PAKISTAN SUSTAINABLE ENERGY SERIES PAKISTAN LEAST-COST ELECTRIFICATION STUDY PAKISTAN LEAST-COST ELECTRIFICATION STUDY ©2024 International Bank for Reconstruction and Development / The World Bank The World Bank 1818 H Street NW Washington DC 20433 Telephone: 202-473-1000 Publication date: June 2024 Disclaimer This work is a product of the staff of the World Bank with external contributions. The findings, interpretations, and conclusions expressed in this work do not necessarily reflect the views of the Executive Directors of the World Bank or the governments they represent. The World Bank does not guarantee the accuracy of the data included in this work and does not assume responsibility for any errors, omissions, or discrepancies in the information, or liability with respect to the use of or failure to use the information, methods, processes, or conclusions set forth. The boundaries, colors, denominations, links/footnotes, and other information shown in this work do not imply any judgment on the part of The World Bank concerning the legal status of any territory or the endorsement or acceptance of such boundaries. The citation of works authored by others does not mean the World Bank endorses the views expressed by those authors or the content of their works. Rights and Permissions The material in this work is subject to copyright. Because The World Bank encourages dissemination of its knowledge, this work may be reproduced, in whole or in part, for noncommercial purposes as long as full attribution to this work is given. Any queries on rights and licenses, including subsidiary rights, should be addressed to World Bank Publications, The World Bank Group, 1818 H Street NW, Washington, DC 20433, USA; fax: 202-522-2625; e-mail: pubrights@ worldbank.org. Report and Cover design: Pi Comm Text Layout: Pi Comm Cover photo: © Dana Smillie / World Bank 2 CONTENTS Acknowledgments............................................................................................................................................................... VI Abbreviations and acronyms........................................................................................................................................VII Executive summary............................................................................................................................................................ IX 1. INTRODUCTION............................................................................................................................................................ 1 1.1 Country context.............................................................................................................................................................. 2 1.2 Existing regulatory framework and key stakeholders................................................................................ 2 1.2.1 Policies and Frameworks............................................................................................................................ 2 1.2.2 Key Stakeholders............................................................................................................................................ 5 1.3 Study objective. .............................................................................................................................................................. 6 1.4 Approach............................................................................................................................................................................ 6 1.5 Report structure. ............................................................................................................................................................ 6 2. METHODOLOGY AND INPUTS. ........................................................................................................................ 7 2.1 Overview of the planning methodology............................................................................................................. 8 2.2 Key model input parameters................................................................................................................................ 10 Local inputs: 2.2.1 Balochistan ..................................................................................................................................................... 14 2.2.2 Khyber Pakhtunkhwa................................................................................................................................. 17 2.2.3 Punjab. ............................................................................................................................................................... 20 2.2.4 Sindh................................................................................................................................................................... 23 3. RESULTS......................................................................................................................................................................... 26 3.1 Balochistan. ................................................................................................................................................................... 27 3.1.1 Reference base-case scenario............................................................................................................. 27 3.1.2 Detailed results for key districts. .......................................................................................................... 33 3.1.3 Summary........................................................................................................................................................... 35 3.2 Khyber Pakhtunkhwa............................................................................................................................................... 36 3.2.1 Reference base-case scenario............................................................................................................. 36 3.2.2 Sensitivity analysis. ..................................................................................................................................... 41 3.2.3 Summary........................................................................................................................................................... 44 3.3 Punjab. ............................................................................................................................................................................. 45 3.3.1 Reference base-case scenario............................................................................................................. 45 3.3.2 Summary........................................................................................................................................................... 52 3.4 Sindh................................................................................................................................................................................. 52 3.4.1 Reference base-case scenario............................................................................................................. 52 3.4.2 Sensitivity analysis. ..................................................................................................................................... 57 3.4.3 Summary........................................................................................................................................................... 63 I PAKISTAN LEAST-COST ELECTRIFICATION STUDY 4. MINI-GRID ASSESSMENT.................................................................................................................................. 64 4.1 Mini-Grid potential in Pakistan............................................................................................................................ 65 4.2 Calibration with LCES.............................................................................................................................................. 68 4.3 Conclusions................................................................................................................................................................... 69 5. RECOMMENDATIONS........................................................................................................................................... 71 5.1 Implementation plan. ................................................................................................................................................ 73 5.2 Business model analysis........................................................................................................................................ 74 5.3 Financial plan. .............................................................................................................................................................. 74 5.4 Sustainability plan...................................................................................................................................................... 75 5.5 Governance................................................................................................................................................................... 76 APPENDIX A: ELECTRIFICATION MODEL (REM). ............................................................................... 77 Techno-economic procedure for least-cost electrification planning .............................................. 79 A.1  A.2 Step 1 – Clustering................................................................................................................................................... 79  tep 2 – F A.3 S  inal decision on the best electrification mode for each cluster........................................................................................................................................ 80 A.4 Different types of input data................................................................................................................................. 82 A.5 Sensitivity analysis (qualitative)......................................................................................................................... 83 A.6 Ground data.................................................................................................................................................................. 83 A.7 Our assumptions........................................................................................................................................................ 84 A.8 Strategic decision-making..................................................................................................................................... 84 APPENDIX B: MINI-GRID PORTFOLIO ASSESSMENT METHODOLOGY.......................... 85 B.1 Step 1 – Identify off-grid villages....................................................................................................................... 86 B.1.1 Identification of the electricity grid...................................................................................................... 86 B.1.2 Identification of potential mini-grid villages.................................................................................... 87 B.2. Step 2 – Extracting village properties. .......................................................................................................... 88 B.2.1 Village size and building ......................................................................................................................... 88 B.2.2 Village address and names.................................................................................................................... 88 B.2.3 Proximity to grid infrastructure and nightlight............................................................................... 88 B.2.4 Accessibility..................................................................................................................................................... 89 B.2.5 Health and education facilities ............................................................................................................ 89 B.2.6 Agricultural data............................................................................................................................................ 89 B.3 Step 3 – Predict mini-grid viability ................................................................................................................... 89 B.3.1 Demand............................................................................................................................................................. 89 B.3.2 Sizing of the main components............................................................................................................ 91 B.3.3 Costing............................................................................................................................................................... 93 B.4 Step 4 – Prioritization of mini-grid sites......................................................................................................... 94 II TABLES Table E.S.1: Summary of results for the least-cost electrification study............................................... XI Table 2.1: Local input parameters for REM. ........................................................................................................ 10 Table 2.2: Network and generation component costing assumptions................................................... 12 Table 2.3: Local input parameters for REM – Balochistan. ......................................................................... 15 Table 2.4: Balochistan’s population data............................................................................................................... 15 Table 2.5: Number of customers (2020, and 2030 estimates) – Balochistan................................... 16 Table 2.6: Demand data for different types of customers – Balochistan. ............................................ 17 Table 2.7: Local input parameters for REM – KP. ............................................................................................ 18 Table 2.8: KP population data...................................................................................................................................... 19 Table 2.9: Number of customers (present – 2020, and 2030 estimates) – KP................................ 19 Table 2.10: Demand data for different types of customers – KP.............................................................. 20 Table 2.11: Electrification rates, 2020, for different utilities in Punjab................................................... 20 Table 2.12: Local input parameters for REM – Punjab.................................................................................. 21 Table 2.13: Punjab population data.......................................................................................................................... 22 Table 2.14: Demand data for different types of customers – Punjab..................................................... 22 Table 2.15: Local input parameters for REM – Sindh..................................................................................... 23 Table 2.16: Sindh population data............................................................................................................................. 24 Table 2.17: Number of customers (2020, and 2030 estimates) – Sindh. ............................................ 24 Table 2.18: Demand data for different types of customers – Sindh. ...................................................... 25 Table 3.1: Summary of results for the least-cost electrification study – Balochistan.................... 27  ummary of least-cost electrification technologies by district for 2030 Table 3.2: S (REM simulation results) – Balochistan ......................................................................................... 30  ummary of results for key districts identified by the Balochistan Provincial Table 3.3: S Energy Department for mini-grid sector development ........................................................... 33 Table 3.4: Summary of results for the least-cost electrification study – KP....................................... 36  ummary of least-cost electrification technologies by district for 2030 Table 3.5: S (REM simulation results) – KP ............................................................................................................ 39  ummary of universal access costs through 2030: Abbottabad district Table 3.6: S (REM simulation results: Base case)................................................................................................ 42  ummary of universal access costs through 2030: Abbottabad district Table 3.7: S (REM simulation results: High grid reliability case). ................................................................. 43  030 comparative results for REM’s sensitivity analysis Table 3.8: 2 for nearly 150,000 candidate customers for Abbottabad district, KP.............................. 43 Table 3.9: Summary of results for the least-cost electrification study – Punjab. ............................. 45 Table 3.10: S ummary of least-cost electrification technologies by district for 2030 (REM simulation results) – Punjab . ............................................................................................... 49 Table 3.11: Summary of results for the least-cost electrification study – Sindh............................... 53 Table 3.12: S ummary of least-cost electrification technologies by district for 2030 (REM simulation results) – Sindh. ................................................................................................... 56 Table 3.13: S ummary of universal access costs through 2030: Tharparkar district (REM simulation results: Base case)............................................................................................. 58 Table 3.14: Scenarios for sensitivity analysis .................................................................................................... 58 III  ummary of universal access costs through to 2030: Tharparkar district Table 3.15: S (REM simulation results: Low off-grid component generation cost)............................. 59  ummary of universal access costs through 2030: Tharparkar district Table 3.16: S (REM simulation results: High-demand case). ......................................................................... 60  ummary of universal access costs through 2030: Tharparkar district Table 3.17: S (REM simulation results: High grid reliability case)................................................................ 61  030 comparative results for REM’s sensitivity analysis Table 3.18: 2 for more than 250,000 candidate customers for Tharparkar district, Sindh............. 62 Table 4.1: Summary of main results – breakdown by region..................................................................... 67 Table 4.2: Demand breakdown for each region................................................................................................. 67 Table B.1: Assumed daily energy demand for different customer types.. ........................................... 90 Table B.2: Daily load profile – percentage of nighttime demand.............................................................. 91 Table B.3: Number of households per commercial customer for each province............................. 91 Table B.4: Lead-acid battery sizing assumptions. ............................................................................................ 92 Table B.5: Solar PV sizing assumptions................................................................................................................ 93 Table B.6: Inverter sizing assumptions................................................................................................................... 93 Table B.7: Costing assumptions................................................................................................................................. 94 IV FIGURES Figure E.S.1.1: Least-cost options for electrification for the four provinces ....................................... X Figure 1.1: Flow of activities for village electrification in Pakistan. ............................................................ 4 Figure 2.1: LCES approach............................................................................................................................................. 8 Figure 2.2: Average hourly demand profile for different types of customers . .................................. 11  ercentage divide between Balochistan REM results for grid extension, Figure 3.1: P mini-grids, and individual systems for customers beyond 500 meters of the grid infrastructure (a) Number of customers (b) Investment cost (CAPEX) (c) Annual energy demand (GWh) (d) Average cost of energy (USD/kWh). ............ 28  istograms of Balochistan grid extension and mini-grid projects (least-cost Figure 3.2: H options) showing CAPEX per customer (USD/connection): The x-axis presents the CAPEX per customer and the y-axis shows the number of electrification projects............................................................................................................................................................ 29  ercentage divide between KP REM results for grid extension, mini-grids, Figure 3.3: P and individual systems (a) Number of customers (b) Investment cost (CAPEX) (c) Annual energy demand (GWh) (d) Average cost of energy (USD/kWh). ............ 37  istogram of KP grid extension and mini-grid projects (least-cost options) Figure 3.4: H showing CAPEX per customer. The x-axis presents the CAPEX per customer and the y-axis shows the number of electrification projects.............................................. 38  ercentage of new customers for different technologies Figure 3.5: P under different scenarios for least-cost electrification........................................................... 44  ercentage divide between Punjab REM results for grid extension, mini-grids, Figure 3.6: P and individual systems (a) Number of customers (b) Investment cost (CAPEX) (c) Annual energy demand (GWh) (d) Average cost of energy (USD/kWh). ............ 48  ercentage divide between Sindh REM results for grid extension, mini-grids, Figure 3.7: P and individual systems (a) Number of customers (b) Investment cost (CAPEX) (c) Annual energy demand (GWh) (d) Average cost of energy (USD/kWh). ............ 53  istogram of Sindh grid extension and mini-grid projects (least-cost options) Figure 3.8: H showing CAPEX per customer (USD/connection). The x-axis presents the CAPEX per customer and the y-axis shows the number of electrification projects........................................................................................................................ 55  omparison of the number of connections for different technologies Figure 3.9: C under different scenarios for the LCES for Tharparkar district......................................... 62 Figure 4.1: Number of potential mini-grid sites in different regions of the country......................... 65 Figure 4.2: Distribution of the average number of connections per province................................... 66 Figure 5.1: Integration of key sectors needed for universal access to sustainable electricity. 72 nput data and inference pipeline designed for electrification planning in REM. ... 78 Figure A.1: I Figure A.2: Example of the structure of clusters (results from the clustering process) .............. 80 Figure A.3: Master-slave decomposition for the optimization of mini-grid generation ................ 81 Figure A.4: Example of minimum-cost electrification solution .................................................................. 81 Figure B.1: Mini-Grid Portfolio Assessment Methodology........................................................................... 86 Figure B.2: Number of potential mini-grid sites in different regions of the country........................ 87 Figure B.3: Convex hull, alpha shape, and minimum spanning tree of a dense cluster............. 88 V PAKISTAN LEAST-COST ELECTRIFICATION STUDY ACKNOWLEDGMENTS This report is based on a two-year study carried out by the World Bank and supported by IIT-Comillas University, Waya Energy, Village Data Analytics, Trama TecnoAmbiental (TTA), and PITCO. The study was commissioned and supervised by Saadia Qayyum (Energy Specialist) with support from Minahil Raza (Energy Specialist), Oliver Knight (Senior Energy Specialist) and Alexandros Korkovelos (Consultant). The report was peer-reviewed by Ashish Shrestha (Energy Specialist), Benjamin Stewart (Senior Geographer), Chiara Rogate (Senior Energy Specialist, Tatia Lemondzhava (Energy Specialist), and Yann Tanvez (Energy Specialist). Comments were also kindly provided by Ian Baring Gould (Program Manager, US National Renewable Energy Laboratory). Final review and editing were carried out by Minahil Raza (Energy Specialist) and Oliver Knight (Senior Energy Specialist). Reja Amatya, Ph.D. (COO – Waya Energy), Nabin Gahire (COO – VIDA), Roger Sallent (Regional Team Leader – TTA), and Usman H. Malik (Power System Team Lead – PITCO) provided support. This study is part of a series of deliverables on sustainable energy commissioned by the World Bank under the Pakistan Sustainable Energy Program (P169313), a multiyear technical assistance program in support of the Government of Pakistan. Funding for this study was generously provided by the Energy Sector Management Assistance Program (ESMAP). ESMAP is a global knowledge and technical assistance program to support low- and middle-income countries in increasing their knowledge and their institutional capacity to achieve environmentally sustainable energy solutions for poverty reduction and economic growth. ESMAP is funded by Australia, Austria, Denmark, the European Commission, Finland, France, Germany, Iceland, Italy, Japan, Lithuania, Luxembourg, the Netherlands, Norway, Sweden, Switzerland, the United Kingdom, and the Rockefeller Foundation, as well as the World Bank. The financial and technical support provided by ESMAP is gratefully acknowledged. The World Bank would like to thank the government of Pakistan, and respective federal agencies, particularly the Central Power Purchasing Agency (CPPA-G), the National Electric Power Regulatory Authority (NEPRA), the National Power Control Center (NPCC), the Planning Commission, and representatives from the provincial governments and electricity distribution companies (DISCOs) for their input and feedback on this study, including provision of data, participation in workshops, and submission of written comments. Input and feedback were also provided by several development partners and private-sector representatives. VI PAKISTAN LEAST-COST ELECTRIFICATION STUDY ABBREVIATIONS AND ACRONYMS AEDB Alternative Energy Development Board CNSE Cost of Non-Served Energy DISCO Distribution Company ESMAP Energy Sector Management Assistance Program FESCO Faisalabad Electric Supply Company GEPCO Gujranwala Electric Power Company GIS Geographic Information Systems GWh Gigawatt hour HH Household HRSL High-Resolution Settlement Layer HV High Voltage HESCO Hyderabad Electric Supply Company ICT Information, Communication, and Technology IGCEP Integrated Generation Capacity Expansion Plan IESCO Islamabad Electric Supply Company K-Electric Karachi Electric Supply Company kWh Kilowatt-hour kWp Kilowatt (peak) LCES Least-Cost Electrification Study LESCO Lahore Electric Supply Company LT Low Tension LV Low Voltage MEPCO Multan Electric Power Company MT Medium Tension MTF Multi-Tier Framework MV Medium Voltage MW Megawatt NEPRA National Electric Power Regulatory Authority NTDC National Transmission and Dispatch Company OSM OpenStreetMap PESCO Peshawar Electric Power Company PPP Public-Private Partnership PV Photovoltaic VII PAKISTAN LEAST-COST ELECTRIFICATION STUDY QESCO Quetta Electric Supply Company RBC Reference Base Case REM Reference Electrification model SEPCO Sukkur Electric Power Company SHS Solar Home System T&D Transmission and Distribution TESCO Tribal Electric Supply Company TSEP Transmission System Expansion Plan VIII PAKISTAN LEAST-COST ELECTRIFICATION STUDY EXECUTIVE SUMMARY The World Bank commissioned this Least-Cost Electrification Study to identify the optimal route for achieving universal access to electricity by 2030 in Pakistan, through expansion of the existing distribution grid and off-grid solutions. A Mini-Grid Portfolio Readiness Assessment was commissioned in parallel to provide further granularity on the potential for mini-grids across the country, and its findings are part of this report. The studies were carried out over a two-year period (2021–2022) by a team of experts, based on methodologies developed and tested in a number of other countries. The results of the Least-Cost Electrification Study are based on a geospatial model that considers four options for delivering universal electrical service by 2030: (i) the densification of the existing electricity grid; (ii) extension of the existing grid; (iii) mini-grids, and (iv) individual off-grid systems. The Mini-Grid Portfolio Readiness Assessment identifies and ranks the current high priority mini-grid sites across Pakistan which can be leveraged by federal and provincial governments, electricity distribution companies, and the private sector to enhance electricity access. The two sets of analysis have been compared and calibrated to inform the outputs and the recommendations contained in this report. This report is intended to help inform the development of government policy on achieving universal electricity access and potential future investments, including actions to support greater private-sector participation in the provision of electricity access. Furthermore, it utilizes data from the Pakistan Energy Survey commissioned by the World Bank in parallel and published in 2023. This Least-Cost Electrification Study should therefore be considered in conjunction with the Pakistan Energy Survey to get a full view of the electricity access situation in Pakistan, and potential policy and investment options. Figure E.S.1 shows prospective consumer numbers for the four options for least-cost electrification across the four provinces. Table E.S.1 provides the overview results for the target year 2030. To achieve universal access to electricity by 2030, Pakistan will need to serve approximately 40 million total electricity connections. The requirement to increase the number of electricity connections takes account of those who do not currently have a formal electricity connection plus population growth. The new connections break down into two categories: those potential customers who are within 500 meters of the existing distribution grid, and those who are beyond 500 meters. Grid densification is the least-cost option for 53 percent of those without electricity access in Pakistan. For customers who are within 500 meters of the existing distribution grid the most cost-effective solution would be grid densification. This involves expanding and strengthening the current grid infrastructure. The geospatial analysis identifies households within this category that are not currently electrified according to the customer count data of the electricity distribution companies (DISCOs) and geospatial analysis. Some of these households may already have informal connections to electricity, and they would need to be included in the grid densification efforts to bring them into the formal system and reduce commercial losses. For the 9.3 million new connections that are located further than 500 meters from the existing grid infrastructure the predominant option is grid extension, followed by off-grid solutions such as mini- grids and individual solar home systems. Grid extension refers to efforts to expand the grid to new areas that are currently more than 500 meters from the existing grid, involving the installation of new poles, wires and transformers, plus new customer connections in the target settlements. The study estimates that grid extension is the least-cost option for 25 percent of total new connections required, and 53 percent of the new connections required beyond 500 meters from the current grid, providing a cost-effective