MINI GRIDS FOR HALF A BILLION PEOPLE Market Outlook and Handbook for Decision Makers MINI GRIDS FOR HALF A BILLION PEOPLE Market Outlook and Handbook for Decision Makers © 2022 | International Bank for Reconstruction and Development / The World Bank 1818 H Street NW, Washington, DC 20433 Telephone: 202-473-1000; Internet: www.worldbank.org Some rights reserved. 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 if full attribution to this work is given. Any queries on rights and licenses, including subsidiary rights, should be addressed to World Bank Publications, World Bank Group, 1818 H Street NW, Washington, DC 20433, USA; fax: +1-202-522-2625; e-mail: pubrights@worldbank.org. Furthermore, the ESMAP Program Manager would appreciate receiving a copy of the publication that uses this publication for its source sent in care of the address above, or to esmap@worldbank.org. This work is available under the Creative Commons Attribution 3.0 IGO license (CC BY 3.0 IGO) http:/ /creativecommons.org/licenses/by/3.0/igo. Under the Creative Commons Attribution license, you are free to copy, distribute, transmit, and adapt this work, including for commercial purposes, under the following conditions: Attribution—Energy Sector Management Assistance Program. 2022. Mini Grids for Half a Billion People: Market Outlook and Handbook for Decision Makers. Washington, DC: World Bank. www.esmap.org/ mini_grids_for_half_a_billion_people. License: Creative Commons Attribution CC BY 3.0 IGO Translations—Add the following disclaimer along with the attribution: This translation was not created by The World Bank and should not be considered an official World Bank translation. The World Bank shall not be liable for any content or error in this translation. Adaptations—Add the following disclaimer along with the attribution: This is an adaptation of an original work by The World Bank. Views and opinions expressed in the adaptation are the sole responsibility of the author(s) of the adaptation and are not endorsed by The World Bank. Third-Party Content—The World Bank does not necessarily own each component of the content contained within the work and does not warrant that the use of any third-party owned individual component or part contained in the work will not infringe on the rights of those third parties. If you wish to reuse a component of the work, it is your responsibility to determine whether permission is needed for that reuse and to obtain permission from the copyright owner. Examples of components can include, but are not limited to, tables, figures, or images. Production Credits Designer | Naylor Design, Inc. Cover Image | Image by vitranc, GettyImages. All images remain the sole property of their source and may not be used for any purpose without written permission from the source. ESMAP’s MISSION The Energy Sector Management Assistance Program reflect the views of the World Bank, its affiliated organiza- (ESMAP) is a partnership between the World Bank and tions, members of its board of executive directors for the 24 partners (https://esmap.org/donors) to help low- countries they represent, or ESMAP. The World Bank and and middle-income countries reduce poverty and boost ESMAP do not guarantee the accuracy of the data included growth through sustainable energy solutions. ESMAP’s in this publication and accept no responsibility whatsoever analytical and advisory services are fully integrated within for any consequence of their use. The boundaries, colors, the World Bank’s country financing and policy dialogue in denominations, and other information shown on any map the energy sector. in this report do not imply on the part of the World Bank Group any judgment on the legal status of any territory or Through the World Bank Group, ESMAP works to acceler- the endorsement of acceptance of such boundaries. ate the energy transition required to achieve Sustainable Development Goal 7 (SDG 7) (https://sdgs.un.org/goals/ The text of this publication may be reproduced in whole goal7) to ensure access to affordable, reliable, sustainable, or in part and in any form for educational or nonprofit and modern energy for all. It helps to shape WBG strate- uses, without special permission, provided the source is gies and programs to achieve the World Bank Group’s Cli- acknowledged. Requests for permission to reproduce por- mate Change Action Plan targets. Learn more at: https:// tions for resale or commercial purposes should be sent to esmap.org. the ESMAP Manager at esmap@worldbank.org. ESMAP encourages dissemination of its work and normally gives ESMAP reports are published to communicate the results permission promptly. The ESMAP Manager would appre- of ESMAP’s work to the development community. Some ciate receiving a copy of the publication that uses this sources cited in this report may be informal documents report for its source, sent to esmap@worldbank.org. not readily available. All images remain the sole property of their source and This work is a product of the staff of the World Bank with may not be used for any purpose without written permis- external contributions. The findings, interpretations, and sion from the source. conclusions expressed in this work do not necessarily MINI GRIDS FOR HALF A BILLION PEOPLE    iii A NOTE TO THE READER This is a big book. It is packed with actionable information • A volume of case studies on the history of mini grids in for decision-makers, and it is the World Bank’s most com- electric power systems, as well as mini grid regulations prehensive and authoritative publication on mini grids to and subsidies in Bangladesh, Cambodia, India (Uttar date. Pradesh), Kenya, Nigeria, and Tanzania. We intend this book as a reference guide to be consulted • Animations, infographics, and videos to present high- when important decisions about mini grids need to be level findings to a wide audience. made at the project, portfolio, or national program level. • Briefs in the Live Wire series that can serve as quick To that end, we have balanced cohesiveness among the reference guides for World Bank operations teams and chapters with each chapter’s ability to stand on its own as other project implementation partners. a resource. Ensuring That Regulations Evolve as Mini Grids – “ The book is structured as follows. The overview presents a Mature” (https://openknowledge.worldbank.org/ global market outlook for mini grids and introduces the 10 handle/10986/31773). building blocks that need to be in place if mini grids are to – “Investing in Mini Grids Now, Integrating with the Main be scaled up in any country. These building blocks also rep- Grid Later: A Menu of Good Policy and Regulatory resent the 10 frontiers for innovation for the sector, where, Options” (https:/ /openknowledge.worldbank.org/ with disruptive digital solutions across all 10 frontiers, the handle/10986/31772). services offered to end users can be raised to a level sub- stantially better than what would be possible with alter- • A roster of experts to provide rapid-response support natives. In the Handbook, the terms “building blocks” and for project implementation. “frontiers” are used interchangeably. Chapters 1–10 pres- The objective of this comprehensive knowledge package ent the 10 building blocks in detail and answer the question is to present road-tested options and examples from the how do we scale up mini grid deployment to connect half a leading edge of mini grid development. Decision-makers billion people by 2030? Chapter 11 is our call to action. can draw on these options and examples to scale up mini This book is part of a comprehensive knowledge package grid deployment in their own contexts. By acknowledging that the World Bank has prepared on mini grids, which con- different national approaches to mini grids and providing sists of the following elements, available online at www. context-specific considerations for implementation, this esmap.org/mini_grids_for_half_a_billion_people: suite of knowledge products offers an adaptive approach to helping countries achieve their electrification targets. • An executive summary, published separately in June 2019 (https://openknowledge.worldbank.org/handle/ 10986/31926). iv   MINI GRIDS FOR HALF A BILLION PEOPLE TABLE OF CONTENTS Acknowledgments   xvi Abbreviations   xix MINI GRIDS BY THE NUMBERS   xxiv MAIN FINDINGS 1 The New Electricity Access Landscape 1 Building Blocks 1 through 10 and Frontiers for Sector Growth: Creating the Environment for Takeoff of Mini Grid Portfolios 4 1. Solar Mini Grid Costs, Design, and Innovation 4 2. Planning national strategies and developer portfolios with geospatial analysis and digital platforms 7 3. Transforming productive livelihoods and improving business viability 7 4. Engaging communities as valued customers 8 5. Delivering services through local and international companies and utilities 9 6. Financing solar mini grid portfolios and end user appliances 9 7. Attracting exceptional talent and scaling skills development 10 8. Supporting institutions, delivery models, and champions to create opportunities 11 9 and 10. Regulating the sector and making it easier to do business 11 A Call to Action 12 References 12 Notes 13 OVERVIEW THE NEW ELECTRICITY ACCESS LANDSCAPE AND THE GROWING SPACE FOR SOLAR MINI GRIDS 14 SDG 7: A Global Agenda Running Behind 14 Access to electricity has increased . . . 14 . . . But progress has been insufficient to meet the goal of universal access 14 More financing is needed, and it must be better targeted 15 Double down on solutions that have the potential for exponential growth curves 16 The Place for Solar Mini Grids 16 What are solar mini grids? 16 The historic role of mini grids in national electrification efforts 18 The role of solar mini grids in universal electrification 22 Where do solar mini grids fit in? 22 How to Scale Solar Mini Grid Deployment to serve Half a Billion People 25 Five drivers: Cost, quality, pace, finance, and enabling environment 25 Ten building blocks 27 Global Market Snapshot, Outlook 2030, and Call to Action 29 Projections 34 Status of the five drivers and ten building blocks 36 Call to action 42 References 43 Notes 43 MINI GRIDS FOR HALF A BILLION PEOPLE    v CHAPTER 1 REDUCING COSTS AND OPTIMIZING DESIGN AND INNOVATION FOR SOLAR MINI GRIDS 45 The Levelized Cost of Mini Grid Electricity 45 The levelized cost of energy from mini grids: Seven analytical cases 46 Modeling assumptions and scenarios 47 Modeling results 48 Effect of expected declines in capital and operating costs by 2030 51 The share of renewable energy 51 Modeling results: LCOE of optimum hybrid vs. 0 percent and 100 percent renewable energy 52 Implications for national utilities of lower mini grid LCOE 52 Implications for national utilities of improving the quality of mini grid services 52 Win-win for mini grids and national utilities 54 Current Status of Selected Mini Grid Capital and Operating Costs 54 Cost per unit of firm power output 54 Investment costs per customer 56 Cost of individual components 57 Replacement costs 66 Operating costs 67 The Outlook for Mini Grid Capital and Operating Costs 67 PV module trends 67 PV inverter trends 68 Battery trends 69 Battery inverter trends 69 Smart meter trends 69 Trends in other capital costs 69 Trends in operating costs 70 The Impact of Economies of Scale 71 Reasoning from First Principles 72 Conclusion 73 Summary of potential cost reductions 73 Government’s role in reducing mini grid costs and catalyzing innovation 74 The importance of coordinated collection of data on mini grid costs 74 References 74 Notes 75 CHAPTER 2 NATIONAL STRATEGIES AND DEVELOPER PORTFOLIOS: THE ROLES OF GEOSPATIAL ANALYSIS AND DIGITAL PLATFORMS 78 Assessing the Market Potential for Mini Grids 79 Simple exploratory spatial data analysis for Sub-Saharan Africa using GRID3 79 The Global Electrification Platform and least-cost electrification analysis for Sub-Saharan Africa 80 Maximal mini grid deployment modeled in the Global Electrification Platform 82 National Electrification Planning 85 Operational experience and the Global Electrification Platform 87 Indicative workflow for the development of a GIS-based national electrification plan 89 Analytical insights and generic observations 91 Lessons learned and challenges ahead 93 Mini Grid Portfolio Planning 94 Overview 94 The workflow phases for mini grid portfolio planning: Spatial data and analytics 95 Lessons learned and next steps 112 References 112 Notes 113 vi   MINI GRIDS FOR HALF A BILLION PEOPLE CHAPTER 3 PRODUCTIVE LIVELIHOODS AND BUSINESS VIABILITY 114 The Multiple Benefits of Connecting Income-Generating Machines and Appliances to Mini Grids 114 Rolling Out Programs to Promote Productive Uses and Stimulate Demand 117 Step 1. Assessing markets and demand: Geospatial analysis superimposed over mini grids, appliances, and finance for end users 118 Step 2. Surveys and workshops build on Step 1 data through community engagement 120 Step 3. Demand analysis for mini grid design and market potential for appliances and associated end-user finance 121 Step 4. Preparation of road shows involving local government, leadership communities, communities, interested appliance providers and end-user financiers, mini grid companies 124 Step 5. Road shows to load centers where mini grid developers, appliance suppliers, and end user financiers explain the value propositions to potential end users based on their current and aspirational living standards 125 Step 6. Rollout of mini grid connections, sales of appliances, and end-user finance 126 Who Organizes PUE Programs? 129 Mini grid developers 129 Government agencies and policy makers 129 Local change agents 130 Timing Productive Use Programs for Maximum Effect 130 What’s Next? 130 References 135 Notes 135 CHAPTER 4 ENGAGING COMMUNITIES AS VALUED CUSTOMERS 136 Why Is Community Engagement Important? 136 Community Engagement throughout the Mini Grid Project Cycle 139 Design and planning phase 139 Promotion and information-sharing phase 141 Financing and procurement phase 142 Implementation and construction phase 143 Operations and maintenance phase 143 Summary of community engagement over the project cycle 144 Important Gender Aspects of the Community Engagement Process 144 Removing Barriers to Scale through Innovations in Community Engagement 146 Country-level program: Smart Power India 146 ICT for community engagement at the developer level: A pilot video hub delivering mini grid stories 147 Conclusion 148 References 148 Notes 149 CHAPTER 5 DELIVERING SERVICES THROUGH LOCAL AND INTERNATIONAL ENTERPRISES AND UTILITIES 150 An Evolving Technology from First-Generation to Third-Generation Mini Grids 151 First- and second-generation mini grids 151 Second-generation mini grids 151 Third-generation mini grids 151 The Business Case for Private-Sector Participation in Third-Generation Mini Grids 152 Private-sector segmentation 155 Mini grid facilitators 165 Facilitating Collaboration Between and Among Local and International Private-Sector Entities 169 Local and International Industry Players Across the Mini Grid Industry Value Chain 170 Component manufacturing 170 Market assessment 170 MINI GRIDS FOR HALF A BILLION PEOPLE    vii Permitting and financing 172 Grid design and procurement 173 Integration and installation 174 Operations and maintenance 175 After sale 177 Profit potential of the solar mini grid value chain 177 Summary 179 References 180 Notes 181 CHAPTER 6 FINANCING SOLAR MINI GRID PORTFOLIOS AND END-USE APPLIANCES 182 What Are the Financial Needs of Mini Grid Developers? 182 What Types of Finance Are Available? 