (of most expensive plant on the system) Life of plant 15 Years Discount rate 10% Annual capital cost 131.5 US$ per kW per Year Fixed O&M cost 2.5% Percentage of capital cost Fixed O&M cost 25.0 US$ per kW per Year Total Annual LRMC Cost 156.5 US$ per kW per Year Reserve Margin 20.0% Planning Criteria LRMC of Generation 195.6 US$ per kW per Year We know, from the discussion above that: LOLE = LRMC / COUE Since we do not know which consumers will suffer during a generation blackout, however, we have used a weighted average of COUEs (referred to as the ‘National COUE’). The ratio between LRMC of generation and the national COUE is derived for each year. The average of these values is then Least Cost Electricity Master Plan (Volume 1) Final Report November 2009 Page 6.11 SECTION 6 GENERATION PLANNING derived providing an estimate of LOLE and LOLP. The calculation of the LOLE and LOLP is presented in Table 6.5 below. Table 6.5: Estimated LOLE Domestic Domestic COUE Non Domestic Non Domestic COUE National COUE LOLE Year Sales (2008 prices) Sales (2008 prices) (2008 prices) (Hours per (GWh) (USD per kWh) (GWh) (USD per kWh) (USD per kWh) Year) 1996 66.4 10.16 92.5 7.7 8.7 22.5 1997 72.5 10.16 90.2 7.8 8.8 22.1 1998 45.2 10.16 71.7 9.8 9.9 19.7 1999 57.5 10.16 82.7 8.7 9.3 21.0 2000 70.2 10.16 102.4 7.1 8.3 23.5 2001 77.8 10.16 112.3 6.6 8.0 24.3 2002 81.5 10.16 111.8 6.8 8.2 23.9 2003 82.7 10.16 124.1 6.3 7.8 25.0 2004 84.1 10.16 130.8 6.2 7.7 25.3 2005 83.6 10.16 137.2 6.1 7.6 25.6 2006 85.4 10.16 146.2 6.0 7.5 26.0 2007 90.8 10.16 154.7 5.9 7.5 26.2 2008 94.1 10.16 148.5 6.4 7.8 25.0 LOLE 23.8 LOLP 0.2722% This analysis indicates that an LOLE of approximately 24 hours per annum (an LOLP of 0.274 per cent) is a reasonable planning criterion to adopt for the Djibouti generation system. 6.3.3 Time horizon For many generating systems the installation and despatch of generating plant is centrally planned so as to minimise the total costs of meeting electricity demand within an acceptable level of reliability. With the objective of minimising total costs, the planning of new generating plant cannot be considered in isolation but must be addressed within the context of the system as a whole. The influence on system costs of commissioning a certain plant is felt throughout the life of the plant and beyond. For this reason a relatively long time horizon improves the accuracy of the modelling as the costs incurred towards the end of the time horizon are so remote that the effects of present-valuing make them insignificant. The time horizon adopted for this study is from 2009 to 2035. 6.3.4 Planning boundary Within a typical electricity system, the costs of generation (investment and production) are the most significant, whereas the costs of transmission investment are less significant. In addition the power and energy losses in transmission are generally small and typically in the order of a few per cent. For Least Cost Electricity Master Plan (Volume 1) Final Report November 2009 Page 6.12 SECTION 6 GENERATION PLANNING these reasons, the optimisation of the generation system is traditionally undertaken independently of the transmission system and this is the case with this study. That said, there will be constant communications between the generation planners and the transmission and distribution planners ensuring that all common issues are harmonised. The optimisation of total system generation against total system demand is undertaken at the generation voltage and for sent-out capacity/demand. By using sent-out capacity rather than generated capacity, all generation options are treated equitably so that the high station auxiliary loads typical of, say, coal fired steam plants in comparison with gas-fired plants are inherently recognized. 6.3.5 Discount rate During the evaluation of projects it is vital to take into account the core financial principle, the time value of money. The time value of money is the notion that money available at the present time is worth more than the same amount in the future, due to its potential earning capacity. For example, a single US dollar is worth more now than it is in the future because that single US dollar can be invested to yield more than one US dollar at a future date (after allowing for inflation). The time value of money is taken into account through the discounting process and a discount rate. Projects under evaluation are likely to incur costs and revenues (a cash flow) over a period of time and when comparing several different projects in order to identify the ‘best’ project, the timing of these cash flows is likely to vary on a project by project basis. In order to compare several different projects, a discount rate is applied to past and future costs to determine their ‘present value’ – the equivalent value in today’s money. By converting all of the project costs into values to one point in time (the present), the time value of money has been taken into account and projects can be evaluated on an equal footing. The estimation of a suitable discount rate to use for discounting for a country is a complicated matter and requires some judgement as the choice of discount rate can effect the decisions made during the planning and evaluation process. Inherently, a low discount rate favours capital intensive projects such as hydroelectric schemes, coal-fired and nuclear power plant whose up-front costs are relatively high compared with its future operating costs. Conversely, high discount rates favour less capital intensive projects such as OCGT and CCGT power plants. The discount rate for an economy as a whole is frequently related to the opportunity cost of capital, either as an expected return on equity or the cost of borrowing, or a combination of both (also known as the Weighted Average Cost of Capital (WACC)). Most companies need (or would like) to earn a rate of return on investments that, at minimum, reflects the cost of capital, otherwise there would be little reason to invest. On the basis that the generation Least Cost Electricity Master Plan (Volume 1) Final Report November 2009 Page 6.13 SECTION 6 GENERATION PLANNING planning exercise in this study is founded on the principles that investment is attracted to the industry through the potential to make reasonable returns without making excessive profit we have assumed the use of a discount rate of 10 per cent in this study. 6.3.6 Cost basis All costs are stated in US dollars (US$) recognising that the dollar is the dominant world currency. All costs incurred now or in the future are stated in constant (real) prices as of 2008. In this way, inflation is essentially stripped out of the analysis (on the assumption that inflation will affect all costs equally) so that future costs can be compared more readily with today’s costs. 6.4 Generation planning scenarios The bulk of our generation expansion has been carried out for the base case demand forecast, as detailed in Section 3 of this report. The scenarios undertaken as part of this study are summarised in below. The investment and operating cost schedules relating to each of the scenarios developed are presented in Appendix F in Volume 2 to this report. Least Cost Electricity Master Plan (Volume 1) Final Report November 2009 Page 6.14 SECTION 6 GENERATION PLANNING Figure 6.4: Base demand forecast scenarios Base Case Demand  Forecast Reference Case "Least Cost Planting  Sensitivity Analysis Programme" Zero Imports Reference Case with  Reference Case with  Reference Case with  Geothermal  Geothermal and Wind  (Scenario 1) Wind Development Only Development Only Development Zero Imports 180 GWh Imports 180 GWh Imports Zero Imports Wind 25%  CF Wind 25%  CF (Scenario 2) (Scenario 4) (Scenario 6) (Scenario 10) 180 GWh Imports Up to 700 GWh Imports 700 GWh Imports Up to  700 GWh Imports Wind 25%  CF Wind 25%  CF (Scenario 3) (Scenario 5) (Scenario 7) (Scenario 11) Up to 700 GWh Imports Up to 700 GWh Imports Wind 25%  CF Wind 35%  CF (Scenario 8) (Scenario 12) Up to 700 GWh Imports Wind 35%  CF (Scenario 9) 6.5 Least cost generation expansion plan (Reference case) As discussed in Section 5 of this report, the key candidate plant considered in this study are: • HFO-fired diesel • Gasoil-fired OCGT • Geothermal • Wind As discussed in Appendix C and D in Volume 2 to this report, the development of geothermal and wind power in Djibouti is in its early phases and further studies, exploration, drilling, analysis, testing and measurements are required to confirm and enable the utilisation of these resources for large- Least Cost Electricity Master Plan (Volume 1) Final Report November 2009 Page 6.15 SECTION 6 GENERATION PLANNING scale power generation. As a result, we have assumed the reference case analysis of this study to be based on the availability of HFO-fired diesel and gasoil-fired OCGT units only. This analysis has been carried out taking into account varying levels of energy import over the interconnector. Given the expectation that EdD are required to build sufficient capacity to operate the system independently from the interconnector, the planting programme will remain unchanged regardless of the level of import over the interconnector. 6.5.1 Planting programme The reference case least cost generation expansion plan (satisfying the planning criteria) is presented in the planting programme shown in Table 6.6 and in the capacity chart shown in Figure 6.5 below. The table indicates that 24 MW of capacity is required in 2013 and a total of 187 MW would be required by the end of the planning period. The development over the planning period comprises ten 12 MW and six 7 MW HFO-fired diesel units, one 15 MW and one 7 MW open cycle gas turbine generating sets. These gas turbine generating sets would replace the retired units at the Marabout power station. On the basis of the planting programme identified below, we have simulated the reference case system assuming 3 alternate energy import scenarios; zero energy imports, 180 GWh of energy imports and 700 GWh of energy imports over the interconnector. The results of these scenarios are discussed below. Least Cost Electricity Master Plan (Volume 1) Final Report November 2009 Page 6.16 SECTION 6 GENERATION PLANNING Table 6.6: Reference case planting programme Year Existing Capacity Capacity Added Total Capacity Peak Demand Reserve Margin MW Net MW Net Type MW Net MW % 2009 97 0 97 68 43 2010 97 0 97 74 31 2011 112 0 112 84 33 2012 112 0 112 100 12 2013 107 24 2xDIESEL HFO 12MW 131 116 13 2014 113 31 1xDIESEL HFO 7MW 144 128 13 2015 98 56 2xDIESEL HFO 12MW 154 138 12 2016 119 56 174 149 17 2017 119 56 174 153 14 2018 119 56 174 158 10 2019 116 63 1xDIESEL HFO 7MW 179 162 11 2020 114 75 1xDIESEL HFO 12MW 189 165 14 2021 111 75 186 168 11 2022 109 87 1xDIESEL HFO 12MW 196 171 15 2023 104 87 191 173 10 2024 104 95 1xGAS TURBINE 7MW 199 176 13 2025 104 95 199 180 11 2026 104 107 1xDIESEL HFO 12MW 211 183 15 2027 104 107 211 187 13 2028 104 107 211 191 11 2029 104 114 1xDIESEL HFO 7MW 218 195 12 2030 82 139 2xDIESEL HFO 12MW 221 198 11 2031 73 151 1xDIESEL HFO 12MW 224 202 10 2032 73 158 1xDIESEL HFO 7MW 231 206 12 2033 67 165 1xDIESEL HFO 7MW 232 210 10 2034 55 187 1xDIESEL HFO 7MW 1xGAS TURBINE 15M 242 214 13 2035 55 187 242 218 11 Figure 6.5: Reference case capacity chart Wind Interconnector New Gas Turbines Marabout Boulaos New Diesel Geothrmal Demand Forecast Reserve Margin 300 50% 45% 250 40% 35% Capacity / Demand (MW) 200 Reserve Margin (%) 30% 150 25% 20% 100 15% 10% 50 5% 0 0% 2009 2010 2011 2012 2013 2014 2015 2016 2017 2018 2019 2020 2021 2022 2023 2024 2025 2026 2027 2028 2029 2030 2031 2032 2033 2034 2035 Least Cost Electricity Master Plan (Volume 1) Final Report November 2009 Page 6.17 SECTION 6 GENERATION PLANNING 6.5.2 Zero energy imports (Scenario 1) The level of generation by plant type, assuming that there are no energy imports over the interconnector throughout the forecast period, is presented in Figure 6.6 below. This Figure indicates that the new diesel capacity (more efficient and reliable than the existing diesel units) occupies the base load duty within the merit order. The new diesel capacity would be followed by the units at the Boulaos power station and the Marabout units (until retired) and the open cycle gas turbines (thereafter) would undertake the peaking duties. This analysis is confirmed in Figure 6.7, whereby the utilisation figures for each plant type are presented. The net present value (NPV) calculated for the capital and operating cost streams for this development plan amounts to US$ 1,084 million over the planning period for this study. Figure 6.6: Energy generation by plant type (180 GWh energy imports) Interconnector New Gas Turbines Marabout Boulaos New Diesel Generation (GWh) 1400 1200 1000 Generation (GWh) 800 600 400 200 0 2009 2010 2011 2012 2013 2014 2015 2016 2017 2018 2019 2020 2021 2022 2023 2024 2025 2026 2027 2028 2029 2030 2031 2032 2033 2034 2035 Least Cost Electricity Master Plan (Volume 1) Final Report November 2009 Page 6.18 SECTION 6 GENERATION PLANNING Figure 6.7: Plant utilisation (180 GWh energy imports) Boulaos Marabout New Gas Turbines New Diesel New Geothermal New Wind 100% 90% 80% 70% 60% 50% 40% 30% 20% 10% 0% The level of system security, assuming no imports from Ethiopia, is presented in Figure 6.8 below. The capacity chart for the reference case (Figure 6.5) indicates a minimum reserve margin of 10 per cent. Figure 6.8 confirms that this level of reserve margin is adequate for the system, whereby the LoLP level for the long term plan remains below the targeted level of 24 hours per year. Figure 6.8 also indicates that the LoLP level in 2012 is expected to exceed the LoLP target. Adding further new units in 2012 is considered to be an unnecessary overinvestment in view of the commissioning of the Ethiopia-Djibouti interconnector and its impact on system reliability/security. This notion is highlighted further in Scenario 2. Least Cost Electricity Master Plan (Volume 1) Final Report November 2009 Page 6.19 SECTION 6 GENERATION PLANNING Figure 6.8: System reliability (180 GWh energy imports) Annual Average LoLP Target LoLP 0.40% 0.35% 0.30% 0.25% LoLP (%) 0.20% 0.15% 0.10% 0.05% 0.00% 2009 2010 2011 2012 2013 2014 2015 2016 2017 2018 2019 2020 2021 2022 2023 2024 2025 2026 2027 2028 2029 2030 2031 2032 2033 2034 2035 6.5.3 180GWh energy imports (Scenario 2) In line with the requirements of the PPA for the Ethiopia-Djibouti interconnector, the generation expansion plan proposed for the reference case enables EdD to meet the forecast peak demand using their own generating units without reliance on the interconnector. Following the commissioning of the Ethiopia-Djibouti Interconnector in 2011 and assuming the minimum stipulated level of import of 180 GWh annual energy imports, Figure 6.9 shows the impact on generation levels by plant type. It can be seen that the Boulaos units would be relegated to peaking duties and while imports over the interconnector are available, the Marabout units may not be required to operate. This analysis is confirmed in the Plant utilisation chart presented in Figure 6.10. The NPV for the capital and operating cost stream for this scenario is calculated at US$ 930 million. When compared to an NPV of US$ 1,084 million for Scenario 1, it is 14 per cent lower and we can easily conclude that energy imports of 180 GWh per annum would potentially save EdD some US$ 154 million over the time span of the planning study. Figure 6.11 shows that energy imports would significantly lower the level of expected LoLP, enhancing the reliability of the EdD electricity network and practically eliminating energy not served. Least Cost Electricity Master Plan (Volume 1) Final Report November 2009 Page 6.20 SECTION 6 GENERATION PLANNING Figure 6.9: Energy generation by plant type (180GWh energy imports) Interconnector New Gas Turbines Marabout Boulaos New Diesel Generation (GWh) 1400 1200 184 187 183 187 187 183 184 186 1000 182 183 184 186 183 184 185 183 183 Generation (GWh) 184 800 182 186 184 184 600 183 184 0 183 400 0 200 0 2009 2010 2011 2012 2013 2014 2015 2016 2017 2018 2019 2020 2021 2022 2023 2024 2025 2026 2027 2028 2029 2030 2031 2032 2033 2034 2035 Figure 6.10: Plant utilisation (180GWh energy imports) Boulaos Marabout New Gas Turbines New Diesel New Geothermal New Wind 100% 90% 80% 70% 60% 50% 40% 30% 20% 10% 0% Least Cost Electricity Master Plan (Volume 1) Final Report November 2009 Page 6.21 SECTION 6 GENERATION PLANNING Figure 6.11: System reliability (180GWh energy imports) Annual Average LoLP Annual Maximum LoLP Target LoLP 0.30% 0.25% 0.20% LoLP (%) 0.15% 0.10% 0.05% 0.00% 2009 2010 2011 2012 2013 2014 2015 2016 2017 2018 2019 2020 2021 2022 2023 2024 2025 2026 2027 2028 2029 2030 2031 2032 2033 2034 2035 6.5.4 700 GWh energy imports (Scenario 3) With the imminent commissioning of new hydro power station in Ethiopia, EEPCO indicated that there 15 could be up to 700 GWh of excess energy being available for export to Djibouti . With EdD maintaining a minimum of 25 MW of own generation at all times, energy imports will have a profound impact on the utilisation of the EdD generating capacity as shown in Figure 6.12 and Figure 6.13, where by utilisation of the new diesel units could be as low as 25 to 30 per cent. The Boulaos generating units would operate at a capacity factor averaging 15 per cent over the planning period. Figure 6.12 in particular indicates that the higher level of energy imports over the interconnector complement EdD’s own energy generation to meet the forecast demand for energy, and displace a significant amount of EdD’s liquid fuel generation. The NPV for the capital and operating cost stream for this scenario is further reduced to US$ 622 million. When compared to an NPV of US$ 1,084 million for Scenario 1, it is some 42 per cent lower and it can be seen that energy imports would potentially save EdD some US$ 462 million over the time span of the planning study. 