solution for expanding access. Where the distances involved are prohibitively costly, or where population densities are much lower, mini-grids and individual solar home systems emerge as the least-cost solution. Mini-grids emerge as the most economical solution for 20 percent of the new connections required, and are especially important in providing electricity access in KP province. There have been encouraging developments in the expansion of mini-grid technologies and business models in other countries, making mini-grids a viable pathway to high-quality electricity access in more far-off locations with relatively high population densities. Of a total number of 3.9 million mini-grid connections required by 2030, from a very low base, 2.1 million would be required in KP province due to its geography and demographic characteristics. Mini-grids in Pakistan are likely to be predominantly solar-powered, although micro hydropower and wind power may be options in some parts of the country. IX PAKISTAN LEAST-COST ELECTRIFICATION STUDY Individual off-grid systems play a relatively small role in reaching universal electricity access in Pakistan, feasible for just 2.5 percent of new connections. While such systems are already a common feature in many households due to the poor quality of grid-provided electricity, or the absence of a grid connection, they are the optimal long-term solution only in situations where households are both remote and relatively dispersed. This is in contrast to countries in Sub-Saharan Africa, where individual off-grid systems play a much larger role. Individual off-grid systems will largely be solar-based, thereby limiting the quantity and quality of power that can be provided, although other technologies may also be viable in specific locations. Figure E.S.1.1: Least-cost options for electrification for the four provinces Number of new electrified custumers through different technologies (2030) Grid Densification Grid Extensions Isolated Systems Mini-grids KP Punjab Balochistan Sindh 0 500.000 1.000.000 1.500.000 2.000.000 2.500.000 3.000.000 3.500.000 4.000.000 4.500.000 5.000.000 Number of electric custumers (connections) X PAKISTAN LEAST-COST ELECTRIFICATION STUDY Table E.S.1: Summary of results for the least-cost electrification study Indicator Grid New Grid Mini-Grids Individual Total Densification Extensions systems BALOCHISTAN: Number of new customers 1.35 million 209,779 465,332 222,538 2.25 million Average cost per kWh of 0.21 0.25 0.33 0.53 demand served ($/kWh) Total CAPEX ($) 0.80 billion 0.175 billion 0.42 billion 0.25 billion 1.6 billion KHYBER PAKHTUNKHWA (KP): Number of new customers 923,105 708,645 2.19 million 99,131 3.93 million Average cost per kWh of 0.17 0.22 0.34 0.53 demand served ($/kWh) Total CAPEX ($) 0.52 billion 0.49 billion 1.8 billion 0.12 billion 2.94 billion PUNJAB: Number of new customers 4.37 million 2.11 million 387,977 108,218 6.98 million Average cost per kWh of 0.15 0.21 0.33 0.52 demand served ($/kWh) Total CAPEX ($) 2.44 billion 1.6 billion 0.4 billion 0.13 billion 4.57 billion SINDH: Number of new customers 3.7 million 1.86 million 879,379 76,024 6.5 million Average cost per kWh of 0.14 0.20 0.32 0.53 demand served ($/kWh) Total CAPEX ($) 2.04billion 1.24 billion 0.83 billion 0.09 billion 4.2 billion Source: REM simulation. All $ references are in United States dollars. The total capital investment required to achieve universal electricity access in these four provinces is estimated to be US$ 13.3 billion. Currently the vast bulk of this capital requirement would fall on public-sector entities, since with the exception of Karachi all the DISCOs are publicly owned. Furthermore, and notwithstanding the need to introduce greater private-sector participation to DISCOs, the initial capital cost of expanding electricity access is usually seen as a public good that justifies an initial subsidy. However, there is strong potential to bring in private capital where commercial business models exist, and some countries are adopting this approach for the roll-out of mini-grids and individual off-grid systems. The estimated capital investment provided here does not include the cost of transmission and generation investment that would be required to supplement this electrification demand, although initially the additional demand from electrifying such consumers is relatively low. XI PAKISTAN LEAST-COST ELECTRIFICATION STUDY An integrated approach including both grid and off-grid systems is recommended as the optimal solution for achieving universal electricity access by 2030. While grid densification and extension are considered the least-cost solution to provide energy access to the majority of the population in Pakistan by 2030, mini-grids and individual off-grid systems will also play a crucial role in achieving universal electricity access. This is because certain areas in Pakistan have sparse and remote populations, and when the impact of non-served energy is taken into consideration, it makes grid extension less cost-effective for providing electricity access. Off-grid solutions allow for localized generation and distribution of electricity, serving a smaller group of customers within a specific area. By establishing these off-grid solutions, communities can leverage local renewable energy sources or other distributed generation options while reducing transmission losses and reliance on the main grid. Pakistan has an abundance of high-priority mini-grid electrification sites across the four provinces, providing an option for achieving accelerated universal electricity access. Mini-grids can play an important role in universal electrification, both in the transitory phase and the long-run. The above-mentioned Mini-grid Portfolio Readiness Assessment, based on distance from the grid (>2.5 km) and the number of households per village, has identified 1,015 current potential high-priority electrification sites distributed across the four provinces. Of these mini-grid potential sites 64 percent converge to the same electrification supply option by 2030 when compared with the least-cost analysis. The remaining 34 percent are currently the potential grid-compatible mini-grid sites, where the grid is likely to reach in the future as it is the least-cost technology in the long run. The development of mini-grids in these areas can significantly expedite the process of achieving universal electricity access in Pakistan. To implement mini-grids in all the identified high-priority clusters, the estimated total investment cost is US$ 360 million. The cost per mini-grid varies depending on the region, with Balochistan having the lowest average cost of US$ 233,213 per mini-grid, while KP has the highest average cost of US$ 467,972 per mini-grid. This cost disparity is primarily due to the variation in the average size of mini-grids in each province. Factors such as improving grid reliability and serving higher demand pushes the least-cost solution in favour of grid extension compared to off-grid solutions, whereas if grid energy costs are high, then off-grid solutions become more favorable. A sensitivity analysis was conducted for two districts, Tharparkar and Abbotabad, to assess the robustness of the results and gain insights into various pathways toward achieving 100 percent electrification. Higher grid reliability in both districts favored grid extension as the least-cost solution for more customers. This is because lower grid reliability penalized grid extension due to the cost of non-served energy. However, improving grid reliability requires DISCOs to make significant investments, which may be challenging due to financial constraints. Additionally, in a scenario with increased demand, grid extension becomes more cost-effective as network components like transformers and wires are fully utilized, resulting in economies of scale. Conversely, higher grid energy costs lead to more mini-grids and fewer grid extensions, especially when the distance between buildings and the grid is short. XII PAKISTAN LEAST-COST ELECTRIFICATION STUDY Recommendations and policy implications The National Transmission and Dispatch Company has already developed the Indicative Generation Capacity Expansion Plan (IGCEP 2022–2031), a least-cost plan for development of new electricity generation capacity, and is finalizing its Transmission System Expansion Plan (TSEP). The National Electricity Plan (2023–2027) has a directive for the development of a strategy for the expansion of the distribution network. Under “Accessibility”, the National Plan directs the DISCOs and respective Provincial Governments to do a GIS-based assessment to identify non-electrified areas and find an optimum solution for electrification. This study together with its GIS-based platform does precisely such geospatial analysis and can support the utilities, provincial energy departments, and the Ministry of Energy (Power Division) to develop the distribution expansion plan accordingly. Combined with the IGCEP and TSEP, this would create an effective national strategy for electrification in Pakistan. To implement a least-cost electrification plan in Pakistan, the following are proposed: 1. A designated entity should take the lead in implementing the least-cost electrification plan and utilize the GIS tool developed for this purpose. It should collaborate with utility companies, provincial energy departments, and the private sector to coordinate efforts in grid expansion and off-grid development. 2. Each DISCO should apply the least-cost electrification methodology and results to its own techno- economic feasibilities. This should be done to evaluate all future grid expansion or off-grid rural or village electrification projects or programs. 3. The GIS-based platform should be integrated into the design and approval process, as stated in the new National Electricity Plan (2023–2027). The Planning and Engineering (P&E) Departments within each DISCO should receive training to effectively utilize the GIS tool in their planning activities. 4. Provincial energy departments play a crucial role in the implementation of off-grid solutions such as mini- grids and individual off-grid systems which are essential in achieving universal electrification. However, before including them in large-scale planning, it is important to pilot these solutions to ensure their success and effectiveness. 5. The mini-grids portfolio assessment can help prioritize villages and sites for mini-grid development. These projects can be funded through public-sector initiatives or by leveraging private-sector financing. 6. It is recommended to make the GIS-based platform publicly available to attract and facilitate private-sector investments. This will enable developers to make informed decisions based on the data provided in the platform. Pakistan’s regulatory framework facilitates the implementation of the least-cost electrification plan as it endorses the integration of comprehensive planning considerations in the National Electricity Policy 2021 and National Electricity Plan, and the formulation of mini-/microgrid licensing regulation that allow for potential future integration with the national grid or independent standalone operations. Furthermore, it is important to align the electrification efforts with national priorities and policies, while also considering the unique needs and priorities of local communities and stakeholders. This approach ensures a targeted and customized approach to address the specific challenges and opportunities in each province. Moreover, financial restrictions and operational implementation constraints should be carefully considered to ensure that the prioritization process is realistic and feasible. • For grid expansion initiatives, the rural electrification projects should be included in the five-year distribution network roll-out plans that the DISCOs are required to submit to the National Electric Power Regulatory Authority (NEPRA) under the Distribution Code. This approach ensures the efficient and effective expansion of the grid infrastructure given the operational and financial constraints of the DISCOs. • For mini-grids, piloting and prioritization will help to determine which projects should be implemented first, how they should be grouped together, what tariffs should be set for each project or location, and how to create a financial plan to ensure their successful implementation. • In addition to addressing the electrification needs, prioritization should also encompass a broader perspective that emphasizes supporting local businesses, agriculture, and industries. By integrating economic stimulation goals into the electrification strategy, the focus extends beyond mere access to electricity and toward empowering local economies. By leveraging the strengths of the regulatory framework and undertaking these subsequent steps, Pakistan can effectively translate the findings of this study into actionable plans and policies. This will facilitate the realization of its electrification goals, promote sustainable development and private investment, and enhance access to reliable electricity throughout the country. XIII PAKISTAN LEAST-COST ELECTRIFICATION STUDY 1. INTRODUCTION 1 PAKISTAN LEAST-COST ELECTRIFICATION STUDY 1.1 COUNTRY CONTEXT Electricity supply is critical for the economic growth and socio-economic development of Pakistan. In February 2016, the government adopted sustainable development goals (SDGs) that underpin its aim to achieve universal access to affordable, reliable, and modern energy services by 2030.1 In Pakistan, the responsibility for electrification rests with the distribution utilities (distribution companies). These are all government-owned except for K-Electric—the only privately-owned and operated utility—which serves Karachi and its surrounding areas. Over the past five years, all of these companies, K-Electric included, have made significant progress in expanding grid connections. The Pakistan Energy Survey, based on a Multi-Tier Framework (MTF) for Energy Access,2 commissioned by the World Bank, shows a nationwide electrification rate of 89.2 percent, with an additional 8.5 percent having access to smaller individual off-grid systems for electrification. Among the grid-connected households, roughly 10 percent are considered informal—essentially illegal—connections, meaning either that they do not pay for the services at all, or pay to the Kunda system,3 but are absent from utility customer lists. On aggregate, 80 percent of households have only Tier 2 access or less, indicating difficulties with affordability, capacity, and availability of supply. According to the survey, reliability of electric supply is a huge challenge in Pakistan, where more than 80 percent of connected households experience 4–14 interruptions per week lasting more than two hours in total. Reliability, affordability, and availability of electricity were identified as key constraints. On account of poor financial and operating performance, most distribution companies in Pakistan struggle to raise the capital required to improve their service or acquire new assets. Some are unable even to recover their operating costs through electricity sales, let alone finance grid expansion. Further to the 18th amendment 4 to Pakistan’s constitution in 2010, provinces are taking on a greater role than the federal agencies in developing, financing and implementing mini-grids and off-grid energy projects. The government of Khyber Pakhtunkhwa has undertaken multiple small-scale hydropower projects within the province, whereas the governments of the two southern provinces, Balochistan and Sindh, have sporadically implemented solar projects. Although the number of operational mini-grids in the country is limited, the existing projects serve as compelling examples of the potential and viability of off-grid solutions in Pakistan. 1.2 EXISTING REGULATORY FRAMEWORK AND KEY STAKEHOLDERS This study examines the current policies, guidelines, and framework for rural electrification in Pakistan, and identifies the main stakeholders. This chapter highlights the important clauses of existing policies and frameworks. It proposes a mechanism for implementation and pinpoints key stakeholders involved in decision- making and initiating projects for affordable, secure, and sustainable energy access for the entire population. 1.2.1 Policies and Frameworks National Electricity Policy, 2021 That government publication provides policy directives for the identification and achievement of major goals and targets for the power sector.5 The Policy defines the main goal for the Power Sector as “Access to affordable, secure and sustainable energy are the broad and overarching goals, attainment of which will crystallize the vision of the Government for the power sector ”. The policy also provides practical guidance for the power sector on planning and implementation to achieve its goals. 1 Sustainable Development Goals Unit, Government of Pakistan. 2019. Classifying Sustainable Development Goals (SDGs) Targets &  Indicators. 2  The Multi-tier Framework (MTF) survey is a tool developed by the World Bank to assess the energy access status across multiple tiers, including reliability, quality, and affordability of electricity services. 3 Kunda refers to electricity theft involving the unauthorized tapping of electrical lines to divert electricity for personal use without paying  for it. 4  The 18th Amendment to the Constitution of Pakistan devolved more powers to the provinces, strengthened parliamentary democracy, and enhanced provincial autonomy, leading to a more decentralized system of government in the country. 5 National Electric Power Regulatory Authority. 2021. “National Electricity Policy 2021.” Accessed March 27, 2024.   2 PAKISTAN LEAST-COST ELECTRIFICATION STUDY The key clauses on rural electrification are: Clause 3.1.1. Accessibility of electric supply to all areas, including rural areas, at affordable rates is the cornerstone of socio-economic development. Making power available, when it is not affordable, has limited value. The Government shall strive to ensure that electricity is accessible to all consumers at rates which are commensurate with their ability to pay, coupled with the development of an efficient and liquid market design. A liquid market design and affordable supply of electricity would also contribute vastly to the financial turnaround and commercial viability of the power sector. Clause 5.8.7. The Government aims for the progressive elimination of load shedding in all areas, including rural areas, for consumers who pay their electricity bills, in accordance with the legal framework of the country. Further, to promote electricity access to areas where grid expansion is financially unviable, off- grid and micro-grid solutions will be explored. Integrated planning shall provide for rural electrification and the provision of electricity to unserved areas of the country. Alternative & Renewable Energy Policy, 2019 The Alternative & Renewable Energy Policy, 2019,6 is applicable to all projects employing alternative or renewable energy for producing power. The policy covers the private sector, public sector, and public–private partnerships. The main targets of the policy are to increase the share of “green” energy in the overall energy mix, provide the least-cost power generation, fast track development of alternative and renewable energy power projects, encourage utilization of indigenous resources, and encourage private-sector investment. Revised Guidelines for Sustainable Development Goals Achievement Programme (SAP) The Sustainable Development Goals Achievement Programme (SAP) is a federal government initiative to achieve Sustainable Development Goals and undertake projects that address pressing, practical issues faced by the community in different socio-economic regions. The Cabinet Division (Development Wing) has approved a Steering Committee on SAP7, and has published Revised Guidelines for SAP which outline the detailed procedures and requirements for the implementation of rural electrification projects. The steering committee has been empowered to oversee the implementation of the program through provincial governments and line ministries or divisions. A dedicated subcommittee has been formed under the steering committee for the identification and approval of projects. The chart below sets out the flow of activities and the role of each stakeholder as described in those revised guidelines. National Electric Power Regulatory Authority Licensing (Microgrid) Regulations, 2022 The National Electric Power Regulatory Authority (NEPRA) published Licensing (Microgrid) Regulations to define the licensing and application process for the establishment of mini-grids in Pakistan. As per NEPRA Regulations, “microgrid8” means a localized energy system that fulfills the following criteria: • It is a self-contained distribution system operating at a voltage not exceeding 33 kV for the distribution of electric power with a peak distribution load not exceeding five megawatts; • It is intended to serve an unserved market; and • It is not connected directly or indirectly to the national grid. A microgrid applicant may be an individual, or a company, a cooperative society, a partnership, or a social welfare organization, in each case duly formed and registered under the applicable law. DISCOs are the main stakeholder in this process. NEPRA has developed a simplified microgrid application form which it has made available on its website. The guidelines mandate that each microgrid shall be designed, constructed, and operated in compliance with the applicable documents, distribution code, applicable laws, and international standards. This is done to ensure that the microgrid is compatible with any future expansion of DISCOs and can be integrated with the distribution network upon approval from NEPRA. 6 National Electric Power Regulatory Authority. 2019. “Alternative and Renewable Energy (ARE) Policy 2019.” Accessed March 27, 2024.  7  Cabinet of Pakistan. “Notification: Special Assistant to the Prime Minister (SAP) Terms of Reference.” June 13, 2022. 8 “ Microgrids” and “mini-grids” are used interchangeably in this report. While the Government of Pakistan documents sometime mention  “microgrids”, this report mostly used “mini-grids”. 3 PAKISTAN LEAST-COST ELECTRIFICATION STUDY Figure 1.1: Flow of activities for village electrification in Pakistan Seeks Approval for Release of Funds Forwarding scheme to Project Manager Application for Construction (DISCO) for preparation Village Electrification Deputy of Feasability report / Cost Estimate Project Manager Construction Applicants Commissioners (DISCO) Forwarding request of the residents along with Feasibility report / Cost Estimate Submission of Progress / Overall Status Report Supervision Steering Committee Sub-Committee Approve the Proposal Rural Electrification schemes and release of Funds Provincial Government / P&D Departments Submission of Status Report Release of Funds Deputy Divisional Commissioner Commissionners Release of Funds Submission of Status Report Project Manager Construction (DISCO) Submission of Completion Report for Approval Opération Division / Consultant Sub-Division Handing over of Completed Schemes 4 PAKISTAN LEAST-COST ELECTRIFICATION STUDY 1.2.2 Key Stakeholders The key stakeholders related to rural electrification and mini-grids are presented below. Ministry of Planning Development & Special Initiatives The Planning Commission is a financial and public policy development institution of the government of Pakistan. Under the aegis of the Ministry of Planning, Development, and Reforms, this Commission undertakes research studies and states policy development initiatives for the growth of the national economy and the expansion of the public and state infrastructure of the country in tandem with the Ministry of Finance (MoF). Ministry of Energy (Power Division) The Federal Ministry of Energy (formerly, Ministry of Water & Power) is the government’s executive arm for all issues relating to electricity generation, transmission & distribution, pricing, regulation, and consumption in the country. It exercises this function through its various line agencies as well as relevant autonomous bodies. It also serves to coordinate and plan the nation’s power sector, formulate policy and specific incentives, as well as liaise with provincial governments on all related issues. Provincial Planning & Development Boards (P&D) In each province, the Planning and Development Board is the main development forum. The board is headed by a Chairman. The Secretary of P&D is the Administrative Head. National Electric Power Regulatory Authority (NEPRA) is responsible for regulating the electricity supply in Pakistan. Distribution Company (DISCO) DISCO means the distribution companies Licensed by NEPRA to engage in the distribution of electric power that are also holders of a Supply License either granted by the Authority or as per provisions of the Act. Provincial Energy Departments Provincial Energy Departments are responsible for the regulation and policy formulation regarding the power sector. After the passing of the 18th Amendment, provinces are fully empowered to develop power projects through the public or private sector. Project Developers Private organizations are interested in setting up mini-grids for the electrification of unserved consumers. The key stakeholders for rural electrification programs are: Public-Sector Schemes • Ministry of Planning Development & Special Initiatives • Ministry of Energy (Power Division) • Provincial Planning & Development Boards (P&D) • National Electric Power Regulatory Authority (NEPRA) • Distribution Companies (DISCOs) Private-Sector Schemes • National Electric Power Regulatory Authority (NEPRA) • Distribution Companies (DISCOs) • Provincial Energy Departments • Project Developers 5 PAKISTAN LEAST-COST ELECTRIFICATION STUDY 1.3 STUDY OBJECTIVE The objective of this study is to provide a geospatial-based least-cost electrification plan that identifies the optimal route for electrification through grid expansion and off-grid services in Pakistan, and recommendations for implementation of policies, actions and investments to achieve universal electricity access. The study also provides detailed analysis on potential mini-grid clusters to support future development of identified mini-grid potential. The study is intended to help inform the World Bank’s engagement with the federal and provincial governments of Pakistan on energy access policy and potential future investments, including actions to support greater private-sector participation in provision of electricity access. 1.4 APPROACH As the majority of the unelectrified population in Pakistan resides in rural sites, a blanket extension of the grid extension would not be a feasible way to achieve universal access to electricity, for both technical and economic reasons. Meanwhile , given the remoteness and sparse population density in many unelectrified communities, off- grid systems, including mini-grids and individual systems, can be a viable route toward universal electricity access. The study analyzed the various options for expanding electrification in Pakistan, with the goal of providing universal access to electricity by 2030. In addition, the work identifies the most cost-effective methods of electrification and evaluates over 1,000 high-priority electrification sites through geospatial analysis. The report includes a summary of the least-cost electrification analysis, mini-grid portfolio assessment, and recommendations for electrification options for Pakistan, taking into consideration policy, institutional, regulatory, and financing aspects. The study was carried out by a team of experts over a period of two years in collaboration with all the key federal and provincial organizations involved in electrification planning including the Ministry of Energy (Power Division), Alternate Energy Development Board (AEDB), Central Power Purchasing Agency (CPPA-G), distribution companies, and provincial energy departments. The study was conducted for Balochistan, Khyber Pakhtunkhwa, Punjab and Sindh provinces. The four provinces represent most of the overall market potential in the country. K-Electric’s jurisdiction was excluded because the entity is privately owned, subject to planning and expansion procedures that would differ from those for national utilities; however, this study’s recommendations could provide useful inputs to K-Electric’s own strategy. The study provides recommendations at a national scale for planning grid and off-grid programs and can help inform rural electrification projects led by the electricity distribution companies (DISCOs), while the provincial energy departments could coordinate and set up appropriate off-grid programs in their respective jurisdictions. The analysis for this study was conducted using the Reference Electrification Model (REM), which can perform techno-economic optimized power supply designs over large geographical areas to determine a least-cost mix of grid and off-grid electrification technologies. The REM designs a grid extension of the existing medium-voltage (11 kV) distribution infrastructure, along with off-grid systems. The study takes into consideration population and demand growth up to 2030. The scenario describing the least-cost electrification plan for universal electricity access is supplemented by sensitivity studies concerning key parameters—namely electricity demand, off-grid component costs, grid reliability, and grid energy cost. The results show the shares of the three technology options for delivering electrical service: individual systems (usually solar home systems, or SHS), mini-grids, and extension of the existing distribution grid. However, it is important to note that the design of any necessary additional high-voltage lines relating to the provision of additional power capacity and stability is outside the scope of this study. 1.5 REPORT STRUCTURE Following this introduction, the second chapter of this report explains the Reference Electrification Model (REM) and how it serves the decision-making processes behind the national perspective offered by this study. In chapter 3, we present the reference base-case scenario for the least-cost electrification study (LCES), plus results, sensitivity analysis and summary conclusion for Balochistan, KP, Punjab, and Sindh respectively. Chapter 4 focuses on mini-grid potential, and chapter 5 goes beyond geospatial planning and least-cost modeling to discuss the steps and framework needed to achieve universal electrification. That chapter offers recommendations for implementing the least-cost plan. 6 PAKISTAN LEAST-COST ELECTRIFICATION STUDY 2. METHODOLOGY AND INPUTS 7 PAKISTAN LEAST-COST ELECTRIFICATION STUDY The objective of the Least-Cost Electrification Study (LCES) is to establish the medium-voltage (11 kV) distribution network least-cost plan, along with off-grid options, that meets the forecast electricity demand for the planning horizon for the year 2030 and provides access to all unelectrified customers by 2030 at the lowest economic cost. In this report, the off-grid options refer to both mini-grids and individual systems. Planning of the infrastructure takes account of demand and customer growth within the 2030 time horizon, while ensuring that the recommended grid extension and off-grid projects can sustain future growth. The plan evaluates the mix of both grid extension and off-grid systems that results in the lowest cost in present value terms. The cost calculations consider both the capital investment needed for building the system as well as the operational cost of running the system. A schematic of the approach is shown below. Figure 2.1: LCES approach 2.1 OVERVIEW OF THE PLANNING METHODOLOGY The main activities were as follows: • Data gathering, inference, and GIS processing: This entails collection, processing, and geo- enrichment of information to estimate relevant attributes of different types of electricity customers in the country. Among these attributes are building location, demand profile, and existing electricity supply. The information gathered during the early stages was used for demand forecast, GIS mapping, and least-cost generation design. • REM analysis: This entails determination of the cost-optimal electrification mode (grid extension at the distribution level, mini-grids, and standalone systems) applicable to the entire country, producing detailed network designs down to the individual connection level for universal access. The model allows decision-makers to incorporate evaluations of quantifiable externalities into the optimization; for instance, by limiting generation from fossil fuels, and promoting renewable energy sources. 