183 Mini Grid Debt and Equity Investors 183 Development Partners’ Investment in Mini Grids 187 Barriers Faced by Private Developers in Accessing Commercial Finance 190 Mini grids are perceived as risky 190 Demand uncertainty remains high 190 Consumers are unable to pay the full costs of supply and lack credit history 190 Equity is inadequate 191 The tenors of assets and liabilities are mismatched 191 Political and regulatory risks create uncertainty 191 Future macroeconomic conditions are uncertain 191 Overcoming These Barriers 192 Developer actions 192 Government and development partner actions 192 Subsidies 197 Facilitating equity 200 Mitigating risk 202 Conclusions, and General Recommendations for Governments 204 References 205 Notes 205 CHAPTER 7 ATTRACTING EXCEPTIONAL TALENT AND SCALING UP SKILLS DEVELOPMENT 206 Mini Grid Development Requires Distinctive Skills and Capacity 207 Preinvestment phase 207 Project construction 207 Project operations and maintenance 207 Identifying Skills Gaps 208 Project developers 209 Utilities 209 Banks and financial institutions 209 Engineers 209 Suppliers 209 Policy makers and regulators 210 Local communities and customers 210 Training and Skill-Building Interventions to Address Skills Gaps 210 National-level training and skill building 210 Project-level training and skills building 213 Community involvement and awareness 215 viii   MINI GRIDS FOR HALF A BILLION PEOPLE Lessons Learned from Effective Training and Skills-Building Programs 218 Database of Training Programs 219 References 220 Notes 222 CHAPTER 8 DELIVERY MODELS AND SUPPORTING INSTITUTIONS 223 Mini Grid Delivery Models 223 The build-own-operate model 223 Public-private models 223 Concession model 224 Utility models with or without private-sector involvement 224 Cooperative model 225 Comparison of various models 225 Institutional Framework 228 The desirable features of a framework to support mini grids 228 Institutions inside the energy sector 231 Institutions outside the energy sector 232 Investors’ Perspective on Institutional Frameworks 233 Role for Development Partners 233 References 234 Notes 234 CHAPTER 9 ENACTING REGULATIONS AND POLICIES THAT EMPOWER MINI GRID COMPANIES AND CUSTOMERS 235 The Importance of Workable Regulations for Scaling Up Mini Grid Development 236 Five Key Regulatory Decisions 238 Regulating entry 239 Regulating retail tariffs 242 Regulating service standards 249 Regulating technical standards 252 Regulating relationships between the main grid and mini grids 255 Combining Regulatory Elements 259 Combining regulatory elements into effective packages 259 Combining regulatory elements in a phased approach 261 Regulatory Innovations to Further Incentivize Private-Sector Investment in Mini Grids 262 Regulation by contract 263 Arbitration-style appeal mechanism to complement a light-handed regulatory approach 264 Investors’ Perspective on Mini Grid Regulations 265 Conclusion and Resources 266 Acknowledgments for the Six-Country Case Studies That Informed This Chapter 267 Bangladesh 267 Cambodia 267 Uttar Pradesh (India) 268 Kenya 268 Nigeria 268 Tanzania 268 References 268 Notes 269 MINI GRIDS FOR HALF A BILLION PEOPLE    ix CHAPTER 10 CUTTING RED TAPE FOR A DYNAMIC BUSINESS ENVIRONMENT 271 Why an Enabling Environment Matters 271 Characteristics of Doing Business as a Mini Grid 272 Long-lived sunk assets 272 Unpredictability and the need for flexibility 272 The political nature of electricity 272 Four Complementary Options to Make It Easier for Mini Grid Developers to Do Business 273 Reducing red tape through standardized, preapproved templates 275 Using technology platforms to connect developers with investors and suppliers and to conduct large-scale mini grid tenders 283 Eliminating duplication of government oversight by delegating authority to a single entity 284 Setting up e-government services to reduce overhead 286 Investors’ Perspectives on the Four Options Presented Above 287 Conclusion 288 References 288 Notes 288 CHAPTER 11 CALL TO ACTION 289 Policy Makers 289 Regulators 289 Development Partners 289 Industry Associations 289 Mini Grid Developers 290 Investors 290 Suppliers 290 Researchers 290 Topics for Future Research 290 Collecting data on installed and planned mini grids 290 Combining mini grids, solar home systems, and main grid extensions into an integrated electrification strategy at the local level 290 Business tactics and strategies for mini grid developers 291 Policies and business environment factors that affect mini grids 291 References 292 x   MINI GRIDS FOR HALF A BILLION PEOPLE BOXES Box 1.1 • The levelized cost of energy for best-in-class mini grids dropped nearly 31 percent from 2018 49 Box 1.2 • Direct current mesh grids 65 Box 2.1 • Data sources for cluster definition in the Nigeria Electrification Project   96 Box 2.2 • Finding the optimal input parameters for DBSCAN   98 Box 3.1 • How IDCOL increases productive uses of energy in solar-hybrid mini grids in Bangladesh 131 Box 3.2 • Rural, productive uses of electricity: Lessons from Ethiopia 133 Box 3.3 • Lessons from a utility-NGO partnership in Indonesia 134 Box 5..1 • Green Village Electric partners with Schneider Electric 155 Box 5.2 • Africa Minigrid Developers Association 160 Box 5.3 • Utility-led rollout of mini grids on the national scale: Case study from Ethiopia 161 Box 5.4 • Prepay is not just for mini grids—Eskom’s “Power for All” scheme 164 Box 6.1 • Women’s limited access to finance 183 Box 6.2 • Nonprice factors in renewable energy projects 194 Box 6.3 • A financial support program in Bangladesh 195 Box 6.4 • A revolving fund for consumer finance in Lao PDR 199 Box 7.1 • Policy and regulatory train-ings provided by Economic Consulting Associates 211 Box 7.2 • Lessons from community training by Trama TechnoAmbiental 216 Box 7.3 • Productive-use training from IEEE Smart Village 217 Box 7.4 • Components of IEEE Smart Village’s comprehensive training program 219 Box 8.1 • The build-own-operate model in Tanzania 224 Box 9.1 • Setting individualized tariff controls 245 Box 9.2 • From laissez-faire to comprehensive regulation: Cambodia’s successful electrification with mini grids 262 Box 9.3 • Independent appeal tribunal in Jamaica 265 Box 10.1 • The effect on investment of permits outside the electricity sector 276 Box 10.2 • World Bank experience with Odyssey Energy Solutions in Nigeria 283 FIGURES Figure MF.1 • A mini grid system (part A) and a containerized solar mini grid (part B) 5 Figure O.1 • Example of a common solar hybrid mini grid setup 17 Figure O.2 • The first, second, and third generations of mini grids 21 Figure O.3 • Matrix of market drivers and building blocks to support them 27 Figure O.4 • Number of people connected to mini grids under business-as-usual and universal access scenarios, 2020–30 34 Figure O.5 • Mini grids installed annually in each of the top 20 electricity-access-deficit countries, 2018–30 37 Figure O.6 • Average mini grid load factor, 2018–30 38 Figure O.7 • Total cumulative investment in mini grids for energy access, 2018–30 39 Figure O.8 • Average RISE score in top 20 electricity-access-deficit countries 40 Figure O.9 • LCOE of best-in-class solar hybrid mini grids 41 Figure 1.1 • Load profiles for 22 percent load factor, 22 percent load factor (sun following), 40 percent load factor, 40 percent load factor (sun following), and 80 percent load factor 47 Figure 1.2 • Economic LCOE calculations for mini grids in 7 cases based on 0 percent subsidy and load profiles described in figure 1.1 50 Figure 1.3 • Comparison of levelized cost of energy of mini grids and utilities in Africa 53 Figure 1.4 • Total economic cost of mini grids per kWfirm as a function of firm power output 55 Figure 1.5 • Mini grid economic costs per customer (left) and costs per customer for mini grids below median cost (right) 56 Figure 1.6 • Average share of component economic costs in total capital costs of mini grids 58 Figure 1.7 • Costs of solar panels (including PV inverters) for mini grids, by year, 2012–21 59 Figure 1.8 • Economic cost trends for the storage capacity ($/kWh) of lithium-ion and lead-acid batteries used in mini grids between 2012 and 2021 60 MINI GRIDS FOR HALF A BILLION PEOPLE    xi Figure 1.9 • Net present value of storage capacity for lithium-ion and lead-acid batteries, 2012–21 61 Figure 1.10 • Unit costs for inverters, energy management systems, and monitoring (blue), and balance of system (orange) 62 Figure 1.11 • Powerhouse innovations can lower costs and expedite deployment 63 Figure 1.12 • Distribution costs per customer, 2012 to 2021 64 Figure 1.13 • Distribution costs per customer as a function of customers served 65 Figure 1.14 • A 30 kWp Mandalay Yoma mini grid in Myanmar (left) and a 40 kWp Winch Energy mini grid in Uganda (right) 66 Figure 2.1 • Scatter plot of settlement population vs population density in Sub-Saharan Africa 79 Figure 2.2 • Sub-Saharan Africa’s addressable market for mini grids 81 Figure 2.3 • Sub-Saharan Africa’s addressable market for mini grids, mapped by settlement population 81 Figure 2.4 • Distribution by country of 429.5 million people served at least cost by mini grids in 58 access-deficit countries 83 Figure 2.5 • Distribution by mini grid size of 429.5 million people served at least-cost by mini grids in 58 countries with severe access deficits 85 Figure 2.6 • Geospatial least-cost rollout plans in Kenya and Rwanda 86 Figure 2.7 • Geospatial least-cost electrification plans for Myanmar and Nigeria by 2030, by technology component 87 Figure 2.8 • The GEP Explorer 88 Figure 2.9 • Typical least-cost electrification planning sequence (best practice) 89 Figure 2.10 • Distance as a function of load size: Break-even grid extension 92 Figure 2.11 • Geospatial portfolio planning sequence 95 Figure 2.12 • Methodology for the generation of population clusters in Nigeria, using the HRSL and OSM data 96 Figure 2.13 • Nigeria’s population clusters: Spatial distribution (left) and size histogram, in hectares (right) 97 Figure 2.14 • Sample outputs from the Digitize Africa building footprint data set 97 Figure 2.15 • Concept of the DBSCAN algorithm 97 Figure 2.16 • Ethiopia’s rural population settlements and mini grid deployment: Bounded DBSCAN clustering 98 Figure 2.17 • Cluster contour delineation: Convex hull (left) and alpha shapes (right) 99 Figure 2.18 • Health facilities and education facilities within 500 meters of village boundary 100 Figure 2.19 • Main grid coverage in Nigeria 100 Figure 2.20 • VIDA GridLight prediction for Ethiopia (blue) compared to Ethiopian Electric Utility data (red) 101 Figure 2.21 • Nighttime lighting in Nigeria 101 Figure 2.22 • Nigerian night lights and 20 km buffer zones 101 Figure 2.23 • Results of prioritization of clusters for mini grid electrification in Nigeria 102 Figure 2.24 • Population clusters falling in protected areas or KBAs flagged for exclusion in Nigeria 103 Figure 2.25 • Requirements for estimating load profile 104 Figure 2.26 • Indicative load profiles for various customer segments in potential mini grid locations 105 Figure 2.27 • Demand curve for a randomly selected candidate site in Ethiopia 105 Figure 2.28 • Sample load profile for a village 106 Figure 2.29 • Output of the voltage drop model 108 Figure 2.30 • Sample outputs of hyperlocal density analysis from Village Data Analytics 108 Figure 2.31 • Indicative flow of financial modeling process of mini grids 109 Figure 2.32 • VIDA interactive platform 110 Figure 2.33 • Mini grid tender preparation in Odyssey 111 Figure 3.1 • The impact of productive electricity uses on the daily load profile and levelized cost of energy 115 Figure B3.1.1 • Share of expected load achieved by selected mini grids in Bangladesh 131 Figure B3.1.2 • Effect of extensive customer awareness campaigns on uptake 132 Figure 4.1 • Typical issues hindering the mini grid development process 138 Figure 4.2 • Typical mini grid project cycle 139 Figure 4.3 • The A-B-C model 140 Figure 5.1 • Tariff-charging type by customer class 163 Figure 5.2 • Tariffs by customer class ($/kWh, charged by consumption) 163 Figure 5.3 • Tariffs by customer class ($/month, charged in flat fee) 163 xii   MINI GRIDS FOR HALF A BILLION PEOPLE Figure 5.4 • Mini grid industry value chain 171 Figure 5.5 • Customer types served by mini grids 172 Figure 5.6 • Average contribution share of each disclosed funding type 173 Figure 5.7 • Generation source of the mini grid 174 Figure 5.8 • Causes behind operators’ worst months of delivering electricity 175 Figure 5.9 • Tiers of daily availability in typical and worst months 176 Figure 5.10 • Tiers of evening availability in typical (left) and worst (right) months 176 Figure 5.11 • Daily peak hour profile 177 Figure 5.12 • Tiers of reliability in typical and worst months 178 Figure 5.13 • Profit potential for facilitator organizations across the value chain, 2019 and 2030 178 Figure 6.1 • Cumulative private-sector investment in mini grid companies 184 Figure 6.2 • Annual number of deals between investors and mini grid companies, 2010–22 185 Figure 6.3 • Top 20 investors in mini grid companies by cumulative investment 185 Figure 6.4 • Trends in debt and equity 186 Figure 7.1 • National-level training needs and relevant stakeholders 211 Figure 7.2 • Project- and portfolio-level training needs and relevant stakeholders 213 Figure B7.3.1 • Future community entrepreneurs and mini grid technicians participate in a classroom discussion at a training program offered by Igniting Africa 217 Figure 8.1 • Roles for developer and utility under different delivery models 226 Figure 8.2 • Sample ecosystem of institutions affecting mini grid developers 229 Figure 9.1 • Regulatory Indicators for Sustainable Energy (RISE) scores for mini grid framework 237 Figure 9.2 • Correlation between mini grid policies and regulations and number of mini grids planned 238 Figure 9.3 • Decision tree for regulating entry 241 Figure 9.4 • Decision tree for regulating tariffs 247 Figure 9.5 • Decision tree for regulating service standards 251 Figure 9.6 • Decision tree for regulating technical standards 254 Figure 9.7 • Decision tree for integration and exit options 258 Figure 10.1 • Nigeria’s Environmental and Social Management System for minimum-subsidy tenders for mini grid development 280 Figure 10.2 • Nigeria’s Environmental and Social Management System for performance-based grants for mini grid development 281 TABLES Table MF.1 • Market drivers and 2030 targets 3 Table MF.2 • Building blocks and identified areas for potential magnitude change 3 Table MF.3 • Sub-Saharan African mini grid markets and their progress across the 10 building blocks 4 Table MF.4 • The levelized cost of energy by load factor, 2018, 2021, and 2030 6 Table O.1 • Benchmarks and price projections, mini grid component costs, 2010–30 24 Table O.2 • SDG 7 and mini grid industry targets, 2020–30 26 Table O.3 • Installed and planned mini grid projects worldwide: A summary 30 Table O.4 • Summary of installed mini grid projects by region 31 Table O.5 • Number of installed and planned mini grids by region 32 Table O.6 • Top-10 lists for key mini grid indicators for installed mini grids 32 Table O.7 • Characteristics of installed and planned mini grids 33 Table O.8 • ESMAP mini grid outlook scenario: A regional breakdown 36 Table O.9 • Top 20 countries with energy access deficits: Doing Business and RISE scores, 2020 40 Table O.10 • The global mini grid sector and its progress across the 10 building blocks 42 Table 1.1 • Representative mini grids from seven cases: An analysis of key characteristics 46 Table B1.1.1 • Estimated and potential levelized cost of mini grid energy, 2018 and 2021 49 Table 1.2 • Optimum renewable energy share for mini grid cases considered 51 MINI GRIDS FOR HALF A BILLION PEOPLE    xiii Table 1.3 • Economic LCOE of hybrid mini grid versus diesel only and renewables only 52 Table 1.4 • Mini grid components: A summary of costs and characteristics 57 Table 1.5 • Average economic costs of key mini grid hardware components, by country 58 Table 1.6 • Performance characteristics of lead-acid and lithium-ion batteries as modeled in HOMER levelized cost of energy calculations 60 Table 1.7 • Mini grid component cost benchmarks and price projections 68 Table 1.8 • Net present value broken down by category with economies of scale 71 Table 1.9 • Change in unit costs with economies of scale, by cost category 72 Table 2.1 • Characteristics of Sub-Saharan African settlements suitable for electrification via mini grid 80 Table 2.2 • Selected electrification results for 2030 retrieved from the Global Electrification Platform, aggregated for 46 countries in Sub-Saharan Africa 82 Table 2.3 • Breakdown of electrification results from bottom-up demand scenario 82 Table 2.4 • GEP scenario codes for each country’s maximum number of new mini grid connections by 2030 83 Table 2.5 • Distribution by mini grid system size of 429.5 million people served at least-cost by mini grids in 58 countries with access deficits 85 Table 2.6 • Maximum cost-justified distance for connecting a customer as a function of the required level of service 92 Table 3.1 • The Mwenga hydro mini grid: Estimated costs and benefits 116 Table 3.2 • Six steps to roll out a PUE program 118 Table 3.3 • Example of PUE program stakeholders identified in the Democratic Republic of Congo, Ethiopia, and Nigeria 119 Table 3.4 • Power requirements, costs, and indicative payback periods of selected income-generating appliances 122 Table 3.5 • Stakeholders that could be involved in road shows and their respective roles 126 Table 4.1 • Potential for community engagement to limit mini grid risks 138 Table 4.2 • Key community engagement activities over the project cycle 144 Table 5.1 • Utility connection rates, 2004–14 153 Table 5.2 • Categories of local and international private-sector players 156 Table 5.3 • Sample mini grid developers 157 Table 5.4 • Profit potential of mini grid operators given certain tariffs and costs of service 158 Table 5.5 • Mini grid developers and large-scale IPPs: A comparison of audited financial results, 2018 159 Table 5.6 • Sample utility mini grid projects 161 Table 5.7 • Sample mini grid experience of EPC companies 166 Table 5.8 • Sample system integrators with technologies in mini grids 166 Table 5.9 • Sample original equipment manufacturers with technologies in mini grids 167 Table 5.10 • Benefits of local and international partnerships 170 Table 6.1 • Types and sources of commercial finance 184 Table 6.2 • Categories of investors in mini grid companies, 2010–22 186 Table 6.3 • World Bank mini grid investment portfolio as of June 30, 2022 188 Table 6.4 • Overcoming barriers to investments in mini grids 193 Table 6.5 • Impact of performance-based subsidies of capital expenses on the levelized cost of energy of a well-designed mini grid 198 Table 7.1 • Conducting a project- or portfolio-level capacity needs assessment 208 Table 7.2 • Selection of tools for resource planning 212 Table 7.3 • Sample tools for financial planning 212 Table 8.1 • Comparative analysis of mini grid delivery models 226 Table 8.2 • Roles of national and international institutions 230 Table 9.1 • Options for regulating entry 239 Table 9.2 • Information requirements for registration, permitting, and licensing of mini grids in five countries 240 Table 9.3 • Options for regulating tariffs 243 Table B9.1.1 • Individualized tariff features in four countries 245 Table 9.4 • Assessment of tariff options 246 Table 9.5 • Options for regulating service standards 250 xiv   MINI GRIDS FOR HALF A BILLION PEOPLE Table 9.6 • Options for regulating technical standards 253 Table 9.7 • Options for preserving value when the main grid arrives 256 Table 9.8 • Effective regulatory packages 260 Table 9.9 • Three phases of evolutionary regulation 261 Table 9.10 • Two innovations in regulation that can further incentivize private-sector investment in mini grids 263 Table 10.1 • Options for making it easier for mini grid developers to do business 274 Table 10.2 • General types of bureaucratic processes that mini grid developers navigate 276 Table 10.3 • Key provisions of power purchase agreements in Cambodia and Tanzania 277 Table 10.4 • Advantages and disadvantages of two