15 Ethiopia – Djibouti Power Interconnection Project, Consultancy services for institutional support Least Cost Electricity Master Plan (Volume 1) Final Report November 2009 Page 6.22 SECTION 6 GENERATION PLANNING In essence, the results of the reference case scenarios 1, 2 and 3 highlight that the higher the amount of energy imports, the greater the EdD fuel savings. Figure 6.12: Energy generation by plant type (700GWh energy imports) Interconnector New Gas Turbines Marabout Boulaos New Diesel Generation (GWh) 1400 1200 1000 703 707 699 Generation (GWh) 704 705 800 701 701 704 705 698 699 698 702 689 674 660 641 623 600 597 574 517 467 406 300 0 183 400 0 200 0 2009 2010 2011 2012 2013 2014 2015 2016 2017 2018 2019 2020 2021 2022 2023 2024 2025 2026 2027 2028 2029 2030 2031 2032 2033 2034 2035 Figure 6.13: Plant utilisation (700GWh energy imports) Boulaos Marabout New Gas Turbines New Diesel New Geothermal New Wind 100% 90% 80% 70% 60% 50% 40% 30% 20% 10% 0% Least Cost Electricity Master Plan (Volume 1) Final Report November 2009 Page 6.23 SECTION 6 GENERATION PLANNING It should also be noted that system reliability in this scenario is even higher than that shown for Scenario 2, which itself was well beyond the minimum requirements for the EdD system. 6.5.5 NPV summary Table 6.7 below presents a summary of the NPV’s for the reference case scenarios assuming varying levels of energy import over the interconnector. This table indicates that energy imports over the Ethiopia-Djibouti interconnector are recommended and that EdD should maximise these imports at every available opportunity. Table 6.7: NPV Summary of reference case scenarios Scenario NPV Import Scenario Geothermal Scenario Wind Scenario No. (US$m) 3 Up to 700 GWh Import No Geothermal No Wind 622 2 180 GWh Import No Geothermal No Wind 930 1 No Imports No Geothermal No Wind 1084 6.6 Sensitivities to the least cost generation expansion plan In addition to the least cost generation expansion plan (i.e. the reference case), we have analysed the inclusion of geothermal and wind power as a sensitivity, so as to determine the future role that these technologies may play in the generation mix and determine whether the development of these technologies should be taken to the next stage. Our sensitivity analysis incorporates the development of a “reference with geothermal only� case, the development of a “reference with wind only� case and the development of a “reference with geothermal and wind� case. For each of these cases, we have also analysed the impact of varying levels of energy import over the interconnector. Again, the planting programme will remain unchanged for each of the above cases regardless of the level of import over the interconnector. We present the results of this analysis in the sub-sections below. 6.6.1 Reference case with geothermal development only The purpose of this scenario is to examine the economic viability of the possible introduction of geothermal power generation to the EdD system. In line with our assessment of geothermal resources in Djibouti (presented in Appendix C), and assuming that the planned and recommended drilling proves to be viable for up to 60 MW of geothermal power generation, we have allowed the Least Cost Electricity Master Plan (Volume 1) Final Report November 2009 Page 6.24 SECTION 6 GENERATION PLANNING optimisation routine to select up to three 20 MW units with 3 year intervals (to allow enough time to drill, test and establish the required geothermal wells for further development), the earliest possible commissioning date being in 2016. The reference case with geothermal development expansion plan is presented in the planting programme shown in Table 6.8 and in the capacity chart presented in Figure 6.14 below. It can be seen that the geothermal units were selected as soon as they were made available; indicating that, for the cost and performance parameters assumed in this study, geothermal generation would be part of the least cost plan for EdD. As previously discussed, geothermal units usually operate supplying the base load due to their inflexible nature, and as can be seen in Table 6.8, these units have displaced an equivalent generating capacity of diesel units (57 MW). The development over the planning period would therefore comprise seven 12 MW and four 7 MW HFO-fired diesel units and one 15 MW open cycle gas turbine generating sets. On the basis of the planting programme identified below, we have simulated 2 alternative energy import scenarios; zero energy imports, and up to 700 GWh of energy imports over the interconnector. The results of these scenarios are discussed below. Table 6.8: Reference case (with geothermal development only) planting programme Year Existing Capacity Capacity Added Total Capacity Peak Demand Reserve Margin MW Net MW Net Type MW Net MW % 2009 97 0 97 68 43 2010 97 0 97 74 31 2011 112 0 112 84 33 2012 112 0 112 100 12 2013 107 24 2xDIESEL HFO 12MW 131 116 13 2014 113 31 1xDIESEL HFO 7MW 144 128 13 2015 98 53 1xDIESEL HFO 7MW 1xGAS TURBINE 15M 152 138 10 2016 119 73 1xGEOTHERMAL 20MW 192 149 29 2017 119 73 192 153 25 2018 119 73 192 158 21 2019 116 93 1xGEOTHERMAL 20MW 210 162 30 2020 114 93 207 165 25 2021 111 93 205 168 22 2022 109 113 1xGEOTHERMAL 20MW 222 171 30 2023 104 113 218 173 25 2024 104 113 218 176 23 2025 104 113 218 180 21 2026 104 113 218 183 19 2027 104 113 218 187 16 2028 104 113 218 191 14 2029 104 113 218 195 12 2030 82 138 2xDIESEL HFO 12MW 220 198 11 2031 73 157 1xDIESEL HFO 7MW 1xDIESEL HFO 12MW 230 202 14 2032 73 157 230 206 11 2033 67 169 1xDIESEL HFO 12MW 236 210 12 2034 55 181 1xDIESEL HFO 12MW 236 214 10 2035 55 188 1xDIESEL HFO 7MW 243 218 11 Least Cost Electricity Master Plan (Volume 1) Final Report November 2009 Page 6.25 SECTION 6 GENERATION PLANNING Figure 6.14: Reference case (with geothermal development only) capacity chart Wind Interconnector New Gas Turbines Marabout Boulaos New Diesel Geothrmal Demand Forecast Reserve Margin 300 50% 45% 250 40% 35% Capacity / Demand (MW) 200 Reserve Margin (%) 30% 150 25% 20% 100 15% 10% 50 5% 0 0% 2009 2010 2011 2012 2013 2014 2015 2016 2017 2018 2019 2020 2021 2022 2023 2024 2025 2026 2027 2028 2029 2030 2031 2032 2033 2034 2035 Zero energy imports (Scenario 4) The level of generation by plant type and the plant utilisation factors assuming zero energy imports over the interconnector are presented in Figure 6.15 and Figure 6.16 below. These figures highlight the notion behind the selection of geothermal plant. The relatively higher capital cost of the geothermal development is offset by the long-term fuel savings that can be achieved and therefore these units are selected at the earliest possible opportunity to primarily displace the costly HFO-fired diesel units as opposed to a system requirement to satisfy the planning criteria. The NPV calculated for the capital and operating cost streams for this development plan amounts to US$ 906 million over the planning period for this study. Compared to the reference case, Figure 6.14 indicates a relatively higher reserve margin in the medium term, between 2016 and 2024. From thereon, the reserve margin gradually reduces to the 10 per cent level previously observed. This observation is further confirmed by the low level of LoLP shown in Figure 6.17. Least Cost Electricity Master Plan (Volume 1) Final Report November 2009 Page 6.26 SECTION 6 GENERATION PLANNING Figure 6.15: Energy generation by plant type (zero energy imports) New Gas Turbines Marabout Boulaos New Diesel New Geothermal Generation (GWh) 1400 1200 1000 Generation (GWh) 800 600 400 200 0 2009 2010 2011 2012 2013 2014 2015 2016 2017 2018 2019 2020 2021 2022 2023 2024 2025 2026 2027 2028 2029 2030 2031 2032 2033 2034 2035 Figure 6.16: Plant utilisation (zero energy imports) Boulaos Marabout Geothrmal New Diesel New Gas Turbines 100.0% 90.0% 80.0% 70.0% 60.0% 50.0% 40.0% 30.0% 20.0% 10.0% 0.0% Least Cost Electricity Master Plan (Volume 1) Final Report November 2009 Page 6.27 SECTION 6 GENERATION PLANNING Figure 6.17: System security (zero energy imports) Annual Average LoLP Target LoLP 0.35% 0.30% 0.25% 0.20% LoLP (%) 0.15% 0.10% 0.05% 0.00% 2009 2010 2011 2012 2013 2014 2015 2016 2017 2018 2019 2020 2021 2022 2023 2024 2025 2026 2027 2028 2029 2030 2031 2032 2033 2034 2035 Up to 700 GWh energy imports (Scenario 5) Energy imports displacing peak thermal generation complements geothermal generation at base load. We have therefore considered this scenario in which the geothermal resources are assumed to be proven, allowing the development of three 20 MW units of geothermal capacity, together with up to 700 GWh of energy imports over the Ethiopia-Djibouti interconnector. For this scenario we have assumed a minimum of 25 MW of diesel generation by EdD until 2016 when the geothermal plant is assumed to be commissioned. Thereon, minimum MW generation by EdD was assumed to incorporate the geothermal plant capacity plus 15 MW of diesel generation. Whilst in theory, there is up to 700 GWh of energy imports available, Figure 6.18 shows that energy imports over the interconnector are likely to be restricted, due to the operational requirements of the geothermal plant, to a value averaging around 400 GWh per annum. Furthermore, utilisation of HFO- fired generating units is unlikely to exceed 25 per cent. As a result, system security is improved in this Scenario. A NPV for the cost streams calculated at US$ 575 million indicates that the least cost long-term generation plan is one that combines geothermal development (base load) with energy imports (peak demand) supplemented by EdD diesel generation. Least Cost Electricity Master Plan (Volume 1) Final Report November 2009 Page 6.28 SECTION 6 GENERATION PLANNING Figure 6.18: Generation by plant type (up to 700 GWh energy imports) New Wind Interconnector New Gas Turbines Marabout Boulaos New Diesel New Geothermal Generation (GWh) 1400 1200 1000 539 516 496 475 455 435 416 397 377 360 343 324 311 Generation (GWh) 299 800 434 421 403 544 520 495 600 516 467 406 300 0 183 400 0 200 0 2009 2010 2011 2012 2013 2014 2015 2016 2017 2018 2019 2020 2021 2022 2023 2024 2025 2026 2027 2028 2029 2030 2031 2032 2033 2034 2035 6.6.2 Reference case with wind development only Appendix D presents and reviews our desktop analysis on the potential wind resources in Djibouti. Our review confirms the potential for wind farm development in Djibouti. The sites of Ghoubet, Ali Sabieh, Djibouti Ville, Egralyta and Bada Wein are expected to be able to generate electricity at costs ranging between 6.64 and 13.39 USc per kWh. The wind monitoring program, managed by CERD, includes the commissioning of several wind monitoring masts across the country and there may be a need for additional data analysis. We therefore, believe that there is a high level of uncertainty surrounding the wind resource. We have developed a number of scenarios that include the possible development of wind power in Djibouti based on the following assumptions: • Due to its intermittent nature, the development of wind power does not add firm capacity to the system generating capacity. • All energy generated by a wind farm must be absorbed by the grid (must take). • The earliest date for commissioning wind power in Djibouti is 2015. • The wind farm is represented by a constant MW output adjusted to reflect the expected capacity factor of the plant. Least Cost Electricity Master Plan (Volume 1) Final Report November 2009 Page 6.29 SECTION 6 GENERATION PLANNING • The annual utilisation factor for wind farms in Djibouti is not expected to exceed 35 per cent and would average around 25 per cent. Based on the above, for a wind farm to be economically viable and form part of the least cost solution, its capital cost expenditure would have to be offset by the fuel cost savings, arising from the displacement of other forms of generation by the wind farm energy generated. The reference case with wind development expansion plan is presented in the planting programme shown in Table 6.9 and in the capacity chart presented in Figure 6.19 below. This planting programme reflects that presented for the reference case, but with an additional 10MW wind power development added in 2015. On the basis of the planting programme identified below, we have developed 4 alternative energy import scenarios: • No imports, Wind @ 25% capacity factor (Scenario 6) • 180 GWh imports, Wind @ 25% capacity factor (Scenario 7) • 700 GWh imports, Wind @ 25% capacity factor (Scenario 8) • 700 GWh imports, Wind @ 35% capacity factor (Scenario 9) The results of these scenarios are discussed below. Least Cost Electricity Master Plan (Volume 1) Final Report November 2009 Page 6.30 SECTION 6 GENERATION PLANNING Table 6.9: Reference case (with wind development only) planting programme Year Existing Capacity Capacity Added Total Capacity Peak Demand Reserve Margin MW Net MW Net Type MW Net MW % 2009 97 0 97 68 43 2010 97 0 97 74 31 2011 112 0 112 84 33 2012 112 0 112 100 12 2013 107 24 2xDIESEL HFO 12MW 131 116 13 2014 113 31 1xDIESEL HFO 7MW 144 128 13 2015 98 56 2xDIESEL HFO 12MW 1xDJOBOUTI VILLE 154 138 12 2016 119 56 174 149 17 2017 119 56 174 153 14 2018 119 56 174 158 10 2019 116 63 1xDIESEL HFO 7MW 179 162 11 2020 114 75 1xDIESEL HFO 12MW 189 165 14 2021 111 75 186 168 11 2022 109 87 1xDIESEL HFO 12MW 196 171 15 2023 104 87 191 173 10 2024 104 95 1xGAS TURBINE 7MW 199 176 13 2025 104 95 199 180 11 2026 104 107 1xDIESEL HFO 12MW 211 183 15 2027 104 107 211 187 13 2028 104 107 211 191 11 2029 104 114 1xDIESEL HFO 7MW 218 195 12 2030 82 139 2xDIESEL HFO 12MW 221 198 11 2031 73 151 1xDIESEL HFO 12MW 224 202 10 2032 73 158 1xDIESEL HFO 7MW 231 206 12 2033 67 165 1xDIESEL HFO 7MW 232 210 10 2034 55 187 1xDIESEL HFO 7MW 1xGAS TURBINE 15M 242 214 13 2035 55 187 242 218 11 Figure 6.19: Reference case (with wind development only) capacity chart Wind Interconnector New Gas Turbines Marabout Boulaos New Diesel Geothrmal Demand Forecast Reserve Margin 300 50% 45% 250 40% 35% Capacity / Demand (MW) 200 Reserve Margin (%) 30% 150 25% 20% 100 15% 10% 50 5% 0 0% 2009 2010 2011 2012 2013 2014 2015 2016 2017 2018 2019 2020 2021 2022 2023 2024 2025 2026 2027 2028 2029 2030 2031 2032 2033 2034 2035 Least Cost Electricity Master Plan (Volume 1) Final Report November 2009 Page 6.31 SECTION 6 GENERATION PLANNING Zero energy imports, 25% wind capacity factor (Scenario 6) We have considered the reference case where EdD is assumed to rely purely on liquid fuel for power generation and compared the cases ‘with’ and ‘without’ a 10 MW wind farm development (with a capacity factor of 25 per cent). The energy allocation chart for the scenario where we add an additional wind farm development is presented in Figure 6.20. The NPV for the case with the wind farm (but no energy imports) was calculated at US$ 1,075 million. This represents a reduction of US$ 9 million in NPV when compared to Scenario 1 (US$ 1,084 million). This indicates that at the cost and performance parameters assumed, the capital cost of a wind power development is likely to be offset by the energy savings. Hence, for this particular scenario, wind power development could form part of the least cost plan for Djibouti. Figure 6.20: Generation by plant type (zero energy imports, 25% capacity factor) Interconnector New Gas Turbines Marabout Boulaos New Diesel New Geothermal New Wind Generation (GWh) 1400 0 0 1200 0 0 0 0 0 0 0 0 0 0 1000 0 0 0 0 0 0 0 0 Generation (GWh) 800 0 0 0 600 0 0 0 400 0 200 0 2009 2010 2011 2012 2013 2014 2015 2016 2017 2018 2019 2020 2021 2022 2023 2024 2025 2026 2027 2028 2029 2030 2031 2032 2033 2034 2035 180 GWh energy imports, 25% wind capacity factor (Scenario 7) We have also considered the case where 180 GWh of energy is imported over the interconnector, with EdD relying on diesel generation only for expanding the generating system and compared the cases ‘with’ and ‘without’ a 10 MW wind farm development (with a 25 per cent capacity factor). The energy allocation chart for the scenario where we add an additional wind farm development is presented in Figure 6.21. Least Cost Electricity Master Plan (Volume 1) Final Report November 2009 Page 6.32 SECTION 6 GENERATION PLANNING The NPV for the case with the wind farm was calculated at US$ 922 million. This represents a reduction of US$ 8 million in NPV compared with the case ‘without’ the wind farm (Scenario 2) (US$ 930 million). This indicates that at the cost and performance parameters assumed, the capital cost of a wind power development is likely to be offset by the energy savings. Hence, for a scenario that is based on liquid fuel generation supplemented by 180 GWh of annual energy imports over the interconnector, wind power development could form part of the least cost plan for Djibouti. Figure 6.21: Generation by plant type (180 GWh energy imports, 25% wind capacity factor) Interconnector New Gas Turbines Marabout Boulaos New Diesel New Wind New Geothermal Generation (GWh) 1400 1200 184 188 183 185 186 183 185 186 1000 182 182 184 185 183 185 187 184 184 Generation (GWh) 183 800 183 186 185 184 600 181 184 0 182 400 0 200 0 2009 2010 2011 2012 2013 2014 2015 2016 2017 2018 2019 2020 2021 2022 2023 2024 2025 2026 2027 2028 2029 2030 2031 2032 2033 2034 2035 Up to 700 GWh energy imports, 25% wind capacity factor (Scenario 8) We have considered the case where up to 700 GWh of energy is imported over the interconnector, with EdD relying on diesel generation only for expanding the generating system and compared the cases ‘with’ and ‘without’ a 10 MW wind farm development (with a 25 per cent capacity factor). Similar to previous scenarios, EdD is assumed to have a minimum 25 MW of base load diesel generation at all times. Following the introduction of a 10 MW wind farm in 2015, this base load generation level is increased to 35 MW to ensure adequate cover for the likely fluctuations in the output of the wind farm. Figure 6.22 below shows the wind energy injected into the EdD network displacing import energy over the interconnector. The marginal cost of wind generation is higher than the marginal cost of energy imports hence, the NPV for the case with the wind farm, calculated at US$ 660 million, represents an Least Cost Electricity Master Plan (Volume 1) Final Report November 2009 Page 6.33 SECTION 6 GENERATION PLANNING increase of US$ 38 million in NPV compared with the case ‘without’ the wind farm (Scenario 3) - US$ 622 million. This indicates that displacing import energy by energy from a wind farm is unlikely to be an economical option. Figure 6.22: Generation by plant type (Up to 700 GWh energy imports, 25% capacity factor) Interconnector New Gas Turbines Marabout Boulaos New Diesel New Geothermal New Wind Generation (GWh) 1400 1200 1000 623 624 619 623 627 Generation (GWh) 621 621 800 623 626 617 622 623 623 608 595 582 563 543 518 495 600 436 468 406 301 0 183 400 0 200 0 2009 2010 2011 2012 2013 2014 2015 2016 2017 2018 2019 2020 2021 2022 2023 2024 2025 2026 2027 2028 2029 2030 2031 2032 2033 2034 2035 Up to 700 GWh energy imports, 35% wind capacity factor (Scenario 9) We have also considered the case where up to 700 GWh of energy is imported over the interconnector, with EdD relying on diesel generation only for expanding the generating system and compared the cases ‘with’ and ‘without’ a 10 MW wind farm development (with a 35 per cent capacity factor). With a higher utilisation factor for the wind farm, the NPV was reduced by US$ 6 million (in comparison to Scenario 8), to US$ 654 million over the forecast period. This still represents an increase of US$ 32 million in NPV compared with the case ‘without’ the wind farm (Scenario 3) - US$ 622 million. 