8 PAKISTAN LEAST-COST ELECTRIFICATION STUDY REM creates a least-cost electrification plan which will provide electricity service to every household and consumer throughout the area of analysis. Historically, in Pakistan—as elsewhere—electrification has focused on the expansion of the existing electricity grid. However, the precarious financial situation of many distribution companies, the lack of incentives for these companies to connect more customers, as well as the diminishing cost of solar photovoltaics (PV) and battery storage have enabled new technologies such as mini-grids and solar home systems (or solar kits) to compete favorably with grid extension under some conditions. Each technology option considered by REM provides specific benefits but is accompanied by trade-offs in cost, environmental impact, and service quality. REM evaluates each technology option with detailed cost calculations before assigning a technology (solar home system, mini-grid, grid extension) to an end-consumer. Trade-offs in reliability and service quality are incorporated into the decision process by assigning a cost- penalty to less-reliable energy service as a cost of non-served energy. REM is a toolbox with several functions, which can be used to support a variety of use cases for electrification planning. The fundamental algorithms include the following four key functions: • Clustering: groups buildings in relevant units of comparison; • Grid extension design: designs a distribution network to connect a group of buildings to the existing distribution network and maintain voltage and frequency stability within the network; • Off-grid supply design: chooses the best off-grid supply components to meet an off-grid load; and • Mini-Grid network design: designs a distribution network for a mini-grid using the Reference Network Model which is built within the REM tool. More details and background information for the model are presented in Annex A1. The key technology options are defined as follows: Technology Definition Option Grid Densification Grid densification is a new grid connection for communities located close to a distribution grid, that is, within 500 m of the current infrastructure. Due to proximity to the electric grid infrastructure, the majority of these densification customers essentially need a low-voltage drop-down connection from the existing medium voltage lines with appropriate step-down transformers. Grid Extension Grid extension is designed using REM for those unelectrified customers who are further than 500 m from the current infrastructure. The grid extension can include medium- and low-voltage line extension from the existing network with transformers and necessary protections. Mini-Grid Mini-Grids are isolated clusters of customers that form a networked system with their own generation assets. Individual System Individual system customers have their own solar home systems with backup diesel generation as needed to provide access. REM designs the generation requirement based on the demand assumption for the individual customer. 9 PAKISTAN LEAST-COST ELECTRIFICATION STUDY 2.2 KEY MODEL INPUT PARAMETERS The table below provides a summary of the key parameters and the data sources used for the LCES. The specific value used for each province varies for some of the key inputs, such as electrification rate, number of existing residential and commercial customers, as well as public buildings. Inputs for each province are discussed later in subsequent sections. Local inputs: Table 2.1: Local input parameters for REM Parameter Value used Source Location of buildings Projected 2030 population and geospatial HRSL layer data 2020 distribution using building location and numbers Census 2017 based on the 2020 HRSL dataset and 2017 Census. Electrification rate Distribution companies’ customer count (2020 data) DISCOs/GridFinder combined with geospatial analysis of proximity to the current grid infrastructure9 Target electrification 100% national electrification rate by 2030 rate Number of existing Based on DISCOs’ commercial customer count DISCOs (2020) commercial customers data Number of new Commercial customers estimated for 2030 based DISCOs (Historical data) commercial customers on historical growth data and the ratio of residential (2030 projection) to commercial customer count from the DISCOs Number of other Schools – GIS data based on the Ministry of Ministry of Education customer types Education’s dataset Health facilities – GIS data (2020) Alhassan Dataset based on Alhassan Dataset High-Resolution Settlement Layer (HRSL) maps produced from satellite imagery have been utilized to identify residential customers. The methodology for determining customer locations has been applied successfully to several electrification planning projects in Africa, Latin America, and Asia. The following steps outline the customer location and demand characterization methodology: • Estimate the existing population within 30 m x 30 m blocks throughout Pakistan (estimates rely on the 2020 dataset).10 • Project population growth within each block for the target planning year (2030) based on past growth rates. Census data has been consulted for the final population growth rate estimate. • Calculate the number of households within every 30 m x 30 m block for the target electrification year. Conversion from population to household number is done using census data on average number of people per household. • Estimate household locations within each block utilizing a uniform distribution for 2020. Additional households for 2030 due to population growth are then distributed on top of the 2020 households as future projections. Classification of urban and residential households is done based on the population density of clusters, and the percentage value of rural and urban households utilized is given by the census data. 9 This methodology does not consider informal connections to the grid, or those households that may be connected to the national grid but  are not directly paying the utilities and do not have meters, and do not show up on the utility customer count. Due to lack of geospatial information regarding these customers, they would be bundled under the grid densification category. 10  acebook Connectivity Lab and Center for International Earth Science Information Network - CIESIN - Columbia University. 2020. High- F Resolution Settlement Layer (HRSL). Source imagery for HRSL © 2020 Digital Globe. 10 PAKISTAN LEAST-COST ELECTRIFICATION STUDY • Merge household locations with the location of additional customer types—in this case: health centers, and schools. This data is based on existing GIS data. Commercial customer location is estimated based on the residential customer clusters. For example: based on the DISCOs data, and projection for 2030, for every 23 residential customers, there is estimated to be one commercial customer for Sindh. • Estimate the electrification status of each customer based on proximity to existing electricity infrastructure (500-meter buffer).11 For the LCES, four key classes of electricity consumer have been identified for Pakistan: (i) Domestic (rural and urban), (ii) Commercial, (iii) Schools, and (iv) Health Centers. One of the major classes, namely Industrial customers, is not included in the analysis given the assumption that most will be in urban grid-connected areas. In Pakistan, the future special economic zones (SEZ), set to house new industrial customers, will be located in areas that are already grid-connected. In the interests of other prospective industrial consumers located further from the current grid, DISCOs have been contacted regarding grid extension when required. Regarding agricultural customers, there is major uncertainty about the locations of the actual irrigation loads.12 Consequently, the base-case scenario for the LCES was created without agricultural irrigation demand, and only looked at agricultural potential for the mini-grid assessment using the VIDA tool. The least-cost geospatial model considers hourly demand profiles. For each customer type a typical daily profile is shown below. The profiles are generated based on daily energy consumption and peak power requirements, which vary slightly for different provinces, and average data is obtained from DISCOs’ customer consumption data. The algorithm mixes three different base profiles with various multipliers: (i) a typical residential profile with a larger evening load, (ii) a typical daytime load profile consistent with customers operational during the daytime, and (iii) a profile with the load distributed over 24 hours. Figure 2.2: Average hourly demand profile for different types of customers 11  ased on feedback from DISCOs, approximate grid coverage distance varies between 100m and 400m for urban and rural areas. 500 B meters is taken into consideration in the study to cover both current grid-connected customers as well as potential densification customers. 12 T  he Spatial Production Allocation Model (SPAM2010) data is the base dataset that provides geospatial information on irrigated land coverage on a 10 km by 10 km pixel. In aggregate, the dataset can provide a good match between the land coverage area shown by SPAM (16 million hectares) and the census data collected through the Provincial Agricultural Departments (18 million hectares). However, the pixel size is too large to get more granular information on the location of the actual agricultural irrigation loads for the least-cost simulation. There is further uncertainty about the type of irrigation: whether rain-fed, or canals, or tube wells. The SPAM data does not break down different types of irrigation systems, so there will be uncertainty in knowing exactly where the pump loads are located even within the cropland areas given that not all croplands are irrigated using pumps. 11 PAKISTAN LEAST-COST ELECTRIFICATION STUDY Costing parameters: Table 2.2: Network and generation component costing assumptions Input Data Value Comments Cost of fuel (diesel) 0.80 USD/L Global Petrol Price database (Nov 2021) Grid electricity cost (grid 0.12 USD/kWh The wholesale market price for energy for utilities supply cost) (see section below) Network component cost 9,912 USD/km (MT line Based on DISCOs’ list of Medium Tension (MT) average using a catalog and Low Tension (LT) standard costings of different types of wires) Costings include poles and line accessories 4,363 USD/km (LT line No meters, service connection, cables, and average using a catalog accessories cost are considered for the network of wires) Additional per customer cost added ($500/ customer as provided by the DISCOs to cover the cost of meter, in-house drop-down connection, accessories, and so forth) PV 500 USD/kWp Standard PV panel costings—includes PV racks (assumption based on current market availability) Battery (Pb-acid) 125 USD/kWh Lead-acid battery costings (assumption based on current market availability) Inverter 190–927 USD/kW Electronics (inverter) costs—range for different sizes* Charge controller 131–481 USD/kW Electronics (charge controller) costs – range for different sizes* Discount rate 10% Based on input from DISCOs * Catalog of electronic components is utilized where cost varies for different sizes. The economy of scale comes into effect here, where larger components have a lower cost per unit. Calculated (adopted) assumptions. Inputs that are adjustable to different scenarios. • Grid energy cost (grid supply cost): Cost of energy estimated at Medium Tension (MT) distribution feeders. The data provided by the distribution companies (DISCOs) show an average cost of energy for the distribution companies to be 8 US cents/kWh. Higher energy costs in the future can result from upstream reinforcements at the generation and transmission levels. Higher energy costs result in more mini-grids and fewer grid extensions (with small influence in individual systems, except when they are special high- demand customers). The effect of this parameter on the solution is gradual and not substantial, especially if building-to-grid distances are large. For the reference base case, 12 US cents/kWh as the energy cost for the grid for 2030 is assumed taking into consideration the need for potential upstream reinforcement investments to achieve universal access. 12 PAKISTAN LEAST-COST ELECTRIFICATION STUDY • Grid reliability: Reliability of the supply of electricity from the existing grid. The reliability metric is expressed as a single overall percentage relating the annual non-served energy to the total annual demand. The transmission and distribution (T&D) losses vary between DISCOs, with an average loss of 17 percent. The National Transmission and Dispatch Company (NTDC) has data on DISCOs’ aggregate technical and commercial (ATC) losses which expresses a combination of technical loss, theft, inefficiency in billing and collection, and default in payment. Since it is impossible to break down the losses due to technical aspects of the network or lack of generation, and commercial losses, single grid reliability is assumed for the base case, adjusting the parameter for sensitivity analysis to show the effect. Utilities in Sindh, Balochistan, and KP have high T&D losses, above 25 percent with PESCO and SEPCO having actual losses above 35 percent.13 The 2022 MTF survey also shows the three provinces making electricity available for only 7–13 hours in 24, roughly half the availability in Islamabad. In Punjab, MEPCO has T&D losses of 15 percent. For these DISCOs, both the System Average Interruption Frequency Index (SAIFI) and System Average Interruption Duration Index (SAIDI) fall considerably short of the target set by the National Electric Power Regulatory Authority (NEPRA), with MEPCO having the worst performance. For these cases, a grid reliability of 75 percent is taken to reflect the current situation. For other DISCOs (in Punjab), namely IESCO, GEPCO, FESCO, and LESCO, the T&D losses are around 10 percent, and also the reliability matrix and performance are better. For these DISCOs, grid reliability of 90 percent is assumed. This parameter can dramatically affect the presence of grid extensions if the cost of non-served energy (CNSE) is significant, since non-served energy and unreliable technology are directly penalized by the CNSE penalty. Strategic decision-making. Inputs that are related to social aspects. • Cost of Non-Served Energy (CNSE): REM is a cost-driven tool, so the lack of quality/reliability of power supply must be translated to cost (it may also be imposed as a constraint in the case of off-grid systems). CNSE is the cost, to consumers, of the energy that is not served. In other words, CNSE represents the cost (that is, the loss of utility) incurred by consumers when there is no electricity at a time when they are planning to use it. The value of the loss of utility depends on the time of the day, the activity being performed at that time, and the economic status of the customer. For simplicity, the model considers a single CNSE value. • There could be multiple ways of estimating the value of CNSE. One of them is to adopt—as a proxy— the cost of an alternative energy solution (for example, diesel generator set, or solar kits) that might be used when electricity is not available for certain uses or functions. CNSE should be set to a value bigger than the typical energy cost in the system. In Pakistan, the cost of energy through diesel generators is estimated to be between 30 and 40 US cents/kWh.14 The energy cost for solar kits, especially in rural areas, can be even greater. A CNSE of 45 US cents/kWh is considered for the 2030 timeline for the modeling of the base-case scenario. A typical value for the cost of non-served energy ranges from 40–80 US cents/kWh based on the experience. REM analyzes the annuity cost for each electrification mode taking into consideration the capital expenditures (CAPEX), operational expenditures (OPEX) including the grid energy supply cost for grid extension, and the CNSE to penalize unreliable systems. It can be misleading when comparing the systems to focus only on the overnight investment cost in the form of CAPEX. A grid-extension option may have a lower CAPEX per connection but could have a substantial OPEX cost due to grid energy supply costs or a high CNSE for unreliable service. However, for the same cluster, a mini-grid’s CAPEX per connection may be higher, but the overall total annuity cost could be lower due to a lower OPEX cost as well as lower CNSE if the system is more reliable than the grid. REM considers all these cost factors and computes the total annuity for each electrification mode to identify the least-cost option. As defined in the prior section, the grid densification customers are identified as those within 500 meters of the existing grid infrastructure and are not considered for REM evaluation. The overnight CAPEX cost estimation for grid densification is based on the average connection cost ($/connection) which is estimated from the REM evaluation of the grid extension projects that are closest to the current 11 kV network. The assumption here is that densification is a type of grid extension defined for those customers who are in very close proximity to the current network. Beyond this cost estimation, no detailed analysis for upstream reinforcement of the current grid infrastructure or load flow analysis has been done on the existing feeders because that would require a more detailed in-depth study of the existing distribution, transmission, and generation plants. For cost calculations of grid-extension projects in REM, the cost of upstream reinforcement regarding current infrastructure as well as the addition of generational capacity is taken into consideration inside a single parameter in the form of the grid supply cost as described earlier. The energy cost as seen at the medium voltage level by the distribution company is assumed to be 12 US cents/kWh, that is, a 50 percent increase considering the current energy cost of 8 US cents/kWh. 13 NEPRA Performance Evaluation Report FY 2020–2021/ FY 2021–2022  14 Economic power generation for an off-grid site in Pakistan, M.K. Abbas, Q. Hasan, IEEE, 2015.  13 PAKISTAN LEAST-COST ELECTRIFICATION STUDY For grid extension systems, the CAPEX consists of network costs including wires, transformers, poles, and accessories. The cost of energy is part of the annual operational cost. For off-grid systems, generation component costs are part of CAPEX. For off-grid systems that meet the demands of individual households, CAPEX to cover the solar set-up only (solar panels, inverters, batteries, and so forth) will suffice, whereas for mini-grids, the cost of solar-based generation plus distribution network will determine the overnight capital cost estimates. In the following section, the specific inputs for the four provinces of interest are discussed separately. 2.2.1 Balochistan The total number of customers has been projected for 2030 based on population growth assumptions and growth in non-residential and commercial customers for Balochistan. The projection is based on 2017 Census data and the Quetta Electric Supply Company (QESCO) historic customer count dataset. For universal access, Balochistan is predicted to have more than 2.6 million electricity service connections by 2030. Among these connections, the existing residential customer count is 477,757 (QESCO, 2020), with an additional 165,219 non- residential connections, giving an electrification rate of 25 percent for the province, excluding the Lesbella district which falls under K-Electric’s jurisdiction. For the least-cost modeling, the districts served by K-Electric are not included in this study. Based on the geospatial analysis, 66 percent of the province’s population is within 500 meters of the current grid infrastructure, and or has a nightlight data signature showing some form of electrification. This would be the combination of the grid-electrified, off-grid households, and those “under-the- grid” households that do not have access. The 2022 MTF survey conducted by the ESMAP group of the World Bank shows a grid electrification rate of 54 percent for Balochistan. Among them, the province has a high informal connection rate of 15 percent and more than 50 percent without electric meters which would not be accurately captured by the utility customer count data. The discrepancy between the electrification rates from the survey and those yielded by geospatial analysis primarily stems from the utilities’ customer count data and informal and off-grid connections. As mentioned previously, due to the lack of proper geospatial information on individual electrified customers, especially the informal ones, it is impossible to distinguish informal connections from formal grid connections. The majority of these informal connections would be classified as grid densification as defined in this study. With the low reliability of services for these customers, grid densification and strengthening would be needed in order to make sure they have proper, formal connections. The overall electrification rate from the utility customer count also does not consider smaller household off-grid systems, which according to the MTF survey cover nearly 43 percent of the population of Balochistan (nationally, the highest proportion). Many of these off-grid systems are small individual systems—for basics, such as lighting—unable to provide higher-tier services. Thus, it makes sense to consider them for the least-cost study. The table below provides a summary of the key parameters and the data sources used for Balochistan’s geospatial least-cost analysis. 14 PAKISTAN LEAST-COST ELECTRIFICATION STUDY Local inputs: Table 2.3: Local input parameters for REM – Balochistan Parameter Value used Source Electrification rate Distribution companies’ customer count (2020 data). QESCO (2020) With a customer count of 477,757 domestic connections, the electrification rate for 2020 is 25% excluding K-Electric jurisdiction (Lasbella district) Number of existing As per DISCOs’ customer count data: QESCO (2020) commercial customers 123,504 Number of new Additional 30,363 commercial customers estimated for QESCO commercial customers 2030 outside the 500-meter buffer, based on historical (2030 projection) growth data and the ratio of residential to commercial (Historical data) customer count from DISCOs Number of other Schools – 12,436 in total (3,874 not electrified) Ministry of Education customer types (2020) Health facilities – 1,480 in total (411 not electrified) Alhassan Dataset Estimated customer 0.9 million (2030 projection) Geospatial analysis count (outside 500 m (HRSL data and buffer from the current Census projection) grid) A methodology as described in the prior section has been utilized in identifying residential customers’ locations using High-Resolution Settlement Layer (HRSL) maps. Table 2.4: Balochistan’s population data Population Average 1998–2017 Households Households (2017) Household Average (2020) (2030) size Annual Growth Rate Balochistan 11,758,858 6.87 3.37 1.89 million 2.66 million (excluding Lasbella) Rural 8,634,055 6.80 3.10 1.38 million 1.87 million Urban 3,124,803 7.06 4.16 0.51 million 0.79 million Source: Pakistan Census 2017 (Waya analysis) 15 PAKISTAN LEAST-COST ELECTRIFICATION STUDY Table 2.5: Number of customers (2020, and 2030 estimates) – Balochistan Estimated New Customers New Customers by 2030 Total Electrified Grid Densification (Beyond 500 m buffer Potential Customers by 2030 of the current grid) Customers 2020 in 2030 Total customers 477,757 1.3 million 0.9 million* 2.67 million (residential) Commercial 123,504 30,363 174,784 School 8,562 3,874 12,436 Health center 1,069 411 1,480 Source: QESCO data & Waya analysis. Note: * Grid Extension or Off-Grid could be the least-cost solution for these customers. Distribution network data was received from the utility namely for the feeders including Sibi, Pishin, Mekran, Loralai, and Khuzdar, as well as interconnecting 66 kV network. Based on the existing network data, nightlight target areas, and population growth projection, by 2030, there would be roughly 1.8 million households that are estimated to be near to (less than 500 meters from) the current grid infrastructure, for whom grid extension and densification would be the least-cost solution. Among them, more than 470,000 households already have grid connections (2020), leaving 1.3 million customers for grid densification.15 REM has identified roughly 0.9 million unelectrified customers who will need to be electrified by 2030 but are beyond 500 meters from the current grid infrastructure. These customers’ locations have been estimated using the HRSL dataset and are input for REM to find the least-cost electrification options for achieving universal access by 2030. Based on QESCO’s data, the current average domestic consumption per customer is about 3.5 kWh/day. According to NTDC’s projection, the average annual growth rate for Balochistan’s residential overall energy demand is roughly 7.5 percent. This growth can be attributed both to a rise in the number of connections and growth in per-customer demand. QESCO’s historic data does not show a big change in the per-connection demand from 2015 to 2020. The current consumption level sits closer to the Tier 3 or Tier 4 level as defined by ESMAP’s Multi-Tier Framework system. For the LCES, we consider a similar level of consumption for the newly electrified urban residential demand, keeping it on track with the current grid-connected customers’ consumption, and have considered a lower Tier 2 (1 kWh/day) as a rural residential demand estimate for 2030. A recent report from the Rockefeller Foundation with Smart Power India has shown a similar level of demand for a rural household—0.79–1.7 kWh daily—based on their experience of mini-grid projects in India.16 Based on rural electrification work around the world,17,18 in general, off-grid connections and rural communities have lower demand, and show slower growth in demand than among grid-connected customers. A high-demand scenario has been explored in sensitivity analysis to show the effect of the parameter on the overall result. Public buildings, that is, schools and hospitals, are assumed to have the same demand throughout the country. The small commercial customer’s demand is based on DISCO’s consumption data using a similar methodology to that used to estimate residential consumption. 15  Informal grid-connected customers not identified in this study would be lumped into the grid densification category as they would be near to the current grid infrastructure. An estimated 66 percent of the population within Balochistan falls within the 500-meter buffer of the grid, and or has nightlight data showing some form of electrification. 16  ural Electrification in India – Customer Behavior and Demand, Smart Power India, Rockefeller Foundation, Feb 2019. R 17 Applied Energy Lab Brief: Consumption Trends, Smart Power Myanmar, 2020.  18  ust Light is Not Enough – Solutions for Improving Energy Access in Rural Sindh, Indus Earth Trust (IET), 2016. J 16 PAKISTAN LEAST-COST ELECTRIFICATION STUDY Table 2.6: Demand data for different types of customers – Balochistan Customer Daily average Annual average Description category consumption consumption (kWh) (kWh) Urban Residential 3.8 1,408 Residential household (HH) with Lights, Cell Phone, Television, and Medium-Power Appliances including cooling fans (ESMAP Tier 3/4) Rural Residential 1 369 Residential HH with Lights, Cell Phone, and some Low-Power Appliances (ESMAP Tier 2) Commercial 3.7 1,363 Commercial shops, stores, restaurants – service sector School 7 2,566 This is equivalent to two computers, five fans, and eight fluorescent light bulbs used for nine hours at a medium-sized school Health center 10.5 3,799 This is equivalent to a medium-sized rural health clinic or a small district health center with basic lighting, autoclave, and laboratory equipment such as a centrifuge, mixer, incubator, refrigerator, and radio19,20 2.2.2 Khyber Pakhtunkhwa For the analysis of Khyber Pakhtunkhwa (KP), both Tribal Electric Supply Company (TESCO) and Peshawar Electric Power Company (PESCO) utilities’ customers are included in the region. The total customer count has been projected for 2030 based on population growth assumptions and growth in non-residential and commercial customers for KP. The projection is based on 2017 Census data and TESCO’s and PESCO’s historic customer count dataset. For universal access, the KP region in Pakistan is predicted to have slightly more than 6.5 million electricity service connections by 2030. Among these connections, the existing utilities’ 2020 residential connection count (TESCO and PESCO combined) is 3,595,814, with an additional 165,219 non-residential connections. Solely based on the census population count and residential utilities connection count, we get an electrification rate of 75 percent for PESCO’s jurisdiction and a 64 percent rate for the TESCO area, giving an aggregate electrification rate of 73 percent. The MTF survey conducted by ESMAP shows a grid electrification rate of roughly 86 percent for KP. The network assumption is solely based on the predictive mapping of the global power system using open data and the tool called GridFinder, which predicts the location of electricity network lines using night-time lights satellite imagery, and OpenStreetMap data.21 The team was not able to get the real network data from TESCO and PESCO due to security concerns. The existing network data is one of the key input requirements for the model as it builds grid extensions from these existing networks, and the analysis of electrified and non-electrified customers is done based on the location of the current network. Based on the network data from GridFinder which takes into consideration nightlight, the GIS analysis was able to identify only 2,749,121 households near the network that is, within 500 meters of the network lines. Based on this number, we get an electrification rate for the entire KP region of 56 percent. One reason for the discrepancy between the electrification rate estimation and the rate calculated from utilities’ connection count or the MTF survey could be if a single residential customer had multiple electrical meters and connections, thus skewing the customer count for the utility data. Also, since the actual network lines from the distribution companies were not available, there could be missing electrified customers who have a connection but do not show up on the nightlight due to low intensity, so GridFinder does not show any network lines through such communities. For REM analysis, we used the inputs based on the assumed network lines, as these are used to determine the electrification status of the households, and to determine future extension networks. 19  Powering Health: Electrification Options for Rural Health Centers 20  ranco et al., A review of sustainable energy access and technologies for healthcare facilities in the Global South, Sustainable Energy F Technologies, and Assessments, 22, 92–105, 2017. 