options for mini grid oversight 285 Table 11.1 • Analytical framework to guide future research on mini grid business models 292 MINI GRIDS FOR HALF A BILLION PEOPLE    xv ACKNOWLEDGMENTS This book is the result of a collaborative effort over more (HOMER). The write-up on the first generation of mini than four years to collect and synthesize the best research, grids is based on a report on the history of mini grids data, and knowledge produced to date on the entire eco- in electric power systems prepared by Morgan Bazil- system that supports mini grids—regulations, financing, ian (Colorado School of Mines), Francesco Fuso Ner- technology, training, gender, productive uses, geospatial ini (KTH), Mark Howells (KTH), Alexandros Korkovelos planning, and more. We would like to acknowledge here (KTH), Lucille Langois (IAEA), Holger Rogner (IIASA), and those individuals and organizations that contributed to Hisham Zerriffi (University of British Columbia), which is this project. available on the companion website to this handbook: www.esmap.org/mini_grids_for_half_a_billion_people. PARTNERSHIP AND FUNDING Guidehouse (formerly Navigant Research) provided valu- This book is part of a much larger comprehensive knowl- able assistance with initial data collection on the mini edge package that was funded by World Bank/ESMAP and grid market in 2018, and we leveraged their global data- the Foreign, Commonwealth and Development Office of the base as a starting point for our data set. We are grateful United Kingdom as one of ESMAP’s key donors in particu- also to World Bank operations teams for helping us col- lar. This comprehensive knowledge package is the result of lect data on mini grids in their respective countries, and ESMAP’s collaboration with a broad set of mini grid sector to Bloomberg New Energy Finance—in particular Itamar stakeholders, including mini grid developers, regulators and Orlandi—for answering our questions and helping us other government officials, financiers, technology provid- understand their earlier outlooks for the mini grid market. ers, researchers, project implementation partners, develop- For the discussion on the Key Performance Indicators for ment partners, and internationally recognized experts. We the mini grid industry, we are grateful for contributions are particularly grateful to the following mini grid industry and guidance from the Africa Minigrid Developers Asso- leaders who not only gave us their time but also helped ciation’s leadership team. ensure our work captured ground-level realities and frontier innovations accurately: Africa Minigrid Developers Associa- CHAPTER 1: LEAD AUTHORS CHRIS GREACEN AND tion (AMDA), Engie, Havenhill, HOMER Energy, Husk Power JON EXEL Systems, INENSUS, Odyssey Energy Solutions, Power Cor- This chapter would not have been possible without the ner, PowerGen, PowerHive, Techno Hill, and Trama Tecno- detailed data provided by mini grid developers. Detailed Ambiental. data on costs and designs are commercially sensitive. The authors thank all mini grid developers who recognized PROJECT MANAGEMENT AND AUTHORSHIP the importance of research on the evolution of mini grid This book was prepared by the Global Facility on Mini Grids costs and provided the data they could. For 32 mini grids team under the overall guidance of ESMAP’s former and in diverse countries across Africa and Asia, the firm Trama current practice managers, Rohit Khanna and Gabriela TecnoAmbiental S.L. (TTA) undertook the difficult work of Elizondo Azuela, respectively. Jon Exel, Tatia Lemondzhava, requesting data and following up to iron out inconsisten- Ashish Shrestha, and Dr. James Knuckles managed the cies as they were discovered. Chris Purcell provided data project and oversaw the book’s development, from incep- for mini grids in Myanmar and Bangladesh; Sunita Chik- tion to publication. Dr. Knuckles was the book’s lead editor. katur Dubey did the same for several mini grids in Ghana. The following individuals and organizations are this book’s The Odyssey team provided extensive data for hundreds co-authors and contributors: of mini grids in Nigeria. The authors thank Andrew Pascale and Ziting Huang for their assistance and expertise in pro- OVERVIEW: LEAD AUTHORS JON EXEL, cessing data and Mr. Pascale for his work on an early draft JAMES KNUCKLES, AND TATIA LEMONDZHAVA of this chapter. Marilena Lazopoulou and Pol Arranz Piera The chapter benefitted from expert inputs from Dr. Ber- from TTA and Chris Purcell were generous and patient nard Tenenbaum, Dr. Chris Greacen, and Peter Lilienthal with our many questions. xvi   MINI GRIDS FOR HALF A BILLION PEOPLE CHAPTER 2: LEAD AUTHORS ASHISH ment deals featured in this chapter. Dr. Chris Greacen and SHRESTHA AND ALEXANDROS KORKOVELOS HOMER Energy conducted modeling and analysis for the The chapter benefitted from expert inputs from Philipp section of this chapter on the impacts of performance-based Blechinger (Reiner Lemoine Institut), Stewart Craine and grants on the cost of solar hybrid mini grid electricity. Yann Monty Craine (Village Infrastructure Angels), Reja Amatya Tanvez and Candice Lanoix provided a great deal of input (MIT), Xiangkun Li and Samuel Booth (NREL), Peter Lil- from the International Finance Corporation. ienthal (HOMER), Rik Wuts (Powerhive), Ibrahim Abada CHAPTER 7: LEAD AUTHORS SUNITA DUBEY AND (Engie), Emily McAteer (Odyssey Energy Solutions), Nabin CASTALIA Raj Gaihre, Philippe Raisin and Tobias Engelmeier (VIDA), and Frankie Eckersley-Carr, Imran Muhammad, Oliver Haas We are grateful for contributions from the Institute of Elec- (Integration). The writeup on least-cost electrification plan- trical and Electronics Engineers and to GIZ, HOMER, and ning was prepared with contributions from Chiara Odetta Trama TecnoAmbiental for their contributions as well. Rogate and Rhonda Lenai Jordan (World Bank). The spatial CHAPTER 8: LEAD AUTHORS SUBODH MATHUR, data analysis for Sub-Saharan Africa using GRID3 data was WITH NICO PETERSCHMIDT AND JOANIS HOLZIGEL carried out by Christopher James Arderne. OF INENSUS CHAPTER 3: LEAD AUTHORS JULIETTE BESNARD, We are grateful to the International Finance Corporation TATIA LEMONDZHAVA, JON EXEL, AND JAMES for important inputs on this chapter, including the Inves- KNUCKLES tors’ Perspective section on institutional frameworks for The chapter benefited from the support of Besnik Hyseni mini grids. (World Bank), the Communities in Kisii and Nyamira Coun- CHAPTER 9: LEAD AUTHORS CASTALIA AND ties (Kenya), CLASP, Erika Lovin (CrossBoundary), Amanda JAMES KNUCKLES DelCore (Factor[e]), GMG Facility Kenya, IDCOL, INENSUS and the World Bank RERED II team, Jon Leary, Ed Brown, This chapter draws on a research project conducted by and Simon Batchelor of Loughborough University, Pact/ ESMAP, Castalia, and Ecoligo, which included 11 field visits Smart Power Myanmar, Adriana Karpinska (Powerhive), and 70 interviews, conducted between August and Sep- SNV Netherlands Development Organisation, Barani Aung tember 2017, with key mini grid sector stakeholders in Ban- (Techno Hill), and Trama TecnoAmbiental. gladesh, Cambodia, India (Uttar Pradesh), Kenya, Nigeria, and Tanzania. Stakeholders in each country included reg- CHAPTER 4: LEAD AUTHORS TATIA LEMONDZHAVA, ulators, developers, rural electrification agencies, and con- WITH FELIX TER HEEGDE OF SNV sumers, as well as utilities and development partners, and We are grateful to Rishabh Sachdeva of Quicksand and we are very grateful for their time and insights. A full list of Elijah Siakpere of the World Bank office in Abuja, Nigeria, acknowledgments for each of the six countries is provided for additional contributions. at the end of chapter 9. We also thank the International Finance Corporation for important inputs on this chapter, CHAPTER 5: LEAD AUTHORS RICKY BUCH, JON EXEL, including the Investors’ Perspective section on mini grid JAMES KNUCKLES, AND TATIA LEMONDZHAVA regulations. Kojo Adom Quagraine led the development of the database CHAPTER 10. LEAD AUTHORS: of investment deals featured in this chapter. Bryan Koo and JAMES KNUCKLES AND CASTALIA Ziting Huang (World Bank) helped manage the nationally representative surveys of mini grid operators conducted by Castalia developed sections of the chapter as part of a the World Bank and contributed to the chapter’s analysis larger project it completed for ESMAP (see chapter 9), and writeup of survey results in Cambodia, Myanmar, and and Elijah Abiodun Siakpere contributed in major ways to Nepal. the write-up of the standardized Environmental and Social Management System in Nigeria. Dr. Chris Greacen and CHAPTER 6: LEAD AUTHORS INTERNATIONAL Anastas Mbawala provided valuable comments on ear- FINANCE CORPORATION, JAMES KNUCKLES, lier drafts of the standardized asset transfer agreement. AND SUBODH MATHUR We are grateful to the International Finance Corporation We are grateful for feedback from David Ross of Statera for important inputs on this chapter as well, including the Capital on earlier drafts of the chapter. Kojo Adom Investors’ Perspective section on factors that make it eas- Quagraine led the development of the database of invest- ier (or harder) for mini grid developers to do business. MINI GRIDS FOR HALF A BILLION PEOPLE    xvii CHAPTER 11: LEAD AUTHORS JON EXEL AND chapters throughout the book. The International Finance JAMES KNUCKLES Corporation also provided extensive comments on several chapters, which strengthened the book as a whole. To all of KEY GENDER ASPECTS THROUGHOUT THE BOOK these reviewers: thank you. Inka Schomer and Mary Dominic (World Bank) provided important contributions on the gender-related aspects STAKEHOLDER ENGAGEMENT of mini grid topics throughout the book, from access to Over the past five years, ESMAP has cohosted events and finance to productive uses to community engagement workshops on mini grids both virtually and in-person in and skills building, among others. Ethiopia, Ghana, Kenya, Myanmar, Nigeria, Spain, Tanzania, the United Kingdom, and the United States. We are par- REVIEW AND CONSULTATION ticularly grateful for the host governments as well as the This handbook underwent a formal World Bank Decision more than 2,000 participants at these events representing Review in February 2022, chaired by Demetrios Papatha- all mini grid stakeholder groups from more than 60 coun- nasiou, Global Director for the Energy & Extractives Global tries for their input, debate, and validation of the knowledge Practice. The peer reviewers were Raihan Elahi (Lead Econ- brought together in this report. The Global Facility on Mini omist, World Bank), Dana Rysankova (Lead Energy Special- Grids team would also like to express its gratitude to the ist and Global Lead for Energy Access, World Bank), Arsh ESMAP communications team, and especially to Lucie Sharma (Senior Energy Specialist, World Bank), Patrick Blyth, Nansia Constantinou, Anita Rozowska, and Janice Thaddayos Balla (Senior Energy Specialist, World Bank), Tuten for their help in synthesizing and tailoring the main and Yann Tanvez (Energy Specialist, International Finance messages of this report for a wide audience, and their over- Corporation). In addition, the following experts carefully all support in its preparation and dissemination. reviewed the book in whole or in part. For their time, exper- EDITING, GRAPHIC DESIGN, AND OTHER tise, and thoughtful comments we are exceptionally grate- CONTRIBUTORS ful. Dr. Bernard Tenenbaum, Castalia, INENSUS, the Rocky Mountain Institute, and Trama TecnoAmbiental reviewed Our editors, Joan O’Callaghan and Steven Kennedy, require the book in its entirety; their comments greatly elevated special praise. We are grateful to have been able to work the quality of the final product. Gabriela Elizondo Azuela with them. Naylor Design, Inc. typeset the manuscript and and Michael Toman (World Bank) provided provided invalu- designed every table, chart, and figure in this book. The able input on the report’s main findings that helped us con- quality and consistency they brought to this report is a tes- vey comprehensive and impactful messages throughout tament to their diligence and expertise. the book. Comments from the United Kingdom’s Foreign, The many people we interviewed or who reviewed this Commonwealth and Development Office, the United States report were gracious with their time and knowledge. They National Renewable Energy Laboratory, the Institute of were patient and accommodating with our requests under Electrical and Electronics Engineers, and the Africa Mini- tight deadlines, and we are sincerely grateful. We alone are grid Developers Association helped us transform several responsible for any errors of fact or interpretation. xviii   MINI GRIDS FOR HALF A BILLION PEOPLE ABBREVIATIONS — not available A-B-C Anchor-Business-Community AC alternating current ACP-EU Africa, Caribbean, and Pacific Group of States–European Union ADB Asian Development Bank AFD Agence Française de Développement (French Development Agency) AfDB African Development Bank AGRITEX Agricultural Technical and Extension Services Ah ampere-hour AMADER Agency for the Development of Domestic Energy and Rural Electrification AMDA Africa Minigrid Developers Association ARE Alliance for Rural Electrification ATP ability to pay BAU business as usual BBBEE Broad-Based Black Economic Empowerment Act of 2003 BERC Bangladesh Energy Regulatory Commission BLS Bureau of Labor Statistics BMZ Federal Ministry for Economic Cooperation and Development BNEF Bloomberg New Energy Finance BOO build-own-operate BRD Development Bank of Rwanda CAPEX capital expenditure CBS Central Bureau of Statistics CE community engagement CEDECAP Centre of Demonstration and Qualification in Appropriate Technologies CELAMeD community engagement, load acquisition and micro-enterprise development CIESIN Center for International Earth Science Information Network CO2 carbon dioxide cofin. cofinanced COGS cost of goods sold COP Conference of the Parties CPI Climate Policy Initiative CREDA Chhattisgarh State Renewable Energy Development Agency CSCs customer service centers DC direct current MINI GRIDS FOR HALF A BILLION PEOPLE    xix DISCO distribution company DRC Democratic Republic of Congo E4I Energy4Impact EAC Electricity Authority of Cambodia ECOWAS Economic Community of West African States ECREEE ECOWAS Centre for Renewable Energy and Energy Efficiency ECS electricity consumer society EDC Electricité du Cambodge (Cambodia Electricity) EDC Enterprise Development Cambodia EPC engineering, procurement, and construction company EEP Energy and Environment Partnership e-MPF European Microfinance Platform ERC Energy Regulatory Commission ESCO energy service company ESIA Environmental and Social Impact Assessment ESMAP Energy Sector Management Assistance Program ESMP Environmental and Social Management Plan ESMS Environmental and Social Management System EUEI PDF European Union Energy Initiative Partnership Dialogue Facility EWURA Energy and Water Utilities Regulatory Authority FCDO Foreign, Commonwealth and Development Office FCV fragility, conflict, and violence FDoE Fiji Department of Energy GEM Global Entrepreneurship Monitor GFMG Global Facility on Mini Grids GIS geographic information system GIZ Deutsche Gesellschaft für Internationale Zusammenarbeit (German Society for International Cooperation) GMG Green Mini-Grid Help Desk GoP Government of Peru GVE Green Village Electric GW gigawatt GWh gigawatt-hours GWp gigawatts-peak ha hectare HOMER Hybrid Optimization of Multiple Energy Resources HR human resources HRSL High Resolution Settlement Layer HV high voltage ICT information and communications technology IDCOL Infrastructure Development Company Limited IDS Institute of Development Studies IEA International Energy Agency xx   MINI GRIDS FOR HALF A BILLION PEOPLE IEEE Institute of Electrical and Electronics Engineers IEG Independent Evaluation Group IFC International Finance Corporation ILO International Labour Organization IMF International Monetary Fund IPCC Intergovernmental Panel on Climate Change IPP independent power producer IQR interquartile range IRENA International Renewable Energy Agency IUCN International Union for Conservation of Nature JPS Jamaica Public Service Company KBA key biodiversity area km kilometer KPI key performance indicator KPLC Kenya Power and Lighting Company kVA kilovolts-ampere kW kilowatts kWfirm kilowatts of firm alternating current output kWh kilowatt-hour kWp kilowatts-peak L liter LCOE levelized cost of energy LCOS levelized cost of storage LED light-emitting diode LF load factor LGA local governmental area li lithium LRP Livelihood Restoration Plan LV low voltage M million MAS Multi-Agent System MDG Millennium Development Goal MFD Maximizing Finance for Development MG mini grid MGA Micro-Grid Academy mi mile MIA Microgrid Investment Accelerator MIT Massachusetts Institute of Technology MLIP Ministry of Labour, Immigration and Population MOHP Ministry of Health and Population MOI Ministry of Infrastructure MRG Minimum Revenue Guarantee MINI GRIDS FOR HALF A BILLION PEOPLE    xxi MTF Multi-Tier Framework MV medium voltage MW megawatt MWh megawatt-hour n.a. not applicable NAPTIN National Power Training Institute of Nigeria n.d. no date NEP Nigeria Electrification Project NERC Nigerian Electric Regulatory Commission NGO nongovernmental organization NIS National Institute of Statistics NPC-SPUG National Power Corporation Small Power Utility Group NREL National Renewable Energy Laboratory (US) O&M operations and maintenance ODI Overseas Development Institute OECD Organisation for Economic Co-operation and Development OEM original equipment manufacturer OPEX operational expenditure OSM OpenStreetMap OUR Office of Utilities Regulation P2PB peer-to-peer business PAYG pay-as-you-go PDR (Lao) People’s Democratic Republic PFI participating financial institution PPA power purchase agreement PPP public-private partnership PRG partial risk guarantee PU productive use PUE productive uses of energy PV photovoltaic PwC PricewaterhouseCoopers RAP Resettlement Action Plan RE renewable energy REA Rural Electrification Agency (Nigeria) REA Rural Energy Agency (Tanzania) REAs rural electrification/energy agencies REF Rural Electrification Fund (Cambodia) REF Renewable Energy Fund REM Reference Electrification Model RENAC Renewables Academy REopt Renewable Energy Integration and Optimization RES4Africa Renewable Energy Solutions for Africa xxii   MINI GRIDS FOR HALF A BILLION PEOPLE RISE Regulatory Indicators for Sustainable Energy RMI Rocky Mountain Institute RoE return on equity SAIDI System Average Interruption Duration Index SAIFI System Average Interruption Frequency Index SEforALL Sustainable Energy for All SDG Sustainable Development Goal SDG7 Sustainable Development Goal 7 SG&A selling, general, and administrative Sh shilling SI system integrator SIDA Swedish International Development Cooperation Agency SM smart meter SNV Netherlands Development Organisation SPD small power distributor SPI Smart Power India SPM Smart Power Myanmar SPP small power producer TANESCO Tanganyika Electric Supply Company TARA Society for Technology and Action for Rural Development TASF Transaction Advisory Services Facility TTA Trama TechnoAmbiental TWp terawatts-peak UCS Union of Concerned Scientists UN United Nations UNDP United Nations Development Programme UNIDO United Nations Industrial Development Organization UNOPS United Nations Office for Project Services USAID United States Agency for International Development USD United States dollar USDA United States Department of Agriculture US DOE United States Department of Energy USG United States Government VEC Village Electrification Committee VIA Village Infrastructure Angels W watt WB World Bank Wfirm watt of firm alternating current output Wdc watt of direct current Wp watts-peak WTP willingness to pay All dollar amounts are US dollars unless otherwise indicated. MINI GRIDS FOR HALF A BILLION PEOPLE    xxiii MINI GRIDS BY THE NUMBERS Where we are today 48 million people connected to 21,500 mini grids, of which half are solar PV, at an investment cost of $29 billion. 