6.6.3 Reference case with geothermal and wind development The purpose of this scenario is to examine the economic viability of the introduction of geothermal and wind power generation to the EdD system. Least Cost Electricity Master Plan (Volume 1) Final Report November 2009 Page 6.34 SECTION 6 GENERATION PLANNING The reference case with geothermal and wind development expansion plan is presented in the planting programme shown in Table 6.9 and in the capacity chart presented in Figure 6.23 below. This planting programme reflects that presented for the reference with geothermal development case, but with an additional 10MW wind power development added in 2015. Table 6.10: Reference case (with geothermal and wind development) planting programme Year Existing Capacity Capacity Added Total Capacity Peak Demand Reserve Margin MW Net MW Net Type MW Net MW % 2009 97 0 97 68 43 2010 97 0 97 74 31 2011 112 0 112 84 33 2012 112 0 112 100 12 2013 107 24 2xDIESEL HFO 12MW 131 116 13 2014 113 31 1xDIESEL HFO 7MW 144 128 13 2015 98 63 1xDIESEL HFO 7MW 1xGAS TURBINE 15M 162 138 10 1 x DJIBOUTIVILLE (WIND) 2016 119 83 1xGEOTHERMAL 20MW 202 149 29 2017 119 83 202 153 25 2018 119 83 202 158 21 2019 116 103 1xGEOTHERMAL 20MW 220 162 30 2020 114 103 217 165 25 2021 111 103 215 168 22 2022 109 123 1xGEOTHERMAL 20MW 232 171 30 2023 104 123 228 173 25 2024 104 123 228 176 23 2025 104 123 228 180 21 2026 104 123 228 183 19 2027 104 123 228 187 16 2028 104 123 228 191 14 2029 104 123 228 195 12 2030 82 148 2xDIESEL HFO 12MW 230 198 11 2031 73 167 1xDIESEL HFO 7MW 1xDIESEL HFO 12MW 240 202 14 2032 73 167 240 206 11 2033 67 179 1xDIESEL HFO 12MW 246 210 12 2034 55 191 1xDIESEL HFO 12MW 246 214 10 2035 55 198 1xDIESEL HFO 7MW 253 218 11 Least Cost Electricity Master Plan (Volume 1) Final Report November 2009 Page 6.35 SECTION 6 GENERATION PLANNING Figure 6.23: Reference case (with geothermal and wind development) capacity chart New Wind New Gas Turbines Marabout Boulaos New Diesel New Geothermal Maximum Demand (MWso) Reserve Margin (%) 300 60% 50% 250 40% 200 Reserve Margin (%) Generation (GWh) 30% 150 20% 100 10% 50 0% 0 -10% 2009 2010 2011 2012 2013 2014 2015 2016 2017 2018 2019 2020 2021 2022 2023 2024 2025 2026 2027 2028 2029 2030 2031 2032 2033 2034 2035 We have developed three geothermal and wind development scenarios. These are • With 180 GWh of imports over the interconnector (25% wind capacity factor), • With up to 700 GWh of imports over the interconnector (25% wind capacity factor); and, • With up to 700 GWh of imports over the interconnector (35% wind capacity factor). We discuss the results of these scenarios below. 180 GWh energy imports, 25% wind capacity factor (Scenario 10) The introduction of a 10 MW wind farm (at 25 per cent capacity factor) to an EdD system that comprises diesel generation, up to 60 MW of geothermal plant and 180 GWh of imports over the interconnector, results in a net present value of US$ 744 million. This is US$ 340 million (31 per cent) lower than the reference case (scenario 1). Figure 6.24 below shows that energy imports are not displaced by renewable energy (in the form of wind and geothermal energy) generated by EdD. Least Cost Electricity Master Plan (Volume 1) Final Report November 2009 Page 6.36 SECTION 6 GENERATION PLANNING Figure 6.24: Generation by plant type (up to 180 GWh energy imports, 25% wind capacity factor) Interconnector New Gas Turbines Marabout Boulaos New Diesel New Wind New Geothermal Generation (GWh) 1400 1200 184 189 183 184 187 183 185 186 1000 182 183 186 187 182 185 186 183 184 Generation (GWh) 800 184 182 185 184 184 600 181 183 0 183 400 0 200 0 2009 2010 2011 2012 2013 2014 2015 2016 2017 2018 2019 2020 2021 2022 2023 2024 2025 2026 2027 2028 2029 2030 2031 2032 2033 2034 2035 Up to 700 GWh energy imports, 25% wind capacity factor (Scenario 11) The introduction of a 10 MW wind farm (at 25 per cent capacity factor) to an EdD system that comprises diesel generation, up to 60 MW of geothermal plant and importing energy over the interconnector up to 700 GWh, results in a net present value of US$ 610 million. This is US$ 474 million (44 per cent) lower than the reference case. However, Figure 6.25 below shows that energy imports do not exceed 460 GWh due to the ‘must take’ nature of the geothermal and wind energies. Least Cost Electricity Master Plan (Volume 1) Final Report November 2009 Page 6.37 SECTION 6 GENERATION PLANNING Figure 6.25: Generation by plant type (up to 700 GWh energy imports, 25% wind capacity factor) Interconnector New Gas Turbines Marabout Boulaos New Diesel New Wind New Geothermal Generation (GWh) 1400 1200 460 440 421 1000 400 381 361 344 321 306 288 272 254 242 231 Generation (GWh) 800 358 342 325 463 440 415 600 435 469 406 300 0 183 400 0 200 0 2009 2010 2011 2012 2013 2014 2015 2016 2017 2018 2019 2020 2021 2022 2023 2024 2025 2026 2027 2028 2029 2030 2031 2032 2033 2034 2035 Up to 700 GWh energy imports, 35% wind capacity factor (Scenario 12) The introduction of a 10 MW wind farm (at 35 per cent capacity factor) to an EdD system that comprises diesel generation, up to 60 MW of geothermal plant and importing energy over the interconnector up to 700 GWh, results in a net present value of US$ 604 million. This represents a US$ 6 million reduction in comparison to Scenario 11. However, this is still some US$ 29 million higher the NPV of Scenario 5 (reference case with geothermal development and up to 700 GWh of energy imports). The generation by plant type for this scenario is presented in Figure 6.26. Least Cost Electricity Master Plan (Volume 1) Final Report November 2009 Page 6.38 SECTION 6 GENERATION PLANNING Figure 6.26: Generation by plant type (up to 700 GWh energy imports, 35% wind capacity factor) Interconnector New Gas Turbines Marabout Boulaos New Diesel New Wind New Geothermal Generation (GWh) 1400 1200 461 439 419 1000 400 379 361 342 322 305 289 273 256 242 230 Generation (GWh) 800 356 345 325 464 439 415 600 439 469 407 300 0 182 400 0 200 0 2009 2010 2011 2012 2013 2014 2015 2016 2017 2018 2019 2020 2021 2022 2023 2024 2025 2026 2027 2028 2029 2030 2031 2032 2033 2034 2035 6.6.4 NPV summary Table 6.11 and Figure 6.27 below presents a summary table of NPVs for all of the scenarios considered for the base case demand forecast, including those related to the reference case.. It is apparent from Table 6.11 that both geothermal and wind power may play an important role in the future of Djibouti’s electricity supply, in many cases offering a lower cost solution to meeting demand than if Djibouti continue to meet demand by burning liquid fuel only. Least Cost Electricity Master Plan (Volume 1) Final Report November 2009 Page 6.39 SECTION 6 GENERATION PLANNING Table 6.11: Summary table for NPVs for scenarios considered Base case demand forecast scenario Scenario NPV Import Scenario Geothermal Scenario Wind Scenario No. (US$m) 5 Up to 700 GWh Import Geothermal (60 MW) No Wind 575 12 Up to 700 GWh Import Geothermal (60 MW) Wind (10 MW - 35% CF) 604 11 Up to 700 GWh Import Geothermal (60 MW) Wind (10 MW - 25% CF) 610 3 Up to 700 GWh Import No Geothermal No Wind 622 9 Up to 700 GWh Import No Geothermal Wind (10 MW - 35% CF) 654 8 Up to 700 GWh Import No Geothermal Wind (10 MW - 25% CF) 660 10 180 GWh Import Geothermal (60 MW) Wind (10 MW - 25% CF) 744 4 No Imports Geothermal (60 MW) No Wind 906 7 180 GWh Import No Geothermal Wind (10 MW - 25% CF) 922 2 180 GWh Import No Geothermal No Wind 930 6 No Imports No Geothermal Wind (10 MW - 25% CF) 1075 1 No Imports No Geothermal No Wind 1084 Figure 6.27: Summary chart of Base demand forecast expansion plans (NPVs) No Imports,  No Geothermal, No Wind No Imports,  No Geothermal, Wind (10 MW � 25% CF) 180 GWh Import, No  Geothermal, No  Wind 180 GWh Import, No  Geothermal, Wind  (10 MW � 25%  CF) No Imports,  Geothermal (60 MW), No  Wind 180 GWh Import, Geothermal (60 MW), Wind (10 MW  ,  � 25% CF) o  Up to  700 GWh  Import , No Geothermal, Wind (10 MW  � 25% CF) Up to  700 GWh  Import , No Geothermal, Wind (10 MW  � 35% CF) Up to  700 GWh  Import, No Geothermal, No Wind Up to  700 GWh  Import, Geothermal (60 MW), Wind  (10 MW � 25% CF) Up to  700 GWh  Import, Geothermal (60 MW), Wind  (10 MW � 35% CF) Up to  700 GWh  Import, Geothermal (60 MW), No Wind 0 200 400 600 800 1000 1200 NPV (US$) Least Cost Electricity Master Plan (Volume 1) Final Report November 2009 Page 6.40 SECTION 6 GENERATION PLANNING 6.7 Additional sensitivities We have additionally undertaken similar studies to explore the generation expansion plan under the high and low demand forecast scenarios. Appendix F in Volume 2 to this report presents our detailed analysis of the high and low generation scenarios carried out as part of this study. Our analysis of higher and lower demand forecasts on the least cost generation plan indicate that the future plant mix is essentially not sensitive to the demand forecast in the sense that oil-fired diesel units remain the least cost option for developing the EdD system, in addition to geothermal and wind generation should these technologies be developed. We have also considered the impact of the following on the generation expansion plans developed for this study: • Variations to the capital costs assumed, • Variations to the discount rate assumed, • Variations to the cost of fuel assumed. These sensitivity studies were based in screening curves analysis and are discussed below. 6.7.1 Capital cost A 25 per cent increase or decrease to the capital cost assumed of diesel plant has no impact on the competitiveness of geothermal plant, i.e. within the range of capital cost assumed for diesel plant, geothermal plant remains an economically attractive option should the resource be developed. An increase of 20 per cent in the capital cost estimate assumed for the geothermal plant may deem the development to be uneconomic. This emphasises the need for the careful staged development of the geothermal resource. Should the capital cost for developing wind energy be 50 per cent higher than the figure assumed in this study, Wind energy is expected to remain an economically attractive option provided its capacity factor exceeds 15 per cent. A 25 per cent increase or decrease to the capital cost estimate assumed for open cycle gas turbine units will have very little impact on the units expected capacity factor. Whilst few units have been selected as part of the least cost generation development plan for EdD, we are aware of certain characteristics which may not be suitable from EdD’s point of view. Diesel generators may well be selected in preference for practical considerations. Least Cost Electricity Master Plan (Volume 1) Final Report November 2009 Page 6.41 SECTION 6 GENERATION PLANNING 6.7.2 Discount rate Should 15 per cent discount rate be selected, then the geothermal development would not be considered an economical option. The generation development plan would be based on the use of diesel units for base load generation. 6.7.3 Fuel cost Should fuel costs be 25 per cent lower than assumed in this study, geothermal development (provided being feasible) would remain the most economical option for base load generation. 6.8 Environmental considerations In undertaking the studies to identify the most appropriate generation technologies for installation as part of the overall master plan for Djibouti, it is important to understand the potential environmental implications of the recommendations made. Ultimately for any electricity grid network that has such a low overall requirement for power (as is the case for Djibouti) it might not be especially practical to employ more efficient and/or less polluting technologies. A good example of this might be the use of combined cycle gas turbines (CCGT) - such plant have better efficiencies than gas turbines operating in simple cycle and would therefore generate less overall pollution, but they are inflexible with regard to their operation and have a limited ability to follow load and provide power quickly to help stabilise the grid at times of frequency loss. The preferred technologies for the purpose of the master plan are identified as being some small gas turbines operating in simple (or ‘open’) cycle with the bulk of generation being provided by small diesel generating plant. This is in no way uncommon for electricity networks such as that in Djibouti where the power demand is not high and there is a need for cheep and flexible forms of electricity generation. The environmental impacts of diesel engine power generation are generally fairly low with the plant having relatively high efficiencies when compared with conventional thermal power plant such as coal or oil burning plant utilising a boiler and steam turbine. Emissions of nitrogen dioxide (NO2) are higher when diesel plant are compared to other forms of generation however appropriate siting and design (e.g. stack height) can be used to ensure that the impacts associated with the installation of additional diesel plant are not of any significance to the local population or wildlife. With regard to the installation of additional gas turbines in Djibouti it should be noted that these will be relatively small in size and would not lend themselves to being installed as combined cycle plant as oppose to simple cycle machines due to limited operational flexibility afforded by CCGT technology. Least Cost Electricity Master Plan (Volume 1) Final Report November 2009 Page 6.42 SECTION 6 GENERATION PLANNING The gas turbines would be about as efficient as the diesel plant installed and would generally have a similar level of environmental impact with regard to air, water and noise emissions etc. A suitably sized CCGT plant for Djibouti would be relatively small and would be expensive to run as it would be diesel-fired. In considering the environmental impact it is of course just as important to consider the socio- economic impact of the plants installation and operation. Importantly for a developing nation, diesel plant and gas turbines are cheap, easy to operate, replace and repair and as such represent a highly appropriate technology choice. In addition, the electricity produced is affordable and is not liable to frustrate economic development through high energy prices. Both diesel engines and gas turbines are considered to be technologies that would be able to help Djibouti to accommodate renewable energy technologies on to the electrical grid were this ever to take place. Diesel engines and gas turbines are uncomplicated technologies with an ability to follow load (ramping up or down to reflect generation by renewables such as wind). The impending commercial operation of the interconnector would serve to reduce the requirement for domestic generation in country. The electricity generated in Ethiopia would be from hydroelectric plant that would reduce the need for Djibouti to rely of fossil fuel forms of generation and would help to limit the operation of any thermal plant installed in country. This would, in turn, reduce carbon dioxide and other pollutant emissions. For reasons of security of supply it is still deemed necessary for Djibouti to posses its own thermal generation such that this could be brought on line when required. Again, it would be very desirable for this generation to be flexible and to be able to follow load requirements - with diesel and gas turbine plant being obvious candidates. It can be concluded from this brief environmental review of the master plan, that the expansion plans identified in this study reflect the economic requirements of Djibouti without giving rise to any likely significant or inappropriate environmental impact. 