21 Gridfinder - Global Energy Infrastructure,” n.d. An open-source tool for predicting the location of electricity network lines, using night-time lights satellite imagery and OpenStreetMap data. 17 PAKISTAN LEAST-COST ELECTRIFICATION STUDY Based on the existing network assumption, and population growth projection, by 2030, there would be roughly 923,100 new customers for grid densification located near to (less than 500 from) the assumed current grid infrastructure. This new customer count stems from population growth from 2020 to 2030. Due to proximity to the electric grid infrastructure, the majority of these densification customers essentially need a low-voltage drop-down connection from the existing medium voltage lines with appropriate step-down transformers. In addition to these unelectrified customers, REM has identified more than three million residential and non- residential customers by 2030 who need to be electrified but are further than 500 meters from the assumed current grid infrastructure. These customers’ locations have been estimated using the HRSL dataset and are input for REM to find the least-cost electrification options for achieving universal access by 2030. The table below provides a summary of the key parameters and the data sources used for the KP analysis. Local inputs: Table 2.7: Local input parameters for REM – KP Parameter Value used Source Electrification rate Based on assumed network data and GIS analysis, only GridFinder/GIS 2,749,121 customers are identified within 500 meters22 analysis of the network lines, giving an electrification rate of 56%. Number of existing As per DISCOs’ customer count data: 390,953 DISCOs (2020) commercial customers Number of new Additional 123,979 commercial customers are estimated DISCOs (Historical commercial customers for 2030 based on historical growth data and the ratio data) (2030 projection) of residential to commercial customer count from the DISCOs. Number of other Schools – 30,992 in total (17,329 not electrified) Ministry of Education customer types (2020) Health facilities – 3,226 in total (971 not electrified) Alhassan Dataset Estimated customer 2.86 million (2030 projection) Geospatial analysis count (outside 500  m (HRSL data and buffer from the current Census projection) grid) A similar methodology as described for all other provinces has been utilized in identifying residential customers’ locations using the High-Resolution Settlement Layer (HRSL) maps. 22 According to the utilities in Pakistan, customers are within 500 meters of the existing 11kV network. For urban customers, the distance is  even less – 100 meters. 18 PAKISTAN LEAST-COST ELECTRIFICATION STUDY Table 2.8: KP population data Population Average 1998-2017 Households Households (2017) Household Average (2020) (2030) size Annual Growth Rate K h y b e r 35,501,964 7.83 2.89 4.88 million 6.49 million Pakhtunkhwa Rural 29,626,670 7.90 2.76 4.04 million 5.27 million Urban 5,875,294 7.52 3.47 0.84 million 1.22 million Source: Pakistan Census 2017 (Waya analysis) Table 2.9: Number of customers (present – 2020, and 2030 estimates) – KP Estimated New Customers New Customers by 2030 Total Potential Existing Grid Densification (Beyond 500 m buffer Customers Customers 2020 by 2030 of the current grid) in 2030 T o t a l 2,749,121 923,105 2.86 million* 6.49 million customers (residential) Commercial 390,953 123,979 514,932 School 13,593 17,329 30,922 Health center 2,255 971 3,226 Source: TESCO/PESCO data & Waya analysis Note: * Grid Extension or Off-Grid can be the least-cost solution for these customers. Based on PESCO’s consumer data, current average domestic consumption per customer is about 4.5 kWh/day. PESCO’s historic data on consumption from 2015 to 2020 does not show a big change in the per-connection demand value. The current consumption level sits closer to the Tier 4 level as defined by ESMAP’s Multi-Tier Framework system. For the LCES, we consider a similar level of consumption for the newly electrified urban residential demand, keeping it on track with the current grid-connected customers’ consumption, and have considered a lower Tier 2 (1 kWh/day) as a rural residential demand estimate for 2030. Public buildings, that is, schools and hospitals, are assumed to have the same demand throughout the country. Similar to residential demand analysis, small commercial consumer demand has been estimated based on the DISCOs’ historic consumption data. 19 PAKISTAN LEAST-COST ELECTRIFICATION STUDY Table 2.10: Demand data for different types of customers – KP Customer Daily avg. Annual average Description category consumption consumption (kWh) (kWh) U r b a n 5.06 1,846 Residential HH with Lights, Cell Phone, Television, Residential and Medium-Power Appliances including cooling fans (ESMAP Tier 4) R u r a l 1 369 Residential HH with Lights, Cell Phone, and some Residential Low Power Appliances (ESMAP Tier 2) Commercial 5 1,825 Commercial shops, stores, restaurants – service sector 2.2.3 Punjab The most populous province in the country has five operating utilities—IESCO, FESCO, GEPCO, LESCO, and MEPCO. Each utility’s jurisdiction has been analyzed separately. The total number of customer counts has been projected for 2030 based on population growth assumptions and growth in non-residential and commercial customers for Punjab. The projection is based on 2017 Census data and utilities’ historic customer count dataset. For universal access, Punjab in total is estimated to have roughly 24 million electricity service connections by 2030. For this analysis, it was impossible to get the electrification rate solely from the utilities’ customer count data for Punjab. For 2020, based on the census dataset, Punjab’s residential household count total was close to 19 million. However, the combined utilities’ domestic connection number was 20.16 million. According to the utilities, 8–10 percent of customers have multiple meters, giving rise to the discrepancy in the number of connections, and households. However, there is no record of the exact numbers of multiple meters. Thus, we have utilized geospatial analysis based on current network data, nightlight data, and the estimated locations of customers to evaluate the electrification rate. Based on the overlap of the network buffer and nightlight information, the geospatial analysis was able to identify 16.1 million households in close proximity to the network, that is, within 500 meters of the grid lines. Based on this number, we get an overall electrification rate for Punjab of 86 percent. The breakdown for utilities is provided below. The MTF survey conducted by ESMAP shows a high grid electrification rate of 97 percent for Punjab. The discrepancy in the electrification rate can be attributed to the overall confidence level (90 percent) and the error margin (7.5 percent) of the survey as well as the limitation of the available network and nightlight data used for the geospatial analysis. Table 2.11: Electrification rates, 2020, for different utilities in Punjab 2020 household count Estimated electrified Electrif ication (based on census 2017) residential customer count rate (2020) (based on geospatial analysis) IESCO 2,013,590 1,887,474 94% FESCO 3,852,115 3,229,877 84% GEPCO 2,854,845 2,618,952 92% LESCO 3,313,242 2,992,152 90% MEPCO 5,965,013 5,373,610 90% Source: Census, 2017 & Waya analysis.23 23 A  ll the utilities in Punjab have validated these electrification rates for 2020, except FESCO. Since the geospatial analysis and available dataset was not able to demonstrate a higher electrification rate, the study has considered these numbers as estimated inputs for the least- cost study. 20 PAKISTAN LEAST-COST ELECTRIFICATION STUDY By 2030, population growth projections suggest that approximately 4.4 million new residential customers within 500 meters of the current grid infrastructure will require electrification. Grid densification would be the most cost-effective option for these future customers. Grid densification involves a new grid connection for communities situated close to a distribution grid. These customers typically require a low-voltage drop-down connection from the existing medium-voltage lines, along with appropriate step-down transformers. In addition to these new unelectrified customers, geospatial analysis indicates that more than 2.6 million residential and non-residential customers will require electrification by 2030, but they are located beyond 500 meters from the current grid infrastructure. These customers’ locations have been estimated using the HRSL dataset and are input for REM to find the least-cost electrification options for achieving universal access by 2030. The table below provides a summary of the key parameters and the data sources used for Punjab’s analysis. Local inputs: Table 2.12: Local input parameters for REM – Punjab Parameter Value used Source Electrification rate Based on the network data, nightlight data, and D I S C O s , GIS analysis, 16 million customers are identified GIS analysis within 500 meters24 of the network lines giving an electrification rate of 86% for the province. Number of existing As per DISCOs’ most recent customer count data: DISCOs (2020) commercial customers 2.44 million Number of new Additional 631,304 commercial customers DISCOs commercial customers estimated for 2030 based on historical growth data (2030 projection) and the ratio of residential to commercial customer (Historical data) count from DISCOs Number of other customer Schools – 95,050 in total (8,293 not electrified) Ministry of Education types (2020) Health facilities – 15,422 in total (245 not electrified) Alhassan Dataset Estimated customer count 2.6 million (2030 projection) Geospatial analysis (outside 500 m buffer from (HRSL data and the current grid) Census projection) A methodology similar to that described for Balochistan and KP has been utilized in identifying residential customers’ locations using the High-Resolution Settlement Layer (HRSL) maps. 24  ccording to the utilities in Pakistan, customers are within 500 meters of the existing 11kV network. For urban customers, the distance is A even less – 100 meters. 21 PAKISTAN LEAST-COST ELECTRIFICATION STUDY Table 2.13: Punjab population data 1998–2017 Average Population Average Households Households Household (2017) Annual (2020) (2030) size Growth Rate Punjab 111,993,023 18.81 million 23.86 million IESCO 12,009,992 5.89 2.75 2.21 million 2.93 million FESCO 22,352,268 6.27 1.98 3.78 million 4.58 million GEPCO 16,120,861 6.57 1.94 2.59 million 3.13 million LESCO 22,430,682 6.37 2.21 3.85 million 5.21 million MEPCO 39,079,220 6.54 2.44 6.38 million 8 million Source: Pakistan Census 2017 (Waya analysis) Based on utilities’ data, the current average domestic consumption per customer is about 5 kWh/day. According to NTDC’s projection, the average annual growth rate for Punjab’s residential overall energy demand is 4.6 to 7.2 percent. This growth can be attributed both to a rise in the number of connections and growth in per- customer demand. Utilities’ historic data from 2015 to 2020 show an average growth in per-customer demand of 1.45 percent. The current consumption level sits closer to the Tier 4 level as defined by ESMAP’s Multi-Tier Framework system. For the LCES, we consider a similar level of consumption for the newly electrified urban residential demand, keeping it on track with the current grid-connected customers’ consumption, and have considered a lower Tier 2 (1 kWh/day) as a rural residential demand estimate for 2030. Public buildings that is, schools and hospitals are assumed to have the same demand throughout the country. The commercial customer’s demand is based on the current consumption level for electrified commercial customers based on DISCO data. Table 2.14: Demand data for different types of customers – Punjab Customer Daily avg. Annual average Description category consumption consumption (kWh) (kWh) Urban Residential 5.77 2,106 Residential HH with Lights, Cell Phone, Television, and Medium-Power Appliances including cooling fans (ESMAP Tier 4) Rural Residential 1 369 Residential HH with Lights, Cell Phone, and some Low Power Appliances (ESMAP Tier 2) Commercial 8 2,951 Commercial shops, stores, restaurants – service sector 22 PAKISTAN LEAST-COST ELECTRIFICATION STUDY 2.2.4 Sindh The total customer count has been projected for 2030 based on population growth assumptions and growth in non-residential and commercial customers. The projection is based on 2017 Census data and DISCOs’ historic customer count dataset. For universal access, Sindh in Pakistan is predicted to have more than eight million electricity service connections by 2030. Among these connections, the 2020 residential customer count of the existing utilities (Hyderabad Electric Supply Company [HESCO] and Sukkur Electric Power Company [SEPCO]) stands at 1.56 million, with an additional 370,000 non-residential connections, giving a roughly 24 percent electrification rate for the province excluding K-Electric consumers. For the least-cost modeling, the Karachi district and areas served by K-Electric are not included,25 and only HESCO and SEPCO utility jurisdictions are considered for this region. The electrification rate is solely based on the customer count data from the DISCOs. The geospatial analysis identifies roughly four million households within 500 meters of the existing grid infrastructure. If all of these households were electrified, the electrification rate would be roughly 60 percent. However, ground data reveals that to be an oversimplification, because there are cases of unelectrified households even in close proximity to the grid. The 2022 MTF survey conducted by the World Bank shows a high grid electrification rate for Sindh of more than 75 percent. Excluding K-Electric customers, the rate is slightly lower, at 70 percent. Among those electrified, the province also has a high informal connection rate of 33 percent, in fact 40 percent of those with electricity lack electric meters and would therefore not be captured by the utility’s customer count data. That largely accounts for the discrepancy between electrification rates. Due to the lack of proper geospatial information regarding these informal connections, it is impossible to distinguish these customers from formal grid connections. The majority of these informal connections would be classified as grid densification as defined in this study. With the low reliability of services for these customers, grid densification and strengthening would be needed to make sure they have proper access. Furthermore, the overall electrification rate based on DISCO customer counts in this study does not consider smaller off-grid systems providing for households. The table below provides a summary of the key parameters and the data sources used for Sindh’s analysis. Local inputs: Table 2.15: Local input parameters for REM – Sindh Parameter Value used Source Electrification rate Distribution companies’ customer count (2020 data). HESCO and SEPCO customer count With a customer count of 1.5 million domestic dataset connections, the electrification rate for 2020 is under 24% excluding K-Electric jurisdiction Number of existing As per DISCOs’ customer count data: DISCOs (2020) commercial customers 290,021 (SEPCO and HESCO combined) Number of new Additional 74,867 commercial customers estimated DISCOs commercial customers for 2030 based on historical growth data and the ratio (2030 projection) of residential to commercial customer count from (Historical data: DISCOs’ data 2010–2020) Number of other Schools – 33,974 in total (15,477 not electrified) Ministry of Education customer types (2020) Health facilities – 2,309 in total (386 not electrified) Alhassan Dataset Estimated customer 2.8 million (2030 projection) Geospatial analysis count (outside 500  m (HRSL data and buffer from the current Census projection) grid) 25 11 kV network data from K-Electric was not available to the team.  23 PAKISTAN LEAST-COST ELECTRIFICATION STUDY Table 2.16: Sindh population data Population Average 1998-2017 Households Households (2017) Household Average Annual (2020) (2030) size Growth Rate Sindh (minus 34,407,172 5.55 2.41 6.6 million 8.2 million K-electric) Urban 11,668,154 5.62 2.74 2.45 million 3.23 million Rural 22,739,018 5.47 2.07 4.17 million 5 million Source: Pakistan Census 2017 (Waya analysis) Table 2.17: Number of customers (2020, and 2030 estimates) – Sindh Estimated New Customers New Customers by 2030 Total Electrified Grid Densification (Beyond 500 m buffer Potential Customers by 2030 of the current grid) Customers 2020 in 2030 Total customers 1.56 million 3.7 million 2.8 million* 8 million (residential) Commercial 290,021 74,867 364,888 School 22,497 15,477 37,974 Health center 1,923 386 2,309 Source: HESCO/SEPCO data & Waya analysis. Note: * Grid Extension or Off-Grid systems could be the least-cost solutions for these customers. By 2030, there would be roughly 1.28 million new customers for grid densification located nearby (less than 500 meters from the current grid infrastructure). This new customer count stems from population growth from 2020 to 2030. In addition, through geospatial analysis, nearly 2.4 million current residential households have been identified that are within 500 meters of the grid but are not electrified.26 These could be considered obvious candidates for grid connection. By 2030, a total of 3.7 million residential households are predicted to be connected to the grid through densification as they are estimated to be in close proximity to the grid. Meanwhile, by 2030 another 2.8 million customers located beyond 500 meters from the current grid infrastructure will need to be electrified. These customers’ locations have been estimated using the HRSL dataset and are input for REM to find the least-cost electrification options for achieving universal access by 2030. 26 Informal grid connected customers not identified in this study would be lumped into the grid densification category as they would be near  to the current grid infrastructure. An estimated 60% of the population in Sindh falls within the 500-meter buffer of the grid. 24 PAKISTAN LEAST-COST ELECTRIFICATION STUDY For Sindh, HESCO and SEPCO’s historical consumption data has been analyzed. The 2020 average domestic consumption per customer is slightly above 7 kWh/day. Based on NTDC’s projection for Sindh, the domestic demand for 2030 will be in the order of 10 kWh/day. According to the 2022 MTF survey, the average daily consumption of electricity for Sindh is 4 kWh/day. This also includes informal connections. According to NTDC’s projection, the average annual growth rate for Sindh’s demand is closer to 5.5%. This energy demand level sits closer between Tier 4 and Tier 5 levels as defined by ESMAP’s Multi-Tier Framework system. This is much higher compared to other Asian or African nations’ averages especially when considering rural demand. The consumption data could be skewed by factors such as theft, meter reading, and billing errors or inefficiencies. For the LCES, the Tier 5 level of consumption has been considered for the urban residential demand, keeping it on track with the current grid-connected customers’ consumption, and we have considered a lower Tier 2 (1 kWh/day) as a rural residential demand estimate for 2030. Similarly, the small commercial customer demand has also been estimated based on DISCO’s consumption data. Demand profiles for schools and health centers have been estimated based on global examples and our experiences with rural electrification. Table 2.18: Demand data for different types of customers – Sindh Customer Daily average Annual average Description category consumption consumption (kWh) (kWh) Urban Residential 7.5 2,759 Residential household with Lights, Cell Phone, Television, and High-Power Appliances including cooling/heating (ESMAP Tier 5) Rural Residential 1 369 Residential household with Lights, Cell Phone, and some Low Power Appliances (ESMAP Tier 2) Commercial 8 2,951 Commercial shops, stores, restaurants – service sector 25 PAKISTAN LEAST-COST ELECTRIFICATION STUDY 3. RESULTS 26 PAKISTAN LEAST-COST ELECTRIFICATION STUDY This report presents a distribution network plan and off-grid design for the four key provinces of Pakistan to achieve universal access to electricity by 2030. The Reference Electrification Model used in the study for geospatial planning is specifically designed for distribution network planning exercises to achieve universal access. The model chooses between grid extension and off-grid options for all unelectrified customers by performing a techno-economic analysis to identify the least-cost electrification solution for each customer. 3.1 BALOCHISTAN 3.1.1 Reference base-case scenario All the parameters are assumed for the reference base case as defined in Section 2. The reference base-case scenario is taken with input parameters and assumptions that are most likely to occur by the target year 2030 based on the current situation in the region. The overall result for the least-cost electrification study is shown in the table below. According to the least-cost plan, in Balochistan the majority of the unelectrified customers in 2030 should be connected via grid extension and densification. Among the unelectrified customers, 20 percent of the new customers (almost half a million households) in the province would be connected via mini-grid solutions as the least-cost option by 2030. Due to the sparse distribution of households, more than 220,000 individual systems are identified as the least-cost option for nearly 10 percent of the unelectrified customers. It would cost approximately US$ 1.6 billion, in 2020 real terms, to achieve the universal access target by 2030. Of this amount, off-grid systems comprise more than US$ 700 million or approximately 44 percent of the total capital investment. In Balochistan, among those customers further than 500 meters from the grid infrastructure, the majority of households will have off-grid solutions as the least-cost option for electrification. Table 3.1: Summary of results for the least-cost electrification study – Balochistan Indicator Individual Mini-Grids New Grid Grid Total systems Extensions Densification Number of new 222,538 465,332 209,778 1.35 million 2.25 million customers Avg. Cost Per kWh 0.53 0.33 0.25 0.21 of Demand Served ($/kWh) Fraction of new 0.1 0.21 0.1 0.6 1 customers Annual Energy Demand 133 260 250 (GWh) Overnight Capital Cost: Total CAPEX ($) 0.25 billion 0.42 billion 0.175 billion 0.80 billion 1.6 billion Total annual OPEX 28 million 17 million 27 million ($/year) Avg. OPEX per 129 36 132 customer ($/year) Avg. CAPEX per 1,120 907 834 590 customer ($/connection) Source: (REM simulation) Waya analysis. All $ references are in United States dollars. 27 PAKISTAN LEAST-COST ELECTRIFICATION STUDY As mentioned previously, the grid densification customers are identified as those within 500 meters of the existing grid infrastructure and are not considered directly for REM evaluation. The overnight CAPEX cost estimation for grid densification is based on the average connection cost ($/connection) which is estimated from the REM evaluation of the grid extension projects that are closest to the current 11 kV network. The figures below show the key results of REM. Grid densification numbers are not included, since they are not a direct output of the REM tool. Figure 3.1: Percentage divide between Balochistan REM results for grid extension, mini-grids, and individual systems for customers beyond 500 meters of the grid infrastructure (a) Number of customers (b) Investment cost (CAPEX) (c) Annual energy demand (GWh) (d) Average cost of energy (USD/kWh) Share of customer Share of investment per electrification mode - REM analysis per electrification mode Mini-grids Individual Systems Grid Extensions Mini-grids Individual Systems Grid Extensions (a) (b) (c) (d) 28 PAKISTAN LEAST-COST ELECTRIFICATION STUDY On aggregate, per-customer grid extension cost is US$ 834. However, owing to sparse habitation, mountainous terrain, and the distributed nature of the utility’s grid, there are more than 126 extension projects which REM evaluated to be at least cost but require more than US$ 1,200 per connection. For many distribution utilities, the high investment cost per customer could be unattractive even though this would in principle offer the swiftest least-cost path to universal access. Therefore, an implementation plan might reasonably relegate these expensive extension projects to the later years. Smaller off-grid systems (that is, individual or mini-grids) could be a near-term solution to providing electricity for these communities. As a near-term solution for current electrification, the current demand may warrant a grid-compatible mini-grid solution with smaller generation assets, which can be connected to local grid networks in the future as demand grows in these areas. Similarly, there are nearly 50 mini-grid projects identified with a per-customer CAPEX cost of more than US$ 2,000. Given that off-grid providers may be reluctant to embrace that scenario, smaller solar kits could be an interim solution. Figure 3.2: Histograms of Balochistan grid extension and mini-grid projects (least-cost options) showing CAPEX per customer (USD/connection): The x-axis presents the CAPEX per customer and the y-axis shows the number of electrification projects. All Grid Extension Projects - Balochistan Province (2030) 180 160 140 120 100 80 60 40 20 0 [590, 640] (640, 690] (690, 740] (740, 790] (790, 840] (840, 890] (890, 940] (940, 990] (990, 1,040] (1,040, 1,090] (1,090, 1,140] (1,140, 1,190] (1,190, 1.240] (1,240, 1,290] (1,290, 1,340] (1,340, 1,390] (1,390, 1,440] (1,490, 1,540] (1,440, 1,490] 29 PAKISTAN LEAST-COST ELECTRIFICATION STUDY All Mini-grid Projects - Balochistan Province (2030) 1400 1200 1000 800 600 400 200 0 [500, 600] (600, 700] (700, 800] (800, 900] (900, 1,000] (1,000,1,100] (1,100, 1,200] (1,200, 1,300] (1,300, 1,400] (1,400, 1,500] (1,500, 1,600] (1,600, 1,700] (1,700, 1,800] (1,800, 1,900] (1,900, 2,000] (2,000, 2,100] (2,100, 2,200] The summary distribution for Balochistan of least-cost technologies by district is shown below in Table 3.2 (the fraction is shown immediately below the total for each district; and likewise for the other provinces in Tables 3.5, 3.10, and 3.12). Table 3.2: Summary of least-cost electrification technologies by district for 2030 (REM simulation results) – Balochistan District Mini-Grids Individual New Grid systems Extensions Awaran 25,445 10,287 0 (0.71) (0.29) (0) Barkhan 6,909 3,741 748 (0.61) (0.33) (0.07) Chagai 19,698 10,274 332 (0.65) (0.34) (0.01) Dera Bugti 5,998 8,420 15,896 (0.2) (0.28) (0.52) Gwadar 18,855 7,865 699 (0.69) (0.29) (0.03) Harnai 2,192 2,676 3,258 (0.27) (0.33) (0.4) 30 PAKISTAN LEAST-COST ELECTRIFICATION STUDY District Mini-Grids Individual New Grid systems Extensions Jaffarabad 11,131 1,282 17,450 (0.37) (0.04) (0.58) Jhal Magsi 19,108 4,992 1,126 (0.76) (0.2) (0.04) Kachhi 2,794 1,953 21,113 (0.11) (0.08) (0.82) Kalat 14,331 8,882 1,131 (0.59) (0.36) (0.05) Kech 65,836 16,097 835 (0.8) (0.19) (0.01) Kharan 12,207 8,137 3,859 (0.5) (0.34) (0.16) Khuzdar 56,679 34,415 3,368 (0.6) (0.36) (0.04) Killa Abdullah 10,484 5,302 10,731 (0.4) (0.2) (0.4) Killa Saifullah 16,420 9,398 950 (0.61) (0.35) (0.04) Kohlu 4,481 8,590 9,355 (0.2) (0.38) (0.42) Lehri 1,314 1,624 4,305 (0.18) (0.22) (0.59) Loralai 29,275 8,393 1,727 (0.74) (0.21) (0.04) Mastung 4,117 3,173 648 (0.52) (0.4) (0.08) Musakhel 2,872 5,731 26,024 (0.08) (0.17) (0.75) Nasirabad 36,659 4,963 15,670 (0.64) (0.09) (0.27) Nushki 1,711 2,776 541 (0.34) (0.55) (0.11) 31 PAKISTAN LEAST-COST ELECTRIFICATION STUDY District Mini-Grids Individual New Grid systems Extensions Panjgur 35,279 13,151 439 (0.72) (0.27) (0.01) Pishin 7,675 3,067 25,805 (0.21) (0.08) (0.71) Quetta 3,328 1,730 2,135 (0.46) (0.24) (0.3) Sheerani 1,666 822 33,797 (0.05) (0.02) (0.93) Sibi 2,563 2,417 812 (0.44) (0.42) (0.14) Sohbaptur 1,334 373 5,451 (0.19) (0.05) (0.76) Washuk 15,972 16,302 353 (0.49) (0.5) (0.01) Zhob 23,324 12,004 807 (0.65) (0.33) (0.02) Ziarat 5,675 3,701 414 (0.58) (0.38) (0.04) Source: (REM simulation) Waya analysis. Note: For each district, the fraction of customers is shown below the total number. Looking closely at the district-level results, the study can help in guiding potential emphasis or implementation for one or the other technology based on the least-cost analysis for planning purposes. For example, in districts such as Dera Bugti, Kachhi, Lehri, Musakhel, Sheerani, or Sohbaptur, the least-cost electrification solution through to 2030 for most customers would be grid extension. Conversely, in Awaran, Barkhan, Chagai, Gwadar, Khuzdar, Killa Saifullah, Loralai, Mastung, Panjgur, Zhob, or Ziarat the majority of customers live further than 500 meters from the current network and would therefore find that off-grid systems are the least-cost solution. These districts should ideally be the first districts that the Provincial Energy Department and off-grid developers consider for deployment of their systems. 32 PAKISTAN LEAST-COST ELECTRIFICATION STUDY 3.1.2 Detailed results for key districts Among the districts, the Provincial Energy Department of Balochistan has already started looking at Awaran, Chaghi, Gwadar, Khuzdar, Panjgur, and Washuk for the development of mini-grid projects. Detailed results as shown in the tables below can support the implementation of mini-grid projects in these districts. Table 3.3: Summary of results for key districts identified by the Balochistan Provincial Energy Department for mini-grid sector development (a) Awaran Indicator Mini-Grids Individual New Grid systems Extensions Number of new customers 25,445 10,287 0 Avg. Cost Per kWh of Demand Served ($/kWh) 0.37 0.64 0 Annual Energy Demand (MWh) 10,695 4,245 0 Overnight Capital Cost: Total CAPEX ($) 20.5 million 8.9 million 0 Avg. CAPEX per customer ($/connection) 808 863 0 (b) Chagai Indicator Mini-Grids Individual New Grid systems Extensions Number of new customers 19,698 10,274 332 Avg. Cost Per kWh of Demand Served ($/kWh) 0.38 0.63 0.58 Annual Energy Demand (MWh) 8,152 4,358 146 Overnight Capital Cost: Total CAPEX ($) 16.4 million 9 million 0.33 million Avg. CAPEX per customer ($/connection) 832 876 1,007 33 PAKISTAN LEAST-COST ELECTRIFICATION STUDY (c) Gwadar Indicator Mini-Grids Individual New Grid systems Extensions Number of new customers 18,855 7,865 699 Avg. Cost Per kWh of Demand Served ($/kWh) 0.35 0.63 0.60 Annual Energy Demand (MWh) 7,910 3,333 290 Overnight Capital Cost: Total CAPEX ($) 14.2 million 6.9 million 0.66 million Avg. CAPEX per customer ($/connection) 757 877 945 (d) Khuzdar Indicator Mini-Grids Individual New Grid systems Extensions Number of new customers 56,679 34,415 3,368 Avg. Cost Per kWh of Demand Served ($/kWh) 0.35 0.60 0.49 Annual Energy Demand (MWh) 28,599 16,340 1,818 Overnight Capital Cost: Total CAPEX ($) 50.8 million 32.7 million 3.2 million Avg. CAPEX per customer ($/connection) 895 950 945 (e) Panjgur Indicator Mini-Grids Individual New Grid systems Extensions Number of new customers 35,279 13,151 439 Avg. Cost Per kWh of Demand Served ($/kWh) 0.34 0.62 0.59 Annual Energy Demand (MWh) 15,400 5,753 190 Overnight Capital Cost: Total CAPEX ($) 26 million 11.8 million 0.4 million Avg. CAPEX per customer ($/connection) 736 898 966 34 PAKISTAN LEAST-COST ELECTRIFICATION STUDY (f) Washuk Indicator Mini-Grids Individual New Grid systems Extensions Number of new customers 15,972 16,302 353 Avg. Cost Per kWh of Demand Served ($/ kWh) 0.34 0.57 0.60 Annual Energy Demand (MWh) 8,667 8,535 150 Overnight Capital Cost: Total CAPEX ($) 14.9 million 16.6 million 0.35 million Avg. CAPEX per customer ($/connection) 933 1,019 995 For the selected districts, the majority of the least-cost solutions are off-grid systems, rather than grid extensions, indicating that these areas would be suitable locations for the expansion of businesses that offer mini-grids or individual systems. Indicators and results of this sort from the least-cost analysis can help utilities and potential off-grid system developers alike to focus on areas where their technologies can be the least-cost option for providing electricity. 3.1.3 Summary In the case of Balochistan, for the 2030 time horizon, roughly 2.25 million customers are estimated to be potential new connections required to achieve 100 percent electrification. Among these, 1.3 million customers are estimated through geospatial analysis to be close to the current grid infrastructure and therefore assigned for grid densification. For the remaining nearly 900,000 customers, the least-cost electrification solutions were identified using REM. However, there may be an overestimation of grid densification customers because the electrification rate is solely based on the utilities’ customer count data and does not consider informal connections, potential households that do not have electric meters or those that pay to third-party vendors such as landlords, neighbors, relatives, local stores, banks, and so forth instead of utilities. For the base reference case, for those customers beyond 500 meters from the grid infrastructure, the mix of grid extension and off-grid systems with a 23:77 percent divide is seen as the least-cost option for achieving a 100 percent electrification rate by 2030 for Balochistan. Including the cost of grid densification, an estimated total budget of US$ 1.6 billion will be needed to provide electricity for all by 2030. District-level breakdown of the results shows some districts such as Kachhi, Musakhel, Sohbaptur, and Sheerani, where more than 70  percent of unelectrified customers are to be grid-connected, showing a clear mandate for achieving universal access through grid extension projects. There are also districts like Awaran, Chagai, Gwadar, Khuzdar, Panjgur, and Washuk, where off-grid systems are the least-cost option for more than 330,000 unelectrified customers. Such districts can be ideal places to build off-grid market potential for the country. The LCES can guide the Provincial Energy Department, the Ministry, and the utility to plan and coordinate different stakeholders’ efforts for future grid extension and development of off-grid projects by providing geospatial analysis with location details, and budget estimates. 