29,400 mini grids planned, 95 percent of them in Africa and South Asia, 99 percent solar PV, connecting more than 35 million people at an investment cost of $9 billion. Where we need to be to reach universal access by 2030 490 million people served at least cost by 217,000 mini grids, almost all solar-powered, requiring an investment of $127 billion. To deploy mini grids at scale, countries must act on 10 Building Blocks: (1) reducing costs and optimizing design & innova- tion for solar mini grids; (2) planning national strategies and developer portfolios with geospatial analysis and digital platforms; (3) transforming productive livelihoods and improving business viability; (4) engaging communities as valued customers; (5) delivering services through local and international companies and utilities; (6) financing solar mini grid portfolios and end user appliances; (7) attracting exceptional talent and scaling skills development; (8) supporting institutions, delivery models, and champions that create opportunities; (9) enacting regulations and policies that empower mini grid companies and customers; (10) cutting red tape for a dynamic business environment. Regional mini grid trends from ESMAP’s database of more than 50,000 mini grid projects in 138 countries Top 5 countries . . . INSTALLED PLANNED INSTALLED PLANNED (mostly first- and second- (mostly third-generation (mostly first- and sec- (mostly third-generation generation mini grids) mini grids) ond-generation mini grids) mini grids) 9,600 South Asia 19,000 South Asia 4,700 Afghanistan 18,900 India 7,200 E ast Asia and Pacific 800 East Asia and Pacific 4,000 Myanmar 2,700 Nigeria 3,100 Africa 9,000 Africa 3,200 India 1,500 Tanzania 1,200 OECD and Central Asia 400 OECD and Central Asia 1,500 Nepal 1,200 Senegal 300 Other 100 Other 1,200 China 600 Ethiopia Current financing Top 3 private-sector developers By installed $29 billion—Cumulative global investment in mini and planned mini grids grids to date 1. Tata Power Renewable Microgrids (10,000 / India) 2. Husk Power (5,000 / India & Africa) $9 billion—Cumulative global investment in Africa 3. OMC Power (5,000 / India) and South Asia in mini grids to date Top 3 utilities By installed and planned mini grids $2.6 billion—Development Partners committed, including AFD, AfDB, FCDO, the Islamic Development 1. RAO (700 / Russia) Bank, GIZ and the World Bank, among others 2. PT Perusahaan Listrik Negara (500 / Indonesia) 3. NPC-SPUG (300 / Philippines) $1.4 billion—World Bank commitment to mini grids in 31 countries through 2027 $500+ million—Private-sector investment in mini Private-sector opportunity grid developers in low-income countries since 2013 $3.3 billion annual profit potential for developers across all mini grids deployed through 2030 25 percent—Average World Bank share of total mini grid investment (government, development $5.8 billion net profit potential across all mini grid partners, and private sector) in client countries component and service suppliers in 2030 alone xxiv   MINI GRIDS FOR HALF A BILLION PEOPLE MINI GRIDS BY THE NUMBERS, continued Cost of a best-in-class Cost of unsubsidized . . . Compared with solar-hybrid mini grid electricity from a best- utilities in Africa today . . . . . . and by 2030 in-class solar hybrid $0.27/kWh average mini grid . . . across 39 utilities $3,659/kWfirm total <$2,500/kWfirm capital expense total capital expense $0.38/kWh (LCOE) 2 of 39 utilities with baseline today cost-recovery tariffs $596/kWp Solar PV $290/kWp Solar PV Module Module $0.28/kWh with income- generating machines to $297/kWh Lithium-ion $137/kWh Lithium- achieve 40 percent load factor batteries ion batteries $0.20/kWh with $265/kW battery income-generating machines inverter and expected 2030 costs Income-generating machinery 3rd generation mini . . . compared with grid service . . . typical utilities < 12 months payback period for more than 130 income-generating machines and 99 percent uptime 40–50 percent other equipment available today uptime Tier 4–5 access 84/100 customer Tier 3–4 access $3.6 billion microfinance needed for 3 million satisfaction rate 41/100 customer machines and other equipment connected satisfaction rate to third-generation mini grids in 2030 Environmental impact Typical third-generation mini grid 10–15 GW solar PV installed by 2030 $0.5– $1.0 million investment 50–110 GWh batteries mostly lithium-ion 200–800 clients connected 60 percent energy savings from energy efficient 800–4,000 people receiving electricity for the first appliances time 1.2 billion tonnes of CO2 emissions avoided 50–100 kWp solar PV installed 200–500 kWh batteries installed What is a mini grid? Mini grids are electric power generation and distribution systems that provide electricity to just a few customers in a remote settlement or bring power to hundreds of thousands of customers in a town or city. They can be fully isolated from the main grid or connected to it but able to intentionally isolate (“island”) themselves from the grid. Mini grids supply power to households, businesses, public institutions, and anchor clients, such as telecom towers and large agricultural processing facil- ities. They are designed to provide high-quality, reliable electricity. A new, “third generation” of mini grids has recently emerged. They incorporate the latest technologies, such as smart meters and remote monitoring systems; and are typically designed to interconnect with the main grid. To be considered in our analysis in the context of this report, a mini grid had to serve multiple customers. Electricity systems that service a single hospital, industrial facility, military base, university campus, mine, or other single entity, were therefore not considered mini grids. We also do not define mini grids in terms of size, although in our detailed analysis of mini grid costs and in our global database of more than 50,000 mini grid projects, the vast majority (90 percent) ranged from 10 kW to 1 MW in installed capacity. Sources and underlying analysis for the figures above are presented throughout the book. MINI GRIDS FOR HALF A BILLION PEOPLE    xxv MINI GRIDS BY THE NUMBERS, continued Key performance indicators for the mini grid industry 2018 2021 2025* Reducing cost (levelized cost of energy [$/kWh] of a best-in-class solar $0.55/kWh $0.38/kWh $0.30/kWh hybrid mini grid) Pace of deployment (mini grids built per key access-deficit country 20–75 150 450 per year) mini grids mini grids mini grids Quality of service (industry-wide standard for reliability of electricity 90–97 percent 99 percent 99 percent supply) uptime uptime uptime Access to finance for mini grids designed to boost access to energy $13 billion $16 billion $25 billion (total cumulative investment) Establish enabling environments (average RISE score for mini grids 59/100 64/100 75/100 framework in top 20 electricity access-deficit countries) Note: * projection with business-as-usual scenario. Mini grid industry progress across all 10 frontiers / building blocks 2018 2021 2025* Reducing costs and optimizing design and innovation for solar mini grids Planning national strategies and developer portfolios with geospatial analysis and digital platforms Transforming productive livelihoods and improving business viability Engaging communities as valued customers Delivering services through local and international companies and utilities Financing solar mini grid portfolios and end user appliances Attracting exceptional talent and scaling skills development Supporting institutions, delivery models, and champions that create opportunities Enacting regulations and policies that empower mini grid companies and customers Cutting red tape for a dynamic business environment Note: * projection with business-as-usual scenario. Dark green = magnitude change has been achieved; light green = irreversible progress towards magnitude change; yellow = needing attention; orange = no significant activities to date. xxvi   MINI GRIDS FOR HALF A BILLION PEOPLE MAIN FINDINGS THE NEW ELECTRICITY ACCESS with 2,500 to 10,000 residents. Finally, for nearly 3,000 settlements, each with 10,000 to 100,000 people, custom LANDSCAPE sizing of mini grids might be more suitable. To achieve Sustainable Development Goal 7 (SDG 7), Internal analysis by the World Bank team based on Multi- 930 million people will have to obtain an electricity con- Tier Framework (MTF) data suggests that users in these nection between 2022 and 2030 (IEA 2021). In 2020, load centers spend on average $5–$20 per month on the global electrification rate reached 91 percent, with alternative forms of energy such as candles, kerosene, the number of people without access dropping to around dry-cell batteries, car batteries, and petrol and diesel fuel 733 million—compared with around 1 billion people in for stand-alone gensets. The introduction of innovative 2016 and 1.2 billion in 2010 (IEA, World Bank, and others technologies in the marketplace (like solar home systems 2022). Nonetheless, the pace of electrification has slowed or mobile phones) has taught us that these new solutions in recent years. Between 2010 and 2018, an average of 130 need to be more than a little better than the current alter- million people gained access to electricity annually. From native. They need to be much better. Why else would con- 2018 to 2020, this number shrank to 109 million per year. sumers take the risk of changing their behaviors? For these While the slowdown is attributed in part to the difficulties in clusters of clients, the service provided by the solar mini reaching the remotest and most vulnerable populations, it grids should be a reliable source for their consumptive activ- was compounded by the devastating effects of the COVID- ities like lighting, charging, and radio/TV. More than that, 19 pandemic. If current policies and efforts are not ramped they need to provide for life-changing productive activities up, only 260 million people are anticipated to be electrified within the current monthly expenditure of $5–$20. From between now and 2030 (IEA 2021), and an estimated 670 the end user’s perspective, a $5–$20 monthly expenditure million people are projected to remain without access, with should cover the cost not only of reliable electricity but also 9 out of 10 of them likely to live in Sub-Saharan Africa (IEA, of transitioning to (and purchasing) electric appliances. So World Bank, and others 2022). over the lifetime of the technology, monthly payments of In Sub-Saharan Africa, nearly 291,000 population clus- about $3–$15 cover the cost of electricity, while monthly ters have profiles favoring the deployment of solar payments fall in the range of $2–$5 for appliances. These mini grids. That is, they are located more than 1km from costs pose a challenge for the mini grid industry if it is to the existing grid network and have a population density fulfil its full market potential. (>1,000 people/km2) that favors decentralized sys- Countries with a comprehensive approach involving tem deployment. More specifically, analysis conducted main grid extensions, mini grids, and solar home sys- internally by the World Bank team—based on spatial tems have achieved the fastest results in electricity distribution of digitalized settlements (GRID3, CIESIN), access (IEA, World Bank, and others 2022). Strong lead- grid network (Arderne C. et al—GridFinder) and popula- ership, supporting policies, and more private financing will tion (WorldPop) over the region—shows that more than be required if electricity access is to reach the remaining 177,000 settlements have a population of 100 to 500 unserved people—including those that depend on frail, people. These settlements could be powered by smaller overburdened urban grids and displaced people and those solar mini grids of up to 20 kilowatts (kW) each. Nearly living in hard-to-reach locations. In Sub-Saharan Africa, 96,000 settlements, each with populations of 500 to electricity services are delivered to end users by 60 utili- 2,500 people, could be powered by medium-sized solar ties, more than 80 solar mini grid companies, and almost mini grids of up to 80 kW. The larger solar mini grids, up to 90 main solar stand-alone-system companies (Balaban- 200 kW, could power more than 15,000 settlements, each yan and others 2021; GOGLA 2022). MINI GRIDS FOR HALF A BILLION PEOPLE    1 Mini grids are not a new phenomenon: nearly all cen- either because the main grid is unreliable or to boost resil- tralized electricity grid systems began as isolated mini ience in the face of climate shocks or severe weather. With grids that were connected to each other over time. more than 160,000 mini grids needed, Sub-Saharan Africa This first generation of mini grids was pivotal to the early accounts for the largest share of mini grids and investment development and industrialization of most modern econ- required to achieve universal access, at a cost of $91 billion omies, including Brazil, China, Denmark, Italy, the Nether- to connect 380 million people. These projections are based lands, Spain, Sweden, the United Kingdom, and the United on country-specific scenarios for the 58 countries with data States. Mini grid systems introduced in the late nineteenth in the Global Electrification Platform (GEP) and ESMAP and early twentieth centuries can be described as the first estimates for countries not included in the GEP. Meanwhile, generation of mini grids. Today a second generation of mini ESMAP estimates that resilience- and renewable-moti- grids is widespread in many low-income countries. These vated mini grids could serve an additional 2–3 million new systems are typically small and isolated, powered by diesel connections globally (serving 6–7 million people) per year, or hydro, and built by local communities or entrepreneurs or the equivalent of 10–15 cities or small regional utilities primarily to provide rural households with access to elec- per year deciding to strengthen their power systems by tricity, especially in areas not yet served by the main grid. developing interconnected micro/mini/metro grids. Tens of thousands of these systems were built, starting in In 2021, the global mini grid market consisted of more the 1980s and ramping up through the 1990s and early than 50,000 installed and planned mini grids in more 2000s. Many of these systems were overtaken by the than 130 countries. Although there is a clear trend toward national grids; the ones that still exist are now prime can- solar as the dominant technology, the overall pace of mini didates for hybridization with solar photovoltaic (PV) sys- grid development is not on track to achieve the 2030 mini tems to reduce the fuel cost. grid market potential. ESMAP identified 21,557 mini grids in Over the past few years, a third generation of solar mini 131 countries and territories, serving more than 48 million grids has emerged. These mini grids, mostly solar PV people. Most of these systems are first- and second-gen- hybrids, are owned and operated by private companies eration mini grids, and approximately half of installed mini that leverage transformative technologies and innovative grids are powered by solar, with hydro and fossil fuels strategies to build portfolios of mini grids instead of one-off accounting for an additional 35 percent and 10 percent, projects. The typical third-generation mini grid is grid-in- respectively. Another 29,353 mini grids are planned for terconnection ready. It also uses energy management development in 77 countries and territories, of which 99 systems, prepay smart meters, and the latest solar hybrid percent will be powered by solar. The trend toward solar has technologies. This third-generation mini grid also incor- been accelerating: more than 10 times as many solar mini porates energy-efficient appliances for productive uses of grids were built per year from 2016 to 2020 than fossil fuel electricity into its business model. These mini grids operate mini grids. Meanwhile, from 2010 to 2014, by comparison, in more favorable business environments, taking advan- about three times as many solar mini grids were built per tage of cost reductions in the latest mini grid component year than fossil fuel mini grids. This is a major acceleration technologies and regulations developed specifically for in solar and deceleration in fossil fuels. But the annual pace private-sector investment. Developers of third-generation of mini grid development worldwide—averaging between mini grids are joining industry associations to speak with 1,300 and 1,900 between 2010 and 2021—would see one voice and drive policies and regulations that favor pri- only 44,800 mini grids serving 80 million people at a total vate-sector investment. investment cost of $37 billion by 2030. This is well short of the 430 million people that could be served at least cost by The Energy Sector Management Assistance Program mini grids in order to achieve universal access. (ESMAP) analysis indicates that a combination of fall- ing costs for key components, the introduction of new Year-on-year gains needed to achieve universal access digital solutions, and early signs of favorable economies will require scaling up private-sector-led mini grid deploy- of scale, has made solar mini grids an option to connect ments from tens to hundreds to thousands of mini grids 490 million people by 2030. Achieving universal access per country per year in each of the top 20 countries with to electricity will require the construction of more than the highest electricity access deficit rates today. Exam- 217,000 mini grids by 2030 at a cumulative investment ples showing this exponential growth are with the introduc- cost of almost $127 billion. Of these totals, mini grids are tion of mobile phones, solar home systems, and electric the least-cost option for 430 million people who would gain vehicles, where the private sector, supported by public poli- access to electricity for the first time at a cost of about $105 cies, provides superior products. Does this mean that more billion. For about $22 billion, an additional 60 million peo- public-sector-led programs cannot be beneficial? Not at ple, mostly in middle- and high-income countries, could be all: these programs provide great benefits to the coun- serviced through an interconnected network of mini grids try and the end users. Yet when one must attain universal 2   MINI GRIDS FOR HALF A BILLION PEOPLE access by 2030, private-sector-led programs should be the The industry is ahead or on track to achieve most of the dominant initiative, across the board, in a country or region. key performance indicators (KPIs), but it lags in terms of number of mini grids installed per key energy access Overarching sector performance indicators and targets deficit country per year, and total cumulative invest- can help benchmark the sector. It is within this context ment. The overall cost of the delivery of electricity services of market dynamics as well as through a collaborative, by mini grids has plunged since 2018, from a levelized cost iterative process, that ESMAP and mini grid industry lead- of energy (LCOE) of a best-in-class solar hybrid mini grid ers—including the Africa Minigrid Developers Association equal to $0.55/kilowatt-hour (kWh) to $0.38/kWh in 2021, (AMDA) and development partners—jointly identified five compared with the $0.45/kWh target for 2021. The quality market drivers and associated targets that will set the sec- of mini grid electricity services is also ahead of pace, with tor on a trajectory to achieve universal electrification and AMDA members achieving uptimes of around 99 percent in its full market potential (table MF.1). 2021 compared with 90–97 percent in 2018 and the 2021 These targets are ambitious but achievable if 10 building objective of 97 percent uptime. A continuation of delivery blocks are in place at the national level. Looking through of high-quality services has improved the average load fac- the lens of innovation and the impact that, for example, dig- tor, from around 22 percent in 2018 to 30 percent in 2021, itization and technological advancement can bring to the ahead of the 2021 objective of 25 percent. Enabling envi- solar mini grid sector, our analysis identified 10 areas that ronments have also improved and are ahead of pace, with stood out where notable magnitude-level improvements the Regulatory Indicators for Sustainable Energy (RISE) can be expected to reach the abovementioned targets of score for mini grids in the top 20 access-deficit countries cost (C), pace (P), quality (Q), finance (F), and enabling rising from 59/100 in 2018 to 64/100 in 2021, on pace to environment (EE). These are identified in table MF.2. achieve 90/100 by 2030, compared with 80/100 as the TABLE MF.1 • Market drivers and 2030 targets Market Driver 2030 Target 1. Reduce the cost of solar hybrid mini grids. $0.20/kilowatt-hour (kWh).  ncrease the pace of deployment through a portfolio 2. I Building around 2,000 projects per key access-deficit country per year approach to mini grid development. by 2030. 3. Provide superior-quality service. Achieving industrywide average uptime of more than 97 percent and industrywide average load factor of 45 percent.  everage development partner funding and government 4. L Attracting approximately $127 billion of investment from development investment to “crowd in” private-sector finance. partners, governments, and the private sector, of which $105 billion for energy access mini grids.  stablish enabling mini grid business environments in 5. E Raising the average Regulatory Indicators for Sustainable Energy (RISE) key access-deficit countries. score in the top 20 electricity-access-deficit countries to 80 out of 100. TABLE MF.2 • Building blocks and identified areas for potential magnitude change Building Blocks / Identified Areas for Potential Magnitude Change Progress 2018–21 2. Planning national strategies and developer portfolios with geospatial analysis and digital platforms (C, P, F, EE) 1. Reducing costs and optimizing design and innovation for solar mini grids (C, P, Q) 9. Enacting regulations and policies that empower mini grid companies and customers (C, P, EE) 5. Delivering services through local and international companies and utilities (C, P, Q, F) 6. Financing solar mini grid portfolios and end-user appliances (P, F, EE) 8. Supporting institutions, delivery models, and champions to create opportunities (C, P, EE) 3. Transforming productive livelihoods and improving business viability (C, Q) 4. Engaging communities as valued customers (C, P, Q) 7. Attracting exceptional talent and scaling skills development (C, P, Q, F, EE) 10. Cutting red tape for a dynamic business environment (C, P, EE) Dark green = magnitude change has been achieved; light green = irreversible progress toward magnitude change; yellow = needs attention. C= cost; EE = enabling environment; F = finance; P = pace; Q = quality. MINI GRIDS FOR HALF A BILLION PEOPLE    3 2030 target. But while the industry has seen a shift from ments. The most progress has been made in geospatial the deployment of mini grids on an individual pilot basis to planning and the costing, design, and innovation of solar their deployment by service providers in portfolios of 5 to hybrid mini grids. The most work, however, is needed to sup- 10 mini grids per month in 2021, the pace across the indus- port these key countries and regions, and others, in commu- try is still behind the 200 mini grids per country per year in nity engagement, scaling up private sector participation and 2021 needed to be on track for achieving 2,000 mini grids utilities to deploy mini grids, and skills development. per country per year in 2030. In addition, while cumulative investment in mini grids for energy access rose from about $13 billion in 2018 to $16 billion in 2021, this is well behind BUILDING BLOCKS 1 THROUGH the $20 billion needed to be on pace to achieve $105 billion cumulative investment in energy access mini grids by 2030. 10 AND FRONTIERS FOR SECTOR GROWTH: Creating the Environment Over the past three years, notable progress has been evident across all building blocks. Most of the advances for Takeoff of Mini Grid Portfolios were seen in the geospatial portfolio planning and workable Building blocks 1 through 10 and the frontiers for sector regulations. Solid progress has also been evident in tech- growth are described in the sections that follow. nology and costing in the private sector and utilities, along with greater access to finance and supporting institutions. BUILDING BLOCK 1. Even though we see advances with productive uses and SOLAR MINI GRID COSTS, DESIGN, AND community involvement (and on attracting exceptional tal- INNOVATION ent and reducing red tape), these are the very areas where transitions must emphasize scale on the ground. For more Solar mini grids consist of specialized components for details on progress by building block, see each of the chap- the generation, distribution, metering, and consumption ters covering these topics. of electricity (figure MF.1). A typical third-generation mini grid comprises a solar hybrid generation system made A cohort of partners has begun to track building blocks up of solar panels, batteries, charge controllers, inverters, by country for in-country coordination and readiness and diesel backup generators. The distribution network to scale. For key access-deficit countries and regions in consists of poles and low-voltage wires; larger mini grids Sub-Saharan Africa, we see that Ethiopia, Kenya, and Nige- sometimes also have medium-voltage systems. Third-gen- ria have taken more steps toward achieving magnitude eration mini grids often use smart meters offering both changes across the 10 building blocks than the other coun- prepaid payment options for consumers and real-time, tries and regions (table MF.3). The Democratic Republic of granular information about energy consumption patterns Congo and the Sahel, despite being large potential markets and system performance. They also use remote-moni- for mini grids, need major support on almost all 10 building toring systems that allow operators to identify technical blocks to prepare the market for scaling up mini grid deploy- TABLE MF.3 • Sub-Saharan African mini grid markets and their progress across the 10 building blocks Building block DRC Ethiopia Kenya Nigeria Sahel 1. Costing, design, and innovation 2. Geospatial planning 3. Income-generating appliances and machines 4. Community engagement 5. Companies and utilities 6. Access to finance 7. Skills development 8. Institutional setup and business models 9. Regulations and policies 10. Cutting red tape Source: ESMAP analysis. Dark green = magnitude change has been achieved; light green = irreversible progress toward magnitude change; yellow = needs attention; orange = no significant activities to date. DRC = Democratic Republic of Congo. 4   MINI GRIDS FOR HALF A BILLION PEOPLE FIGURE MF.1 • A mini grid system (part A) and a containerized solar mini grid (part B) A AC load in village AC appliances Poletop hardware Smart meter Generator Distribution line AC bus Residential Smart meter AC/DC inverter Service drop Pole PV array Charge Battery block Commercial controller Smart Solar-hybrid generation system Distribution system meters Efficient productive loads AC = alternating current; DC = direct current; PV = photovoltaic. B solar battery system only, solar-biomass-based systems, AC (alternating current) or DC (direct current) systems, and high digital solution integration. The Levelized Cost of Energy of a Solar Mini Grid In 2021, the LCOE in a best-in-class, third-generation mini grid was $0.38/kWh at a 22 percent load factor, or a 31 percent reduction from 2018. This trajectory is fueled by the falling expenditures for preparation, capital, and Source: © SustainSolar. Used with permission by SustainSolar. Further operations, combined with more income-generating uses permission required for reuse. of electricity and more efficient economies of scale. The combination of expected cost reductions and higher load issues before they affect energy services and rectify prob- factors (from 22 percent to 40 percent) caused by produc- lems quickly and inexpensively, thus improving the quality tive use is expected to bring the LCOE of third-generation of customer service. Many developers of third-generation mini grids to $0.20/kWh by 2030 (table MF.4). mini grids encourage and incentivize customers to use effi- Preparation costs have been reduced by more than an cient household appliances as well as efficient machines order of magnitude to $2,300 per mini grid; however, and equipment for income-generating activities, and pro- there is limited progress in efficiency. The introduction of vide or facilitate access to financing options to help cus- geospatial and other digital technologies have decreased tomers manage upfront costs. the cost of preparation and planning by an order of magni- Smaller solar mini grids with an installed capacity of about tude. In the past, the unit cost per site was more or less the 100 kW or less are more and more standardized, ranging same, irrespective of the number of sites—about $30,000 from prefabricated components to containerized mini per site—because each one required a high level of on-site grids. Larger systems with an installed capacity of more analysis. Today, portfolios of mini grids can be prepared to than 250 kW remain designed and delivered on an individ- the point where they are ready for full feasibility assessment ual system basis. Irrespective of the installed capacity, ser- and community engagement at a cost of about $2,300 per vice providers have chosen their mini grid design around site in 2021, based on the World Bank’s recent experience one main technical and business approach, for example, in Ethiopia, Nigeria, and South Sudan. MINI GRIDS FOR HALF A BILLION PEOPLE    5 TABLE MF.4 • The levelized cost of energy by load second-generation mini grids today have a load factor of factor, 2018, 2021, and 2030 around 22 percent, indicative of low levels of income-gen- erating uses of electricity. However, third-generation mini Levelized cost of energy (US$/kWh) grids provide high-quality, reliable electricity services that Load factor (percent 2018 2021 2030 can support income-generating loads, such as agricul- 22 0.55 0.38 0.29 tural milling. If mini grids can achieve a 40 percent load 40 0.42 0.28 0.20 factor through strong daytime consumption by local busi- Source: ESMAP analysis. nesses and commercial clients, the costs of producing Note: The 2018 LCOE data are for a best-in-class 294-kWfirm solar hybrid electricity drop 25 percent compared with a load factor mini grid in Bangladesh serving more than 1,000 customers (more than of 22 percent. For an 80 percent load factor—achieved 5,000 people). LCOE data for 2021 are based on a representative mini by inclusion of a water pump with storage tank and an grid synthesized from average costs and consumption levels in three mini grids in Myanmar, Nigeria, and Ethiopia commissioned in 2020 or 2021. anchor load, such as a telecommunications tower—LCOE The 2030 LCOE is for a “best-in-class” mini grid based on projected com- reduction is 37 percent. ponent costs in 2030. A detailed description of the underlying analysis is provided in chapter 1. kWh = kilowatt-hour. Implications for national power sectors As a result of declining LCOE, increasing income-generat- ing uses of electricity, and the mainstreaming of geospa- A best-in-class solar hybrid mini grid costs about $3,700/ tial planning, solar mini grids can have transformational kWfirm,1 and the falling trend is expected to continue effects on power sectors. They are on track to provide through 2030, bringing capital expenditure (CAPEX) to power at lower cost than many utilities by 2030. At $0.40/ below $2,500/kWfirm. Components used for generating kWh, mini grid LCOE would be less than the LCOE of national and distributing electricity account for 66 percent of total utilities in 7 out of 39 countries in Africa. At $0.20/kWh, mini capital costs. The components with the largest share of grid LCOE would be less than the LCOE of national utilities in overall CAPEX were batteries (15 percent). PV modules 24 African countries (Trimble and others 2016). This would (10 percent), inverters/energy management systems (9 make mini grids the least-cost solution for grid-quality elec- percent), and distribution grids (poles, wires; 27 percent). tricity for more than 60 percent of the population in Africa in Meanwhile, component costs vary widely across countries a scenario assuming that national utilities do not dramati- and regions, mostly as a result of a combination of taxes cally change their operations—with major implications for and duties, differences in margins charged by wholesalers the allocation of both public and private investment funds. and distributors, and other costs incurred in doing busi- However, scaling up mini grids does not mean scaling back ness that vary from country to country. Downward trends the main grid. On the contrary, solar mini grids enhance the in component costs mean that the up-front investment economic viability of expanding the main grid. By designing cost of solar and solar hybrid mini grids fell from about the system from the beginning to interconnect with the $8,000–$10,000/kWfirm in 2010 to $3,900/kWfirm in main grid and by promoting income-generating uses of 2018 and less than 3,700/kWfirm in 2021. Looking ahead, electricity through effective community engagement and the expected decreases in component costs associated training, third-generation