6.9 Summary Comparison and analysis of the generation expansion planning results detailed in this Section of the report leads to the following conclusions: • The least cost generation expansion programme derived for this study assumes the availability of HFO-fired diesel and gasoil-fired OCGT only. The least cost planting programme indicates the following: Least Cost Electricity Master Plan (Volume 1) Final Report November 2009 Page 6.43 SECTION 6 GENERATION PLANNING - 24 MW of capacity is required in 2013 and a total of 187 MW would be required by the end of the planning period. - The development over the planning period would comprise ten 12 MW and six 7 MW HFO-fired diesel units, one 15 MW and one 7 MW open cycle gas turbine generating sets. - The current mix of unit sizes remains appropriate for the medium and long- term development of the system. • Our analysis of the impact of varying level of energy imports on the system NPV indicates the following: - Energy imports over the Ethiopia-Djibouti interconnector are recommended. - As long as available, EdD should maximise energy imports. • Our sensitivity analysis of the potential role for geothermal and wind power indicates the following: - The long-term development of the EdD system relying on liquid fuel only represents the most expensive alternative. - Any development option (geothermal and wind) that displaces energy generated by diesel plant is economically attractive. - Subject to confirmation of the adequacy of the geothermal resource, the development of geothermal power plant is recommended. - At the cost and performance parameters assumed for this study, geothermal power generation forms part of the long-term least cost expansion plan for EdD. - The least cost development scenario would be one in which energy imports are maximised combined with geothermal development. - For the cost and performance parameters assume for wind power generation in this study, wind power generation is an economical option, displacing energy that would be generated by HFO-fired diesel plant. Least Cost Electricity Master Plan (Volume 1) Final Report November 2009 Page 6.44 SECTION 6 GENERATION PLANNING - The displacement of import energy over the interconnector by wind energy is not recommended. - The development of geothermal and wind technologies would be worthwhile in Djibouti and their development should be taken to the next stage. - The development of these resources will be faced by many issues and significant effort and cooperation will be required between the GoD, EdD, financing institutions, contractors and consultants in order to realise the potential of this resource. • Our analysis of higher and lower demand forecasts on the least cost generation plan indicate that the future plant mix is essentially not sensitive to the demand forecast in the sense that oil-fired diesel units remain the least cost option for developing the EdD system, in addition to geothermal and wind generation should these technologies be developed. The expansion plans identified in this study reflect the economic requirements of Djibouti without giving rise to any likely significant or inappropriate environmental impact. Least Cost Electricity Master Plan (Volume 1) Final Report November 2009 Page 6.45 SECTION 7 TRANSMISSION PLANNING SECTION 7 TRANSMISSION PLANNING 7 TRANSMISSION PLANNING 7.1 General This section of the report documents the work undertaken to produce a transmission expansion plan for Djibouti for the period from 2009 to 2035. The transmission expansion plan forms part of the Least Cost Electricity Master-plan. The transmission expansion plan identifies the required expansion of the EdD transmission system over the planning period to accommodate the power generation levels and demand forecast detailed in this report. Load flow, fault level and transient stability studies were conducted with the objective of identifying the required transmission system reinforcements to accommodate the forecast growth in levels of demand and generation whilst meeting the planning criteria. 7.2 Transmission planning criteria 7.2.1 Security of supply The transmission network provides the link between generators and load centres allowing the bulk transfer of power from producers to consumers. Consequently, outages in the transmission network can have a widespread impact in terms of demand lost and number of customers affected. As a result, the economic cost of an outage in the transmission network can be large and it may be justifiable to provide redundancy in order to cope with the most frequent outages without loss of load. For planning purposes, it is common among utilities to design the transmission network so that there is no loss of supply in the event of a planned or unplanned outage, and allowing for operator action, that there would be no risk of overloading transmission plant above its nominal rating. This is known as the “N-1� criterion, where “N� represents the number of intact network elements and “1� represents a power system element suffering an outage. The transmission network is planned such that an outage of any component (e.g. overhead lines, transformers, etc) or any generating unit can be accommodated. It should be noted that there might be some loss of supply in the event of multiple or overlapping outages. Transmission circuits are generally reliable (provided they are adequately maintained) and an N-1 criterion is usually found to provide acceptable reliability. The use of an N-1 security criterion is considered appropriate for planning the Djibouti transmission system. Thermal ratings of equipment should not be exceeded during the outage of an item of plant and voltages should remain within limits indicated below. Least Cost Electricity Master Plan (Volume 1) Final Report November 2009 Page 7.1 SECTION 7 TRANSMISSION PLANNING 7.2.2 Voltage limits Maximum voltage levels should always remain within the capability of the plant and equipment in the transmission system. Voltage regulation should be as follows: • Under normal operating conditions, the voltage at any given node should not be less than 0.95 per unit and should not exceed 1.05 per unit. • Under N-1 operating conditions the voltage at any given node should not be less than 0.90 per unit and should not exceed 1.05 per unit. In all cases consideration will be given to the effect of voltage control equipment on voltage levels, including transformer tap changers, reactor and capacitor switching, and static VAR compensators as appropriate in the transmission system studies. Consideration will also be given to ensure that voltage step changes following switching operations of lines and/or items of plant remain within acceptable limits. 7.2.3 Frequency limits System frequency should be maintained between +/-2.5 per cent of 50Hz. 7.2.4 Circuit parameters Following completion of the inter-connector project, the EdD transmission system will comprise 230 kV, 63 kV and 20 kV circuits. The type, thermal rating and electrical parameters of the overhead lines and underground cables that will be in use in Djibouti are shown in Table 7.1 and Table 7.2 respectively. The thermal rating is based on the following conditions: Ambient air temperature 40 °C Ground temperature 30 °C Ground thermal resistivity 2.0 Km/W Maximum conductor temperature 75 °C Intensity of solar radiation 1200 watts/m2 Wind velocity 0.5 m/s Least Cost Electricity Master Plan (Volume 1) Final Report November 2009 Page 7.2 SECTION 7 TRANSMISSION PLANNING Table 7.1: Overhead line parameters Circuit Surge Circuit Circuit Thermal Voltage Single/ Susceptance Impedance Conductor Resistance Reactance Rating (kV) Twin (micro- Loading (ohm/km) (ohm/km) (MVA) siemens/km) (MW) 230 Ash Twin 0.110 0.399 2.87 290 142 63 Ash Single 0.220 0.399 2.92 40 11 63 Aster 366 Single 0.109 0.371 3.15 65 12 20 Aster 148 Single 0.269 0.362 3.23 11 1.2 2 20 34 mm Single 1.186 0.411 2.84 5 1.1 Table 7.2: Underground cable parameters Cross Susceptance Thermal Resistance Reactance Voltage (kV) section Type (micro- rating 2 (ohm/km) (ohm/km) (mm ) siemens/km) (MVA) 63 400 3 x 1 core Al 0.092 0.116 69.08 36 20 150 3 x 1 core Al 0.243 0.116 81.64 6.5 20 185 3 x 1 core Al 0.194 0.116 81.64 7.5 Full utilisation of transmission lines to their thermal limit is not always possible. For longer lines, the short circuit level at the sending end and voltage drop along the line limit the level of power transfer. The voltage drop along the line increases in direct proportion to the length and non-linearly with active and reactive power flow. Lagging reactive power flow in particular creates a significant voltage drop along the line. The surge impedance loading (SIL) of a line in MW represents the level of loading at unity power factor at which the line exhibits a flat voltage profile. At this loading level, the voltage drop across the line inductive reactance is just balanced by the voltage rise due to capacitive charging current. The SIL therefore provides an indication of the level of loading that a transmission line could take without a significant voltage drop. However, this provides only a rough approximation as it neglects line resistance and assumes a load of unity power factor. In Table 7.1 it may be observed that the SIL is well below the thermal rating for each of the line types shown. The 230 kV Twin Ash line will have a thermal rating in Djibouti of approximately 290 MVA per circuit, whereas the SIL for the same line is 142 MW. Figure 7.1 shows the reduction in receiving end voltage for a 280 km 230 kV Twin Ash overhead line as levels of power transfer are increased. In this case line resistance is taken into account and a Least Cost Electricity Master Plan (Volume 1) Final Report November 2009 Page 7.3 SECTION 7 TRANSMISSION PLANNING sending end voltage of 1 per unit (p.u.) was assumed. The maximum power that can be transferred over this distance without capacitive support at the receiving end is 60 - 140 MW (depending on the load power factor) to maintain the receiving end voltage above 0.9 p.u. This is well below the thermal rating of the line and will be even less than this if the short circuit level at the sending end is critically low. The distances and power levels may be increased if there is a strong grid at each end of the circuit or if there is significant generation capacity at each end to support the voltage. Shunt compensation can also be used to increase power flows by supporting the voltage at the receiving end. There are, however, limits to the levels of shunt compensation that may be used without compromising voltage stability. Figure 7.1: Power-Voltage curve for 230 kV 280 km Twin Ash line Power - Vr curve 1.1 1 0.9 0.8 0.85 pf Vr (per unit) 0.88 pf 0.7 0.91 pf 0.6 0.94 pf 0.5 0.97 pf 0.4 1 pf 0.3 Thermal rating V min 0.2 0.1 0 0 50 100 150 200 250 300 350 Delivered Power (MW) - and Line ratings (in MVA) Figure 7.2 shows the reduction in receiving end voltage for a 72 km, 63 kV Single Ash transmission line. The maximum power that can be transferred over this distance without capacitive support at the receiving end is 10 - 20 MW depending on the power factor of the load. This is well below the 40 MW thermal rating of the line. Least Cost Electricity Master Plan (Volume 1) Final Report November 2009 Page 7.4 SECTION 7 TRANSMISSION PLANNING Figure 7.2: Power-Voltage curve for 63 kV 72 km Single Ash line Power - Vr curve 1.1 1 0.9 0.8 0.85 pf Vr (per unit) 0.7 0.88 pf 0.91 pf 0.6 0.94 pf 0.5 0.97 pf 0.4 1 pf 0.3 Thermal rating V min 0.2 0.1 0 0 10 20 30 40 50 Delivered Power (MW) - and Line ratings (in MVA) 7.2.5 Short-circuit levels Short circuit levels should remain within the capability of plant. As the network expands and becomes heavily interconnected with more added generation capacity, short circuit levels will increase over existing values. In line with commonly adopted practice, for networks and planning studies of this type three phase short circuit levels will be considered in this study. 7.2.6 Transient stability Generation and generation groups should retain stability with the system for a three-phase fault which is cleared within 120ms. Stability should be retained post-fault with the faulted circuit no longer in service. The network should not exhibit any poorly damped natural frequencies that could give rise to sustained oscillations between machines or machine groups. 7.2.7 Load power factor Due to significant air conditioning demand during the peak summer months, the load was assumed to comprise 70 per cent at 0.8 power factor and 30 per cent at 0.9 power factor, resulting in a composite power factor of 0.83. This power factor was applied to the loads lumped on the 20 kV busbars and was therefore assumed to include losses between the point of load and the 20 kV busbars. Least Cost Electricity Master Plan (Volume 1) Final Report November 2009 Page 7.5 SECTION 7 TRANSMISSION PLANNING 7.3 Distribution of demand Table 7.3 shows the distribution of demand across the 63/20 kV substations in the Djibouti town region. This demand distribution was applied in the EEPCO/EDD Ethiopia-Djibouti Power Interconnection Study dated January 2008 and indicates that much of the demand growth will be accommodated at the new PK-12 and Palmeraie substations to the west of Djibouti town. Table 7.3: Distribution of demand Substation 2008 2009 2010 2011 2012 2015 2020 2025 2030 2035 Marabout 40% 40% 35% 35% 35% 30% 25% 20% 20% 20% Boulaos 60% 60% 55% 55% 55% 50% 40% 30% 30% 30% PK-12 10% 10% 10% 20% 30% 40% 40% 40% Palmeraie 5% 10% 10% 10% Following completion of the inter-connector project and the transmission line from PK-12 to Ali Sabieh, the south of the country will be interconnected with the Djibouti town system. There are no committed projects associated with the electrical interconnection of the north with Djibouti town, however this will be considered within the study, possibly associated with the potential geothermal or wind generation projects. Table 7.4 shows the assumed distributed demand forecast for the whole country based on the High Case demand forecast (Section 3.8). The demand forecast for the North and South regions was based on the same level of growth as Djibouti town. The distributed demand forecast is shown graphically in Figure 7.3. Table 7.4: Distributed demand forecast (High Case) Region Substation 2008 2009 2010 2011 2012 2015 2020 2025 2030 2035 Marabout 22.2 26.6 25.6 28.9 34.4 41.0 41.7 37.3 42.3 48.1 Boulaos 33.3 39.9 40.2 45.4 54.1 68.3 66.8 55.9 63.5 72.1 Djibouti PK-12 0.0 0.0 7.3 8.2 9.8 27.3 50.1 74.6 84.7 96.2 Town Palmeraie 0.0 0.0 0.0 0.0 0.0 0.0 8.3 18.6 21.2 24.0 Total 56 67 73 82 98 137 167 186 212 240 Ali Sabieh 1.1 1.3 1.5 1.6 2.0 2.7 3.3 3.7 4.2 4.8 South Dikhil 0.3 0.4 0.4 0.5 0.6 0.8 1.0 1.1 1.3 1.4 Total 1.4 1.7 1.9 2.1 2.5 3.5 4.3 4.8 5.5 6.2 Total Demand (MW) 57 68 75 85 101 140 171 191 217 247 Least Cost Electricity Master Plan (Volume 1) Final Report November 2009 Page 7.6 SECTION 7 TRANSMISSION PLANNING Figure 7.3: Distributed demand forecast (High Case) Marabout Boulaos PK-12 Palmeraie Ali Sabieh Dikhil 300.0 250.0 200.0 Demand (MW) 150.0 100.0 50.0 0.0 2008 2009 2010 2011 2012 2015 2020 2025 2030 2035 Table 7.4 and Figure 7.3 show the peak demand levels based on the High Case load forecast. These peak demand levels occur during the wet season in Ethiopia. The dry season peak demand levels were assumed to be 93 per cent of the wet season levels. 