35 PAKISTAN LEAST-COST ELECTRIFICATION STUDY 3.2 KHYBER PAKHTUNKHWA 3.2.1 Reference base-case scenario All the parameters are assumed for the reference base case as defined in Section 2. Similar to Balochistan province, PESCO’s overall T&D losses and grid reliability matrices are also very poor. The reliability assumed for the base-case scenario is 75 percent. The overall result for the least-cost electrification study is shown in the table below. In the KP region, slightly more than half of the new unelectrified customers should be connected via mini-grids for the least-cost solution. Nearly 100,000 individual systems are the least-cost option for roughly two percent of the unelectrified customers. Grid extension and densification is the least-cost option for more than 1.6 million new customers. It would cost approximately US$ 2.95 billion, in 2020 real terms, to achieve the universal access target by 2030. Of this amount, off-grid systems comprise US$ 1.9 billion, or approximately 65 percent of the total cost. Table 3.4: Summary of results for the least-cost electrification study – KP Indicator Individual Mini-Grids New Grid Grid Total systems Extensions Densification Number of new customers 99,131 2,198,268 708,645 923,105 3.93 million Avg. Cost Per kWh of 0.53 0.34 0.22 0.17 Demand Served ($/kWh) Fraction of new customers 0.02 0.56 0.18 0.24 1 Annual Energy Demand 62.6 1,160 876 (GWh) Overnight Capital Cost: Total CAPEX ($) 0.12 billion 1.8 billion 0.49 billion 0.54 billion 2.95 billion Total annual OPEX 12.85 million 103.6 million 143 million ($/year) Avg. OPEX per customer 130 47 125 ($/year) Avg. CAPEX per customer 1,232 825 696 565 ($/connection) Source: (REM simulation) Waya analysis. All $ references are in United States dollars. For grid extension systems, the CAPEX consists of network costs including wires, transformers, poles, and accessories. The cost of energy is part of the annual operational cost. For off-grid systems, generation component costs are part of CAPEX. For individual systems, solar home systems are considered to meet the demand, whereas, for the mini-grids, the cost of solar-based generation plus distribution network determines the overnight capital cost estimates. The figures below show the key results of REM. Grid densification numbers are not included, since they are not a direct output of the model, and these show results for unelectrified customers who are beyond the 500-meter buffer from the current infrastructure. 36 PAKISTAN LEAST-COST ELECTRIFICATION STUDY Figure 3.3: Percentage divide between KP REM results for grid extension, mini-grids, and individual systems (a) Number of customers (b) Investment cost (CAPEX) (c) Annual energy demand (GWh) (d) Average cost of energy (USD/kWh) Share of customer Share of investment per electrification mode - REM analysis per electrification mode 24% 20% 5% 3% 75% 73% Mini-grids Individual Systems Grid Extensions Mini-grids Individual Systems Grid Extensions (a) (b) Annual energy demand (GWh/year) Average cost of energy (USD/kWh) Grid Extensions Grid Extensions Individuals Systems Individuals Systems Mini-grids Mini-grids 0 200 400 600 800 1000 1200 1400 0 0.1 0.2 0.3 0.4 0.5 0.6 (c) (d) Even though the average per-customer grid extension cost is less than US$ 700, there are more than 140 extension projects which REM evaluated as the least cost but require more than US$ 1,200 per connection. For many distribution utilities, the high investment cost per customer could make grid extension unattractive, even when in the interests of achieving universal access that would be the least-cost option for electrification. Therefore, an implementation plan might reasonably relegate these expensive extension projects to the later years. Smaller off-grid systems (that is, individual or mini-grids) could be a near-term solution to providing electricity for these communities. As a near-term solution for current electrification, the current demand may warrant a grid-compatible mini-grid solution with smaller generation assets, which can be connected to local grid networks in the future as demand grows in these areas. Similarly, there are 46 mini-grid projects identified with a per-customer CAPEX cost of more than US$ 2,000. Given that off-grid providers may be reluctant to embrace that scenario, smaller solar kits could be an interim solution. 37 PAKISTAN LEAST-COST ELECTRIFICATION STUDY Figure 3.4: Histogram of KP grid extension and mini-grid projects (least-cost options) showing CAPEX per customer. The x-axis presents the CAPEX per customer and the y-axis shows the number of electrification projects. All Grid Extension Projects - KP (2030) All Mini-grid Projects - KP (2030) The summary distribution of least-cost technologies by district for the KP region is shown below. 38 PAKISTAN LEAST-COST ELECTRIFICATION STUDY Table 3.5: Summary of least-cost electrification technologies by district for 2030 (REM simulation results) – KP Estimated Individual New Grid District Mini-Grid electrification systems Extensions rate (2020) Abbottabad 144,780 1,222 1,670 62% (0.98) (0.01) (0.01) Bajaur 107,992 2,300 23,727 17% (0.81) (0.02) (0.18) Bannu 73,874 988 3,541 72% (0.94) (0.01) (0.05) Batagram 92,137 2,555 482 28.5% (0.97) (0.03) (0.01) Buner 112,674 2,850 2,079 45% (0.96) (0.02) (0.02) Charsadda 53,767 441 40,277 73% (0.57) (0) (0.43) Dera Ismail Khan 73,152 7,988 19,097 59% (0.73) (0.08) (0.19) FR Bannu 2,919 1,320 110 24% (0.67) (0.3) (0.03) FR Dera Ismail 5,862 1,988 332 33% Khan (0.72) (0.24) (0.04) FR Kohat 0 113 25,089 10% (0) (0) (1) FR Lakki Marwat 347 266 1,476 7% (0.17) (0.13) (0.71) FR Peshawar 0 26 9,957 7% (0) (0) (1) FR Tank 1,814 1,548 0 19% (0.54) (0.46) (0) Hangu 50,743 1,236 2,581 54.5% (0.93) (0.02) (0.05) Haripur 47,122 2,471 3,418 68% (0.89) (0.05) (0.06) Karak 53,239 3,498 2,868 49% (0.89) (0.06) (0.05) 39 PAKISTAN LEAST-COST ELECTRIFICATION STUDY Estimated Individual New Grid District Mini-Grid electrification systems Extensions rate (2020) Khyber 19,993 2,620 62,221 43% (0.24) (0.03) (0.73) Kohat 21,889 1,851 58,002 54% (0.27) (0.02) (0.71) Kohistan 8,968 5,427 110,252 22% (0.07) (0.04) (0.88) Kurram 77,514 4,530 35,044 18% (0.66) 0.04 (0.3) Lakki Marwat 38,627 3,361 10,204 57.5% (0.74) (0.06) (0.2) Lower Dir 155,347 2,271 2,800 41% (0.97) (0.01) (0.02) Malakand PA 48,407 955 9,051 57% (0.83) (0.02) (0.15) Mansehra 90,053 4,735 3,596 57% (0.92) (0.05) (0.04) Mardan 162,903 1,030 7,358 65% (0.95) (0.01) (0.04) Mohmand 55,006 3,033 23,580 25% (0.67) (0.04) (0.29) North Waziristan 41,969 7,244 11,364 36% (0.69) (0.12) (0.19) Nowshera 57,772 1,002 2,091 83% (0.95) (0.02) (0.03) Orakzai 14,454 992 37,590 13.5% (0.27) (0.02) (0.71) Peshawar 20,008 912 3,245 98% (0.83) (0.04) (0.13) Shangla 83,471 2,821 11,446 28% (0.85) (0.03) (0.12) South Waziristan 27,028 4,692 52,254 19% (0.32) (0.06) (0.62) Swabi 135,829 1,251 5,039 61% (0.96) (0.01) (0.04) 40 PAKISTAN LEAST-COST ELECTRIFICATION STUDY Estimated Individual New Grid District Mini-Grid electrification systems Extensions rate (2020) Swat 115,055 5,763 58,036 53% (0.64) (0.03) (0.32) Tank 23,466 1,747 2,569 50.5% (0.84) (0.06) (0.09) Tor Ghar 0 88 41,442 25.5% (0) (0) (1) Upper Dir 109,247 5,850 24,627 28% (0.78) (0.04) (0.18) Source: (REM simulation) Waya analysis. Note: For each district, the fraction of customers is shown below the total number. Looking closely at the district-level results, the study can help in guiding potential emphasis or implementation for one or the other technology based on the least-cost analysis for planning purposes. For example, in districts such as FR Kohat, FR Lakki Marwat, FR Peshawar, Khyber, Kohat, Kohistan, Orakzai, or Tor Ghar, the least- cost electrification solution through to 2030 for most customers would be grid extension. Even though the assumed network lines are available only in some parts of the districts (for example in the Kohistan district, the network line runs along the Indus River, and the N35 road) the population density map shows that most of the communities are near the network line, and extending the grid would be the least-cost option. Conversely, in Abbottabad, Bannu, Batagram, Buner, Hangu, Lower Dir, Mansehra, Mardan, Nowshera, or Swabi the majority of customers would see that off-grid systems offer the least-cost solution. These districts should ideally be the first districts that off-grid developers consider for deploying their systems. Detailed results at the district level from REM analysis can support the implementation of mini-grid projects in these districts. 3.2.2 Sensitivity analysis Grid reliability is one of the key value drivers for the overall result of the least-cost simulation. A sensitivity analysis was carried out for the Abbottabad district to showcase its effect. Given PESCO’s large T&D losses of roughly 38 percent (2020–2021), grid reliability is a huge issue, and depending on the future goals of the utility, off-grid solutions may be more cost-effective in many parts of the province as shown by the base-case scenario. The sensitivity analysis allows the stakeholder and decision-makers to evaluate the robustness of the current reference plan presented in this report, plan for several different scenarios, and provide insight into the diverse possible pathways to achieving 100 percent electrification. Reference base case Abbottabad district’s results are shown in detail in the three tables below from the reference base case (RBC) analysis with input assumptions as discussed in Section 2.2. For this district, the majority of the new unelectrified customers are to be connected via off-grid systems in accordance with the base case assumptions. Of the new customers 98 percent would be electrified with mini-grids as the least-cost option, with one percent of the households being part of new grid extensions and the remaining one percent having individual systems. 41 PAKISTAN LEAST-COST ELECTRIFICATION STUDY Table 3.6: Summary of universal access costs through 2030: Abbottabad district (REM simulation results: Base case) Indicator Individual Mini-Grids New Grid Total systems Extensions Number of new customers 1,222 144,780 1,670 147,672 Fraction of customers 0.01 0.98 0.01 1 Avg. Cost Per kWh of Demand 0.53 0.36 0.57 Served ($/kWh) Annual Energy Demand (GWh) 0.75 61.2 0.7 62.6 Overnight Capital Cost: Total CAPEX ($) 1.4 million 106.4 million 1.56 million 109.4 million Avg. CAPEX per customer 1,148 735 937 ($/customer) Source: (REM simulation) Waya analysis. All $ references are in United States dollars. Scenario 1 (Higher grid reliability) In this sensitivity analysis, the grid reliability is increased from 75 percent to 90 percent. The higher reliability of the grid can make grid extension solutions economically very attractive as the least-cost option. With higher grid reliability, there is a significant increase in the number of grid extension systems compared to the base case. Among unelectrified customers in the district 60 percent will have grid extension as the least-cost solution. Increasing the reliability of the network will require significant investment. The solution and the necessary investment cost for increasing the overall reliability of the grid will vary depending on the source of the reliability issue—technical losses at the transmission or distribution level, generation shortage, commercial losses due to institutional inefficiencies, theft, and so forth. It is beyond the scope of this study to be able to estimate the cost requirements for increasing the reliability of the grid in Pakistan. 42 PAKISTAN LEAST-COST ELECTRIFICATION STUDY Table 3.7: Summary of universal access costs through 2030: Abbottabad district (REM simulation results: High grid reliability case) Indicator Individual Mini-Grids New Grid Total systems Extensions Number of new customers 485 59,285 87,902 147,672 Fraction of customers 0 0.4 0.6 1 Avg. Cost Per kWh of Demand 0.46 0.356 0.347 Served ($/kWh) Annual Energy Demand (GWh) 0.38 25 37.2 62.6 Overnight Capital Cost: Total CAPEX ($) 0.65 million 41.8 million 57.3 million 99.75 million Avg. CAPEX per customer 1,333 706 652 ($/customer) Source: (REM simulation) Waya analysis. All $ references are in United States dollars. Summary of sensitivity results With higher grid reliability, the ratio of off-grid to grid extension solutions for the least-cost option flips, whereby grid extension would be the cheapest solution for the majority of the unelectrified population . There remain 40  percent of the unelectrified population for whom mini-grids would be the least-cost solution even with a highly reliable grid. These clusters could be supported with the long-term establishment of off-grid systems to provide access. Table 3.8: 2030 comparative results for REM’s sensitivity analysis for nearly 150,000 candidate customers for Abbottabad district, KP Indicator Reference Base Case S1: Higher grid reliability Number of new grid-connected customers 1,670 87,902 Number of SHS customers 1,222 485 Number of Mini-Grid customers 144,780 59,285 Total capex ($) 109.4 million 99.75 million Source: Waya analysis 43 PAKISTAN LEAST-COST ELECTRIFICATION STUDY Figure 3.5: Percentage of new customers for different technologies under different scenarios for least-cost electrification Reference Base Case Higher Grid Reliability 1% 1% 40% 98% 60% 0% Grid Extensions Individual Systems Mini-grids Grid Extensions Individual Systems Mini-grids Increasing the reliability of the existing grid requires additional investment and commitment to maintenance from the utility as well as provincial, and federal level support. Such projects can take a long time to complete. One way to future-proof the investment, especially in mini-grid systems today, could be to work on grid- compatible systems such that in the future when the demand grows and the grid is more reliable, the grid could be connected with the mini-grid systems without having to replace the entire infrastructure. 3.2.3 Summary A distribution network layout from GridFinder was used in the analysis. Based on the geospatial analysis utilizing the location of customers and assumed network layout, the province’s electrification rate for 2020 was calculated to be 56 percent. In the case of KP, to achieve 100 percent electrification for the 2030 time horizon, roughly 3.9 million customers are predicted to be potential new connections. Among these, 0.9 million customers are estimated through geospatial analysis to be close to the current grid infrastructure and therefore assigned for grid densification. For the remaining 3 million customers, the least-cost electrification solutions were identified using REM. For the base reference case, the mix of grid extension and off-grid systems with a 24:76 percent divide is seen as the least-cost option for achieving a 100 percent electrification rate by 2030 for KP for the customers analyzed using REM. Including the cost of grid densification, an estimated total budget of US$ 2.95 billion is needed to provide electricity for all by 2030. District-level breakdown of the results shows some districts, such as FR Kohat, FR Peshawar, Kohistan, and Tor Ghar, where more than 90 percent of unelectrified customers are to be grid-connected, showing a clear mandate for achieving universal access through grid extension projects. There are also districts like Abbottabad, Batagram, Buner, Lower Dir, Mardan, Nowshera, and Swabi, where off-grid systems are the least-cost option for more than 90 percent of the unelectrified customers. Even though many of these districts have a high population density, given the input assumptions, especially around the available network layout and grid reliability, off-grid systems are the least-cost option for most of the customers. Sensitivity analysis around key parameters—above all grid reliability—demonstrates the latter factor’s decisive influence on the overall least-cost options. With an increase in reliability from 75 to 90 percent, the majority of unelectrified connections would have grid extension as the least-cost option instead of mini-grid systems. If the grid reliability upgrade costs are to be under US$ 10 million, it would make sense to look closely at grid extension as the least-cost option for the majority of unelectrified customers. The LCES can be utilized as a guide by the Provincial Energy Department, the Ministry, and the utility to plan for future grid extension and development of off-grid projects as it provides geospatial analysis with location details and budget estimates. 44 PAKISTAN LEAST-COST ELECTRIFICATION STUDY 3.3 PUNJAB 3.3.1 Reference base-case scenario All the parameters are assumed for the reference base case as defined in Section 2. One of the key parameter differences for Punjab compared to the rest of the study concerns grid reliability. Except for MEPCO, the reliability for other DISCOs is assumed to be 90 percent. As described in Section 2, the reliability of the rest of the DISCOs in Punjab (excluding MEPCO) is much better than elsewhere in Pakistan. MEPCO’s reliability is the same as those assumed for Balochistan and KP at 75 percent. The overall result for the least-cost electrification study is shown in the table below. The majority of connections to be electrified by 2030 fall under grid densification; in fact, more than 90 percent of the new unelectrified customers should be connected via grid extension and densification for the least-cost solution. Among the total unelectrified customers, five percent of the new customers (387,977 customers) are to be connected via mini-grid solutions as the least-cost option. It would cost approximately US$ 4.57 billion, in 2020 real terms, to achieve the universal access target by 2030. Of this amount, off-grid systems comprise US$ 0.53 billion, or approximately 12 percent of the total capital investment. Utility level breakdown of results is also provided below. Table 3.9: Summary of results for the least-cost electrification study – Punjab Indicator Individual Mini-Grids New Grid Grid Total systems Extensions Densification Number of new customers 108,218 387,977 2,114,787 4,372,161 6.98 million Avg. Cost Per kWh of Demand 0.52 0.33 0.21 0.15 Served ($/kWh) Fraction of new customers 0.015 0.05 0.30 0.63 1 Annual Energy Demand (GWh) 70 242 2,452 Overnight Capital Cost: Total CAPEX ($) 0.13 billion 0.4 billion 1.6 billion 2.44 billion 4.57 billion Total annual OPEX ($/year) 14 million 16.2 million 289 million Avg. CAPEX per customer 1,239 1,043 765 560 ($/connection) Source: (REM simulation) Waya analysis. All $ references are in United States dollars. 45 PAKISTAN LEAST-COST ELECTRIFICATION STUDY FESCO (Bhakkar, Chiniot, Faisalabad, Jhang, Khushab, Mianwali, Sargodha, Toba Tek Singh) Indicator Individual Mini-Grids New Grid Total systems Extensions Number of new customers 20,657 18,446 806,236 845,339 Avg. Cost Per kWh of Demand 0.52 0.35 0.21 Served ($/kWh) Annual Energy Demand (GWh) 13.2 11.3 982 Overnight Capital Cost: Total CAPEX ($) 25 million 21 million 629.6 million 675.6 million Total annual OPEX ($/year) 2.6 million 0.62 million 116.7 million Avg. CAPEX per customer 1,223 1,139 781 ($/connection) Source: (REM simulation) Waya analysis. All $ references are in United States dollars. GEPCO (Gujranwala, Gujrat, Hafizabad, Mandi Bahauddin, Narowal, Sialkot) Indicator Individual Mini-Grids New Grid Total systems Extensions Number of new customers 3,551 2,110 346,357 352,018 Avg. Cost Per kWh of Demand 0.48 0.35 0.19 Served ($/kWh) Annual Energy Demand (GWh) 2.6 1.35 506.5 Overnight Capital Cost: Total CAPEX ($) 4.8 million 2.5 million 249.5 million 256.8 million Total annual OPEX ($/year) 0.46 million 0.07 million 57.5 million Avg. CAPEX per customer 1,349 1,215 720 ($/connection) Source: (REM simulation) Waya analysis. All $ references are in United States dollars. 46 PAKISTAN LEAST-COST ELECTRIFICATION STUDY IESCO (Islamabad, Attock, Chakwal, Jhelum, Rawalpindi) Indicator Individual Mini-Grids New Grid Total systems Extensions Number of new customers 10,174 1,274 172,766 184,214 Avg. Cost Per kWh of Demand 0.48 0.34 0.21 Served ($/kWh) Annual Energy Demand (GWh) 7.5 0.98 223 Overnight Capital Cost: Total CAPEX ($) 13.7 million 1.76 million 139 million 154.5 million Total annual OPEX ($/year) 1.3 million 0.0 4 26.7 million million Avg. CAPEX per customer 1,349 1,349 805 ($/connection) Source: (REM simulation) Waya analysis. All $ references are in United States dollars. LESCO (Kasur, Lahore, Nankana Sahib, Okara, Sheikhupura) Indicator Individual Mini-Grids New Grid Total systems Extensions Number of new customers 2,029 2,246 496,313 500,588 Avg. Cost Per kWh of Demand 0.53 0.32 0.21 Served ($/kWh) Annual Energy Demand (GWh) 1.25 1.5 488 Overnight Capital Cost: Total CAPEX ($) 2.4 million 2.6 million 337 million 342 million Total annual OPEX 0.26 million 0.07 million 58 million ($/year) Avg. CAPEX per customer 1,187 1,187 680 ($/connection) Source: (REM simulation) Waya analysis. All $ references are in United States dollars. 47 PAKISTAN LEAST-COST ELECTRIFICATION STUDY MEPCO (Bahawalnagar, Bahawalpur, Dera Ghazi Khan, Khanewal, Layyah, Lodhran, Multan, Muzaffargarh, Pakpattan, Rahim Yar Khan, Rajanpur, Sahiwal, Vehari) Indicator Individual Mini-Grids New Grid Total systems Extensions Number of new customers 71,807 353,901 293,115 718,823 Avg. Cost Per kWh of Demand 0.53 0.33 0.31 Served ($/kWh) Annual Energy Demand (GWh) 45 227 252 Overnight Capital Cost: Total CAPEX ($) 80 million 376 million 260 million 716 million Total annual OPEX ($/year) 9.3 million 15.4 million 30.1 million Avg. CAPEX per customer ($/ 1,224 1,035 887 connection) Source: (REM simulation) Waya analysis. All $ references are in United States dollars. The figures below show the key results of REM. Grid densification numbers are not included, since they are not a direct output of the model. Figure 3.6: Percentage divide between Punjab REM results for grid extension, mini-grids, and individual systems (a) Number of customers (b) Investment cost (CAPEX) (c) Annual energy demand (GWh) (d) Average cost of energy (USD/kWh) Share of customer Share of investment per electrification mode - REM analysis per electrification mode 15% 19% 4% 6% 75% 81% Mini-grids Individual Systems Grid Extensions Mini-grids Individual Systems Grid Extensions (a) (b) 48 PAKISTAN LEAST-COST ELECTRIFICATION STUDY Annual energy demand (GWh/year) Average cost of energy (USD/kWh) Grid Extensions Grid Extensions Individuals Systems Individuals Systems Mini-grids Mini-grids 0 500 1000 1500 2000 2500 3000 0 0.1 0.2 0.3 0.4 0.5 0.6 (c) (d) The summary distribution of least-cost technologies by district for Punjab is shown below. The Islamabad capital territory is included as it is part of IESCO’s jurisdiction. Table 3.10: Summary of least-cost electrification technologies by district for 2030 (REM simulation results) – Punjab District Mini-Grids Individual New Grid systems Extensions Islamabad 0 50 24,118 (0) (0) (1) Attock 586 3,493 39,119 (0.01) (0.08) (0.91) Bahawalnagar 47,535 12,043 17,094 (0.62) (0.16) (0.22) Bahawalpur 44,767 11,369 19,409 (0.59) (0.15) (0.26) Bhakkar 9,930 8,360 151,192 (0.06) (0.05) (0.89) Chakwal 512 3,952 22,305 (0.02) (0.15) (0.83) Chinot 569 204 80,789 (0.01) (0) (0.99) Dera Ghazi Khan 51,841 12,891 67,696 (0.39) (0.1) (0.51) Faisalabad 0 169 98,573 (0) (0) (1) Gujranwala 314 341 123,783 (0) (0) (0.99) 49 PAKISTAN LEAST-COST ELECTRIFICATION STUDY District Mini-Grids Individual New Grid systems Extensions Gujrat 124 903 68,885 (0) (0.01) (0.99) Hafizabad 487 290 33,419 (0.01) (0.01) (0.98) Jhang 1,815 2,105 109,410 (0.02) (0.02) (0.97) Jhelum 85 1,338 23,835 (0) (0.05) (0.94) Kasur 499 542 128,697 (0) (0) (0.99) Khanewal 6,500 2,127 15,232 (0.27) (0.09) (0.64) Khushab 3,151 6,195 47,785 (0.06) (0.11) (0.84) Lahore 52 22 25,371 (0) (0) (1) Layyah 20,115 6,402 14,920 (0.49) (0.15) (0.36) Lodhran 2,638 1,003 13,624 (0.15) (0.06) (0.79) Mandi Bahauddin 798 996 28,848 (0.03) (0.03) (0.94) Mianwali 1,774 2,305 123,831 (0.01) (0.02) (0.97) Multan 18,739 1,621 8,100 (0.66) (0.06) (0.28) Muzaffargarh 51,111 6,162 32,330 (0.57) (0.07) (0.36) Nankana Sahib 371 383 79,193 (0) (0) (0.99) Narowal 112 460 17,470 (0.01) (0.03) (0.97) Okara 945 491 176,618 (0.01) (0) (0.99) 50 PAKISTAN LEAST-COST ELECTRIFICATION STUDY District Mini-Grids Individual New Grid systems Extensions Pakpattan 21,840 1,488 9,315 (0.67) (0.05) (0.29) Rahim Yar Khan 5,734 4,786 45,141 (0.1) (0.09) (0.81) Rajanpur 32,593 8,317 29,972 (0.46) (0.12) (0.42) Rawalpindi 91 1,341 63,389 (0) (0.02) (0.98) Sahiwal 41,084 1,156 11,596 (0.76) (0.02) (0.22) Sargodha 869 705 133,173 (0.01) (0.01) (0.99) Sheikhupura 379 591 86,434 (0) (0.01) (0.99) Sialkot 275 561 73,952 (0) (0.01) (0.99) Toba Tek Singh 338 614 61,483 (0.01) (0.01) (0.98) Vehari 19,404 2,442 8,686 (0.64) (0.08) (0.28) Source: (REM simulation) Waya analysis. Note: For each district, the fraction of customers is shown below the total number. Looking closely at the district-level results, the study can help in guiding potential emphasis or implementation for one or the other technology based on the least-cost analysis for planning purposes. The analysis for Punjab shows a clear case of grid-extension preference in areas where the existing network is more reliable. Among the utilities, MEPCO has the least reliable grid system. Districts within MEPCO’s jurisdiction have the highest number of off-grid least-cost systems. Among the districts, Vehari, Sahiwal, Pakpattan, Muzaffargarh, Multan, Bahawalnagar, and Bahawalpur have mini-grid systems as the least-cost option for more than 50 percent of their unelectrified customers. For other DISCOs, in populous districts such as Islamabad, Faisalabad, Gujranwala, Lahore, Okara, Mianwali, or Sargodha the least-cost solution through to 2030 for most customers would be grid extension. These types of indicators can help utilities as well as the Provincial Energy Department and potential off-grid system developers to focus on areas where their technologies can be the least-cost option for providing electricity. 51 PAKISTAN LEAST-COST ELECTRIFICATION STUDY 3.3.2 Summary In the case of Punjab, for the 2030-time horizon, roughly 7 million customers are estimated as potential new connections to be electrified to achieve 100 percent electrification. Among these, 4.4 million customers are estimated through geospatial analysis to be close to the current grid infrastructure and therefore assigned for grid densification. For the remaining 2.6 million customers, the least-cost electrification solutions were identified using REM. For the base reference case, the mix of grid extension and off-grid systems with an 81:19 percent divide is seen as the least-cost option for achieving a 100 percent electrification rate by 2030 for Punjab for the customers analyzed using REM. Including the cost of grid densification, an estimated total CAPEX budget of US$ 4.57 billion is needed to provide electricity for all by 2030. District-level breakdown of the results shows populous districts within utilities with high grid reliability such as Islamabad, Faisalabad, Lahore, Mianwali, and Okara. Here, more than 98 percent of unelectrified customers are to be grid-connected, showing a clear mandate for achieving universal access through grid extension projects. There are also districts, especially within MEPCO’s jurisdiction, where off-grid systems are the least-cost option for more than 350,000 unelectrified customers. Such districts can be ideal places to build off-grid market potential for the country. The deployment of mini-grids that could be connected to the grid if and when a reliable grid arrives would be a good strategy for Punjab overall. The LCES can guide the Provincial Energy Department, the Ministry, and the utility to plan for future grid extension and development of off-grid projects by providing geospatial analysis with location details, and budget estimates. 3.4 SINDH 3.4.1 Reference base-case scenario For the base case, all the parameters are assumed as defined in Section 2. This base case sets the bottom line for the sensitivity analysis for the least-cost study to different probable scenarios described in the following sections. The overall result for the least-cost electrification study is shown in the table below. According to the simulation, the majority of the new unelectrified customers should be connected via grid extension and densification for the least-cost solution. Among the total unelectrified customers, 14 percent of the new customers (955,403) are to be connected via off-grid solutions as the least-cost option. It would cost approximately US$ 4.2 billion CAPEX, in 2020 real terms, to achieve the universal access target by 2030. Of this amount, off-grid systems comprise US$ 922 million, or approximately 22 percent of the total capital investment cost. 52 PAKISTAN LEAST-COST ELECTRIFICATION STUDY Table 3.11: Summary of results for the least-cost electrification study – Sindh Indicator Individual Mini-Grids New Grid Grid Total systems Extensions Densification Number of new customers 76,024 879,379 1.86 million 3.7 million 6.5 million Avg. Cost Per kWh of Demand 0.53 0.32 0.20 0.14 Served ($/kWh) Fraction of new customers 0.01 0.135 0.28 0.57 1 Annual Energy Demand (GWh) 48 574.8 2,715 Overnight Capital Cost: Total CAPEX ($) 0.09 billion 0.83 billion 1.24 billion 2 billion 4.2 billion Total annual OPEX ($/year) 9.85 million 49.95 million 261 million Avg. OPEX per customer 130 57 140 ($/year) Avg. CAPEX per customer 1,224 943 666 553 ($/connection) Source: (REM simulation) Waya analysis. All $ references are in United States dollars. The figures below show the key results of REM. Grid densification numbers are not included, since they are not a direct output of the model. Figure 3.7: Percentage divide between Sindh REM results for grid extension, mini-grids, and individual systems (a) Number of customers (b) Investment cost (CAPEX) (c) Annual energy demand (GWh) (d) Average cost of energy (USD/kWh) Share of customer Share of investment per electrification mode - REM analysis per electrification mode 31% 38% 58% 66% 3% 4% Mini-grids Individual Systems Grid Extensions Mini-grids Individual Systems Grid Extensions (a) (b) 53 PAKISTAN LEAST-COST ELECTRIFICATION STUDY Annual energy demand (GWh/year) Average cost of energy (USD/kWh) Grid Extensions Grid Extensions Individuals Systems Individuals Systems Mini-grids Mini-grids 0 500 1000 1500 2000 2500 3000 0 0.1 0.2 0.3 0.4 0.5 0.6 (c) (d) The total CAPEX cost estimation provided by REM considers both network and generation equipment costs (in the case of an off-grid system) and equipment installation costs. REM does not cover soft costs such as project planning & management, engineering & design, and other costs like land acquisition for generation or rights- of-way for distribution network in its financial evaluation. More detailed costs per project should be estimated during feasibility studies. In Sindh, even though the average per-customer grid extension cost is less than US$ 700, there are extension projects which REM evaluated as the least cost but require more than US$ 1,500. For many distribution utilities, the high investment cost per customer could make grid extension unattractive, even when for achieving universal access that would be the least-cost option for electrification. Smaller off-grid systems (that is, individual or mini- grids) could be a near-term solution to providing electricity for these communities. One also has to take into consideration that REM has planned the distribution network expansion and off-grid systems looking at the 2030 horizon. As a near-term transitory solution for current electrification, the current demand may warrant a grid-compatible mini-grid solution with smaller generation assets, and lower connection costs, which can be connected to local grid networks in the future as demand grows in these areas. Similarly, there are about 20 mini-grid projects identified with a per-customer CAPEX cost of more than US$ 2,000. Given that off-grid providers may be reluctant to embrace that scenario, smaller solar kits could be an interim solution. 