mini grids can provide early eco- with current best practices can reduce up-front investment nomic growth, so that significant load already exists by the costs to less than $2,500/kWfirm by 2030. time the main grid arrives, and customers have a greater Mini grid operating expenditure (OPEX) averages around ability to pay. New regulatory frameworks give developers $80 per customer per year. Costs are expected to decline viable options for what happens when the main grid arrives, because of technological advances over the next decade. and reductions in the cost of components enable develop- Staff costs on average account for 76 percent of operations ers to build grid-interconnection-ready systems, while still costs, but economies of scale and new remote-controlled, keeping tariffs affordable. prepay smart meters and remote-monitoring technologies Supporting solar mini grids therefore goes hand in hand have slashed labor costs per mini grid. Replacement costs with strengthening the power sector. Interconnecting have also fallen as more developers invest in lithium-ion third-generation mini grids with the main grid can increase (Li-on) batteries, which have about twice the number of the resource diversity and overall resilience and efficiency charging cycles before failure compared with conventional of the power system. However, this presents a couple of lead-acid batteries, and the costs of power electronics, such operational challenges that are better addressed in a com- as PV inverters and battery inverters, are also decreasing. prehensive strategy for developing the sector, for example, Further cost reductions per kWh are derived from for governments through their electrification strategies increasing income-generating uses of electricity, which to allow for utilities, mini grids, and off-grid companies to can decrease the LCOE by 25 percent or more. Most deliver services in the country, as well as for utilities to be 6   MINI GRIDS FOR HALF A BILLION PEOPLE able to introduce the practical technical functions to sup- turning to implementation of these plans. For example, sev- port power system operations and planning with multiple eral national electrification plans have incorporated buffer mini grids connected to the distribution grid, such as short- zones for grid extension (for example, 15 kilometers from and long-term forecasting and other procedures. the existing grid), which during implementation has created situations where the utility is delayed in certain geographical Experiences with interconnected mini grid collabora- areas and the mini grid and solar companies are not allowed tions are emerging, for example, in Nigeria and India, to sell products and services because they are prohibited to and are providing valuable lessons. These interconnected do so under the plan. It is important that during the prepa- mini grids are built to serve different market segments: ration of these plans, the different stakeholders are carefully rural and peri-urban towns and villages, large urban mar- consulted so that the underlying assumptions are based in ketplaces, commercial and industrial (C and I) installa- reality. Countries that are using advanced geospatial analy- tions, and separate urban residential communities. Early sis to develop national electrification plans include Angola, evidence seems to indicate that these interconnected mini Cambodia, the Democratic Republic of Congo, Ethiopia, The grids can create “win-win-win” economic outcomes for the Gambia, Haiti, India, Kenya, Liberia, Mozambique, Myanmar, three key parties. The arrangement can eliminate or reduce Nigeria, Pakistan, Rwanda, Sierra Leone, Somalia, Tanzania, financial losses for distribution companies (DISCOs) that Togo, Uganda, and Zambia, among others. are forced to sell electricity at non-cost-recovering retail tariffs. Interconnections also allow DISCOs to earn new Geospatial analysis is also being used as part of a port- revenues through bulk power sales to the mini grid as well folio planning approach for mini grid development. This as rental revenues from the leasing of some or all of the would complement a comprehensive national least-cost DISCO’s existing distribution system to the mini grid. For electrification planning framework or, in the absence of such the mini grid operator, a physical connection to the contig- a framework, identify portfolios of mini grid sites where grid uous DISCO offers the possibility of purchasing bulk power, extension is expected to be limited or unlikely because of whether on a firm or an “as available” basis from the inter- political considerations, insolvency of the DISCOs, and so connected DISCO or an upstream supply source. This can forth. Geospatial portfolio planning, which is already being lead to lower operating and capital costs (for example, a used by a number of established mini grid companies, util- lower LCOE) for the interconnected mini grid than if oper- ities, and governments in Sub-Saharan Africa, slashes the ates in a pure stand-alone mode. And for the mini grid’s preinvestment cost associated with preparing sites for mini customers, this should lead to lower tariffs than would be grid development compared with traditional approaches, possible than if the mini grid operated in a totally isolated which rely on the deployment of multidisciplinary teams to mode. Finally, it is well documented that mini grids, whether villages to explore the scope for mini grid electrification. interconnected or isolated, routinely achieve high levels of reliability for their customers than DISCOs do for theirs BUILDING BLOCK 3. (Tenenbaum, Greacen, and Shrestha 2022 forthcoming). Transforming productive livelihoods and improving business viability BUILDING BLOCK 2. Because of their reliability, third-generation mini grids Planning national strategies and developer can support income-generating uses of mini grid elec- portfolios with geospatial analysis and digital tricity, which creates an everyone-wins scenario for mini platforms grid developers, rural entrepreneurs, communities, and Countries are using geospatial analysis to develop national utilities over time. Increasing income-generating national electrification plans that delineate areas for uses of electricity reduces the LCOE (see table MF.4), which mini grids. Through a geospatial approach to national elec- increases the developer’s margins and therefore financial trification planning, the existing grid network is mapped and viability. Entrepreneurs and small businesses benefit from its attributes are digitalized. The supply of and demand for switching from expensive diesel generators to affordable electricity are geolocated and overlaid with supporting data, mini grid electricity. In one of the most comprehensive including demographic, social infrastructure, and economic assessments of productive-use appliances and equipment data. Spatial modeling then delivers a least-cost plan that to date, ESMAP identified more than 130 machines and identifies the optimal ranges for grid, mini grid, or off-grid appliances available today that had payback periods of less technologies. Even though these national plans now typically than 12 months. Communities benefit from the jobs cre- include all options for electrification—grid extension, mini ated and increased economic activity. The growth of rural grids, and off-grid—they still rely on chosen input assump- economies also benefits national utilities once intercon- tions that can result in a more advantaged position of one nection to the main grid is considered, because it increases solution over the other. Furthermore, the chosen param- customers’ ability to pay higher tariffs and creates a strong eters can also exclude large groups of customers when base of demand for electricity. MINI GRIDS FOR HALF A BILLION PEOPLE    7 But demand uncertainty remains a key area of risk for end user financiers, and mini grid companies. Road shows both developers and financiers. In ESMAP surveys of are the next step, where mini grid developers, appliance mini grids presented in this book—the detailed survey of suppliers, end user financiers visit load centers to explain more than 400 mini grids in Africa and Asia (see chapter the value propositions to potential end users. The final step 1), the high-level survey of installed and planned mini grids is the roll-out of mini grid connections, sales of appliances globally (see the overview), and the detailed nationally and end user finance. representative surveys of mini grid operators (see chap- A number of digital tools are emerging that also allow for ter 5)—demand per customer varied widely from one mini a more efficient and lower-cost planning and rollout of grid to another, and from one country to another. Most mini productive uses activities in conjunction with the arrival grids had demand per customer of between 5 and 35 kWh of electricity from mini grids. The above-mentioned geo- per month, but all else held equal there is a sevenfold differ- spatial tools that help to identify and prioritize mini grid ence in revenue expectations from customers consuming portfolios are now also used to share the associated mar- 5 kWh per month and those consuming 35 kWh per month. ket intelligence with the appliance providers and end user Developers use demand estimates as key inputs not only financiers. The information supports these companies to to inform the designs of their mini grids, but also to secure make an informed decision if their products have a suf- external financing. The uncertainties around future demand ficient addressable market. The tools also support the growth therefore represent a key risk area for both develop- mini grid developers, appliance providers, and end user ers and financiers. Mitigating this risk requires concerted financiers to coordinate visits to these communities, so efforts to increase the daytime use of income-generating that their collective, potential clients can learn how elec- appliances and machines, and financing mechanisms that tricity with the appropriate, affordable appliances can help de-risk some of the demand uncertainty. alter their lifestyle and business prospects for the better. Increasing the uptake of productive-use equipment Often nongovernmental organizations and social change requires access to approximately $3.6 billion in afford- organizations operate in these peri-urban, rural areas and able consumer finance and a proactive involvement of can play an important role in coordinating these efforts these financiers and appliance providers. Assuming an on the ground. average up-front cost of $1,200 and 15 appliances per mini grid for 200,000 new mini grids by 2030, approx- BUILDING BLOCK 4. imately $3.6 billion in microfinance will be needed for Engaging communities as valued customers the purchase of 3 million productive-use appliances by Community engagement strategies can help increase 2030. Although they have relatively high up-front costs, productive uses of electricity and stimulate demand for most productive-use appliances and equipment provide mini grid services. Experience from successful mini grid opportunities to generate or increase revenue. Financing developers indicates that community engagement begins the up-front purchase cost of the appliances—by the mini by raising awareness before moving to adoption, produc- grid operator via on-bill financing or by a third party, such tive operation, and word-of-mouth marketing. Community as a microfinance organization—is a good way to increase engagement requires a flexible approach; a clear under- productive uses of mini grid electricity. Both financing standing of the local socioeconomic and cultural charac- pathways have benefits and drawbacks for the mini grid teristics; and tailoring of promotional tools, materials, and operator, and both require the operator to develop new channels.2 business model capabilities. The benefits of prioritizing access to female-led house- Drawing from existing research and the World Bank’s holds and small businesses and increasing the partici- recent experience with productive uses programs across pation of women in management positions in mini grid Africa, Asia, and Latin America and the Caribbean, businesses are clear. Mini grids can greatly boost women’s ESMAP has identified six steps to roll out initiatives productivity, particularly in labor-intensive agricultural and that support the uptake of income-generating appli- food processing activities that women dominate. Women ances in towns served by mini grids. Step one is a mar- are 9–23 percent more likely to gain employment outside ket/demand assessment with geospatial analysis overlying the home following electrification (Smith 2000). Electrifica- mini grids, appliances, and end use finance. Step 2 is com- tion lowers fertility levels, through greater exposure to tele- munity engagement to confirm and improve data collected vision (Buckley 2012). Electrifying health clinics for lighting during Step 1 through survey(s) and workshops. Step 3 is and the refrigeration of medication is especially beneficial a demand analysis for mini grid design and market poten- for maternal health. Mini grid projects can create jobs for tial for appliances and associated end user finance. Step women while shaping new community decision-making and 4 is preparation of roadshows involving local government, leadership models by placing women in leadership roles. community leaders, interested appliance providers and 8   MINI GRIDS FOR HALF A BILLION PEOPLE Innovations in community engagement are emerging centers for mini grid components will be solar PV, battery that can reduce costs and improve effectiveness. One storage, and distribution infrastructure and technologies example from a few years ago was the smartphone app like smart meters. As the costs of solar PV and battery and accompanying online YouTube-like platform called Mini storage continue to fall, the fraction of energy produced by Grid Stories, developed by Quicksand Design Studio with solar PV and batteries will approach 100 percent, resulting support from ESMAP. Following simple on-screen instruc- in the profit potential for diesel dropping to nearly zero over tions, mini grid customers and staff of mini grid companies the next decade. ESMAP analysis also indicates a profit used the free smartphone app to create short videos—on potential for mini grid developers that could exceed $3.3 how a customer uses electricity in her small business, for billion on an annual basis for all third-generation mini grids example—and uploaded them to a Mini Grid Stories web- deployed between 2022 and 2030. It is important to note site, where the videos could be viewed, shared, and down- that financial support packages, including subsidies from loaded. The approach was inspired by the success of the governments and development partners, will be needed to agricultural web-based platform Digital Green, which uses unlock this profit potential, particularly over the next few videos for agricultural extension work, which was 10 times years to set the market on the trajectory of rapid scale-up. more cost efficient than traditional community engage- Public funds enabled high-income countries to achieve uni- ment services on a cost-per-adoption basis (Abate and versal electricity access; the same will be true for electricity others 2018). Another example is Smart Power India (SPI), access-deficit countries today. supported by the Rockefeller Foundation. This India-based, Even in countries in which the government leads mini Indian-led organization intermediates between key stake- grid development, the private sector is a key partner in holders, including developers, national and local govern- mini grid initiatives. Public-private partnerships are often ment entities, and community organizations (Rockefeller an effective way of distributing responsibilities to optimize Foundation 2017). SPI’s approach is called “Community government and private-sector capacities. They enable Engagement, Load Acquisition and Micro-enterprise Devel- mini grid operators that do not have substantial financial opment” (CELAMeD). With SPI support, developers have resources to enter the market. In addition, major opportu- crafted communication and marketing strategies to inform nities for partnership between local and international firms consumers about the benefits of renewable energy and exist across the mini grid industry value chain. Local enti- catalyze the growth of rural businesses (SPI 2017). ties are best positioned to focus on the aspects of the value BUILDING BLOCK 5. chain that require knowledge of local rules and regulations Delivering services through local and international or require coordination with the customer being served by companies and utilities the mini grid; international companies are best suited to perform tasks that can be replicated across geographic Connecting 490 million people by 2030 will require utili- boundaries. Recent local-international partnership agree- ties and private companies to develop and operate more ments include Caterpillar and Powerhive in Africa, ABB and than 210,000 mini grids. National utility companies in Husk Power in India, Mitsui and OMC in India, ENGIE and Kenya, Madagascar, the Philippines, Russia, and many other Mandalay Yoma Energy in Myanmar, and Schneider Elec- countries are already important developers of mini grids. tric with both EM-ONE and GVE in Nigeria. Private-sector developers—including Tata Power Renew- able Microgrids, Engie Energy Access, Havenhill, PowerGen, Industry associations can facilitate collaboration and OMC Power, Green Village Electric (GVE), and Husk Power, deal making between local and international entities. among many others—are developing large portfolios of mini AMDA comprises more than 40 developers, each operating grids. In a well-established market, private-sector-led