7.4 Generation dispatch The generation dispatch for the least-cost generation plan is provided in Table 7.5, which shows the dispatch for the following conditions: • High case demand forecast wet season peak with imports from Ethiopia. • High case demand forecast peak dry season (93 per cent of wet season peak) with no imports from Ethiopia. Least Cost Electricity Master Plan (Volume 1) Final Report November 2009 Page 7.7 SECTION 7 TRANSMISSION PLANNING Table 7.5: Least cost generation dispatch (High Case demand forecast) Dry season peak Wet season peak Unit 2009 2011 2015 2020 2035 2009 2011 2015 2020 2035 Interconnector 51.5 113.9 144.8 172.2 Diesel HFO Plant 7MW 6.8 6.8 47.5 Diesel HFO Plant 12MW 46.6 58.2 139.7 56.4 Boulaos G23 5.7 5.7 5.6 5.3 5.7 5.7 5.6 5.3 Boulaos G31 4.1 4.1 4.0 3.8 Boulaos G32 4.1 4.1 4.0 3.8 Boulaos G22 13.8 13.8 13.8 13.5 12.8 13.8 13.8 13.8 13.5 12.8 Boulaos G25 12.6 12.4 12.4 12.1 12.6 12.4 6.7 7.4 Boulaos G21 9.0 9.0 13.8 13.3 9.0 1.2 Boulaos G12 5.8 5.8 5.7 5.6 5.8 Boulaos G13 4.4 4.3 4.3 4.2 4.4 Boulaos G14 4.4 4.3 4.3 4.2 4.4 Boulaos G15 4.4 4.3 4.3 4.2 4.4 Boulaos G16 4.9 4.8 4.8 4.7 4.9 Boulaos G17 4.4 5.8 5.7 5.6 5.8 Boulaos G18 0.6 5.7 5.7 3.2 Boulaos G11 2.5 5.7 3.8 Boulaos G24 1.8 Marabout M1 Marabout M2 Marabout M3 Marabout M4 Marabout M5 Marabout M6 Total 63.6 78.9 130.6 159.7 230.0 68.2 84.6 140.0 171.2 246.7 Summary Inter-connector 51.5 113.9 144.8 172.2 Candidate plant 53.4 65.0 187.2 56.4 Boulaos 63.6 78.9 77.2 94.7 42.8 68.2 33.1 26.2 26.5 18.1 Marabout Total 63.6 78.9 130.6 159.7 230.0 68.2 84.6 140.0 171.2 246.7 7.5 Network model A detailed model of the EdD power system was developed including generation and transmission infrastructure. The key parameters on which the model was based are indicated in Section 4.4 and Least Cost Electricity Master Plan (Volume 1) Final Report November 2009 Page 7.8 SECTION 7 TRANSMISSION PLANNING this section of the report. The model was developed using the power systems analysis program DigSilent (Version 14). The model incorporates the projected increase in demand derived from the High Case demand forecast and the corresponding generation dispatch based on the least cost generation expansion plan. The timing and nature of the transmission reinforcement requirements for the Base and Low Case scenarios can be identified from the results obtained for the High Case by referencing the demand of interest to the year where that level of demand is indicated. Both wet and dry season peak demand conditions were modelled. Load flow, fault level and transient stability studies were conducted with the objective of identifying the required transmission system reinforcements to accommodate the forecast growth in levels of demand and generation whilst meeting the planning criteria. The main emphasis is on the load flow studies which are used primarily to identify the transmission circuit and reactive power requirements. Fault level studies are used to determine the prospective short circuit currents and thereby identify the switchgear rating requirements. Transient stability studies are used to confirm that the generation on the system will remain in synchronism following major, credible disturbances. 7.6 Network expansion Transmission network development to meet the load forecast and least-cost generation expansion plan while complying with the planning criteria is shown in Table 7.6. The table indicates the year in which the transmission reinforcement is required and distinguishes between the dry season peak when no power is imported from Ethiopia and the wet season peak, when power import is at a maximum. 7.6.1 Isolated regions Following completion of the inter-connector project, Division South including Ali Sabieh and Dikhil will be inter-connected with Djibouti town via 63 kV transmission line. It is therefore assumed that Ali Sabieh and Dikhil will be supplied with electricity from the main network and that the local generators in these towns will then operate in standby mode. Division North including Tadjoura and Obock will remain isolated in the base case network expansion, however separate consideration will be given to the feasibility and costs associated with inter- connection of this region with the rest of the system. The various reinforcements are briefly described below. Least Cost Electricity Master Plan (Volume 1) Final Report November 2009 Page 7.9 SECTION 7 TRANSMISSION PLANNING Table 7.6: Transmission Network Development Year Season Peak No. Transmission reinforcement Dry 2011 Wet 1 2nd Marabout-Boulaos 63 kV cable circuit Dry 2 2nd PK-12 63/20 kV 40 MVA transformer 3 2nd and 3rd Marabout 63/20 kV 36 MVA transformers 1 3rd PK-12 230/63 kV 63 MVA transformer 2015 2 3rd Boulaos 63/20 kV 36 MVA transformer 3 2nd PK-12 - Palmeraie OHL double circuit (AAAC Aster conductor) Wet 4 2nd Palmeraie-Boulaos cable circuit (2 x 400 mm2) 5 2nd Palmeraie-Marabout cable circuit (2 x 400 mm2) 6 Boulaos reactive compensation - 40 MVAr capacitor bank (4 x 10 MVAr) 1 Palmeraie 63/20 kV substation Dry 2 3rd PK-12 63/20 kV 40 MVA transformer 2020 1 PK-12 reactive compensation - 30 MVAr capacitor bank Wet 2 Marabout reactive compensation - 10 MVAr capacitor bank Dry 1 4th PK-12 63/20 kV 40 MVA transformer 2035 1 PK-12 20 kV reactive compensation - additional 20 MVAr capacitor bank/SVC Wet 2 PK-12 63 kV reactive compensation - 90 MVAr capacitor bank/SVC 7.6.2 2009 Load flow studies were conducted for the 2009 network to establish the starting point for the network expansion. Figure G.1 in Appendix G shows the load flow plot for the 2009 peak. All of the generation is provided at Boulaos, with 27.5 MW being transmitted across the 63 kV circuit to supply Marabout substation. 7.6.3 2011 Dry Season By 2011, the inter-connector and associated transmission infrastructure will have been completed. Figure G.2 shows the load flow plot for the dry season peak period when there is no power import over the inter-connector. Least Cost Electricity Master Plan (Volume 1) Final Report November 2009 Page 7.10 SECTION 7 TRANSMISSION PLANNING Power flows are within the thermal rating of equipment and voltages are within the planning limits for normal conditions. An outage of the PK12-Boulaos 63 kV circuit will cause the Marabout-Boulaos circuit to overload during the peak. Operation of three of the Marabout generators during this outage will relieve the circuit. Marabout and PK-12 63/20 kV substations do not comply with N-1 security of supply as they each have just one 63/20 kV transformer. Much of the Marabout load could however be transferred to Boulaos through 20 kV switching in the event of an outage of the Marabout 63/20 kV transformer. Wet Season Figure G.3 shows the load flow plot for the 2011 wet season peak. During an outage of a 230/63 kV transformer at PK-12, the remaining transformer will exceed its continuous thermal rating and it will be necessary to make use of the 20 per cent overload rating of the transformer during the peak period. The load on the Marabout 63/20 kV transformer will reach its continuous rating during the peak period. An outage of the PK12-Marabout 63 kV circuit will cause the Marabout-Boulaos circuit to overload during the peak. Operation of one of the Marabout generators during this outage will relieve the circuit. 7.6.4 2015 Dry Season Figure G.4 shows the load flow plot for the 2015 dry season peak. The new generation includes four 12 MW and one 7 MW HFO diesel units are assumed to be connected at PK12-63 kV via step-up transformers. The following transmission reinforcements are required to meet the 2015 dry season peak: • A 2nd Marabout-Boulaos 63 kV circuit will be required in case of outage of the PK12- Marabout 63 kV circuit. • A 2nd PK-12 63/20 kV transformer. • A 2nd and 3rd additional Marabout 63/20 kV transformer. Least Cost Electricity Master Plan (Volume 1) Final Report November 2009 Page 7.11 SECTION 7 TRANSMISSION PLANNING Wet Season Figure G.5 shows the load flow plot for the 2015 wet season peak. The following transmission reinforcements are required: • A 3rd 63 MVA 230/63 kV PK-12 transformer. • A 3rd Boulaos 63/20 kV transformer. • A 2nd additional PK-12 – Marabout and PK-12 – Boulaos 63 kV circuit. • 4 x 10 MVAr mechanically switched capacitor bank at Boulaos 20 kV. 7.6.5 2020 Dry Season Figure G.6 shows the load flow plot for the 2020 dry season peak. The following transmission reinforcements are required: • Establishment of Palmeraie 63/20 kV substation. • A 3rd PK-12 63/20 kV transformer. Wet Season Figure G.7 shows the load flow plot for the 2020 wet season peak. The following transmission reinforcements are required: • 3 x 10 MVAr mechanically switched capacitor bank at PK-12 20 kV. • 2 x 5 MVAr mechanically switched capacitor bank at Marabout 20 kV. 7.6.6 2035 Dry Season Figure G.8 shows the load flow plot for the 2035 dry season peak. The following transmission reinforcements are required: • A 4th PK-12 63/20 kV 40 MVA transformer. Wet Season Figure G.9 shows the load flow plot for the 2035 wet season peak. The following transmission reinforcements are required: • Additional 2 x 10 MVAr mechanically switched capacitor bank at PK-12 20 kV. • 3 x 30 MVAr reactive compensation by means of SVC at PK-12 63 kV. Least Cost Electricity Master Plan (Volume 1) Final Report November 2009 Page 7.12 SECTION 7 TRANSMISSION PLANNING 7.7 Short-circuit studies All generating plant contributes to the current that flows in the event of a short circuit fault on the power system. Generating plant, which is close to the fault location, will provide greatest contribution and those remote from the fault will provide very limited contribution to the total fault current. Generators equidistant from a short circuit will provide fault current contribution approximately in proportion to their rating. The short circuit current must be interrupted by one or more circuit breakers, which must have sufficient fault current interruption capability to interrupt the total prospective fault current at the point on the system at which they are located. As the levels of generation increase on a power system due to increased demand, so too will the short-circuit levels. When the prospective short circuit levels reach the switchgear rating at a particular location, it will be necessary to replace it with switchgear of a higher rating. Short circuit studies were undertaken to determine the short circuit levels on the busbars over the planning period. The short circuit levels were compared with switchgear ratings to determine whether switchgear replacement or upgrading is likely to be required. Table 7.7 shows the calculated three-phase short-circuit levels across the system for each of the years considered over the planning period. The table indicates that the three-phase short-circuit levels are well within the switchgear rating for all 20 kV and 63 kV busbars. While the single phase to earth short-circuit levels may be higher than the three-phase levels, these will also be well within the switchgear rating. Table 7.7: Short-circuit levels Voltage Switchgear Short-circuit level (kA) Substation (kV) rating (kA) 2009 2011 2015 2020 2035 Boulaos 20 25 10.5 12.6 16.7 15.5 13.1 Marabout 20 25 4.4 5.5 9.9 11.8 13.5 PK-12 20 25 6.1 9.1 13.2 18.6 Ali Sabieh 20 25 2.5 2.5 2.8 3.1 Boulaos 63 25 3.1 5 5.5 7.5 9.8 Marabout 63 25 3 4.8 5.4 7.4 9.7 PK-12 63 25 5.3 5.5 7.7 11.5 Ali Sabieh 63 25 1.6 1.6 2.1 2.7 Palmeraie 63 25 7.5 10 PK-51 63 25 3.6 5.9 Lac Assal 63 25 2.8 4 Least Cost Electricity Master Plan (Volume 1) Final Report November 2009 Page 7.13 SECTION 7 TRANSMISSION PLANNING 7.8 Transient stability studies 7.8.1 General Transient stability is concerned with the period up to a few seconds following a major disturbance such as a short circuit and considers the ability of the generators to regain synchronism once the short circuit has been cleared. During the short circuit there will be a mismatch between the mechanical power supplied to the generators and the electrical power delivered to the system. In particular, the terminal voltage of generators close to the fault location will drop sharply and therefore the electrical power output will reduce. The mechanical power will initially be unchanged from the pre- fault condition and therefore the machines close to the fault location will accelerate. The initial rate of change of speed of a machine will depend on the magnitude of the power mismatch and the combined inertia of the turbine/engine and generator. As the speed of the generators close to the fault increase, so too does the power angle (or rotor angle) with respect to the other generators on the system. When the fault is cleared, if the power angle between two sets of generators has deviated too far, the generators will not regain synchronism and pole slipping would occur, causing generators to trip. In many cases, complete system blackout will result if the generators do not regain synchronism. The critical fault clearance time is the maximum fault duration before generator pole slipping starts to occur. On transmission systems, short circuits will generally be cleared by protection within 100 ms. Transiently stability is mostly dependent on the inertia of the machines and the system impedance between machines or groups of machines. The inertia determines the rate of change of speed of a machine when there is a power mismatch, such as during a fault. The larger the inertia, the slower the rate of change of speed of a machine relative to the rest of the machines on the grid and therefore the less likely it is that the machine will lose synchronism with the other machines. The lower the impedance of the circuits between machines, the greater the synchronising power and therefore the less likely that loss of synchronism will occur. Transient stability studies were conducted to assess the stability of the EdD transmission system. In each case, a three-phase short circuit was applied as this is more onerous in terms of transient stability than a single phase to earth or two-phase short circuit. The studies were conducted on the 2011 and 2035 networks. Typical data was used to model the generators and controllers including automatic voltage regulators (AVRs) and governors. This is sufficiently accurate for the purpose of long term planning studies. The results of the transient stability studies are presented graphically in Appendix H. These plots show electrical power and generator rotor angle for a number of generators across the system. In each case, the measured parameters are indicated on the plots. Least Cost Electricity Master Plan (Volume 1) Final Report November 2009 Page 7.14 SECTION 7 TRANSMISSION PLANNING 7.8.2 2011 The following disturbances were modelled for both the dry and wet season conditions: • Three phase short-circuit at Boulaos 63 kV busbar, cleared after 200 ms by tripping Boulaos-PK-12 circuit. • Three phase short-circuit at PK-12 63 kV busbar, cleared after 200 ms by tripping Boulaos-PK-12 circuit. • Three phase short-circuit at Boulaos 20 kV busbar, cleared after 200 ms by tripping Boulaos 63/20 kV transformer. The results are shown graphically in Figures H.1 to H.6 in Appendix H. In all cases, the generators regained synchronism following fault clearance and therefore transient stability was maintained with critical fault clearance time >200 ms. 7.8.3 2035 The following disturbances were modelled for both the dry and wet season conditions: • Three phase short-circuit at Boulaos 63 kV busbar, cleared after 200 ms by tripping Boulaos-PK-12 circuit. • Three phase short-circuit at PK-12 63 kV busbar, cleared after 200 ms by tripping Boulaos-PK-12 circuit. • Three phase short-circuit at Lac Assal 63m kV busbar, cleared after 200 ms by tripping Lac Assal-PK-51 circuit. The results are shown graphically in Figures H.7 to H.12 in Appendix H. In the event of a short-circuit at Lac Assal pole-slipping of the Lac Assal generators occurred. It was found necessary to reduce the fault clearance time to 110 ms in order to achieve rotor angle stability following fault clearance. The following factors contribute to the reduced critical fault clearance time in this case: • Relatively high impedance circuits between the geothermal generators and the rest of the system. • Lack of rapid turbine control of geothermal generators. • Operating the geothermal generators close to rated output. Although the fault clearance time is reduced for a fault near Lac Assal generators, it is still within the capability of standard switchgear to clear a fault faster than the critical clearance time of 110 ms. Least Cost Electricity Master Plan (Volume 1) Final Report November 2009 Page 7.15 SECTION 7 TRANSMISSION PLANNING 7.8.4 Impact of wind power on system stability Wind power is intermittent in nature and therefore, for a stable power supply, the wind power generation must be complemented by a controllable source of power that is able to respond rapidly to changes in demand and changes in generated wind power. In the case of Djibouti, the reciprocating engines are able to provide this dynamic response provided they are operated with sufficient spinning reserve. There is an economic trade-off between the provision of spinning reserve and load shedding in the event of a drop in wind generation. The Ethiopia-Djibouti inter-connector will remove the need to provide spinning reserve within Djibouti as any fluctuations in demand or generation will be absorbed by the Ethiopian system via the Interconnector. Provided the inter-connector is in service, we do not therefore foresee significant stability issues associated with the introduction of wind power in Djibouti. 