54 PAKISTAN LEAST-COST ELECTRIFICATION STUDY Figure 3.8: Histogram of Sindh grid extension and mini-grid projects (least-cost options) showing CAPEX per customer (USD/connection). The x-axis presents the CAPEX per customer and the y-axis shows the number of electrification projects. All Grid Extension Projects - Sindh Province (2030) All Mini-grid Projects - Sindh Province (2030) The summary distribution of least-cost technologies by district for Sindh is shown below. 55 PAKISTAN LEAST-COST ELECTRIFICATION STUDY Table 3.12: Summary of least-cost electrification technologies by district for 2030 (REM simulation results) – Sindh District Mini-Grids Individual New Grid systems Extensions Badin 53,450 2,892 74,792 (0.41) (0.02) (0.57) Dadu 13,943 3,160 57,203 (0.19) (0.04) (0.77) Ghotki 34,108 2,239 50,794 (0.39) (0.03) (0.58) Hyderabad 1,506 274 17,907 (0.08) (0.01) (0.91) Jacobabad 15,152 549 186,563 (0.08) (0.01) (0.93) Jamshoro 16,477 9,474 7,675 (0.49) (0.28) (0.23) Kashmore 31,558 1,399 190,438 (0.15) (0.01) (0.86) Khairpur 198,919 9,221 122,974 (0.61) (0.03) (0.38) Larkana 1,095 382 133,878 (0.01) (0) (0.99) Matiari 2,189 858 5,729 (0.25) (0.1) (0.65) Mirpur Khas 32,490 624 44,842 (0.42) (0.01) (0.58) Naushahro Feroze 12,019 1,234 61,407 (0.16) (0.02) (0.82) Qambar Shahdadkot 16,249 1,385 275,914 (0.06) (0.01) (0.94) Sanghar 73,829 4,951 49,556 (0.58) (0.04) (0.39) Shaheed Benazirabad 16,051 1,531 30,876 (0.33) (0.03) (0.64) Shikarpur 11,259 442 212,573 (0.06) (0.01) (0.95) 56 PAKISTAN LEAST-COST ELECTRIFICATION STUDY District Mini-Grids Individual New Grid systems Extensions Sujawal 27,436 3,401 66,037 (0.28) (0.04) (0.68) Sukkur 52,074 3,734 127,830 (0.28) (0.02) (0.70) Tando Allah Yar 3,727 449 18,829 (0.16) (0.02) (0.82) Tando Muhammad Khan 0 78 26,440 (0.0) (0.0) (1) Tharparkar 205,703 18,404 31,784 (0.81) (0.08) (0.13) Thatta 34,855 5,904 53,668 (0.37) (0.06) (0.57) Umerkot 25,290 3,439 19,029 (0.53) (0.07) (0.40) Source: (REM simulation) Waya analysis. Note: For each district, the fraction of customers is shown below the total number. Looking closely at the district-level results, the study can help in guiding potential emphasis or implementation for one or the other technology based on the least-cost analysis for planning purposes. For example, in districts such as Hyderabad, Jacobabad, Larkana, Qambar Shahdadkot, and Shikarpur the least-cost electrification solution through to 2030 for most customers would clearly be grid extension. Similarly, the Tharparkar district has the majority of off-grid systems customers for the least-cost solution with more than 205,000 customers for mini-grids, and should ideally be the first district that off-grid developers consider for deploying mini-grids. In terms of customer fraction, the least-cost electrification scenario for more than 60 percent of the unelectrified population in districts such as Jamshoro, Khairpur, Sanghar, and Umerkot will entail off-grid systems. These types of indicators can help utilities as well as potential off-grid system developers to prioritize and focus on areas where their technologies can be the least-cost option for providing electricity. 3.4.2 Sensitivity analysis To showcase the effect of key-value drivers on the overall result, a sensitivity analysis was carried out for the Tharparkar district. In the base-case scenario, this district has the majority off-grid systems, and any drastic change in the results due to key parameter change could be easily highlighted. The sensitivity analysis shows the robustness of the results, especially if the majority of systems would still be off-grid in the case of different input conditions. The key parameters varied are rural residential demand, off-grid generation component costings, namely for photovoltaic (PV) and battery, the energy cost (wholesale price) for distribution companies at a medium voltage level, and the grid reliability. These sensitivity assessments demonstrate the robustness of the reference plan presented in this report, plan for several different scenarios, and provide insight into the diverse possible pathways to achieving 100 percent electrification. A comparative analysis of all the sensitivity cases against the base case is summarized at the end of this section. 57 PAKISTAN LEAST-COST ELECTRIFICATION STUDY Reference Base Case Tharparkar district’s results are shown in detail below from the reference base case (RBC) analysis with input assumptions as discussed in Section 2.2. For this district, the majority of the new unelectrified customers are to be connected via off-grid systems as shown in the table below for the base-case assumptions. As the least- cost option, more than 80 percent of the new customers would be electrified with mini-grids, with 13 percent of the households being part of new grid extensions and the remaining seven percent having individual systems. Table 3.13: Summary of universal access costs through 2030: Tharparkar district (REM simulation results: Base case) Indicator Individual Mini-Grids New Grid Total systems Extensions Number of new customers 18,404 205,703 31,784 255,891 Fraction of customers 0.07 0.80 0.13 1 Avg. Cost Per kWh of Demand 0.54 0.31 0.283 Served ($/kWh) Annual Energy Demand (GWh) 11.4 139.4 25.4 176.3 Overnight Capital Cost: Total CAPEX ($) 22.4 million 196 million 22.6 million 241 million Avg. CAPEX per customer 1,216 954 710 ($/customer) Source: (REM simulation) Waya analysis. All $ references are in United States dollars. Table 3.14: Scenarios for sensitivity analysis Scenario 1 Cheaper off-grid component cost + RBC (everything else the same) PV cost: 300 USD/kWp, Battery cost: 75 USD/kWh Scenario 2 Higher demand + RBC (everything else the same) Rural household demand: 2 kWh/day (equivalent to 759 kWh annual demand) Scenario 3 Higher wholesale grid energy cost + RBC (everything else the same) Energy cost: 20 US cents/kWh Scenario 4 Higher grid reliability + RBC (everything else the same) Grid reliability: 90% 58 PAKISTAN LEAST-COST ELECTRIFICATION STUDY Scenario 1 (Lower off-grid generation component cost) In the base-case scenario, the cost of off-grid systems (PV and battery) is reflective of the current benchmark cost and market availability for these components in Pakistan. Here, in this scenario, the technology trend is forecast based on the experience curve, and economy of scale utilization. It may not be feasible for near-term implementation—without market maturity—to achieve the low cost of components assumed here. This analysis captures the overall effect of off-grid generation component costs on the least-cost solutions. In this scenario, the PV cost is assumed to be 300 USD/kWp (0.6x base-case scenario), and the storage system is assumed to have a cost of 75 USD/kWh (0.6x base-case scenario). In a direct comparison to the base case, for the assumed low off-grid component costing, the vast majority of the unelectrified customers have mini-grids as the least-cost option for universal access with few individual systems, and even fewer grid extensions. Compared to the base case, the grid extension systems dwindle from 31,784 to a mere 1,414 connections. The low-cost generation components make off-grid solutions attractive for achieving universal access. Table 3.15: Summary of universal access costs through to 2030: Tharparkar district (REM simulation results: Low off-grid component generation cost) Indicator Individual Mini-Grids New Grid Total systems Extensions Number of new customers 18,992 235,485 1,414 255,891 Fraction of customers 0.07 0.92 0.01 1 Avg. Cost Per kWh of Demand 0.40 0.214 0.39 Served ($/kWh) Annual Energy Demand (GWh) 13.3 162 0.85 176.2 Overnight Capital Cost: Total CAPEX ($) 18.9 million 196.2 million 1.18 million 216.3 million Avg. CAPEX per customer 995 833 839 ($/customer) Source: (REM simulation) Waya analysis. All $ references are in United States dollars. Scenario 2 (Higher demand) In this sensitivity analysis, domestic demand—in effect, rural demand—was increased to twice that of the base- case scenario, to consider annual consumption of nearly 800 kWh per year per rural customer. This analysis captures the overall effect of the demand on the least-cost solutions. With higher demand, there is an increase in the number of grid extension systems compared to the base case. In this case 68 percent of unelectrified customers in the district will have grid extension as the least-cost solution. With higher demand, the off-grid system requirement increases, making it more expensive. However, grid extension starts to become cheaper on a per kWh basis with higher demand, owing to more intensive usage of network components such as transformers and wires, increasing economies of scale. High demand also moves customers away from solar home systems into either mini-grids or grid connection due to higher system costs. 59 PAKISTAN LEAST-COST ELECTRIFICATION STUDY Table 3.16: Summary of universal access costs through 2030: Tharparkar district (REM simulation results: High-demand case) Indicator Individual Mini-Grids New Grid Total systems Extensions Number of new customers 10,719 72,046 173,126 255,891 Fraction of customers 0.04 0.28 0.68 1 Avg. Cost Per kWh of Demand 0.44 0.29 0.24 Served ($/kWh) Annual Energy Demand (GWh) 10 71.5 181.45 263 Overnight Capital Cost: Total CAPEX ($) 17.75 million 87.48 million 122.48 million 227.7 million Avg. CAPEX per customer 1,656 1,214 707 ($/customer) Source: (REM simulation) Waya analysis. All $ references are in United States dollars. Scenario 3 (Higher grid energy cost) This scenario envisages a situation where the grid energy cost at the distribution level is higher than in the base-case scenario. With an assumption of 20 US cents/kWh, the wholesale energy cost at medium voltage is more than double the current energy cost (8 US cents/kWh). The increase in energy cost in the future can be attributed to additional generation and network capacity as well as reinforcement of the current network to support an increase in demand and customer count. Even with this higher energy cost assumption, there is no appreciable change in the ratio of grid extension and off-grid solutions among the unelectrified customers. That is because the off-grid energy cost is more than 30 US cents/kWh on average for the mini-grids and individual systems, which is still higher than the average cost due to grid extension—28 US cents/kWh in this case. Scenario 4 (Higher grid reliability) In this sensitivity analysis, the grid reliability is increased from 75 percent to 90 percent. The higher reliability of the grid can make grid extension solutions economically very attractive as the least-cost option. With higher grid reliability, there is a significant increase in the number of grid extension systems compared to the base case; 90 percent of unelectrified customers in the district will have grid extension as the least-cost solution. Increasing the reliability of the network will require significant investment. The solution and the necessary investment cost for increasing the overall reliability of the grid will vary depending on the source of the reliability issue—technical losses at the transmission or distribution level, generation shortage, commercial losses due to institutional inefficiencies, theft, and so forth. It is beyond the scope of this study to be able to estimate the cost requirements for increasing the reliability of the grid in Pakistan. However, this sensitivity analysis does show grid reliability to be a crucial parameter that affects the least-cost technology distribution. 60 PAKISTAN LEAST-COST ELECTRIFICATION STUDY Table 3.17: Summary of universal access costs through 2030: Tharparkar district (REM simulation results: High grid reliability case) Indicator Individual Mini-Grids New Grid Total systems Extensions Number of new customers 6,574 16,764 232,553 255,891 Fraction of customers 0.03 0.07 0.9 1 Avg. Cost Per kWh of Demand 0.55 0.34 0.27 Served ($/kWh) Annual Energy Demand (GWh) 3.8 9.2 163.3 176.3 Overnight Capital Cost: Total CAPEX ($) 7.5 million 16.1 million 165.7 million 189.3 million Avg. CAPEX per customer 1,145 961 712 ($/customer) Source: (REM simulation) Waya analysis. All $ references are in United States dollars. Summary of sensitivity results A comparative analysis is shown in the figure and table below. These provide a summary of the sensitivity results. In the case of lower off-grid component costs for PV and battery, there is an increase in the number of mini-grid systems for the least-cost solutions compared to the base-case scenario. With higher demand and higher grid reliability, the ratio of off-grid to grid extension solutions for the least-cost option flips, such that grid extension would be the cheapest solution for the majority of the unelectrified population. 61 PAKISTAN LEAST-COST ELECTRIFICATION STUDY Figure 3.9: Comparison of the number of connections for different technologies under different scenarios for the LCES for Tharparkar district Table 3.18: 2030 comparative results for REM’s sensitivity analysis for more than 250,000 candidate customers for Tharparkar district, Sindh Indicator Reference S1: Lower S2: Higher S3: Higher S4: Higher Base Case cost demand grid energy grid off-grid cost reliability Number of new grid- 31,784 1,414 173,126 31,784 232,553 connected customers Number of SHS 18,404 18,992 10,719 18,404 6,574 customers Number of Mini-Grid 205,703 235,485 72,046 205,703 16,764 customers Total capex ($) 241.1 million 216.3 million 227.7 million 241.2 million 189.3 million Source: Waya analysis 62 PAKISTAN LEAST-COST ELECTRIFICATION STUDY In the future, lower off-grid component cost, higher demand, and higher grid reliability should all be attainable. As demonstrated by these sensitivity runs, the least-cost solution could switch between off-grid and grid extension depending on the input parameters. The base-case scenario is the most likely outcome given today’s insights and knowledge. One way to future-proof the investment, especially in mini-grid systems could be to work on grid-compatible systems such that in the future when the demand grows and the grid is more reliable, the grid could be connected with the mini-grid systems without having to replace the entire infrastructure. The distributed generation resources of the off-grid system could also provide resiliency to the electricity infrastructure, and improved reliability by providing grid support when the grid is operating. Regulations should be developed to allow future grid connection, especially if different parties are going to be responsible for mini-grid development and grid development. 3.4.3 Summary In the case of Sindh, for the 2030-time horizon, roughly 6.5 million customers are estimated as potential new connections to be electrified to achieve 100 percent electrification. Among these, 3.7 million customers are estimated through geospatial analysis to be close to the current grid infrastructure and therefore assigned for grid densification. For the remaining 2.8 million customers, the least-cost electrification solutions were identified using REM. Note that there may be an overestimation of grid densification customers given that the electrification rate is solely based on the utilities’ customer count data and does not consider informal connections, potential households that do not have electric meters or those that pay to third-party vendors such as landlords, neighbors, relatives, local stores, banks, and so forth instead of utilities.. For the base-case scenario, the mix for Sindh of grid extension and off-grid systems with a 66:34 percent divide is seen as the least-cost option for achieving universal access by 2030 for the customers analyzed using REM. Including the cost of grid densification, an estimated total budget of US$ 4.2 billion is needed to provide electricity for all by 2030. District-level breakdown of the results shows some districts where more than 90 percent of unelectrified customers are to be grid-connected, showing a clear mandate for achieving universal access through grid extension projects. There are also districts like Tharparkar and Jamshoro, where off-grid systems are the least-cost option for more than 222,000 unelectrified customers. Such districts can be ideal places to build off-grid market potential for the country. The LCES can guide the Provincial Energy Department, the Ministry, and the utility to plan for future grid extension and development of off-grid projects by providing geospatial analysis with location details, and budget estimates. The technology trend and experience curve for PV and battery show potential for a great reduction in cost and can benefit the off-grid sector greatly. By contrast, with higher demand and higher grid reliability, more grid extension systems will be the least-cost option for the unelectrified population as shown by the sensitivity analysis. The reference base-case scenario described in the report presents the most likely outcome based on the current information and presents an integrated approach for achieving universal access through a mix of grid and off-grid systems. 63 PAKISTAN LEAST-COST ELECTRIFICATION STUDY 4. MINI-GRID ASSESSMENT 64 PAKISTAN LEAST-COST ELECTRIFICATION STUDY In addition to the least-cost electrification study, the study also included an assessment of mini-grid potential in Pakistan. The objective of this aspect of work was to provide a detailed analysis of potential mini-grid clusters to support the future development of mini-grids in the country. The outputs of this assignment will help inform the World Bank’s engagement with the government of Pakistan and the provincial energy departments on policy development and potential investments, including actions to support greater private-sector participation in the provision of electricity access. This chapter presents the mini-grid portfolio assessment for the four key provinces in Pakistan: Sindh, Balochistan, Khyber Pakhtunkhwa (KP), and Punjab. The assessment was conducted alongside the LCES. Thus, the team has been able to include the grid extension or off-grid potential for the evaluated high-priority sites. The LCES results indicate mini-grids being the least-cost solution for electrifying more than four million customers by 2030, which is equivalent to 21 percent of the total unelectrified population by that time. Low access to electricity, especially in rural areas, provides an excellent opportunity to deploy decentralized mini- grid-based electricity solutions across Pakistan. Grid reliability also affects the least-cost viability of the sites, as discussed in the previous chapters. Thus, mini-grids can be an effective transitory as well as a permanent solution in providing access. This chapter aims at providing a prefeasibility level assessment for these potential mini-grid sites across the country. 4.1 MINI-GRID POTENTIAL IN PAKISTAN Village Data Analytics (VIDA) has been used to identify potential high-priority electrification sites in the country and extract key information from them. This information was then used to predict high-level mini-grid viability for each of the sites. The location, key information, and viability indicators are published in the electrification platform. This information has been compared and calibrated against the national least-cost electrification planning study. The clustering algorithm identified 1,015 potential high-priority electrification sites in the country, distributed across the four provinces as follows: 159 in Sindh, 185 in Balochistan, 331 in Khyber Pakhtunkhwa (KP), and 340 in Punjab. These are the only high-priority electrification sites (with higher density and large size) that are likely to be viable for development in the current scenarios. Several smaller potential mini-grid sites have not been included in this assessment. Figure 4.1: Number of potential mini-grid sites in different regions of the country Sindh Balochistan 15,7% 18,2% Punjab 33,5% Khyber Pakhtunkhwa 32,6% 65 PAKISTAN LEAST-COST ELECTRIFICATION STUDY In Balochistan, Awaran and Pishin have the largest number of mini-grid sites (29 and 23 respectively). In Khyber Pakhtunkhwa (KP) and Bajaur have the most (46 and 37 respectively). In Punjab, Bhakkar has 118 and Mianwali has 44 mini-grid sites. In Sindh, Tharparkar and Khairpur have 59 and 39 respectively.  In Pakistan, a total of 284,190 connections were found within these high-priority sites, of which 33,708 were in Balochistan, 135,465 in KP, 66,065 in Punjab, and 48,952 in Sindh. The average number of connections for the 1,015 high-priority communities was 280.  The average number of connections per mini-grid is higher in Sindh and KP (323 and 445 respectively) while Punjab and Balochistan have fewer users per mini-grid. Mini- Grids in the KP region provide access to the largest share of the population, covering almost 900,000 people. Punjab, Sindh, and Balochistan follow with approximately 430,000 people, 318,000 people, and 219,000 people respectively. These numbers are based on the 2021–2022 population density dataset. Of these sites 48 percent are small (less than 200 buildings) while 37 large sites have more than 750 buildings.  Figure 4.2: Distribution of the average number of connections per province Average number of connections Balochistan Khyber Pakhtunkhwa Punjab Sindh Province As per the data extrapolated from the DISCOs, Punjab is the region with the highest concentration of businesses. This is reflected in the demand distribution between the type of users—residential versus small commercial— each accounting for 47 percent of the total demand. In the other regions, the greatest share of demand is allocated to residential users, while businesses consume between 10 percent and 24 percent. Schools and health centers are more frequently encountered in Balochistan. Overall, health clinics make a small contribution to demand, ranging between 0.1 percent and 1 percent of the total. The demand assumptions for these high- priority sites are consistent with the LCES assumptions. First-order approximation for technical and financial evaluations was done for these high-priority sites based on solar mini-grid generation. 66 PAKISTAN LEAST-COST ELECTRIFICATION STUDY Table 4.1: Summary of main results – breakdown by region Balochistan KP Punjab Sindh Total No. of mini-grids sites 185 331 340 159 1,015 Avg. PV capacity (kWp) 75 214 153 148 158 Avg. Number of connections 191 445 220 323 304 Avg. Cost per connection 1,343 1,165 1,531 1,097 1,310 (USD/connection) Avg. CAPEX Cost per mini-grid 233,213 467,972 317,404 343,365 355,227 (USD/cluster) Total Estimated CAPEX Cost 43.1 154.9 107.9 54.6 360.5 (US$, millions) Total population covered 219,108 880,515 429,470 318,157 1,847,250 Total demand (kWh/day) 45,265 191,187 140,586 70,955 447,993 The total investment cost of implementing these high-priority mini-grids powering all the identified clusters is estimated at US$ 360 million. KP accounts for approximately 43 percent of the total amount, Punjab around 30 percent, and Sindh and Balochistan 15 percent and 12 percent respectively. The average cost per mini-grid differs for each region. The lowest is in Balochistan (US$ 233,213 per mini-grid), while the highest is in KP (US$ 467,972 per mini-grid). This simply reflects the average size of the mini-grids (smallest in Balochistan; largest in KP). Sindh and KP have a lower average cost per connection, Balochistan is in line with the average and Punjab is the most expensive. The average cost of connection is influenced by several factors such as the average number of connections (mini-grids with a larger number of connections will tend to be cheaper), the population density (densely populated villages can be electrified with a more compact power distribution infrastructure, whereas sparsely populated villages require longer distribution lines), or the share of domestic customers (higher shares of residential customers, who have low energy demands, decrease the average cost per connection). Table 4.2: Demand breakdown for each region Demand Residential Commercial Education Health clinics Streetlights Balochistan 74.5% 10.5% 5% 1% 9% K h y b e r 71% 22% 2.5% 0.1% 4.4% Pakhtunkhwa Punjab 47% 47% 1.9% 0.1% 4% Sindh 69% 24% 3.3% 0.2% 3.5% Total 63.4% 29% 2.7% 0.2% 4.6% 67 PAKISTAN LEAST-COST ELECTRIFICATION STUDY Overall, this assessment of potential mini-grid sites in Pakistan shows that there is a good market for mini-grids across the country. To proceed further and deepen the knowledge of the viability of mini-grids in the country the government may use the information available in the VIDA platform to select high-priority electrification sites for implementation. 4.2 CALIBRATION WITH LCES Given that the least-cost electrification study also identifies areas that are best electrified by mini-grids, it is important to understand if there are any differences between the two tasks. The least-cost electrification study is designed to be utilized as a planning tool with a long-term horizon (with 2030 as the target date for universal access), whereas the mini-grid portfolio methodology looks at identifying settlements with high viability for mini- grid electrification today. There are subtle differences between the two methodologies. For example, the LCES considers population growth to estimate the 2030 customer count, whereas the VIDA tool considers today’s population count to build the mini-grid details. Mini-Grid assessment accounts for a more detailed analysis of the sites by including parameters such as streetlighting loads, cost of the powerhouse, and development costs. Since the VIDA tool does not directly calculate the cost of grid extension for comparison, the high- priority electrification sites identified by VIDA are not necessarily the least-cost option for electrification through mini-grids. Because of this, we might see some settlements for which the least-cost electrification solution by 2030 would be grid extension, but nevertheless potentially justify the deployment of mini-grids today, because they are large, dense, and unelectrified. This is an especially feasible option in places where grid extension is slow, and where DISCOs may not be able to extend the grid in the next 1–2 years. In these cases, developing mini-grids that can be interconnected with the local grid if it is extended would provide improved services to local customers. The reliability of the grid also affects the least-cost choice between grid extension and off- grid systems. In areas where the grid is not reliable today, mini-grid systems could be a transitory solution that provides power today and could be connected with the nearby grid in the future when the grid system is improved.   In the case of Pakistan, 652 out of 1,015 mini-grid sites identified by VIDA are also identified to have mini-grid as the least-cost electrification method. This is an overlap of 64.2 percent. Most of the remaining sites (346 sites out of 1,015 or 34.1 percent) would at least cost be electrified by grid extension. These are sites that are relatively close to the grid and are in areas with high grid density. An example is shown below of mini-grid sites that would be electrified at least cost by grid extension.  The majority of these least-cost grid extension sites identified by VIDA as high-potential mini-grids are in Punjab province. Since most of the DISCOs in Punjab have high grid reliability, grid extension does become the least-cost solution. However, sensitivity analyses for the LCES have shown a higher number of mini-grids in the province if the utilities were to have lower reliability. In cases where the difference between grid extension and off-grid systems is small, or external parameters such as reliability affect the least-cost option, the use of mini-grids as a potential transitory solution with the possibility of grid interconnection in the future makes sense. The analysis shows great potential for grid-compatible mini-grids in Pakistan. While these sites are more than 2.5km from the grid and dense enough for VIDA to identify them as potential mini-grid sites for investment today, these are nevertheless sites that the grid is likely to reach before long, making that the least-cost solution soon enough. Hence, these sites can be developed as mini-grids today but can then be connected by a grid when the required infrastructure for connection is built. Only 17 out of the 1,015 sites identified by VIDA have solar home systems (SHS) as the least-cost electrification method. These are relatively smaller sites that are next to the grid infrastructure, where the cost of setting up a transformer and distribution is higher when compared to providing individual connections with SHS. Some of the sites, especially those in Lahore and Peshawar districts are sprinkled closer to larger electrified areas, and may already be grid-connected, but for lack of the full grid infrastructure data, these households were considered as unelectrified for the study. A quick ground survey would verify the electrification status of these smaller sites. To summarize, 64 percent of the mini-grid sites in VIDA and LCES converge to the same electrification supply option. Further, 34 percent of the sites are potential grid-compatible mini-grid sites, where the grid is likely to reach in the future as it is the least-cost technology in the long run. Of the VIDA mini-grid sites, 1.6 percent are relatively small sites where SHS might be the cheapest option for deployment if the demand is primarily residential. 68 PAKISTAN LEAST-COST ELECTRIFICATION STUDY 4.3 CONCLUSIONS This assessment of potential mini-grid sites in Pakistan shows that there is a good market for mini-grids across the country. This study also provides information on high-level demand and viability characteristics.  The government, using the results of this mini-grids assessment, can roll out a larger mini-grids program either at a national or provincial level. Given the sizable number of high-priority sites identified across the country, the government should consider attracting private-sector investment for some or all of these sites. The following next steps are recommended: • Using the information available in the electrification platform, the Ministry and the Provincial Energy Departments can select high-priority electrification sites for implementation.  • This can then be followed up with collection of on-ground data, including verification of buildings, collection of data on local renewable resources, a detailed study of demand, and so forth. • In the future, it would be important to keep the electrification planning and analysis updated with on-ground data and design information and continue to deepen its analytical usefulness for the federal government. Then, to meet the infrastructure needs resulting from this analysis, it is of utmost importance that the government take a series of key decisions that will shape the way the off-grid market will evolve in the country. The critical aspects to address include: • Institutional framework. Define the roles and responsibilities of each stakeholder, and especially of the various government entities in relation to national electrification and its respective policy definition, regulatory drafting, oversight, licensing, tariff definition, project planning, implementation of projects, and so forth. In a highly decentralized country such as Pakistan, it is important to define the roles of each level of administration.  • Piloting of mini-grid models. Currently, mini-grids are still at an early stage of development in Pakistan. It is critical to pilot and test different mini-grid models (such as a community-owned model, fee-for-service model, and so forth) before integrating them in large-scale systems planning. • Approach to off-grid electrification planning. For the market to develop rationally, a clear approach for the mini-grid sector must be defined. Inevitably, there are pros and cons associated with the various approaches (see below), but the status quo—no approach—leaves a leadership vacuum that can only frustrate any attempt to address the universal access challenge. - Top-down approach. Projects are identified and prioritized by the Government following a centralized approach. Projects are put out to tender for implementation according to a Master Plan. This approach implies a significant role for the government in determining the timing and location of mini-grid projects. Therefore, it requires a responsible agency to be adequately staffed, with the capacity and resources to successfully implement the approach. - Bottom-up approach. The government relies on nongovernmental parties to identify and propose potential projects. The government’s role is to develop eligibility requirements and determine, based on these criteria, whether proposed projects move forward. If necessary, the government can arbitrate between parties that end up competing over site development. - Hybrid approach. Combination of the above. The preferred approach can be differentiated on specific criteria, such as project size or viability of the projects.  This decision is also very relevant to how the geospatial analysis is to be used. Under a top-down approach, the geospatial platform would be used by the government to develop its master plan and stipulate which projects to implement at which time. By contrast, under a bottom-up approach, in which the private sector can propose projects, the geospatial platform must be made publicly available so that developers can base their decisions on its data. In Pakistan’s context, the government might well adopt a hybrid approach, whereby large mini-grid sites can be tendered out for development by the private sector, while medium or smaller sites can be developed by the public sector. 69 PAKISTAN LEAST-COST ELECTRIFICATION STUDY • Delivery Model. The delivery model defines the level of involvement of the private sector and includes aspects such as the ownership (state-owned, private, community), and in the case of hybrid approaches (public–private partnerships) defines conditions such as concession terms, duration, and so forth. It should also clarify the government’s stance on licensing preferences, specifically regarding the granting of exclusive power retail rights and the associated conditions. Another aspect to be defined is the scope of services being covered by the delivery model, especially whether indoor wiring and appliances are contemplated as part of the service provision. This is very relevant in rural remote contexts where access to appliances and after-sales services can be challenging. Finally, it is also important to define whether the delivery model covers only mini-grids or if standalone systems can be included as part of an integrated solution. Often the least-cost electrification approach in a community is not based on a single technology but a combination of: - Mini-Grids: for spatially clustered customers with higher demands and greater ability to pay; and - Standalone solar home systems: for more thinly distributed customers with lower demands. • Tariff Determination. Identify potential strategies applicable to the local context regarding tariff setting, tariff structure, and tariff oversight for electrification projects. The chosen strategy must describe the tariff policy (whether tariffs should be uniform across the country), the tariff calculation (how tariffs are calculated and set), and tariff-setting for bundled ‘energy-as-a-service’ supplies. • Subsidies. Off-grid electrification is typically not economically viable, and therefore subsidies are generally required for its implementation. However, this subsidy should be very strategic and market-based. In the past, there have often been excessive subsidy levels in certain government schemes, leading to the unsustainable installation of systems, with inadequate ownership and maintenance services, resulting in their deterioration. Therefore, the right level of subsidy which just bridges the viability gap is a crucial component for long-term sustainability. • Licensing procedures. Define the approval process for rural electrification projects. Typically, a tiered approach is used based on project categorization, offering lighter procedures for smaller projects and more comprehensive procedures for larger ones. Determine requirements for developers, contractors, operators, and so forth, including requirements and timelines for document submission.  • Technical regulation. Define the need for and extent of technical regulation, including but not limited to household wiring and safety standards, power distribution code, standards for components, and so forth. Technical regulation also applies to Service Quality and Power Quality, which typically define the service categorization. • Social and Environmental regulation. Assess existing regulations that could affect electrification projects and identify gaps, opportunities to streamline procedures, and so forth. Assess if the sector might benefit from the preparation of some sector-specific guidelines or capacity building, with special attention to electronic waste management or decommissioning procedures. Electrification of rural areas comes at the expense of generating a new stream of waste (appliances, batteries, and so forth) that must be managed properly. 70 PAKISTAN LEAST-COST ELECTRIFICATION STUDY 5. RECOMMENDATIONS 71 PAKISTAN LEAST-COST ELECTRIFICATION STUDY The output of the least-cost electrification study with grid extensions, mini-grids, and individual off-grid systems can feed into the design of a detailed electrification strategy, providing the basis for rigorous work on financial, business, regulatory, and longer-term policy aspects, as briefly described below. In this chapter, we want to look beyond the geospatial plan, toward the necessary steps and framework to enable the accomplishment of universal electrification. The following outlines the general activities needed to implement an electrification strategy. • Identify the forces which push forward the electrification process in Pakistan, the drivers of the different actors (government, power sector, Development Finance Institutions, beneficiaries, and so forth), especially the technical and business models—including financial innovations—which can accelerate the process. • Analyze the necessary framework to be established at the government, private, and even local levels. • Identify the steps needed to take forward the electrification strategy. The figure below showcases key sectors at a macro and micro level that need to work together for achieving universal access to sustainable electricity. Figure 5.1: Integration of key sectors needed for universal access to sustainable electricity Source: Waya Energy A review of the existing institutional structure, policies, and regulations shows that a regulatory framework for rural electrification projects is already in place. However, there is a need to clearly define and improve upon the implementation framework for these policies and regulations. 72 PAKISTAN LEAST-COST ELECTRIFICATION STUDY Based on the information gathered, it is recommended that the Ministry of Energy (Power Division) at the federal level take responsibility for the LCES results and GIS-based platform created for this project. They should work with utility companies, Provincial Energy Departments, and the private sector to coordinate efforts in both grid expansion and off-grid development. The Ministry of Energy (Power Division) can direct each DISCO to utilize the LCES methodology and results that align with its own techno-economic capabilities, and to evaluate all future grid expansion or off-grid village electrification projects. This GIS analysis should be a crucial component of the design and approval process as stated in the new National Electricity Plan (2023–2027). A training program should be created for the Planning and Engineering (P&E) Departments within each DISCO to utilize the GIS tool in their planning activities. Expansion of the private-sector rural electrification schemes will primarily be demand-driven (based on applications by residents). As already stated above, NEPRA has specified the relevant micro-grid regulations for electrifying currently unserved areas. Provincial Energy Departments can play a key role in private-sector expansion by sharing data on public demand and soliciting proposals from interested parties on setting up electrification projects. 5.1 IMPLEMENTATION PLAN The least-cost electrification results allow the establishment of a project pipeline, detailed by electrification mode (grid extension and mini-grids) per province and district and concession area according to: • National or per-actor priorities; • Financial restrictions; and • Operative implementation restrictions. The key activities for creating an actionable implementation plan would be to: • Identify the implementation monitoring mechanisms and quality control; • Determine project packaging, especially operative lots27 (Figure 52) according to: - Minimum investment size required or other business model requirements, - Optimal logistics for implementation, operation, and management, - Sustainability and financial considerations at the lot level, - Mainstreaming of financial, regulatory, and legal requirements; • Pilot off-grid projects, then if appropriate scale them up; • Assess the assignment of projects to different actors, preparation of auctions or other mechanisms, the definition of sustainable tariffs per project, per lot or per actor, and finally develop a detailed financial plan; • Establish the need for complementary investments, upstream reinforcements in distribution required by the new connections, and investment in additional generation or transmission lines; and • Assess the opportunity for grid-connected mini-grids, as an alternative to centralized generation, and suitable approaches, particularly the opportunity for private actors, or Public–Private partnerships (PPPs) to support utilities. 27  eographic areas with multiple potential mini-grid installations that would allow economics of scale in the development and operation of G multiple mini-grid power systems. 73 PAKISTAN LEAST-COST ELECTRIFICATION STUDY 5.2 BUSINESS MODEL ANALYSIS The project-specific definitions within the geospatial plan enable the assessment of the cost of service and the calculation of the revenue requirements essential for establishing a sustainable business model at both the individual project and investment package levels. Whether guided by government-regulated tariffs (for grid extension, or even for mini-grids or standalone systems) or through privately established service fees or pay- as-you-go models, these cost-of-service calculations allow the determination of the financial gap between the ability to pay, and the affordability of each customer type and the actual cost of service associated with the supply (including system costs, O&M, or overheads). This financial gap might require a direct or indirect subsidy, grants, and investment scheme, whose specifics need to be determined in the financial plan, but the initial step involves establishing the business plan associated with each supply technology. This must include: • A proposal on business models: economic, technical, and environmental sustainability; technological aspects and scalability; affordability, tariffs, and subsidies; and other performance indices; and • A proposal to focus the process on the improvement of living conditions and development as the objective of electrification, and participation of local communities. 5.3 FINANCIAL PLAN To successfully implement an electrification plan, it is important to have a workable financial plan to support it. The LCES provides an estimate of the investment needed for various technologies to achieve universal access, along with the operating cost requirements. To develop a financial plan, the following tasks must be conducted to assess the financial sector: • Identify public, private, and international cooperation, reimbursable and non-reimbursable, and other sources of capital and financing mechanisms currently employed; • Detail the procedures for the acquisition and use of capital from the identified sources and mechanisms; • Identify the pros and cons of the use of current capital sources and financing mechanisms; • Analyze the recovery of investment costs in the current schemes (what portion of the investment is recovered via tariffs and how much ends up being absorbed by the government, or other agents); • Evaluate sources of capital and financing mechanisms used in international markets, especially by nearby and neighboring countries; and • Identify promising sources and mechanisms that could be applied in Pakistan. Once the framework relevant to Pakistan is established, a financial plan can be developed based on the results of the geospatial least-cost plan, the implementation roadmap for 2023–2030 and beyond, along with the revenue requirement determined by the cost of service for the whole plan, the collection of fees from the grid, mini- grids, and standalone system customers (including various categories of customer: domestic, productive and community), the possible subsidy schemes, grants, and other financing mechanisms, conducting a sensitivity analysis to assess the impact of various approaches on the financial plan. This would require developing proposals for financing mechanisms including, at a minimum: • Put forward a proposal to improve the financial mechanism via the public budget to make the current model more efficient; • Put forward one or several proposals involving private investment (PPPs, among others); • Identify the key actors and the activities to be carried out by these actors for the development and execution of the proposed mechanisms; • Detail the sources of capital and the procedures for the execution of the financing mechanisms; • Rank the proposed mechanisms in order of priority for implementation in the short- and medium-term, considering the following characteristics for Pakistan’s context: implementation model and feasibility, sources of capital, sustainability, scalability, and cost of capital, among others to be considered; • Propose a financing model, detailing structure, assumptions, revenue model, cost models, and capital structure; • Assess the financial impact of the electrification plan in the power sector (grid and off-grid); demonstrate any necessity to modify the initial geospatial or implementation plan prior to evaluation of the overall plan’s feasibility; 74 PAKISTAN LEAST-COST ELECTRIFICATION STUDY • Assess the need to reinforce the operative capacities of the electrification actors, including the incumbent government bodies, regulators, private actors, international organization offices, and the financial sector; • Avoid donor fragmentation—that is, support the alignment of donor programs behind one approved plan and one or multiple procurement modalities (see example of Nigeria or Rwanda); this might necessitate some additional prior analysis of overall CAPEX and subsidy requirements; and • Formulate an adequate tariff policy (linked to both the availability of subsidies and the population’s purchasing capacity). 5.4 SUSTAINABILITY PLAN An adequate institutional framework, including articulation of responsibilities between national and provincial levels, would be required to ensure successful implementation of the national electrification strategy to achieve universal access. This complex interface could be managed through memorandums of understanding (MoUs) and the use by provincial governments of nationally standardized templates and processes, coupled with provincial-level capacity building. A sustainability plan from an integrated grid and off-grid perspective will surely involve some reform of the distribution segment of the Pakistan power sector to achieve compatibility with pertinent legislation, economic efficiency (impact on tariffs), good performance in the supply of electricity (reliability, quality, losses), a clean energy transition, and full electrification of Pakistan by 2030. An Integrated Distribution Framework (IDF) emphasizes the use of financially viable business models for the distribution of electricity to end-consumers by all modes of electrification. Its key principles include: • A commitment to universal access that leaves no one behind. This requires permanence of supply and the existence of a utility-like entity with ultimate responsibility for providing access in a defined territory; • Efficient and coordinated integration of on- and off-grid solutions (that is, grid extensions, mini-grids, and standalone systems): this requires integrated planning and appropriate business models for all types of consumers in a defined service territory; • A financially viable business model for distribution: this will typically require some form of distribution concession to provide legal security and ensure the participation of external and mostly private investors, as well as subsidies for viability gap funding; and • A focus on development to ensure that electrification produces broad socio-economic benefits: this principle links expanded access to the delivery of critical public services (for example, health, education) and to multiple economically beneficial end-uses. Specific steps toward a sustainable grid and off-grid distribution design might include: • A strategic review of ongoing political support, policy framework, and current regulations, with a view to their adequacy and integrity for the regulatory oversight required under a program to systematically scale up access to electricity, aiming for universal access in areas within and outside grid service areas and involving utilities, regulators, and service providers; • A proposal for policy planning: inclusiveness, social orientation, public–private participation, and vision for the future; • Specification of any legal, normative, or regulatory modifications needed for the implementation of the proposed financing mechanisms; and • A proposal that includes oversight and certification mechanisms to verify subsidy payments that the government of Pakistan designates and implements for the expansion of access and supply of services by public or private agents, and to ensure the government’s capacity to fulfill its role in regulatory oversight and compliance control over utilities and other entities involved in the delivery of electric services. 75 PAKISTAN LEAST-COST ELECTRIFICATION STUDY 5.5 GOVERNANCE Given that governance entails multifaceted processes, the following should be borne in mind: • Usually, a geospatial plan results in a long negotiation process between incumbent ministries, agencies, development finance institution, and the power sector; • A business plan and a geospatial electrification plan are complementary. Having established in principle what the best techno-economic solution is, the federal government would need to make sure implementation is feasible; and • Transparency is paramount—publication of the plan and ideally the underlying data—not just as an end in itself, but also to stimulate and support private-sector development through confidence in the overall market depth. 76 PAKISTAN LEAST-COST ELECTRIFICATION STUDY APPENDIX A: ELECTRIFICATION MODEL (REM) 77 PAKISTAN LEAST-COST ELECTRIFICATION STUDY The REM 28 developed jointly by the Massachusetts Institute of Technology (MIT; USA) and Universidad Pontifical Comillas Institute for Research in Technology (IIT; Spain) supports the electrification planner when they decide which technology to use to supply electricity in any given territory (cell, village, or group of individual customers). REM makes use of geospatial computer-based techno-economic optimization algorithms to produce plans for distribution network expansion and off-grid systems for universal energy access. It allows decision-makers and planners to apply some policy objectives and to test assumptions easily by using computing power, while leaving the final decision-making in the hands of the human planner. The outputs provided by REM rely on a combination of ground data, a carefully selected set of assumptions, and strategic decision-making. As such, the tool should be used in close interaction with the lead agency and planning department. For our master electrification planning projects, we have developed a pipeline of highly scalable machine learning techniques to comprehensively characterize both demand and supply characteristics over large areas to produce inputs for REM. As shown in Figure A.1, this pipeline includes building extraction from satellite imagery, load localization, demand classification, and electrification status estimation. Figure A.1: Input data and inference pipeline designed for electrification planning in REM Source: Waya 28 P. Ciller, D. Ellman, C. Vergara, A. González-García, S. Lee, C. Drouin, M. Brusnahan, Y. Borofsky, C. Mateo, R. Amatya, R. Palacios,  R.Stoner, F. de Cuadra, and I. Pérez-Arriaga, “Optimal Electrification Planning Incorporating On- and Off-Grid Technologies: The Reference Electrification Model (REM),” Proceedings of the IEEE, Special issue, 2019. 78 PAKISTAN LEAST-COST ELECTRIFICATION STUDY  ECHNO-ECONOMIC PROCEDURE FOR LEAST-COST A.1 T ELECTRIFICATION PLANNING REM is designed as a toolbox with several functions that can be used to support a variety of use cases. The functions can be grouped into three categories: 1) fundamental algorithms, 2) structuring functions, and 3) auxiliary scripts/functions. The first category corresponds to those which solve a particular problem independently from the use case, while the second includes code that performs simple operations such as comparisons or interpolations, which tend to be related to particular use cases. The third category deals with file input/output. To date, the fundamental algorithms that have been developed include the following four key functions: • Clustering: Groups buildings in relevant units of comparison; • Off-grid supply design: Chooses the best off-grid supply components to meet an off-grid load; • Mini-Grid network design: Designs a distribution network for a mini-grid using the Reference Network Model; and • Grid extension design: Designs a distribution network to connect a group of buildings to the existing distribution network. REM performs two major steps: (1) individual demand clustering using a bottom-up approach, and (2) a final decision on the best electrification mode for each cluster. Before step 1, to avoid multiple detailed evaluations of the optimal generation mix for each one of the many cluster combinations, REM calculates optimal generation designs for representative off-grid systems and stores the corresponding data in a look-up table. To find these generation designs, REM minimizes costs (both investment and operation costs plus a penalty for demand that is not met) using an optimization strategy with a master/slave strategy. The master optimization makes decisions about the design variables, using a direct pattern search approach, and the slave optimization performs a simulation with a load-following dispatch strategy for each representative mini-grid. When the clustering or final design algorithm REM requires information about a generation design that is absent from the look-up table, it performs interpolation with the nearest available designs. The generation technologies that REM presently considers for off-grid systems are solar, batteries, and diesel generation. The cost of the charge controllers and inverters is also included (if needed). A.2 STEP 1 – CLUSTERING REM groups many buildings into potential electrical subsystems. This step is very important because it will condition the spatial distribution of off-grid and on-grid systems. Since the actual number of clustering possibilities is unmanageable, REM implements the following strategy: • A systematic bottom-up greedy algorithm, based on local decisions, builds clusters at two hierarchical layers, as shown in Figure A.2. The first layer is built assuming only off-grid solutions, then a second layer adds the possibility of grid extension; • Local clustering decisions depend on the balance between savings (size-related economies of scale) and extra costs (network investments to connect buildings); • Economies of scale result from size-related administrative or maintenance costs, network components/ designs, and generation components/designs. The results depend critically on accurate inputs or estimations of these size-related factors; • Since the number of possible local decisions may be huge in real-life problems, some simplifications have been made: - Connectivity options are limited to the subset of the most promising grouping solutions identified through graph theoretical methods. - Extra network costs are estimated by simplified representations of the networks. - Generation costs are obtained by interpolating in the look-up table calculated in the generation sizing block of REM. 79 PAKISTAN LEAST-COST ELECTRIFICATION STUDY Figure A.2: Example of the structure of clusters (results from the clustering process) Source: Waya STEP 2 –  A.3  FINAL DECISION ON THE BEST ELECTRIFICATION MODE FOR EACH CLUSTER After the clusters have been identified, the costs of the electrification options at different layers are calculated for each cluster as follows: • For the grid extension option: - Find the nearest viable connection points to the existing grid. - Design the lowest-cost distribution network to connect the buildings to the grid. This is done using the Reference Network Model (RNM). - Calculate the final cost, considering: . Cost of energy purchased from the grid. . Network costs (investment, maintenance, and losses). . Cost of non-served energy because of imperfect grid reliability. . Administrative and connection costs. • For the mini-grid option: - Calculate the cost of generation. This cost is either obtained by interpolating in the look-up table calculated in the generation sizing block of REM or performing a full optimization process for this particular cluster, considering the hourly dispatch for the total consumption of the customers within a specific mini-grid. - Calculate the network cost (RNM tool,29 but adapted to a mini-grid case). - Calculate the final cost, considering: . Generation costs (investment, operation, maintenance, and non-served energy). . Network costs (investment, maintenance, and losses). . Administrative and connection costs. 29 RNM (Reference Network Model) is a distribution network optimization model developed by IIT-Comillas. RNM is used as an internal  module within REM to compute the networks for grid extensions and mini-grids. 80 PAKISTAN LEAST-COST ELECTRIFICATION STUDY Figure A.3: Master-slave decomposition for the optimization of mini-grid generation Source: Waya • For the single-building option: The model can be used to design solar kits as the available solution for individual customers. In such cases, if the peak load of the customer is less than a certain limit, then this cost corresponds to that of the solar kit specified by the client for those consumers. If the consumption is higher (isolated productive or community loads), we proceed as in the mini-grid case, except for the network design and cost steps (since no network is needed). • After the cost of the three electrification modes is calculated, the lowest-cost combination of clusters and their electrification modes is selected. • The figure below shows a case in which the optimum solution is made of one grid extension (groups “A” and “B” of customers), one microgrid (group “C”), and a set of individual systems (group “D”). There is also the possibility of biasing solutions to favor some specific delivery technology, by overriding the minimum- cost criteria to some extent. Therefore, it is important to notice that no heuristic rule is applied for the determination of the electrification mode (grid or off-grid), either based on the distance to the grid or density of load. Instead, a thorough calculation of the actual implementation cost is performed for each delivery mode (grid extension, mini-grid or solar kit/ standalone system) before the final decision is made, based on pure techno-economic criteria and subject to policy constraints. Figure A.4: Example of minimum-cost electrification solution Source: Waya 81 PAKISTAN LEAST-COST ELECTRIFICATION STUDY A.4 DIFFERENT TYPES OF INPUT DATA As mentioned above, the outputs provided by REM rely on a combination of ground data, a carefully selected set of assumptions, and strategic decision-making. The classification of inputs into these three groups is not always clear, so the following will have to serve as a reasonable example. Ground data. Inputs that are considered fixed data (although they may represent plans or states of the system): • Location of buildings. It is necessary to know the latitude and longitude of all buildings in the study area, as well as the type of building if different types of demand profiles are used. • Existing distribution feeders, and therefore the distances from buildings to the existing grid. The location of the existing distribution feeders and transformers must be obtained for the study area. In the absence of this data, it is possible to estimate it with the RNM tool (greenfield mode), provided that the already electrified customers are given. • Energy resources (solar power availability, diesel cost if available and allowed, micro-hydro sites). The availability of different energy resources in a given area is necessary to determine the suitability of different types of generation. • Topography data (altitude map and penalized areas). RNM and REM use these data to design networks and incorporate restrictions in the clustering step. Calculated assumptions. Inputs that are adjustable to different scenarios. • Grid energy cost. Cost of energy estimated at distribution level feeders. • Grid reliability. Reliability of the supply of electricity from the existing grid. The reliability metric can be expressed either as a single overall percentage relating the annual non-served energy to the total annual demand or broken up into time-of-day-related percentages (off-peak reliability, peak reliability, and so on). The per-unit cost of non-served energy (CNSE) is closely related to that of reliability. Since REM includes the cost of the non-supplied energy as an operation cost, the higher the CNSE, the more reliable the design provided by REM will be. • Demand profiles: Following the classification of buildings, the demand for each building needs to be characterized. To design electrification solutions for unelectrified buildings it is necessary to estimate how much electricity each building might consume if it had access to electricity. Since the model will try to meet the specified demand at the lowest cost, more detail about demand at each load point is likely to influence the results. • Network components (catalog of lines and transformers). Power capacity and cost characteristics are the most relevant parameters. • Mini-Grid generation components (catalog of PV panels, batteries, diesel generators, and power conversion equipment). Strategic decision-making. Inputs that are related to social aspects and the electrification business model. • Administrative cost. This is the general management cost of the system, and it has different values for the different electrification modes, reflecting economies of scale that depend on the size of the systems. These costs are calculated differently for off-grid and grid extension systems. In REM, the administrative cost of a system only depends on its number of consumers. The input parameters that REM requires to calculate this cost are the administrative cost of small and medium-size mini-grids, as defined by their number of consumers, and at the asymptote the administration cost per customer of the central grid. - In an off-grid system, the administrative cost is approximated with an analytic expression, which is calculated with the input parameters just described. Specifically, the model uses an exponential function to ensure that the per-consumer administrative cost is a decreasing function of the number of consumers. - The administrative cost of a grid extension is calculated with a constant per-consumer cost (it does not depend on the size of the system) equal to the input parameter assigned to a large mini-grid. This ensures that the administrative cost of a grid extension is always lower than the administrative cost of a mini-grid. 82 PAKISTAN LEAST-COST ELECTRIFICATION STUDY • CNSE: REM is a cost-driven tool, so the lack of quality/reliability of power supply must be translated to cost (it may also be imposed as a constraint in the case of mini-grids). CNSE is the cost, to consumers, of the energy that is not served. CNSE represents the cost (that is, the loss of utility) incurred by consumers when there is no electricity at a time when they were planning to use it. The value of the loss of utility depends on the time of the day, the activity being performed at that time, and the economic status of the customer. There could be multiple ways of estimating a value for CNSE. One of them is to adopt—as a proxy—the cost of an alternative energy solution (for example, kerosene) that might be used when electricity is not available for some of the uses that electricity can provide. • Minimum mini-grid quality/reliability of power supply. In addition to representing reliability using CNSE, REM may include a minimum reliability target as a constraint in the case of mini-grids, expressed in terms of the percentage of energy served annually. • The discount rate is needed to translate upfront investments into annuities. It is related to the business/ investment models. Different rate values and different economic lifetimes for components can be set for different electrification modes. • Grid connection criteria. REM can be made to include some social or political strategies that tend to favor some supply modes over others, for instance, to promote grid connection of customers who are not too far from the grid (and not far from other customers). Cost minimization can therefore be subject to constraints derived from energy policies. A.5 SENSITIVITY ANALYSIS (QUALITATIVE) The different inputs described in the previous section are analyzed here with regard to their influence on the expected results. The proposed classification and terms are respected, but two intermediate results are also used in the explanations, namely economies of scale and size. • Economies of scale are the basis for local clustering decisions. They are also critical in the detailed design step. Since they may derive from different sources or inputs, they will be mentioned explicitly in the explanations. • Size of clusters (or subsystems), in terms of the number of buildings connected within the cluster. Economies of scale are closely related to size, but we will try to identify the two effects individually when possible. The expected results of REM are affected by the different input data elements in the following way: A.6 GROUND DATA • Location of buildings (in terms of population density). Higher population density produces larger clusters. They will result in larger mini-grids (if far from the grid) and more grid extensions (if not far from the grid). The influence of size in grid extensions is closely related to the components of the network catalogs. • Existing distribution feeders (in terms of distances to buildings). Part of the connection costs are proportional to distance, so systems of at least a minimum size and proximity to the grid will tend to become grid extensions (here the availability of small network components is critical). • Energy resources (solar power availability; diesel cost if available and allowed). High solar power availability and low diesel prices favor mini-grids. • Topography data (altitude map and penalized areas). Adverse terrain characteristics produce smaller systems and even individual solutions. 