7.9 Transmission losses The calculated peak transmission losses on the EdD system over the planning period were as shown in Table 7.8. The losses increase from 0.9 per cent in 2009 to 2.9 per cent in 2035. These losses are at an acceptable level for a transmission system of this type. Table 7.8: Transmission losses Peak losses Year Season (% of (MW) peak demand) 2009 0.58 0.9% Dry 0.67 0.8% 2011 Wet 1.19 1.4% Dry 0.84 0.6% 2015 Wet 1.88 1.3% Dry 3.68 2.3% 2020 Wet 4.62 2.7% Dry 6.28 2.7% 2035 Wet 7.28 2.9% 7.10 Transmission expansion costs The costs of transmission expansion to meet the High Case demand forecast and least cost generation plan are shown in Table 7.9. These are based on the unit costs for transmission infrastructure shown in Table 7.10. In each case, the costs include the switchgear associated with the transmission reinforcement project. Least Cost Electricity Master Plan (Volume 1) Final Report November 2009 Page 7.16 SECTION 7 TRANSMISSION PLANNING Table 7.9: Transmission costs Capital Costs Incurred in Year ($000) Year Season Peak No. Transmission reinforcement 2015 2020 2035 Dry 2011 Wet 1 2nd Marabout-Boulaos 63 kV cable circuit 2,600 Dry 2 2nd PK-12 63/20 kV 40 MVA transformer 1,550 3 2nd and 3rd Marabout 63/20 kV 36 MVA transformers 3,100 1 3rd PK-12 230/63 kV 63 MVA transformer 4,200 2015 2 3rd Boulaos 63/20 kV 36 MVA transformer 1,550 3 2nd PK-12 - Palmeraie OHL double circuit (AAAC Aster conductor) 2,296 Wet 4 2nd Palmeraie-Boulaos cable circuit (2 x 400 mm2) 2,500 5 2nd Palmeraie-Marabout cable circuit (2 x 400 mm2) 2,950 6 Boulaos reactive compensation - 40 MVAr capacitor bank (4 x 10 MVAr) 1,000 1 Palmeraie 63/20 kV substation 10,000 Dry 2 3rd PK-12 63/20 kV 40 MVA transformer 1,550 2020 1 PK-12 reactive compensation - 30 MVAr capacitor bank 750 Wet 2 Marabout reactive compensation - 10 MVAr capacitor bank 300 Dry 1 4th PK-12 63/20 kV 40 MVA transformer 1,550 2035 1 PK-12 20 kV reactive compensation - additional 20 MVAr capacitor bank/SVC 500 Wet 2 PK-12 63 kV reactive compensation - 90 MVAr capacitor bank/SVC 9,300 Total ($000) 21,746 12,600 11,350 Table 7.10: Transmission equipment unit costs Unit Unit cost ($000) 230/63 kV 63 MVA transformer Each 1500 230 kV switchgear bay Each 2000 63 kV swgr bay Each 700 63/20 kV 15 MVA tx Each 600 63/20 kV 40 MVA tx Each 800 63 kV Double Circuit OHL (Ash) km 80 63 kV Double Circuit OHL (Aster) km 112 63 kV 1 x 400 mm2 cable km 250 63 kV 2 x 400 mm2 cable km 450 20 kV Reactive compensation (MSC) MVAr 20 63 kV Reactive compensation (MSC/SVC) MVAr 80 20 kV Cable km 60 20 kV Line km 45 20 kV swgr panel Each 50 Least Cost Electricity Master Plan (Volume 1) Final Report November 2009 Page 7.17 SECTION 7 TRANSMISSION PLANNING 7.10.1 Interconnection of Division North A brief study has been conducted to assess the technical requirements and costs associated with inter-connection of Division North including the towns of Tadjoura and Obock with the main transmission system. This could be achieved (following grid connection of the geothermal generation at Lac Assal) by installing a 63 kV circuit from Ghoubet to Tadjoura (30 km) and then installing a 20 kV circuit from Tadjoura to Obock (50 km). The approximate costs associated with this scheme, including the necessary 63/20 kV and 20/0.4 kV substations are as follows: Grid connection of Tadjoura US$ 4.4 million Grid connection of Obock US$ 2.6 million (assuming prior grid connection of Tadjoura) 7.11 Summary The transmission expansion plan detailed for this study is summarised below: • The transmission network was developed to meet the N-1 planning criterion, meaning that thermal ratings of equipment should not be exceeded during the outage of an item of plant and voltages should remain within the planning limits. • A detailed model of the EdD power system was used to conduct load flow, short- circuit level and transient stability studies. • Seasonality of demand and use of the interconnector was considered. • The studies indicate that transmission network reinforcement including additional 63 kV circuits and 63/20 kV transformers will be required from 2015. • Short-circuit levels were found to be within the switchgear rating over the planning period. • Transient stability was found to be acceptable for the cases considered. Least Cost Electricity Master Plan (Volume 1) Final Report November 2009 Page 7.18 SECTION 8 DISTRIBUTION PLANNING SECTION 8 DISTRIBUTION PLANNING 8 DISTRIBUTION PLANNING 8.1 General In this section of the report we consider the design and performance of the distribution system in Djibouti. We then assess the distribution system requirements to meet the load growth over the planning period. 8.2 Performance of existing distribution network 8.2.1 MV feeder loading Table 8.1 shows feeders’ peak loads (September 2008) and their thermal rating. A number of the feeders comprise several types of conductor, each having different thermal rating as indicated. Table 8.1: MV feeder loads Length Peak Peak Load Main Substation MV Feeder (m) Type Size Rating (MVA) Load (A) (MVA) Telex 5,315 UG 185/150 7.5/6.5 139 4.8 Esperey 6,972 UG 150 6.5 105 3.6 Artois 2,854 UG 150 6.5 58 2.0 Pharmacie 2,585 UG 150 32 1.1 Marabout Batignolles 6,125 UG 185 7.5 100 3.5 Conteneur 4,850 UG 185/150 7.5/6.5 50 1.7 Total 70,108 UG/OH 185/150/34.4/148 7.5/6.5/5/11 230 8.0 Z. Franche 2,930 UG 150 6.5 151 5.2 Rocade 2,940 UG 185/150 7.5/6.5 38 1.3 Kempinski UG 150 6.5 65 2.3 Sogik 8,288 UG 185/150 7.5/6.5 154 5.3 Cooperation 11,072 UG 185/150 7.5/6.5 109 3.8 Sirage 7,799 UG 185/150 7.5/6.5 165 5.7 190LGTS 36,596 UG/OH 185/150/34.4 7.5/6.5/5 153 5.3 Boulaos LYS 5,460 UG 185/150 7.5/6.5 42 1.5 Lettelier 5,583 UG 185/150 7.5/6.5 168 5.8 O. Kamil 10,650 UG 185/150 7.5/6.5 198 6.9 G. Batal 53,528 UG 185/150/34.4 7.5/6.5/5 208 7.2 Bowling 8,605 UG 185/150 7.5/6.5 54 1.9 Total length 252,260 76.9 The peak loads are within the thermal rating of the feeders apart from the Total feeder, where the thermal rating is slightly exceeded during peak load conditions. It should be noted however that the thermal ratings indicated are approximate only and may in reality be higher or lower than indicated Least Cost Electricity Master Plan (Volume 1) Final Report November 2009 Page 8.1 SECTION 8 DISTRIBUTION PLANNING depending on ambient conditions and details of installation. Also, the peak load period may be relatively short and the off-peak load considerably lower than the peak. Then, depending on the thermal time constant of the cable, the thermal limits on which the rating is based may not be exceeded despite a peak load in excess of the rating. With load growth, it will be necessary to either replace certain feeder sections with larger cross- section cables (240 mm2) or to install additional MV feeders and redistribute the load. 8.2.2 Voltage performance Within Djibouti town, all areas are served by underground medium voltage feeders which exhibit good voltage regulation (less than 2 per cent voltage drop expected). However, supply to the village of Arta is by means of a very long overhead circuit and calculations indicate that the voltage drop across the MV feeder to Arta may be as high as 20 per cent during peak load conditions which is clearly unacceptable. The voltage at Arta may be improved by the following means: • Power factor correction at Arta by means of capacitor banks. • Specification of distribution transformers with off-nominal turns ratio, i.e. 20/0.41 kV or 19/0.4 kV. • An additional overhead line circuit to share the load. An additional overhead line circuit will be costly due to the length of the circuit and therefore a combination of power factor correction and distribution transformers with off-nominal turns ratios is likely to be the most effective means of improving the voltage at Arta. 8.3 Losses Losses are important in the planning, designing and operating of a distribution system. A reduction in losses either reduces capital investment for generation and transmission or releases more power for demand and energy sales. Some of the benefits derived from a reduction in losses are: • Deferment of future investments; • Reduction in fixed costs; • Lower operating costs; • Increases in energy sales; • Improvement in voltage levels. Least Cost Electricity Master Plan (Volume 1) Final Report November 2009 Page 8.2 SECTION 8 DISTRIBUTION PLANNING 8.3.1 Review of technical & non-technical losses study A Technical and Non-Technical Losses Study was recently conducted by consultants for EdD. The study identified total losses for 2008 of 27 per cent comprising 12 per cent technical and 15 per cent non-technical losses. The study concluded that the technical losses could be reduced to around 7 per cent by balancing the loads across the three phases and by improving power factor. The report stated that measurements conducted between August and September 2008 identified: • Phase imbalance on approximately 1/3 of distribution transformers. • Poor load power factor of most MV customers. EdD has introduced reactive power charging for industrial customers which should result in reduced losses and should also increase the amount of active power that may be delivered by the generators. Some long lines contribute to relatively high losses but most 20 kV lines are loaded to less than 60 per cent of rating and therefore require no new investment. It is understood that EdD are addressing voltage issues at Arta by installing 20/0.41 kV or 19/0.4 kV distribution transformers. The report also notes that a 6 MVAr capacitor bank could be used to improve the voltage at Arta. The report concluded that re-configuration of the MV network would not result in significant loss reduction as most of the network is composed of adequately sized 150 mm2 feeders. Whilst our analysis agrees with the overall level of losses, we believe that the split is around 15 per cent technical losses and 10 to 12 per cent non-technical losses. As previously discussed, technical losses of around 15 per cent is not considered excessive for a system of EdD’s size and equipment. However, the level of non-technical losses is certainly high and EdD is urged to investigate measures and actions that would lead to a rapid reduction in the level of non-technical losses. Reduction of non-technical losses could be through: • Improvement in billing systems • Installation of pre-payment meters. Least Cost Electricity Master Plan (Volume 1) Final Report November 2009 Page 8.3 SECTION 8 DISTRIBUTION PLANNING 8.4 Design considerations Long term development of the distribution system should include consideration of the following basic design criteria: • Safety • Continuity of supply • Voltage within statutory limits • Minimum capital cost • Ease and minimum cost of maintenance • Ease of expansion to meet load growth • Standardisation of equipment. 8.5 Network configuration The level of inter-connection of the MV distribution feeders particularly within Djibouti town leads to good flexibility and supply security. The ring-main arrangement allows each urban feeder to be supplied from either end. A faulty section of circuit or a substation may be isolated and supply restored in the time taken for manual switching. The more lightly loaded rural areas may be served by single radial feeders. In this case restoration of supply following a fault will be in the time taken to repair the faulted section. It is recommended that expansion of the network is based on the same configuration. 8.6 Optimum conductor size The economic loading of a feeder is generally considerably less than the thermal rating. For a particular level of loading, the optimum conductor size for a distribution feeder may in theory be selected by evaluating the total costs of a number of standard conductor sizes and selecting the lowest cost option. The costs of a distribution feeder include the cost of losses and the capital cost. For a particular section of feeder, the annual costs of several different standard sized conductors may be compared and the least cost option selected. In general however, provided the peak feeder demand may be met and the voltage regulation is within the allowable limits, then the smallest standard section feeder which meets these criteria is likely to provide the lowest cost solution. This is due to the fact that the load on the feeder will be considerably less during off-peak periods and therefore the average load will be much less than the thermal rating of the feeder. Least Cost Electricity Master Plan (Volume 1) Final Report November 2009 Page 8.4 SECTION 8 DISTRIBUTION PLANNING 8.7 Average incremental cost of distribution Distribution expansion typically requires replication of the existing distribution system. This means that the incremental cost of distribution will approximately equal the average cost. The average cost of distribution may be calculated by dividing the current replacement cost of the distribution assets by the maximum demand. In this case, the marginal cost of distribution capacity is calculated from the replacement cost of the complete distribution system in Djibouti town and Arta. This should however reflect any anticipated changes in design practice. It has been assumed that the utilisation level and therefore the level of redundancy of the distribution assets will remain at the present level. The system considered includes the 20 kV feeder panels at Boulaos and Marabout, 20 kV distribution circuits, distribution substations and LV feeders. The local costs associated with the service line from the LV distribution main to the customer’s meter and the meter are not included in the calculation and should be accounted for separately. It has been assumed that the distribution systems in the outlying towns will be expanded and reinforced in a similar manner to Djibouti town (depending on load growth in those towns) and that the marginal cost of distribution capacity in these towns will be similar to that for Djibouti town. 8.7.1 Asset quantities The asset quantities were obtained through discussions with EdD and are intended to be representative of the complete distribution system for Djibouti town as of 2009. These include the MV feeder circuit breakers at Boulaos and Marabout, all MV distribution circuits, distribution substations including both ‘cabine’ and H61 type, and all LV feeder circuits. These quantities are set out in Table 8.2. 8.7.2 Unit costs The unit costs shown in Table 8.2 include for supply and installation, as well as an allowance for project management, site survey and design. These are intended to reflect the present day costs for expansion of the distribution system. The distribution substation costs include MV switchgear, transformer, LV switchgear and building in the case of the ‘cabine’ type. For the H61 type, the costs include the transformer and LV fuses. The breakdown of costs for distribution substations is shown in Table 8.2. 8.7.3 Marginal cost of distribution The total replacement cost of the distribution assets in Djibouti town may be calculated from the asset quantities and unit costs as shown in Table 8.2. Least Cost Electricity Master Plan (Volume 1) Final Report November 2009 Page 8.5 SECTION 8 DISTRIBUTION PLANNING The marginal cost of distribution (US$ per kW) may then be derived from the total replacement cost and the present system maximum demand. The peak demand for 2009 is expected to be around 68 MW and this figure is used in the calculation of marginal costs. The marginal or average incremental cost of distribution network expansion is estimated at $694/kW as shown in Table 8.2. Table 8.2: Marginal cost calculation 2009 Unit 2009 Unit 2009 Total Cost Cost Cost Quantity (DJF 000's) (US$ 000's) ( US$ 000) 20 kV Feeder Circuit Breakers Boulaos 17,504 98.9 10 989 Marabout 17,504 98.9 10 989 20 kV Feeders Underground Cable (150mm2 + 185mm2) 10,053 56.8 140 7,952 OHL (148 mm2) 8,742 49.4 51 2,519 OHL (34 mm2) 4,917 27.8 77 2,139 Pole Mounted Substations 50 kVA 913 5.2 1 5 100 kVA 1,149 6.5 6 39 160 kVA 1,356 7.7 7 54 Ground Mounted Substations 250 kVA 14,406 81.4 22 1,791 400 kVA 14,681 82.9 113 9,372 630 kVA 15,182 85.8 98 8,406 800 kVA 15,474 87.4 12 1,049 1000 kVA 15,901 89.8 10 898 LV Feeders LV Underground Cable (3*150mm2+70mm2) 7,649 43.2 100 4,322 LV ABC (3*70mm2+54mm2) 4,753 26.9 248 6,660 Total cost (US$ 000's) 47,183 Maximum demand (MW) 68 Average incremental cost ($/kW) 694 8.8 Annual load related and non-load related expenditure. The annual load related investment on distribution infrastructure is then simply calculated by applying the average incremental cost of distribution equipment to the load forecast as shown in Table 8.3. Assuming an average distribution asset life of 40 years, the average annual non-load related expenditure is calculated by assuming that on average, 1/40th of the existing asset base will be replaced each year over the planning period. The total annual expenditure on distribution is then the sum of the load related and non-load related expenditure. Least Cost Electricity Master Plan (Volume 1) Final Report November 2009 Page 8.6 SECTION 8 DISTRIBUTION PLANNING Table 8.3: Distribution expenditure 2010 2011 2012 2013 2014 2015 2016 2017 2018 2019 2020 2021 2022 Maximum demand (MW) 74.5 83.8 99.7 116.3 128.2 137.6 148.6 153.3 158.1 161.6 165.2 167.9 170.6 Load growth (MW) 6.5 9.3 15.9 16.6 11.9 9.4 11 4.7 4.8 3.5 3.6 2.7 2.7 Load related expenditure ($ 000) 4,510 6,453 11,033 11,518 8,257 6,522 7,633 3,261 3,331 2,429 2,498 1,873 1,873 Non-load related expenditure ($ 000) 1,180 1,180 1,180 1,180 1,180 1,180 1,180 1,180 1,180 1,180 1,180 1,180 1,180 Total expenditure ($ million) 5.7 7.6 12.2 12.7 9.4 7.7 8.8 4.4 4.5 3.6 3.7 3.1 3.1 2023 2024 2025 2026 2027 2028 2029 2030 2031 2032 2033 2034 2035 Maximum demand (MW) 173.5 176.4 179.9 183.4 187 190.8 194.6 198.5 202.3 206.2 210.2 214.3 218.5 Load growth (MW) 2.9 2.9 3.5 3.5 3.6 3.8 3.8 3.9 3.8 3.9 4 4.1 4.2 Load related expenditure ($ 000) 2,012 2,012 2,429 2,429 2,498 2,637 2,637 2,706 2,637 2,706 2,775 2,845 2,914 Non-load related expenditure ($ 000) 1,180 1,180 1,180 1,180 1,180 1,180 1,180 1,180 1,180 1,180 1,180 1,180 1,180 Total expenditure ($ million) 3.2 3.2 3.6 3.6 3.7 3.8 3.8 3.9 3.8 3.9 4.0 4.0 4.1 8.9 Summary The distribution expansion plan of this system is summarised below: • The distribution system in Djibouti town is well designed and is capable of high levels of reliability. Most circuits are loaded well within their thermal rating and voltage regulation is good within Djibouti town. • Expansion of the distribution system should follow present practice; however it may be necessary to replace certain feeders with 240 mm2 cables to meet demand growth. • Low voltage levels are experienced at Arta due to the length of the 20 kV feeders to Arta. • Shunt compensation and specification of transformers with off-nominal turns ratios may be used to improve voltage levels at Arta. Least Cost Electricity Master Plan (Volume 1) Final Report November 2009 Page 8.7 SECTION 9 ISOLATED SYSTEMS SECTION 9 ISOLATED SYSTEMS 9 ISOLATED SYSTEMS The EdD power system comprises Djibouti town, Division North and Division South respectively. The three regions are not electrically interconnected at present although Division South will be interconnected with Djibouti town as part of the Interconnector project. Figure 9.1 below shows the relative location of Tadjoura and Obock (Division North), Dikhil and Ali Sabieh (Division South) and Djibouti-Ville. The Ethiopia-Djibouti interconnector will connect to the main grid network of Djibouti-Ville at the substation PK12 from which 63kV circuits will feed the towns of Dikhil and Ali Sabieh, connecting Division South to the main EdD grid in 2011. Supply and demand for electricity in Division South are therefore included in our analysis of the main grid network. However, due to the relative remoteness of both Tadjoura and Obock, there are currently no plans to connect Division North to the main grid (see Section Figure 9.1 for an estimate of costs). In this section of the report, separate electricity development plans are detailed for both Tadjoura and Obock. Figure 9.1: Map of Djibouti Least Cost Electricity Master Plan (Volume 1) Final Report November 2009 Page 9.1 SECTION 9 ISOLATED SYSTEMS 9.1 Tadjoura Tadjoura is the oldest town in Djibouti and is situated on the Gulf of Tadjoura. It is home to a population of around 25,000 people. Tadjoura has an airstrip and is linked by ferry with Djibouti City. It is known for its whitewashed buildings and nearby beaches. 9.1.1 Demand Forecast Table 9.1 below presents the historical maximum demand and energy data for the isolated network in 16 Tadjoura . This data covers the period between 2002 and 2007 only. The maximum demand in Tadjoura in 2007 was approximately 880 kW and the system load factor was approximately 50 per cent. Table 9.1: Historical electricity statistics (Tadjoura) Tadjourah 2002 2003 2004 2005 2006 2007 Maximum Demand (MW) 0.58 0.68 0.75 0.80 0.85 0.88 Maximum Demand Growth (%) 17.2% 10.3% 6.7% 6.3% 3.5% Energy Generated (GWh) 3.455 3.867 4.002 3.405 3.867 Losses (GWh) 0.503 0.573 0.570 0.241 0.131 Losses (%) 17.0% 17.4% 16.6% 7.6% 3.5% Energy Delivered (GWh) 2.953 3.295 3.432 3.164 3.736 Load Factor (%) 58.0% 58.9% 57.1% 45.7% 50.2% The data available for this isolated network is not as complete or reliable as that of the main grid and therefore the type of demand forecast technique that can be used to determine future demand for electricity is limited. Figure 9.2 below presents historical sales and the historical level of maximum demand in Tadjoura for the period 2002 to 2007. To derive a reasonable demand forecast for this isolated network we have adopted a historical trend 17 analysis technique . This technique has been applied to both historical maximum demand and historical sales (energy delivered) data. In order to derive a forecast of both sales and maximum demand, best-fit trend lines have been fitted to the historical data and extended for a period of 28 years to cover the period 2008 to 2035. The 16 The maximum demand, generation sent out and generation delivered data presented in Table 9.1 are provided in the Annual Reports of EdD for the years 2002 to 2007. Loss data, load factor and maximum demand growth have been derived using the data made available. 17 Trend analysis is the identification of trends in historical data using the Microsoft Excel trend line function and then using the equation of the trend line to determine a future value. Least Cost Electricity Master Plan (Volume 1) Final Report November 2009 Page 9.2 SECTION 9 ISOLATED SYSTEMS equation of the best-fit trend line can then be used to determine the future level of demand and sales. Figure 9.3 below presents the trend line analysis used for Tadjoura and details the forecasting equation of the trend line. Figure 9.2: Historical maximum demand and generation (Tadjoura) Maximum Demand (MW) Energy Delivered (GWh) 1.0 5.0 0.9 4.5 0.8 4.0 0.7 3.5 Maximum Demand (MW) Energy Delivered (GWh) 0.6 3.0 0.5 2.5 0.4 2.0 0.3 1.5 0.2 1.0 0.1 0.5 0.0 0.0 2001 2002 2003 2004 2005 2006 2007 2008 Least Cost Electricity Master Plan (Volume 1) Final Report November 2009 Page 9.3 SECTION 9 ISOLATED SYSTEMS Figure 9.3: Maximum demand and generation trend analysis (Tadjoura) Maximum Demand (MW) Energy Delivered (GWh) Linear (Maximum Demand (MW)) Expon. (Energy Delivered (GWh)) 3.0 15.0 2.5 12.5 y =  0.05886x  � 117.22248 R² =  0.96329 2.0 10.0 Maximum Demand (MW) y =  1.20908E� 37e 4.29932E�02x Energy Delivered (GWh) R² = 5.94948E�01 1.5 7.5 1.0 5.0 0.5 2.5 0.0 0.0 2000 2005 2010 2015 2020 2025 2030 2035 2040 By applying the respective trend line equations, a maximum demand and sales forecast can be derived. To derive a forecast of sent out generation, we have assumed a level of losses of 15 per cent throughout the forecast period. These losses have been added to the sales forecast to derive the sent out energy. Table 9.2 below presents the demand forecast for the isolated Tadjoura network. Least Cost Electricity Master Plan (Volume 1) Final Report November 2009 Page 9.4 SECTION 9 ISOLATED SYSTEMS Table 9.2: Demand forecast (Tadjoura) Maximum Maximum Generation Demand Sales Losses Load Factor Year Demand Sent Out Growth (GWh) (%) (%) (MWso) (GWh so) (%) 2007 0.88 3.736 3.5% 3.867 50.2% 2008 0.96 9.4% 3.761 15.0% 4.424 52.5% 2009 1.02 6.1% 3.926 15.0% 4.618 51.6% 2010 1.08 5.8% 4.098 15.0% 4.821 50.9% 2011 1.14 5.4% 4.278 15.0% 5.033 50.4% 2012 1.20 5.2% 4.466 15.0% 5.254 50.1% 2013 1.26 4.9% 4.662 15.0% 5.485 49.8% 2014 1.32 4.7% 4.867 15.0% 5.726 49.7% 2015 1.37 4.5% 5.081 15.0% 5.978 49.6% 2016 1.43 4.3% 5.304 15.0% 6.240 49.7% 2017 1.49 4.1% 5.537 15.0% 6.514 49.8% 2018 1.55 3.9% 5.780 15.0% 6.801 50.0% 2019 1.61 3.8% 6.034 15.0% 7.099 50.3% 2020 1.67 3.7% 6.300 15.0% 7.411 50.7% 2021 1.73 3.5% 6.576 15.0% 7.737 51.1% 2022 1.79 3.4% 6.865 15.0% 8.077 51.6% 2023 1.85 3.3% 7.167 15.0% 8.431 52.2% 2024 1.90 3.2% 7.482 15.0% 8.802 52.8% 2025 1.96 3.1% 7.810 15.0% 9.189 53.4% 2026 2.02 3.0% 8.153 15.0% 9.592 54.2% 2027 2.08 2.9% 8.512 15.0% 10.014 54.9% 2028 2.14 2.8% 8.885 15.0% 10.454 55.8% 2029 2.20 2.8% 9.276 15.0% 10.913 56.7% 2030 2.26 2.7% 9.683 15.0% 11.392 57.6% 2031 2.32 2.6% 10.109 15.0% 11.893 58.6% 2032 2.38 2.5% 10.553 15.0% 12.415 59.7% 2033 2.43 2.5% 11.016 15.0% 12.960 60.8% 2034 2.49 2.4% 11.500 15.0% 13.530 62.0% 2035 2.55 2.4% 12.006 15.0% 14.124 63.2% 9.1.2 Generation plan At present there are 6 x 600 kVA high speed gas-oil fired diesel generating units located in Tadjoura. Each unit has an available generating capacity of approximately 0.36 MW. Two of these units were installed in 2005 and the remaining four were installed in 2008. It is assumed that each of these units will operate for 25 years and that no capacity additions are currently planned. As derived in Section 9.1.1 above, demand for electricity in Tadjoura is forecast to increase from 1,022 kW in 2009 to 1,375 kW in 2015, to 1,963kW by 2025 and to 2,552 kW by 2035. The relationship between the available capacity of the existing plant and the forecast of peak demand is shown in Figure 9.4 below. It can be seen that the balance between supply and demand is positive Least Cost Electricity Master Plan (Volume 1) Final Report November 2009 Page 9.5 SECTION 9 ISOLATED SYSTEMS for the years 2009 to 2028, indicating that additional generating capacity may not be required until then. However, when allowance is made for the need to carry reserve capacity, plant will need to be added to the system earlier than 2028. Figure 9.4: Supply / demand power balance (Tadjoura) Existing Available   Capacity  (MW) Additional Available  Capacity  (MW) Demand Forecast (MW) Reserve Margin (%) 3.0 150% 2.5 100% 2.0 50% Capacity  /  Demand  (MW) Reserve  Margin  (%) 1.5 0% 1.0 �50% 0.5 �100% 0.0 �150% It has been assumed for this study that any isolated system should operate to an “N-1� planning criterion (i.e. demand can be met with the largest unit out of service). On this basis, we have derived the generation plan presented in Table 9.3 below. This generation plan indicates the need of a further 18 9 high speed diesel units (fired on gas-oil) over the forecast period, the first being required in 2023 . Figure 9.5 shows the supply/demand power balance when capacity is added to the system as per Table 9.3. 18 It should be noted that although not considered for this study, localised wind, solar and geothermal power developments could be considered as possible generation expansion options in Tadjoura. As with large-scale power generation however, the availability of these resources for even small-scale generation needs to be confirmed before these resources can be considered for the area. Least Cost Electricity Master Plan (Volume 1) Final Report November 2009 Page 9.6 SECTION 9 ISOLATED SYSTEMS Table 9.3: Generation plan (Tadjoura) Least Cost Electricity Master Plan (Volume 1) Final Report November 2009 Page 9.7 SECTION 9 ISOLATED SYSTEMS Figure 9.5: Supply / demand power balance after capacity additions (Tadjoura) Existing Available   Capacity  (MW) Additional Available  Capacity  (MW) Demand Forecast (MW) Reserve  Margin (%) 3.5 140% 3.0 120% 2.5 100% Capacity  /  Demand  (MW) Reserve  Margin  (%) 2.0 80% 1.5 60% 1.0 40% 0.5 20% 0.0 0% 2009 2010 2011 2012 2013 2014 2015 2016 2017 2018 2019 2020 2021 2022 2023 2024 2025 2026 2027 2028 2029 2030 2031 2032 2033 2034 2035 Assuming a capital cost of US$ 130,000 for a 600 kVA diesel generator, and a CAPEX phasing programme of 50 per cent in year Yn-1 and 50 per cent in year Yn, the total capital cost of investment in generating facilities in Tadjoura is estimated to be approximately US$ 145,000 (at 10 per cent discount rate, in 2008 prices). It should be noted that the capital cost of the generating plant includes for a building extension (a simple steel framed sheet metal clad building with slab floor) and associated civil works to house the new generating sets. 9.2 Obock Obock is a small port town in Djibouti, located on the northern shore of the Gulf of Tadjoura where it opens out into the Gulf of Aden. The population in Obock is estimated to be approximately 8,300. The town is home to an airstrip and has ferries to Djibouti City, while mangroves lie nearby. 9.2.1 Demand Forecast Table 9.4 below presents the historical maximum demand and energy data for the isolated network in Obock. Only generation sent out and generation delivered data presented in Table 9.4 are provided in the Annual Reports of EdD for the years 2003 to 2007. In order to derive the historical level of maximum demand we have assumed a load factor of 54 per cent. This assumption is based on the Least Cost Electricity Master Plan (Volume 1) Final Report November 2009 Page 9.8 SECTION 9 ISOLATED SYSTEMS average load factor experienced in Tadjoura for the period 2002 to 2007. The maximum demand in Tadjoura in 2007 is therefore assumed to be approximately 420 kW. Table 9.4: Historical electricity statistics (Obock) Obock 2002 2003 2004 2005 2006 2007 Maximum Demand (MW) 0.309 0.373 0.374 0.384 0.418 Maximum Demand Growth (%) 0.0% 20.7% 0.2% 2.6% 8.9% Energy Generated (GWh) 1.664 1.918 1.893 1.814 1.984 Losses (GWh) 0.203 0.154 0.125 - 0.008 Losses (%) 13.9% 8.7% 7.1% 0.0% 0.4% Energy Delivered (GWh) 1.462 1.765 1.768 1.814 1.976 Load Factor (%) 54.0% 54.0% 54.0% 54.0% 54.0% To derive a reasonable demand forecast for this isolated network we have used historical trend analysis techniques. This technique has been applied to both historical maximum demand and historical sales (energy delivered) data. Figure 9.6 below presents historical sales and the historical level of maximum demand in Obock for the period 2003 to 2007. Figure 9.6: Historical maximum demand and generation (Obock) Maximum Demand (MW) Energy Delivered (GWh) 0.5 5.0 0.4 4.0 Maximum Demand (MW) Energy Delivered (GWh) 0.3 3.0 0.2 2.0 0.1 1.0 0.0 0.0 2002 2003 2004 2005 2006 2007 2008 In order to derive a forecast of both sales and maximum demand, best-fit trend lines have been fitted to the historical data and extended for a period of 28 years to cover the period 2008 to 2035. The Least Cost Electricity Master Plan (Volume 1) Final Report November 2009 Page 9.9 SECTION 9 ISOLATED SYSTEMS equation of the best-fit trend line can then be used to determine the level of future demand and sales. Figure 9.7 below presents the trend line analysis used for Obock and details the trend line equation. Figure 9.7: Maximum demand and generation trend analysis (Obock) Maximum Demand (MW) Energy Delivered (GWh) Linear (Maximum Demand (MW)) Linear (Energy Delivered (GWh)) 1.2 6.0 1.0 5.0 y =  0.02280x  � 45.34556 R² =  0.83843 0.8 4.0 y =  0.10781x   � 214.39353 Maximum Demand (MW) Energy Delivered (GWh) R² =  0.83843 0.6 3.0 0.4 2.0 0.2 1.0 0.0 0.0 2000 2005 2010 2015 2020 2025 2030 2035 2040 By applying the respective trend line equations, a maximum demand and sales forecast can be derived. We have assumed a level of losses of 15 per cent throughout the forecast period. To derive a forecast of sent out generation, these losses have been added to the sales forecast. Table 9.5 below presents the demand forecast for the isolated Obock network. Least Cost Electricity Master Plan (Volume 1) Final Report November 2009 Page 9.10 SECTION 9 ISOLATED SYSTEMS Table 9.5: Demand forecast (Obock) Maximum Maximum Generation Demand Sales Losses Load Factor Year Demand Sent Out Growth (GWh) (%) (%) (MWso) (GWh so) (%) 2007 0.42 1.976 0.4% 1.984 54.0% 2008 0.44 4.5% 2.089 15.0% 2.458 64.2% 2009 0.46 5.2% 2.197 15.0% 2.584 64.2% 2010 0.48 5.0% 2.305 15.0% 2.711 64.2% 2011 0.51 4.7% 2.412 15.0% 2.838 64.1% 2012 0.53 4.5% 2.520 15.0% 2.965 64.1% 2013 0.55 4.3% 2.628 15.0% 3.092 64.1% 2014 0.57 4.1% 2.736 15.0% 3.219 64.1% 2015 0.60 4.0% 2.844 15.0% 3.345 64.0% 2016 0.62 3.8% 2.951 15.0% 3.472 64.0% 2017 0.64 3.7% 3.059 15.0% 3.599 64.0% 2018 0.66 3.6% 3.167 15.0% 3.726 64.0% 2019 0.69 3.4% 3.275 15.0% 3.853 64.0% 2020 0.71 3.3% 3.383 15.0% 3.980 63.9% 2021 0.73 3.2% 3.490 15.0% 4.106 63.9% 2022 0.76 3.1% 3.598 15.0% 4.233 63.9% 2023 0.78 3.0% 3.706 15.0% 4.360 63.9% 2024 0.80 2.9% 3.814 15.0% 4.487 63.9% 2025 0.82 2.8% 3.922 15.0% 4.614 63.9% 2026 0.85 2.8% 4.030 15.0% 4.741 63.9% 2027 0.87 2.7% 4.137 15.0% 4.867 63.9% 2028 0.89 2.6% 4.245 15.0% 4.994 63.9% 2029 0.92 2.6% 4.353 15.0% 5.121 63.8% 2030 0.94 2.5% 4.461 15.0% 5.248 63.8% 2031 0.96 2.4% 4.569 15.0% 5.375 63.8% 2032 0.98 2.4% 4.676 15.0% 5.502 63.8% 2033 1.01 2.3% 4.784 15.0% 5.628 63.8% 2034 1.03 2.3% 4.892 15.0% 5.755 63.8% 2035 1.05 2.2% 5.000 15.0% 5.882 63.8% 9.2.2 Generation plan At present there are 3 x 400 kVA high speed gasoil fired diesel generating units located in Obock. Each unit has an available generating capacity of approximately 0.24 MW. All three units were installed in 2005. It is assumed that each of these units will operate for 25 years. An additional 2 x 600 kVA high speed gasoil-fired diesel generating units have been ordered and are expected to be fully operational at the beginning of 2010. It is assumed that these units will also operate for 25 years. As derived in Section 9.2.1 above, demand for electricity in Obock is forecast to increase from 460 kW in 2009 to 596 kW in 2015, to 824 kW by 2025 and to 1,052 kW by 2035. The relationship between Least Cost Electricity Master Plan (Volume 1) Final Report November 2009 Page 9.11 SECTION 9 ISOLATED SYSTEMS the available capacity of the existing plant and the forecast of peak demand is shown in Figure 9.8 below. It can be seen that the balance between demand and supply is positive for the years 2009 to 2029, indicating that new generating units may not be required. Even when allowance is made for the need to carry reserve capacity, the need for new generating units is unlikely to arise earlier than 2029. This is because 3 units are scheduled to retire simultaneously at the end of 2029. Figure 9.8: Supply / demand power balance (Obock) Existing Available   Capacity  (MW) Additional Available  Capacity  (MW) Demand Forecast (MW) Reserve  Margin (%) 1.6 250% 1.4 200% 1.2 150% 1.0 100% Capacity  / Demand  (MW) Reserve  Margin  (%) 0.8 50% 0.6 0% 0.4 �50% 0.2 �100% 0.0 �150% 2009 2010 2011 2012 2013 2014 2015 2016 2017 2018 2019 2020 2021 2022 2023 2024 2025 2026 2027 2028 2029 2030 2031 2032 2033 2034 2035 It has been assumed for this study that any isolated system should operate to an “N-1� planning criterion (i.e. demand is met with the largest unit out of service). On this basis, we have derived the generation plan presented in Table 9.6 below. This generation plan indicates the need of a further 4 600 kVA high speed diesel units (fired on gas-oil) over the forecast period, the first being required in 19 2030 . Figure 9.9 shows the supply/demand power balance when capacity is added to the system as per Table 9.6. 