83 PAKISTAN LEAST-COST ELECTRIFICATION STUDY A.7 OUR ASSUMPTIONS • Grid energy cost. Higher energy costs result in more mini-grids and fewer grid extensions (with small influence in individual systems, unless the latter have high-demand users). The effect of this parameter on the solution is gradual and modest if buildings-to-grid distances are large. If distances are short, then the impact of this parameter may be dramatic (since the connection cost thresholds depend mainly on transformers, rather than lines). • Grid reliability. This parameter dramatically affects the presence of grid extensions if CNSE is significant, since non-served energy is directly penalized by CNSE. In contrast with mini-grids, grid reliability is a user input that cannot be mitigated by extra investments. • Demand profiles. The profile of critical energy is important since its CNSE value is usually high. More critical energy means a higher cost of grid extensions (if reliability is not 100 percent). In mini-grids, it may just result in more reliable designs, but the cost may not be affected much, since non-served energy costs are replaced by generation costs. The effect of the demand growth rate is similar to the effect of smaller components in the catalogs (network and generation). Higher demands produce bigger sizes and better economies of scale in the system, and therefore they favor bigger mini-grids and more grid extensions. Per unit cost ($/kWh) usually decreases as demand increases. • Network components (catalogs of lines and transformers). Larger and less expensive components tend to produce bigger mini-grids and more grid extensions since they allow more connections with the same savings. • Mini-Grid generation components. Smaller and less expensive components tend to produce bigger mini-grids and more grid extensions, due to the bottom-up clustering strategy (initial decisions are possible). Beyond that, what is relevant is the presence of economies of scale in generation components. Diesel generators have economies of scale, both in investment costs and operational costs (efficiency). In the case of batteries and PV panels, in which big systems are made of many small components, economies of scale should be modeled explicitly in realistic terms. Economies of scale produce bigger mini-grids, and indirectly they may even produce more grid extensions (since large systems are more likely to be connected to the grid). A.8 STRATEGIC DECISION-MAKING • Administrative costs. These may have different fixed values for different electrification modes (grid extension or mini-grids) or also include per-unit costs as a decreasing function of size, to reflect economies of scale in the administration of larger systems, for instance in fee collection tasks. The influence of economies of scale is relevant: as mentioned above, they favor big mini-grids and more grid extensions in the final solution. As these costs are estimated, they are reflected separately in the final solution. • CNSE. CNSE should be set to a value bigger than the typical energy cost in the system. CNSE is closely related to grid reliability in the case of grid extensions since non-served energy is directly penalized by CNSE. In the case of mini-grids, the effect is not so dramatic, since non-served energy costs are replaced by generation costs. Also, in mini-grids, the use of CNSE as a reliability driver may be replaced (or coordinated) with the use of minimum reliability constraints (next input described). • Minimum mini-grid quality/reliability of power supply. This constraint may be imposed in the case of mini-grids, in terms of the percentage of energy served. The effect is to guarantee a minimum reliability level for every off-grid solution, despite the cost. • The discount rate needed to translate upfront investments to annuities: The effect is obviously to change the annual costs, imposing a shorter or longer recovery of the investment. Since different values may be set for different electrification modes, it may bias the final solution one way or another (grid extensions, mini-grids, or individual systems). • Grid connection criteria. They may bias the grid connection of priority customers that are not too far from the grid (and not far from other customers) so that the effect on the final solution is directly predictable in qualitative terms. The quantitative effects are quite interesting, since they may be used to estimate the actual cost of the particular criteria applied. 84 PAKISTAN LEAST-COST ELECTRIFICATION STUDY APPENDIX B: MINI-GRID PORTFOLIO ASSESSMENT METHODOLOGY 85 PAKISTAN LEAST-COST ELECTRIFICATION STUDY The team used the following methodology to identify the high-priority electrification sites across the country, extract key information about the identified sites, and predict high-level mini-grid viability and demand information. All the data about the villages, their characteristics, and their high-level viability are then visualized in VIDA software.  Figure B.1: Mini-Grid Portfolio Assessment Methodology Off-grid villages Demand and viability VIDA uses machine learning A machine learning algorithm algorithms to identify remote predicts the estimated demand villages in a user-selected area and off-grid viability factors using satellite imagery. for every village. IDENTIFY EXTRACT PREDICT PRIORITIZE Villages characteristics Off-grid villages For every identified village, VIDA Based on the extracted automatically extracts village and predicted data VIDA characteristics belonging to categories scores every village. such as demographics, density, grid Villages in the selected access, potential high value customers, geography are ranked water access, agriculture analysis, etc. based on this score. The sections below provide a detailed description of the methodology used and key findings.  B.1 STEP 1 – IDENTIFY OFF-GRID VILLAGES B.1.1 Identification of the electricity grid Often, in the context of developing countries, there is no reliable information about the location of the power grid. Wherever that is the case, VIDA uses its proprietary GridLight algorithm to predict the location of the grid based on nightlight imagery from NASA. The algorithm determines where so-called targets are located. These are locations that emit enough light at night to be considered fully electrified. The algorithm then uses a shortest-path algorithm to predict where the medium-voltage grid will be. To guide the prediction, the algorithm considers high voltage lines, settlement data, road, and topographic data. The GridLight algorithm is built from earlier work by World Bank.30 It has been tested in Nigeria against ground truth. The accuracy level was 82 percent. In the case of Pakistan, we received several data sets from utilities that we considered to be reliable. But the data set did not include details of grid data for the entire country. Owing to several security and accessibility issues, we got access to grid data only from Sindh, Punjab, and Balochistan. Given the lack of data for northern parts of the country, we augmented the grid data with nightlight imagery-based grid prediction algorithms. Hence, the VIDA GridLight algorithm was used to augment the grid data provided by the utility. For this task, VIDA’s GridLight ingested NASA’s VIIRS nightlight imagery from mid-2019 till July 2020. The resulting data, with VIDA’s GridLight and on-ground data, provided a source of information about the existing grid infrastructure in the country that was utilized as the input data for the modeling work. 30 “Gridfinder - Global Energy Infrastructure,” n.d.  86 PAKISTAN LEAST-COST ELECTRIFICATION STUDY B.1.2 Identification of potential mini-grid villages Next, VIDA detects the high-priority electrification sites as “off-grid communities”. A village is labeled “off-grid” if it is at least 2.5 km from the nearest national grid and it does not show light at night. The underlying assumption is that such communities could be more efficiently electrified by distributed (off-grid) energy solutions (such as mini-grids) than by grid extension. Thus, settlements within 2.5 km of the existing grid line are not considered for this analysis. Comparison with the least-cost electrification study results further help to identify those sites for which off-grid systems would be cheaper than grid extension. In addition, sites within 500 meters of a “nightlight target”—an area with regular nightlight detected by VIDA’s GridLight algorithm, and thus likely to have reliable electricity access—were excluded. Village identification was based on high-resolution settlement layer (population) data and a clustering algorithm. The first task was to define parameters for what constitutes a village, in terms of size and density. The VIDA team used a simple machine-learning module to learn the best parameter that constitutes a village. The algorithm was taught by hand-labeled village data from Pakistan. This was done because there are barely any mini-grid sites in the country with which to teach the algorithm. Hence, the VIDA team hand-labeled about 50 sites and then trained the model to predict across the country.   Using VIDA’s clustering algorithm, VIDA identified 1,015 potential mini-grid sites in the country, distributed across the four provinces as follows: 159 in Sindh, 185 in Balochistan, 331 in Khyber Pakhtunkhwa, and 340 in Punjab.  Figure B.2: Number of potential mini-grid sites in different regions of the country Sindh Balochistan 15,7% 18,2% Punjab 33,5% Khyber Pakhtunkhwa 32,6% For all identified settlements, VIDA then drew boundaries based on an alpha-shape generating algorithm to identify the exact contour of the settlement. This helps to accurately estimate the size, area, and density of the settlements (by comparison, a simple convex hull-based village boundary always overestimates the village area and underestimates density). Figure B.7 shows the comparison between the convex hull and the alpha shape. 87 PAKISTAN LEAST-COST ELECTRIFICATION STUDY Figure B.3: Convex hull, alpha shape, and minimum spanning tree of a dense cluster Convex hill Alpha shape Minimal spanning tree In addition, VIDA generated a minimum spanning tree (MST), connecting each building within the cluster/ settlement with its nearest neighbor. This resulted in a single line that is the shortest path to connect every household. The minimum spanning tree provides a measure of the density of the settlement and also provides a rough estimate of the distribution network required to connect every building in the settlement. B.2. STEP 2 – EXTRACTING VILLAGE PROPERTIES In the next step, VIDA’s algorithm extracted key information for each of the identified villages. Details of the extracted information are described below.  B.2.1 Village size and building  VIDA identified information on the village area, the number of households within the village, population, and density. Given that the actual building, and footprints were not available, VIDA calculated the number of households using the population and average size of houses in terms of population (6.5 people per household as provided by the census 2017). Based on the number of total households in the village and the total area, VIDA labeled villages as “large”, “medium” or “small”; and as “high”, “medium” or “low density”. In Pakistan, a total of 284,190 households were found within the mini-grid sites, of which 33,708 were in Balochistan, 135,465 in Khyber Pakhtunkhwa, 66,065 in Punjab, and 48,952 in Sindh. On average, the 1,015 villages have 280 households each.  Also, the analyses reveal that about 48 percent of the sites are small (less than 200 buildings) while 37 large sites have more than 750 buildings.  B.2.2 Village address and names VIDA identifies the address of the sites up to the district level. Also, using a reverse geocoding algorithm and the village location data, VIDA determined the names of the identified villages from OpenStreetMap (OSM) data. For roughly 56 percent of the villages, names could be extracted from the OSM database. For the remaining villages, the names can be added following on-ground visits to the locations. B.2.3 Proximity to grid infrastructure and nightlight VIDA automatically identifies the distance (in km) from the center of the village to the nearest existing grid line. Based on that distance, VIDA automatically labels villages as either “close to grid” (up to 5 km – 51 percent of all sites), “mid-range” (5 km to 10 km – 15 percent of all sites), or “far from the grid” (more than 10 km – 24 percent of all sites). In Pakistan, the sites are on average 8.5 km away from the grid, more specifically around 14.5 km in Balochistan and 7 km in the remaining provinces. 88 PAKISTAN LEAST-COST ELECTRIFICATION STUDY B.2.4 Accessibility VIDA used OSM data to visualize roads within and around the villages. In addition, the distance (in km) was calculated to the closest regional hub, thus a larger settlement of economic importance as categorized by OpenStreetMap.  B.2.5 Health and education facilities  VIDA integrated quality-checked datasets on health care and educational facilities as social load data within and around the villages. They are counted within the identified village hulls, and a filter on the existence of any of these facilities is available. In the select provinces in Pakistan, 81 sites have both unelectrified educational facilities and health care facilities. 775 sites have at least one unelectrified school. 87 sites have at least one unelectrified health care facility. B.2.6 Agricultural data VIDA ingests agricultural data to identify the key agricultural activities around the potential mini-grid sites. For agricultural production data, VIDA uses IFRI’s Spatial Production Allocation Model (SPAM), which identifies the five most widely grown regional products, the regional agricultural area (in hectares), and regional crop yield (in kg/hectare). Because of the resolution of the SPAM data (10x10km), we do not identify village-level agriculture; instead we calculate the productivity of the region (district level). All the villages in the district are then associated with this data.  B.3 STEP 3 – PREDICT MINI-GRID VIABILITY  The output of the VIDA analysis has been complemented with an exercise to estimate the demand of the communities of each cluster. Only clusters with no nightlight visible (from space) were selected. The demand estimation has been followed by a high-level sizing of the main mini-grid technical components to provide communities with access to reliable and clean electricity. The associated cost has also been estimated. This has been done with a geospatial data enrichment process and through a spreadsheet-based model.  The following sections describe the inputs used, the assumptions made, and the methodology used for this purpose.  B.3.1 Demand For the demand estimation, different customer types have been assigned a reference daily demand value (kWh/ day) per connection. 89 PAKISTAN LEAST-COST ELECTRIFICATION STUDY Table B.1: Assumed daily energy demand for different customer types. Customer type Balochistan KP Punjab Sindh Remarks Rural Residential (kWh/connection/ Aligned with LCES 1 1 1 1 day) assumptions Aligned with LCES Commercial (kWh/connection/day) 3.75 5 8 8 assumptions31 Aligned with LCES Schools (kWh/connection/day) 7 7 7 7 assumptions Health facilities (kWh/connection/ Aligned with LCES 10.5 10.5 10.5 10.5 day) assumptions One 50W lamp every 120m of Street lighting (kWh/km/day) 3.75 3.75 3.75 3.75 the distribution network The electricity demand for rural residential, commercial, schools, and health facilities is computed by multiplying the estimated number of each user type by its reference daily energy demand. The demand of each community has been calculated as the sum of the total demand of the customer types assumed: Residential, Commercial, Schools, Health Facilities, and Streetlighting. The rationale behind the selection of the assumed energy demand is based on the following observations: • Typical average residential demand levels seen in rural mini-grids in African countries, South-East Asia, and the Pacific region is in the range 0.25–1.20 kWh/day/household. • The energy needs of households in rural areas are expected to be lower compared to customers currently connected to the national grid. Therefore, it is considered that the actual demand levels coming from DISCO customer databases might not be fully representative of the future energy needs of rural communities. • In general, based on the experience, the rural households’ availability to pay will tend to be lower than that of DISCO customers. Thus, they are assumed to be unable to afford more than lower service levels based on the typical mini-grid tariffs, even assuming that the CAPEX will probably be partially or fully subsidized. • For a prior study,32 the results of surveys carried out in 20 villages in the Sindh region of Pakistan in 2016 show total daily consumption per household ranging from 0.4 to 0.6 kWh/day. This is based on the actual energy needs and the assessed availability to pay, provided that 65–75 percent of the CAPEX is subsidized. (The villages were willing to pay the remainder with a five percent down payment and a loan to be paid off over five years). • References from nearby countries and the our own experience in different regions indicate that it is a common problem to encounter lower than forecasted electricity demands, and this results in mini-grids oversizing. Some regional references show the following: - Myanmar: The technical brief “Consumption Trends” from Smart power Myanmar (2020) revealed that the average household consumption in eight mini-grids in Myanmar “was 0.67 kWh/day as opposed to the 1.62 kWh/day forecasted”. “Average consumption per commercial connection was 0.70 kWh/day as opposed to the 2.66 kWh/day forecasted, with approximately 75 percent of commercial connections consuming less than 2 kWh/day”.  - India: According to the report “Rural Electrification in India – Customer Behavior and Demand”, the average demand for a rural household is 1.3 kWh/day, but in villages with less than 547 HH (page 68) it is 0.79 kWh/day (95 percent of the mini-grid clusters in Sindh and Balochistan identified by VIDA and TTA have less than 547 HH).  31 The commercial demand is estimated based on current and historical electricity consumption data for DISCOs’ small commercial  customers. The demand estimation is the same for the LCES. More explanation on demand estimation is provided in the LCES report. 32 Just Light is Not Enough – Solutions for Improving Energy Access in Rural Sindh, Indus Earth Trust (IET), 2016.  90 PAKISTAN LEAST-COST ELECTRIFICATION STUDY The daily consumption trends (and the corresponding load profiles) are defined for each type of customer. This information is important, not least for battery sizing. For example, residential customers mostly use electricity during the evening, as the main use is for lighting purposes, while schools’ night consumption would be minimal. The table below indicates the percentage of nighttime demand (occurring when the solar PV array is not generating electricity). Table B.2: Daily load profile – percentage of nighttime demand Customer Archetype Night demand Residential 80% Commercial 40% Schools 5% Health facilities 20% Street lighting 100% The number of households for each village as well as the number of health facilities and schools was calculated by the VIDA analysis. As the exact number of commercial customers is not available for every village, a ratio of households to commercial customers has been extrapolated based on DISCO data. The value changes from province to province, ranging from eight to 28 households for each commercial customer. The table below shows the ratios for each province. Table B.3: Number of households per commercial customer for each province Ratio Balochistan KP Punjab Sindh Remarks Households per commercial Based on LCES 28 12 8 23 customer assumptions Finally, the streetlight demand is related to the length of the distribution network: it has been assumed that a 50W lamp is installed every 120m of the distribution network and that it is lit for nine hours per day.  B.3.2 Sizing of the main components The main assumptions taken for the sizing of the components are listed below. The sizing exercise focuses on the key components of the generation assets as well as the distribution network. Mini-Grids have been assumed to be powered by Solar PV (no backup fuel Genset) and equipped with a Lead- Acid Battery Energy System Solution (BESS). Alternative design approaches could have been considered for these key assumptions. For the sake of simplicity, we opted for simulating one of the many different design approaches that are feasible, reasonable, and to some extent representative of the others. LCES results also indicated that, for the majority off-grid systems, full solar-based systems were more cost-effective than having a diesel-based backup system with greater reliability. It should be noted that the preliminary sizing exercise is presented here for high-level energy planning purposes, not as a substitute for feasibility studies and engineering designs required at a later stage. During the implementation phase, developers will have to refine the preliminary system designs offered here. The sizing of components, for instance, will have to be reviewed using more granular and validated site-specific information. 91 PAKISTAN LEAST-COST ELECTRIFICATION STUDY The selection of solar PV as a generation technology for the modeling responds to the universality of the solar resource. As opposed to hydro or wind, the solar resource is not site-specific, it will be similar across the country and can be predicted with acceptable levels of accuracy without the need for intensive field measurements. That is why solar is not only a good reference for simulation purposes but will also be the technology that ends up being implemented in most cases. Should developers seek other renewable energy sources it will mean that the project is located on a site with good resources for these alternatives, and therefore might benefit from achieving lower costs. Nonetheless, it is considered appropriate to take solar PV as the standard option and starting point. In the case of backup generation, the options are whether or not to include fossil-fuel-based gensets and if so, to what extent. There is no such thing as a single hybrid design that can be replicated anywhere, because the optimal version depends on several factors that are project specific. One determining aspect is fuel cost. Cheaper fuel costs will favor hybrid options with higher contributions from the fuel-based gensets, whereas higher fuel costs will favor higher renewable fractions. Other key factors to consider are power service reliability and system resilience. Projects pursuing very high-reliability standards (for example, power supply to a major health center) will necessarily have to include a fuel genset to cover those periods when solar PV generation and stored energy together cannot satisfy the demand. Otherwise, the solar PV and/or the BESS would have to be oversized. It must be noted though that the incorporation of fuel gensets comes at the expense of decreased resilience, because the resulting systems are dependent on the fuel supply, which very often is unreliable in remote areas. Should developers seek designs incorporating fuel gensets, the CAPEX figures provided in this report would increase slightly. In the case of BESS, lead-acid or lithium-ion (Li-ion) are usually the main options. Li-ion batteries require higher upfront costs than lead-acid batteries. On the other hand, in general terms Li-ion batteries offer better performance (higher depth of discharge and increased number of cycles). Should developers opt for Li-ion batteries, the investment costs would be higher than those shown in this analysis for the same sizing of the plants. Over the duration of the project, the increased costs could be offset thanks to extended lifetimes (lower levelized cost of energy). To size the components, the solar resources for each village site were used: • GTI: Global irradiation for optimally tilted surface (annual average) [kWh/m²/day] • GTI: Global irradiation for optimally tilted surface (month with the lowest irradiation) [kWh/m²/day] • OPTA: Optimum tilt to maximize yearly yield [°] The capacity of the Battery Energy Storage System (BESS) is calculated considering the need to store and dispatch all the demand that is not directly supplied by the PV during the day. To do this, the fraction of the demand that has to be supplied through the BESS is calculated by multiplying the total daily demand of each archetype by its respective night demand factor. An ideal Depth of Discharge (DoD) is calculated in such a way that the target lifespan of the battery is at least eight years. Considering the typical number of equivalent cycles guaranteed by reputable manufacturers of good-quality lead-acid batteries, it is possible to estimate an Ideal Night Depth of Discharge (DoD) that guarantees a target lifespan of eight years. Therefore, the battery capacity is sized to meet the night demand without being discharged over the ideal night DoD at night. Table B.4: Lead-acid battery sizing assumptions Lead-acid batteries Unit of measurement Value Max. Depth of Discharge % 70% Cycles to failure # of equivalent cycles guaranteed at the specified DoD 1,500 Target lifespan years 8 92 PAKISTAN LEAST-COST ELECTRIFICATION STUDY The sizing of the solar PV array was done starting from the global irradiation for the optimally tilted surface at each site and assuming a Performance Ratio (PR) indicated in the following table. The PR is the ratio between the Final Yield Yf and the Reference Yield Yr and includes capture losses, array losses, and system losses. Table B.5: Solar PV sizing assumptions Solar PV Value Performance Ratio (yearly average) 50% The battery inverter must be able to supply the peak power demand and therefore be sized accordingly. The Load Factor (LF) specified in the following table has been used to size the inverter. The LF is defined as the average load divided by the peak load over a specified period, and it gives an approximate idea of the shape of the load profile. Table B.6: Inverter sizing assumptions Inverter Value Load factor 28% The estimated length of the mini-grid low-voltage distribution network has been fine-tuned by refining the length of the Minimum Spanning Tree (MST) automatically generated by the VIDA algorithm. This was done by calibrating the VIDA’s outputs with the network length of real mini-grid designs with similar population and population density. B.3.3 Costing The Capital Expenditures (CAPEX) of each mini-grid have been estimated considering the cost of generation and distribution assets, the connection cost, and the development cost. The unitary cost of the main components has been assumed based on experience in similar projects globally and within the region. Development cost has been estimated assuming lots of 10 sites per lot. The cost includes feasibility study, environmental and social impact assessment (ESIA), generation and distribution license acquisition, company foundation and establishment, acquisition of capital, including due diligence, set up of village and customer relationships, land, land acquisition, and other costs. 93 PAKISTAN LEAST-COST ELECTRIFICATION STUDY Table B.7: Costing assumptions33 Generation assets unit Cost Solar PV USD/kWp 550 Solar Charge controller (if DC-coupling) / PV inverter (if AC-coupling) USD/kWp 150 Battery storage (lead-acid) USD/kWh 150 Battery inverter USD/kVA 350 Powerhouse - Unit cost USD/m2 300 Powerhouse - Fixed cost (fixed space required) USD 7,500 Powerhouse - Variable cost (space for battery storage) USD/kWh of battery storage 26 EMS/remote monitoring system USD 1,500 Balance of system % of CAPEX w/o BOS 5% Distribution assets unit Cost LV lines USD/km 8,000 Connection unit Cost Connection costs USD/connection 100 Development unit Cost Development costs USD/site 25,000 B.4 STEP 4 – PRIORITIZATION OF MINI-GRID SITES Following the grid and village identification, the extraction of village characteristics, and the prediction of mini- grid viability, all identified sites are prioritized along with a user-customized score.  In the case of Pakistan, we used the total number of households per village to prioritize the sites. The biggest sites are at the top of the list. This parameter can be changed at any time to reflect different user groups, strategies, and decision-making processes. 33 Cost assumptions are based on input from local stakeholders, DISCOs, and based on current market availability in Pakistan.  94 PAKISTAN SUSTAINABLE ENERGY SERIES PAKISTAN LEAST-COST ELECTRIFICATION STUDY June 2024