19 It should be noted that although not considered for this study, localised wind, solar and geothermal power developments could be considered as possible generation expansion options in Obock. As with large-scale power generation however, the availability of these resources for even small-scale generation needs to be confirmed before these resources can be considered for the area. Least Cost Electricity Master Plan (Volume 1) Final Report November 2009 Page 9.12 SECTION 9 ISOLATED SYSTEMS Table 9.6: Generation plan (Obock) Least Cost Electricity Master Plan (Volume 1) Final Report November 2009 Page 9.13 SECTION 9 ISOLATED SYSTEMS Figure 9.9: Supply / demand power balance after capacity additions (Obock) Existing Available   Capacity  (MW) Additional Available  Capacity  (MW) Demand Forecast (MW) Reserve  Margin (%) 1.6 240% 1.4 210% 1.2 180% 1.0 150% Capacity  / Demand  (MW) Reserve  Margin  (%) 0.8 120% 0.6 90% 0.4 60% 0.2 30% 0.0 0% 2009 2010 2011 2012 2013 2014 2015 2016 2017 2018 2019 2020 2021 2022 2023 2024 2025 2026 2027 2028 2029 2030 2031 2032 2033 2034 2035 Assuming a capital cost of US$ 130,000 for a 600 kVA diesel generator, and a CAPEX phasing programme of 50 per cent in year ‘n-1’ and 50 per cent in year ‘n’, the total capital cost of investment in generating facilities in Obock is estimated to be approximately US$ 54,000 (at 10 per cent discount rate, in 2008 prices). It should be noted that the capital cost of the generating plant includes for a building extension (a simple steel framed sheet metal clad building with slab floor) and associated civil works to house the new generating sets. 9.3 Distribution Distribution in Division North including Tadjoura and Obock comprises approximately 16.6 km of medium voltage over-head lines, 6.5 km of low voltage over-head lines and 12 distribution substations. The distribution arrangement is understood to be similar to that employed in the more rural areas of the main Djibouti distribution system. Indicative costs for development of the northern subdivision distribution system to meet demand growth may be obtained by referencing the costs for development of the main distribution system. On this basis, the annual investment in distribution equipment for the northern subdivision is estimated at approximately US$ 100,000 per annum. Least Cost Electricity Master Plan (Volume 1) Final Report November 2009 Page 9.14 SECTION 9 ISOLATED SYSTEMS 9.4 Summary Our analysis of the isolated system can be summarised as follows: • The EdD power system comprises Djibouti town, Division South, and Division North. • Division South will be interconnected with Djibouti town when the interconnector is completed. It is therefore included in our analysis of the main grid network. • There are no plans to connect Division North to the main grid. • We have developed expansion plans for both Tadjoura and Obock separately. • For Tadjoura: - Demand for electricity in Tadjoura is anticipated to increase from 1 MW in 2008 to approximately 1.4 MW in 2015, eventually rising to 2.5 MW by 2035. - At present, there are currently 6 x 600 kVa high speed gasoil-fired diesel generating units located in Tadjoura. Each unit has an available capacity of approximately 360 kW. - On the basis of an “N-1� planning criterion, we estimate the addition of 9 x 600 kVA high speed diesel generating units (fired on gas-oil) over the forecast period, the first being required in 2023. - Assuming a capital cost of US$130,000 for a new 600 kVA diesel generator and a CAPEX phasing programme for 50 per cent in year N-1 and 50 per cent in year N, the total capital cost of investment (NPV) in generating facilities is estimated to be approximately US$145,000 (at a 10 per cent discount rate, 2008 prices) • For Obock: - Demand for electricity in Obock is anticipated to increase from 440 kW in 2008 to approximately 600 kW in 2015, eventually rising to 1,050 kW by 2035. - At present, there are currently 3 x 400 kVa high speed gasoil-fired diesel generating units located in Obock. Each unit has an available capacity of approximately 240 kW. Least Cost Electricity Master Plan (Volume 1) Final Report November 2009 Page 9.15 SECTION 9 ISOLATED SYSTEMS - An additional 2 x 600 kVA high speed gasoil-fired diesel generating units have been ordered and are expected to begin operations at the beginning of 2010. Each unit has an available capacity of approximately 360 kW. - On the basis of an “N-1� planning criterion, we estimate the addition of 4 x 600 kVA high speed diesel generating units (fired on gas-oil) over the forecast period, the first being required in 2030. - Assuming a capital cost of US$130,000 for a new 600 kVA diesel generator and a CAPEX phasing programme for 50 per cent in year N-1 and 50 per cent in year N, the total capital cost of investment (NPV) in generating facilities is estimated to be approximately US$54,000 (at a 10 per cent discount rate, 2008 prices) • Distribution in Division North comprises approximately 16.5 km of MV overhead lines, 6.5 km of LV overhead lines and 12 distribution substations. • Indicative costs for development of the northern distribution system to meet demand growth are in the region of US$100,000 per annum. Least Cost Electricity Master Plan (Volume 1) Final Report November 2009 Page 9.16 SECTION 10 CONCLUSIONS SECTION 10 CONCLUSIONS 10 CONCLUSIONS The World Bank appointed Parsons Brinckerhoff (PB) to undertake engineering consultancy services for the preparation of an electricity sector least cost master plan for Djibouti. The main objective of the assignment is to define the least-cost investment program for the development of Djibouti’s electric generation, transmission and distribution system for the next 25 years, particularly taking into consideration the country’s resources and recent economic and sector developments. 10.1 Future development options There are no known oil or gas resources in Djibouti, hence, at present, the Republic of Djibouti imports 85 per cent of its energy needs as hydrocarbon products, and produces only 15 per cent through indigenous wood and charcoal. The entire EdD electricity production is based on thermal powered plants fired on imported HFO and gasoil. In the second quarter of 2010 a transmission line between Ethiopia and Djibouti is expected to be commissioned. This interconnector will go a long way to reducing Djibouti’s reliance on the imported fuel products and thus reducing production costs. The amount of energy provided by the interconnector however, is limited, both in terms of total energy that can be supplied from Ethiopia to Djibouti as well as the time of day and season. Confronted with an ever increasing petroleum demand and a rapid growth in urbanisation; the Government of Djibouti (GoD) have also decided to direct the country’s energy policy towards the development of renewable energy resources (geothermal and wind in particular). Future development options have been identified as: • Energy imports over the Ethiopia-Djibouti interconnector; • HFO-fired medium speed diesel engines; • Gas-oil fired open cycle gas turbines; • Development of geothermal power plant; and, • Development of wind farms. The development of geothermal and wind power in Djibouti is in its early phases and further studies, exploration, drilling, analysis, testing and measurements are required to confirm and enable the utilisation of these resources for large-scale power generation. As a result, it is assumed that the reference case analysis of this study is based on the availability of HFO-fired diesel and gasoil-fired OCGT units only. This analysis has been carried out taking into account varying levels of energy import over the interconnector. Least Cost Electricity Master Plan (Volume 1) Final Report November 2009 Page 10.1 SECTION 10 CONCLUSIONS In addition to the least cost generation expansion plan (i.e. the reference case), we have analysed the inclusion of geothermal and wind power as a sensitivity, so as to determine the future role that these technologies may play in the generation mix and determine whether the development of these technologies should be taken to the next stage. 10.2 Generation Plan Our generation planning analysis leads to the following conclusions: • The least cost generation expansion programme derived for this study assumes the availability of HFO-fired diesel and gasoil-fired OCGT only. The least cost planting programme indicates the following: - 24 MW of capacity is required in 2013 and a total of 187 MW would be required by the end of the planning period. - The development over the planning period would comprise ten 12 MW and six 7 MW HFO-fired diesel units, one 15 MW and one 7 MW open cycle gas turbine generating sets. - The current mix of unit sizes remains appropriate for the medium and long- term development of the system. • Our analysis of the impact of varying level of energy imports on the system NPV indicates the following: - Energy imports over the Ethiopia-Djibouti interconnector are recommended. - As long as available, EdD should maximise energy imports. • Our sensitivity analysis of the potential role for geothermal and wind power indicates the following: - The long-term development of the EdD system relying on liquid fuel only represents the most expensive alternative. - Any development option (geothermal and wind) that displaces energy generated by diesel plant is economically attractive. Subject to confirmation of the adequacy of the geothermal resource, the development of geothermal power plant is recommended. Least Cost Electricity Master Plan (Volume 1) Final Report November 2009 Page 10.2 SECTION 10 CONCLUSIONS - At the cost and performance parameters assumed for this study, geothermal power generation forms part of the long-term least cost expansion plan for EdD. - The least cost development scenario would be one in which energy imports are maximised combined with geothermal development. - For the cost and performance parameters assume for wind power generation in this study, wind power generation is an economical option, displacing energy that would be generated by HFO-fired diesel plant. - The displacement of import energy over the interconnector by wind energy is not recommended. - The development of geothermal and wind technologies would be worthwhile in Djibouti and their development should be taken the stage. - The development of these resources will be faced by many issues and significant effort and cooperation will be required between the GoD, EdD, financing institutions and any contractors and consultants in order to realise the potential of this resource. • Our analysis of higher and lower demand forecasts on the least cost generation plan indicate that the future plant mix is essentially not sensitive to the demand forecast in the sense that oil-fired diesel units remain the least cost option for developing the EdD system, in addition to geothermal and wind generation should these technologies be developed. • The expansion plans identified in this study reflect the economic requirements of Djibouti without giving rise to any likely significant or inappropriate environmental impact. 10.3 Transmission Plan Load flow, fault level and transient stability studies were conducted with the objective of identifying the required transmission system reinforcements to accommodate the forecast growth in levels of demand and generation whilst meeting the planning criteria. The costs of transmission expansion to meet the demand forecast and least cost generation plan are shown in Table 10.1. The costs include for the switchgear associated with the transmission reinforcement project. Least Cost Electricity Master Plan (Volume 1) Final Report November 2009 Page 10.3 SECTION 10 CONCLUSIONS Table 10.1: Transmission costs Capital Costs Incurred in Year ($000) Year Season Peak No. Transmission reinforcement 2015 2020 2035 Dry 2011 Wet 1 2nd Marabout-Boulaos 63 kV cable circuit 2,600 Dry 2 2nd PK-12 63/20 kV 40 MVA transformer 1,550 3 2nd and 3rd Marabout 63/20 kV 36 MVA transformers 3,100 1 3rd PK-12 230/63 kV 63 MVA transformer 4,200 2015 2 3rd Boulaos 63/20 kV 36 MVA transformer 1,550 3 2nd PK-12 - Palmeraie OHL double circuit (AAAC Aster conductor) 2,296 Wet 4 2nd Palmeraie-Boulaos cable circuit (2 x 400 mm2) 2,500 5 2nd Palmeraie-Marabout cable circuit (2 x 400 mm2) 2,950 6 Boulaos reactive compensation - 40 MVAr capacitor bank (4 x 10 MVAr) 1,000 1 Palmeraie 63/20 kV substation 10,000 Dry 2 3rd PK-12 63/20 kV 40 MVA transformer 1,550 2020 1 PK-12 reactive compensation - 30 MVAr capacitor bank 750 Wet 2 Marabout reactive compensation - 10 MVAr capacitor bank 300 Dry 1 4th PK-12 63/20 kV 40 MVA transformer 1,550 2035 1 PK-12 20 kV reactive compensation - additional 20 MVAr capacitor bank/SVC 500 Wet 2 PK-12 63 kV reactive compensation - 90 MVAr capacitor bank/SVC 9,300 Total ($000) 21,746 12,600 11,350 A brief study has been conducted to assess the technical requirements and costs associated with the interconnection of Division North, including the towns of Tadjoura and Obock with the main transmission system. This could be achieved (following grid connection of the geothermal generation at Lac Assal) by installing a 63 kV circuit from Ghoubet to Tadjoura (30 km) and then installing a 20 kV circuit from Tadjoura to Obock (50 km). The approximate costs associated with this scheme, including the necessary 63/20 kV and 20/0.4 kV substations are as follows: Grid connection of Tadjoura US$ 4.4 million Grid connection of Obock US$ 2.6 million (assuming prior grid connection of Tadjoura). 10.4 Distribution Plan The annual load related investment on distribution infrastructure has been calculated by applying the average incremental cost of distribution equipment to the load forecast as shown in Table 10.2. Assuming an average distribution asset life of 40 years, the average annual non-load related expenditure is calculated by assuming that on average, 1/40th of the existing asset base will be Least Cost Electricity Master Plan (Volume 1) Final Report November 2009 Page 10.4 SECTION 10 CONCLUSIONS replaced each year over the planning period. The total annual expenditure on distribution is then the sum of the load related and non-load related expenditure. Table 10.2: Distribution expenditure 2010 2011 2012 2013 2014 2015 2016 2017 2018 2019 2020 2021 2022 Maximum demand (MW) 74.5 83.8 99.7 116.3 128.2 137.6 148.6 153.3 158.1 161.6 165.2 167.9 170.6 Load growth (MW) 6.5 9.3 15.9 16.6 11.9 9.4 11 4.7 4.8 3.5 3.6 2.7 2.7 Load related expenditure ($ 000) 4,510 6,453 11,033 11,518 8,257 6,522 7,633 3,261 3,331 2,429 2,498 1,873 1,873 Non-load related expenditure ($ 000) 1,180 1,180 1,180 1,180 1,180 1,180 1,180 1,180 1,180 1,180 1,180 1,180 1,180 Total expenditure ($ million) 5.7 7.6 12.2 12.7 9.4 7.7 8.8 4.4 4.5 3.6 3.7 3.1 3.1 2023 2024 2025 2026 2027 2028 2029 2030 2031 2032 2033 2034 2035 Maximum demand (MW) 173.5 176.4 179.9 183.4 187 190.8 194.6 198.5 202.3 206.2 210.2 214.3 218.5 Load growth (MW) 2.9 2.9 3.5 3.5 3.6 3.8 3.8 3.9 3.8 3.9 4 4.1 4.2 Load related expenditure ($ 000) 2,012 2,012 2,429 2,429 2,498 2,637 2,637 2,706 2,637 2,706 2,775 2,845 2,914 Non-load related expenditure ($ 000) 1,180 1,180 1,180 1,180 1,180 1,180 1,180 1,180 1,180 1,180 1,180 1,180 1,180 Total expenditure ($ million) 3.2 3.2 3.6 3.6 3.7 3.8 3.8 3.9 3.8 3.9 4.0 4.0 4.1 10.5 Isolated systems The EdD power system is currently split into three regions to supply Djibouti town, Division North and Division South respectively. The three regions are not electrically interconnected at present however, Division South will be interconnected with Djibouti town as part of the Ethiopia-Djibouti Interconnector project. Due to the relative remoteness of both Tadjoura and Obock, there are currently no plans to connect Division North to the main grid. Assuming a capital cost of US$ 130,000 for a 600 kVA diesel generator, and a CAPEX phasing programme of 50 per cent in year Yn-1 and 50 per cent in year Yn, the total capital cost of investment in generating facilities in Tadjoura is estimated to be approximately US$ 145,000 (at 10 per cent discount rate, in 2008 prices). Assuming a capital cost of US$ 130,000 for a 600 kVA diesel generator, and a similar CAPEX phasing programme,, the total capital cost of investment in generating facilities in Obock is estimated to be approximately US$ 54,000 (at 10 per cent discount rate, in 2008 prices). Indicative costs for development of the Division North distribution system to meet demand are based on referencing the costs for development of the main distribution system. On this basis, the annual investment in distribution equipment for the northern subdivision is estimated at approximately US$ 100,000 per annum. 10.6 Investment plan The investment plan relating to the least cost development plan (i.e. the reference case) for the EdD network in terms of generation, transmission and distribution for the main grid and for the isolated Least Cost Electricity Master Plan (Volume 1) Final Report November 2009 Page 10.5 SECTION 10 CONCLUSIONS system (Obock and Tadjoura) is set out Table 10.3 below.. Expenditure is detailed for the years 2010 to 2020 inclusive. Investment plans are also provided for the least cost development options, should: (i) geothermal be available, and, (ii) both wind and geothermal be available for large-scale power generation. As such, Table 10.4 sets out the expenditure profile for Scenario 5 and Table 10.5 sets out the expenditure for Scenario 11. Least Cost Electricity Master Plan (Volume 1) Final Report November 2009 Page 10.6 SECTION 10 CONCLUSIONS Table 10.3: Investment plan for least cost development plan (Scenario 1, 2 and 3) Least Cost Electricity Master Plan (Volume 1) Final Report November 2009 Page 10.7 SECTION 10 CONCLUSIONS Table 10.4: Investment plan for liquid fuel & geothermal development (Scenarios 4 and 5) Least Cost Electricity Master Plan (Volume 1) Final Report November 2009 Page 10.8 SECTION 10 CONCLUSIONS Table 10.5: Investment plan for liquid fuel, geothermal and wind farm developments (Scenarios 10, 11 and 12) Least Cost Electricity Master Plan (Volume 1) Final Report November 2009 Page 10.9