initia- a portfolio of commercially viable mini grids in Sub-Saha- tives have a better chance of reaching exponential growth— ran Africa. AMDA helps its members present a unified voice something that is needed to reach universal access by 2030. and facilitates deals between developers and suppliers. By National utilities—including the Ethiopian Electric Utility collecting data from their members, associations can pres- (EEU), the Kenya Power and Lighting Company (KPLC), ent data-driven opportunities to investors as well as suppli- and Engie—also see an expanding role for mini grids based ers of specialized products and services. on their organizational cost-benefit analysis.3 BUILDING BLOCK 6. The mini grid industry offers major profit potential to Financing solar mini grid portfolios and end user private-sector equipment and service suppliers and appliances developers alike, but financial support packages are Private investors—both domestic and international— needed to unlock this potential. ESMAP analysis projects are financing third-generation mini grids and driving that the annual profit potential across the mini grid value innovation in financing mechanisms. Private financiers chain will be almost $5.8 billion by 2030.4 The largest profit invested more than $500 million in developers building MINI GRIDS FOR HALF A BILLION PEOPLE    9 mini grids in low-income countries between 2012 and may require an expanded support package, as women 2022, according to ESMAP’s analysis of publicly available often face additional barriers to accessing finance. data on more than 100 unique deals between developers Performance-based grants have become a mainstream and investors. Impact investors and commercial inves- subsidy mechanism, and can greatly lower the cost of tors, as well as local and national banks, have developed mini grid electricity to allow mini grid services to be equity, debt, and blended finance options to help devel- affordable to a larger group of end users. According to an opers scale up their mini grid business. Acumen, Bamboo ESMAP analysis, a 40 percent capital cost grant reduces Capital Partners, CrossBoundary Energy Access, ElectriFi, the LCOE of a best-in-class third-generation mini grid from InfraCo Africa, and Shell Foundation are just a few exam- $0.38/kWh to $0.28/kWh in a scenario with very low pro- ples of recent investors in mini grids. ductive uses of electricity. In scenarios where productive Development partners, including the World Bank, have uses increase the mini grid’s load factor to 40 percent, the increased funding for mini grids, from millions of dol- same 40 percent capital cost grant reduces the LCOE from lars in the 2000s to billions of dollars in 2018. A group $0.28/kWh to $0.22/kWh. of 15 major international donors and development part- Performance-based grants for mini grids based on a ners, including the World Bank, has collectively commit- percentage of the developer’s cost to connect new ted approximately $2.6 billion just to mini grid investment customers are often less than the implicit or explicit (that is, excluding funding for technical assistance and subsidy that the main grid receives for each new con- research). The World Bank has committed more than nection. A survey of 39 national utility companies in Africa $1.4 billion to mini grids over the next five to seven years, showed that utilities received explicit or implicit subsi- through 50 projects in 42 countries (41 projects approved dies that enabled them to sell electricity at prices that by the World Bank Board and at least 9 under preparation). were on average 41 percent—and up to 80 percent—less The investment plans of this portfolio include the deploy- than the utilities’ unsubsidized LCOE (Trimble and others ment of 3,000 mini grids by 2027, with the expectation 2016; Kojima and Trimble 2016). This would indicate that of bringing electricity to more than 11 million people. This many national utilities in Africa receive implicit subsidies investment commitment is expected to crowd in close to $1 that are more than 40 percent of the connection cost. billion of cofinancing from private-sector, government, and With national utility connection costs often exceeding development partners. $2,000 in rural areas (Trimble and others 2016; Blimpo In countries where the World Bank has an investment and Cosgrove-Davies 2019), it is therefore likely that many commitment in mini grids, the Bank’s investment rep- national utilities in Africa receive implicit cost subsidies in resents on average about 25 percent of the total invest- excess of $800 per connection. To put this in perspective, ment in mini grids in each country from governments, the a performance-based grant equivalent to 40 percent of a private sector, and development partners. On a demand typical third-generation mini grid developer’s connection basis, the World Bank will continue to provide support for costs would be about $400–$900 per connection. well-designed, new energy access projects that include Performance-based grants should be applied with cau- mini grid investments. In the broader context, the upscal- tion, however, as relying exclusively on final output ing of financing in the sector will need the involvement of makes it difficult for developers to finance their up-front the World Bank, development partners, and governments, capital costs. Therefore, it is reasonable to designate some at least at the same level of engagement over the next five intermediate results—such as purchase orders or the years, to create the leverage for exponential private-sector arrival of goods on site—as a basis for early subsidy pay- involvement. In the longer run, the percentages of public ments. Capital cost subsidies can also dilute the benefits of funds compared with overall investment should taper off increasing productive uses of electricity. Although the com- with the growth of private-sector investment. bined impact of grants and productive uses on the LCOE Different financing packages—consisting of different is typically greater than either on its own, their cumulative combinations of equity, debt, subsidy, and risk-shar- impact can increase the LCOE when OPEX costs are large ing mechanisms—are required for different types of relative to CAPEX.5 mini grid developers. In response, governments and their development partners are preparing packages of financial BUILDING BLOCK 7. support for mini grid developers that help them overcome Attracting exceptional talent and scaling skills barriers and finance the scale-up of mini grid deployments. development Larger international and local firms tend to have greater Scaling up mini grid deployments will be possible only access to equity and debt; smaller, mostly local firms usu- if human capital keeps pace with financial capital. Inno- ally do not. Female-led enterprises and project developers vative technologies and initiatives have emerged to train 10   MINI GRIDS FOR HALF A BILLION PEOPLE the stakeholders needed to support a thriving mini grid • Governments that recognize mini grids as a desirable industry. ESMAP has identified more than 50 training pro- and viable electrification option. grams for key stakeholder groups in the mini grid ecosys- • Government institutions that support mini grid develop- tem, including developers, financiers, policy makers, and ment through their actions and decisions. regulators. Many of these courses leverage new technolo- gies. For example, LED Safari’s flexible curriculum design • Flexible institutional frameworks able in principle to and remote web-based training enables developers and support various mini grid delivery models. governments to create high-quality, reputable certification • Frameworks that minimize duplication of oversight and programs. Comprehensive training programs that follow a conflicting roles. train-the-trainer approach, such as the Institute of Electri- cal and Electronics Engineers’ Smart Village’s Comprehen- BUILDING BLOCKS 9 AND 10. sive Training Program, can provide training to thousands Regulating the sector and making it easier to do of people. These programs seek to create a skilled, knowl- business edgeable ecosystem of stakeholders that can support the No single approach to regulating mini grids works best rapid scale-up of mini grids. in all settings, and regulation has costs as well as ben- Capacity needs assessments are a critical early step in efits. ESMAP has developed a series of decision trees that designing training and skills-building initiatives. They present options for how to regulate mini grids and the con- reveal gaps in key areas, including technical expertise, ditions under which each option is suitable. The decision management skills, institutional capacity, policy frame- trees are not prescriptive. They can provide guidance to works, partnerships, knowledge, and implementation help regulators and policy makers make informed deci- know-how. Needs assessments generally follow a four-step sions in five regulatory areas: market entry, tariffs, tech- process—(1) identifying key actors, (2) determining the nical specifications, service standards, and what happens capacity needs of a project or portfolio, (3) assessing exist- when the main grid arrives in the service area of a mini grid. ing capacity, and (4) identifying capacity gaps—that uses a Several countries are developing mini-grid-specific reg- mixed-methods approach using existing data or data col- ulatory frameworks that support private-sector invest- lected from key interviews with respondents and commu- ment. Across Asia and Africa, countries such as Bangla- nity members, focus group discussions, and surveys. desh, Cambodia, India, Kenya, Nigeria, Rwanda, Tanzania,6 and Zambia have developed regulatory frameworks for BUILDING BLOCK 8. mini grids that address key issues. Supporting institutions, delivery models, and champions to create opportunities The goal of a regulatory framework for mini grids should be to promote good service at the lowest cost-recov- National-level institutions are supporting the scale-up of ery tariffs. Pursuit of this goal throughout the stages of mini grids as a key element of electrification strategies. development of a country’s mini grid sector—taking into Haiti’s Ministry of Public Works has developed a special account subsidies and the broader national electrification unit, the Energy Cell, to implement a World Bank-sup- strategy—requires a regulatory framework that is predict- ported national mini grids program. Nigeria’s Rural Elec- able but flexible enough to evolve as the market does. trification Agency is implementing the largest mini grid program in Africa, targeting 850 mini grids by 2025, out of Meanwhile, innovative solutions that cut down on red an estimated potential market of 10,000 sites. Regulatory tape and make it easier for mini grid developers to do agencies in Nigeria, Rwanda, Zambia, and several other business are emerging, and include the following: countries have teams dedicated to mini grids. Ministries, • Standardized templates for key bureaucratic processes national utilities, and rural electrification agencies are col- that affect mini grids, including standardized power pur- laborating on national electrification plans, as with Kenya, chase agreements, which define the terms under which Myanmar, Nigeria, and Rwanda mentioned earlier. mini grid developers sell electricity to the main grid, ESMAP’s research identified four characteristics of an and standardized environmental and social manage- institutional framework that can support mini grids, ment systems, which identify when mini grid developers given the diversity in potential mini grid delivery mod- obtain environmental approvals. els. The most common delivery models for mini grids • Technology platforms to connect developers with inves- are build-own-operate, public-private partnerships, con- tors and suppliers and to run large-scale mini grid ten- cessions, utility models with and without private-sector ders, greatly boosting market efficiencies. involvement, and cooperative models. Strong institutional frameworks that can accommodate diverse delivery mod- • Formal delegation of mini grid industry oversight author- els are characterized by: ity to a single entity—usually the local government or a MINI GRIDS FOR HALF A BILLION PEOPLE    11 government agency that provides grants or subsidies measure the global mini grid industry’s progress against to mini grid developers (such as a rural electrification the 10 building blocks and 5 market drivers outlined above. agency)—in countries where the absence of a formal regulator increases the risk that mini grid developers face multiple layers of government oversight. REFERENCES • Introduction of e-government to reduce overhead cost Abate, G., T. Bernard, S. Makhija, and D. Spielman. 2018. “Accelerating for business registration, land and building permits, and Technical Change through Video-Mediated Agricultural Extension: environmental approvals. Evidence from Ethiopia.” Working Paper, Cornell University, Ithaca, NY. http:/ /barrett.dyson.cornell.edu/NEUDC/paper_421.pdf. Africa population count data: Linard, C., Gilbert, M., Snow, R.W., Noor, A CALL TO ACTION A.M. and Tatem, A.J., 2012, Population distribution, settlement patterns and accessibility across Africa in 2010, PLoS ONE, 7(2): Connecting half a billion people to mini grids by 2030 e31743. (WorldPop—https:/ /www.worldpop.org) is a monumental task that requires unprecedented lev- Balabanyan, Ani, Yadviga Semikolenova, Arun Singh, and Min A Lee. els of investment, innovation, and commitment from 2021. “Utility Performance and Behavior in Africa Today.” World development partners, governments, and the mini grid Bank, Washington, DC. https://openknowledge.worldbank.org/han- dle/10986/36178. industry. This book calls for action by stakeholders across the mini grid value chain. Key recommendations are for the Blimpo, M., and M. Cosgrove-Davies. 2019. Electricity Access in Sub-Sa- haran Africa: Uptake, Reliability, and Complementary Factors for Eco- following actors: nomic Impact. Africa Development Forum Series. Washington, DC: • Policy makers to leverage the latest geospatial analysis World Bank. technology to develop national electrification plans that BNEF (Bloomberg New Energy Finance). 2019. “1Q 2019 Frontier Power Market Outlook.” can guide investment in mini grids, main grid extension, and solar home systems, as well as develop initiatives Buckley, A. 2012. “Best Practice Community Engagement for Infra- structure Projects: Building Community Ties That Dig Deeper.” Pub- that promote productive uses of electricity and build lic Infrastructure Bulletin 1 (8). human capital. C. Arderne, C. Zorn, C. Nicolas, and E. E. Koks, “Predictive mapping of • Development partners to work with government coun- the global power system using open data,” Sci. Data, vol. 7, no. 1, p. terparts and the private sector to create enabling 19, Dec. 2020, doi: 10.1038/s41597-019-0347-4. (https:/ /gridfinder. org/) environments for mini grids through investments in CIESIN (Center for International Earth Science Information Network), portfolios of projects and technical assistance for devel- Columbia University, and Novel-T. 2020. “GRID3 Central African oping workable regulations and strengthening institu- Republic Settlement Extents Version 01, Alpha.” Palisades, NY: tions. Geo-Referenced Infrastructure and Demographic Data for Devel- opment (GRID3). Source of Building Footprints ‘Ecopia Vector • Regulators to adopt an evolving, light-handed approach Maps Powered by Maxar Satellite Imagery’ .” Accessed June 1, 2022. for a maturing mini grid sector, providing at each stage https://doi.org/10.7916/d8-y2ax-p859. of development clear guidance on market entry, retail GOGLA. 2022. “Our Members.” Accessed May 18, 2022. https://www. tariffs, service standards, technical standards, and gogla.org/about-us/our-members. arrival of the main grid. IEA (International Energy Agency), IRENA (International Renewable • The mini grid industry and its associations to work Energy Agency), UNSD (United Nations Statistics Division), World Bank, and WHO (World Health Organization). 2021. Tracking SDG 7: toward increasing the pace of deployment, retaining The Energy Progress Report 2021. World Bank, Washington, DC. superior-quality service delivery of third-generation mini https://www.iea.org/reports/tracking-sdg7-the-energy-progress- grids, and reducing the cost of these systems through report-2021. innovation to reach a value proposition that is affordable IEA. 2021. World Energy Outlook 2021. Paris. https://www.iea.org/ to the end users. reports/world-energy-outlook-2021 • National utilities to adopt an openness to partnerships IEA, IRENA, UNSD, World Bank, and WHO. 2022. Tracking SDG 7: The Energy Progress Report 2022. World Bank, Washington, DC. https:// with the third-generation mini grid industry on the basis www.iea.org/reports/tracking-sdg7-the-energy-progress-report- that the systems are grid-integration ready, which can 2022. provide for more financially viable grid expansion pro- Kojima, M., and C. Trimble. 2016. Making Power Affordable for Africa and grams for the utility in the long run. Viable for Its Utilities. Washington, DC: World Bank. Finally, there is a clear need for accurate, up-to-date, and Rockefeller Foundation. 2017. “Smart Power for Rural Development: Transforming Lives through Energy Access.” https:/ /www.rocke- widely available data to inform any type of initiative that fellerfoundation.org/report/smart-power-rural-development-bro- supports mini grids. To this end, we strongly recommend chure/. the development of a global tracking tool to monitor and 12   MINI GRIDS FOR HALF A BILLION PEOPLE SPI (Smart Power India). 2017. Smart Power Connect Magazine, Vol. 2 The importance of tailoring the community engagement approach 2.  (May). https://smartpowerindia.org/wp-content/uploads/2021/07/ to the local context was emphasized in an interview with Havenhill smartpowerindia_magazine_may_2017.pdf. Synergy Ltd., a Nigerian mini grid developer operating several solar Smith, J. 2000. Solar-Based Rural Electrification and Microenterprise hybrid mini grids in the Kwali and Kuje local government areas of Development in Latin America: A Gender Analysis. Golden, CO: Nigeria. National Renewable Energy Laboratory. https:/ /www.nrel.gov/docs/ RAO Energy in Russia, TANESCO in Tanzania, JIRAMA in Madagas- 3.  fy01osti/28995.pdf. car, and KPLC in Kenya are utility companies that operate dozens of Tenenbaum, B., C. Greacen, and D. Vaghela. 2018. Mini Grids and the mini grids nationwide. These mini grids are typically diesel powered Arrival of the Main Grid: Lessons from Cambodia, Sri Lanka, and (or, in the case of JIRAMA, hydro powered). They tend to be large, Indonesia. ESMAP Technical Report 013/18. Washington, DC: World typically on the order of several hundred kilowatts to a few mega- Bank. watts. Some utilities (in Niger, for example) have started to hybridize their diesel systems with solar PV panels. Tenenbaum, B., C. Greacen, and A. Shrestha. Forthcoming 2022. Undergrid Mini Grids in Nigeria and India: Interconnected and Non- Rather than provide a definitive number, this analysis is designed 4.  Interconnected. to understand the relative profit potential among different mini grid value chain stakeholders. Such an analysis can be used to determine Trimble, C., M. Kojima, I. Perez Arroyo, and F. Mohammadzadeh. 2016. the viability of establishing business lines focused on the mini grid “Financial Viability of Electricity Sectors in Sub-Saharan Africa: Qua- market. The data reflect the profit potential after all variable produc- si-Fiscal Deficits and Hidden Costs.” Policy Research Working Paper tion and manufacturing costs are taken into consideration. Detailed 7788, World Bank, Washington, DC. http:/ /documents.worldbank. assumptions and methodology are documented on the companion org/curated/en/182071470748085038/pdf/WPS7788.pdf. website to this handbook: www.esmap.org/mini_grids_for_half_a_ billion_people. On average, CAPEX accounted for about 65 percent and OPEX for 5.  NOTES about 35 percent of the fully cost-recovering tariff. While the mini grid regulations in Tanzania are some of the most 6.  Firm power output means that the peak load for which the system 1.  advanced in Africa, issues concerning implementation and enforce- was designed can be supplied by the mini grid any second of the day ment, as well as elements within the regulations themselves, have throughout the year. In solar hybrid mini grids, we approximate firm recently restricted private-sector investment in mini grids. power output as the sum of the generator capacity and 25 percent of the PV array capacity. For a more detailed description of this metric and the rationale for using it, please see chapter 1. MINI GRIDS FOR HALF A BILLION PEOPLE    13 OVERVIEW: The New Electricity Access Landscape and the Growing Space for Solar Mini Grids SDG 7: A GLOBAL AGENDA RUNNING . . . BUT PROGRESS HAS BEEN INSUFFICIENT TO MEET THE GOAL OF UNIVERSAL ACCESS BEHIND These achievements notwithstanding, progress has fallen In September 2015, the United Nations (UN) General far short of what is needed. According to the latest Track- Assembly adopted Resolution 70/1, which introduced a ing SDG 7: The Energy Progress Report, taking into account new global path for sustainable development. The 2030 population growth and recent slowdowns in access as Agenda laid out 17 ambitious Sustainable Development a result of the COVID-19 pandemic, a total of 930 million Goals (SDGs), to be achieved by 2030 (UN 2015). people will need to gain access to electricity over the next eight years if universal access is to be achieved. However, The SDGs focus on key economic and social development under current and planned policies and taking into account issues, such as education, health, and climate change. Rec- the effects of the pandemic, 670 million people are still pro- ognizing that access to basic energy services is a prereq- jected to remain without access in 2030 (IEA, World Bank, uisite for poverty alleviation, sustainable livelihoods, and and others 2022). economic growth, one of the goals (SDG 7) aims to ensure access to affordable, reliable, sustainable, and modern Meanwhile, adjusted for global population growth rates, the energy for all. Its targets include universal access to elec- annual pace of access has been steadily decreasing since tricity, clean fuels and clean cooking technologies, a dou- 2018. While the annual rate of grown in energy access was bling of the rate of improvement in energy efficiency, and a 0.8 percent between 2010 and 2018, it fell to 0.5 percent substantial increase in the share of renewables in the global in 2018–20. Furthermore, these increases have been con- energy mix. centrated in a handful of countries and very unevenly dis- tributed across regions, between rural and urban areas and ACCESS TO ELECTRICITY HAS INCREASED . . . across socioeconomic groups. Impressive advances have been made in closing the elec- tricity access gap in recent decades. Between 2010 and 2020, the share of the global population with access to electricity grew from 83 percent to 91 percent, as 1.3 billion Between 2010 and 2020, about 1.3 billion people gained access during this time period (IEA, World people gained access to electricity, but 733 Bank, and others 2022).1 million people are still currently without access. After taking into account population growth, The pace of electrification accelerated from 2000 to 2018 about 930 million people will need to gain access but has since tapered off: between the years 2000 and to electricity by 2030 to achieve SDG 7. However, if 2010, 100 million people gained access every year, ramping the current pace of electrification, current policies, up to 130 million people per year between 2010 and 2018. and current population trends continue, as many as But in the final two years of the decade, 2018–20, the num- 670 million people are predicted to remain without ber of new people gaining access dropped to 109 million access to any source of electricity by 2030. per year (IEA, World Bank, and others 2022). 14   MINI GRIDS FOR HALF A BILLION PEOPLE Countries that pursue a comprehensive approach to elec- MORE FINANCING IS NEEDED, AND IT MUST BE trification through main grid extension, mini grids, and solar BETTER TARGETED home systems achieved the fastest gains. In most of the A major cause of the present gap in electricity access is countries with the fastest gains in electrification between lack of financing. Current commitments to all electrifica- 2010 and 2020—including Bangladesh, Cambodia, Kenya, tion projects in the 20 highest-access-deficit countries— Myanmar, Nepal, Rwanda, and Tanzania—national electri- which account for 560 million people—are estimated at fication strategies leveraged a combination of main grid, $32 billion a year. This is 78 percent of the $41 billion a year mini grid, and solar home system investments. Nigeria is needed to achieve universal access by 2030, and a 27 per- another recent example of a country that has developed a cent decline from 2018, when the figure reached $43.6 bil- comprehensive national electrification strategy and imple- lion (SEforALL and CPI 2021). mentation plan. This comprehensive approach is the only way to connect the 930 million people that will need access This financing has been distributed highly unevenly, both to electricity by 2030. across the group and within different customer categories. As such, most financing targets provision of electricity ser- Meanwhile, slightly fewer than 76 percent, or 560 million vices to nonresidential customers. As of 2021, only $12.9 people living without electricity are concentrated in 20 billion—less than a third—of the funds committed in the countries with the highest absolute deficit in energy access 20 high-deficit countries—were aimed at households; the (IEA, World Bank, and others 2022).2 Closing the gaps in rest targeted commercial, industrial, and public consumers these countries is therefore essential to achieving the goal (SEforALL and CPI 2021). of universal access by 2030. Within these countries, Kenya and Uganda made the greatest gains since 2010, expanding Forecasts by the International Energy Agency (IEA) indi- access by more than 3 percentage points a year between cate that 95 percent of the additional investment in elec- 2010 and 2020 (IEA, World Bank, and others 2022). trification must target Sub-Saharan Africa if the world is to reach universal access by 2030 (IEA 2021). At present, Fragile and conflict-affected countries also require sub- however, investments in these countries are estimated at stantial support if SDG 7 is to be achieved by 2030. The 39 only approximately 15 percent of what would be required countries on the World Bank’s list of fragile and conflict-af- for them to reach full electricity access (IEA 2021). One fected countries account for well over half of the global notable example of this gap is the Democratic Republic access deficit—nearly 57 percent. The access rate in these of Congo, the country with the second-highest number of countries from 2010 to 2020 rose only by 11 percent—from people without access to electricity on the continent (72 44 to 55 percent (IEA, World Bank, and others 2022). million), which by 2021 saw only approximately $18 million The rural-urban divide in energy access is also stark. In per year committed to electricity access, compared to the 2020, global access rates were almost 97 percent in urban nearly $3 billion estimated to be needed annually to reach areas but just 83 percent in rural areas. Given that 80 per- universal electrification (SEforALL and CPI 2021). This is cent of the world’s unelectrified population reportedly lives in contrast to the progress made by India, where by 2019 in rural areas, identifying electrification solutions that meet the government declared the country to have reached a 99 rural needs is essential to reaching universal access (IEA, percent electrification rate, moving it from third to seven- World Bank, and others 2022). teenth place in the list of highest-access-deficit countries (IEA, World Bank, and others 2022). Encouragingly for the global sustainable development Populations without access to electricity agenda, electrification investments in the highest-ac- tend to be concentrated geographically. cess-deficit countries appear to be firmly shifting from Just 20 countries account for almost 76 percent fossil fuels to renewables. While grid-connected fossil of the global population without access to elec- fuels received over $21 billion in investments in 2018— tricity; fragile and conflict-affected countries col- compared to the $17 billion invested in grid-connected lectively account for well over half of the global renewables—by 2019, the numbers nearly flipped, with access deficit; and more than 80 percent of the grid-connected renewables receiving over $14 billion in world’s unelectrified population lives in rural areas. investments, compared to the under $8 billion in grid-con- As a result, identifying electrification solutions that nected fossil fuels. Much of this shift can be attributed to meet rural needs, particularly in these key electric- the firmer commitments to renewables made by the gov- ity access-deficit countries, is essential to reaching ernments in some of the key high-access-deficit coun- universal access to electricity by 2030. tries—such as Pakistan and Bangladesh—which saw them end approvals for new coal-powered projects. At MINI GRIDS FOR HALF A BILLION PEOPLE    15 Recognizing the need for increasing the impact of each dollar of international public financing and the private sec- Current financing commitments to energy tor’s growing relevance in development finance, in 2018 access in the 20 highest-access-deficit the World Bank Group adopted a new approach of Maxi- countries are estimated at $32 billion a year—just mizing Finance for Development (MFD). Otherwise known 78 percent of what is needed to achieve universal as the “cascade” approach, MFD is aimed at pursuing pri- access by 2030. Forecasts indicate that 95 percent vate-sector solutions for reaching development goals and of the additional investment has to be directed to reserving the limited public funds for key areas where the Sub-Saharan Africa, with only 15 percent channeled engagement of the private sector is not optimal or possible to the continent so far, and largely to nonresidential (World Bank and IMF 2017). customers. The guidelines for implementing the “cascade” follow a decision tree approach, designed to determine whether a new project has a sustainable private-sector solution that the same time, at 0.9 percent ($294 million) of the total limits public debt and contingent liabilities. It encourages electricity investments in the group of countries in 2019, the use of nonlending World Bank Group instruments— financing flows toward mini grid and off-grid renewable such as support for policy and regulatory reforms or solutions remain far behind what is needed to reach uni- de-risking mechanisms—to promote such private solu- versal access (SEforALL and CPI 2021). tions whenever feasible. When viewed through the lens of private and public sources DOUBLE DOWN ON SOLUTIONS THAT of funds, the financing trend appears quite uneven. By HAVE THE POTENTIAL FOR EXPONENTIAL 2018, the flow of international private financing into energy GROWTH CURVES access in the 20 highest-access-deficit countries reached nearly $11.5 billion—a major ramp-up from less than $3 Another reason for the financing gap is that electrifica- billion in 2013. However, in 2019 the number shrank to tion programs have traditionally focused on extending the under $7.5 billion. A parallel trend can be noted in domes- national grid. Doing so is often both expensive and slow in tic private financing for energy access—while it followed a remote settlements and areas with low population densi- steady growth pattern since 2013 and peaked at over $14 ties and low demand for electricity. Developing electrifica- billion in 2018; in 2019 it came down to a little over $8.5 tion models that complement grid extension is therefore billion (SEforALL and CPI 2021). Similarly, public invest- critical to achieving SDG 7. ments—both international and domestic—appear to have Mini grids and off-grid systems are two practical and com- come down slightly from a peak of over $18 billion in 2018 plementary approaches to grid extension. Recent tech- to under $15.9 billion in 2019. nological breakthroughs, the emergence of innovative Some countries are bucking the global trend by devel- business models, and enabling regulations and policies oping, in collaboration with development partners, com- have made mini grids and off-grid systems affordable, scal- prehensive support packages for all three electrification able options for expanding electricity services along expo- pathways—main grid extensions, solar home systems, nential growth curves. and mini grids. Support packages for mini grids consist of subsidies—increasingly in the form of performance-based grants —as well as debt facilitation and risk-sharing THE PLACE FOR SOLAR MINI GRIDS mechanisms, alongside private-sector debt and equity. The objective of these support packages is to increase WHAT ARE SOLAR MINI GRIDS? the affordability of mini grid electricity and incentivize pri- While there is no unanimously accepted definition of mini vate-sector investment, while ensuring that public funds grids, they are commonly described as power generation are deployed appropriately and efficiently. For example, and distribution systems built to provide electricity in areas performance-based grants are increasingly favored by that have not been reached by the main grid or whose private-sector developers and investors. The Energy Sec- costs of a grid-based connection are prohibitive. Mini grids tor Management Assistance Program (ESMAP) analysis typically supply electricity to local communities, covering presented in chapter 6 shows that these subsidies can domestic, commercial, and industrial demand. They range reduce the cost of electricity by almost 50 percent, but at in size, from systems that provide electricity to just a few the same time can dampen the effect of productive-use customers in a remote settlement to systems that bring programs and put additional pressure on developers to power to tens of thousands of customers (usually groups secure major up-front funding. 16   MINI GRIDS FOR HALF A BILLION PEOPLE of households, businesses, and public institutions) in a countries for installations of similar capacity size but serv- town or city (Tenenbaum, Greacen, and Vaghela 2018). ing somewhat different purposes and customers. In OECD Most mini grids are powered by alternating current (AC).3 countries, the term microgrid is used to refer to systems that are almost always connected to the main grid, but that Mini grids can be either fully isolated from the main grid or can operate in an “island” mode to achieve exceptionally connected to it in some capacity—to feed excess energy high levels of reliability, to supply power for applications for into it, take energy from it whenever needed, or both. Mini which a power outage would prove extremely costly or haz- grids that are connected to the main grid generally have the ardous, such as industrial processes, military or medical capacity to intentionally isolate—or “island”—themselves facilities, and data farms. Such microgrid systems are also from it. This means that they are able to disconnect and frequently individually designed as one-off bespoke proj- reconnect to it, ideally without disturbing power quality, ects with no intention for scale. By contrast, in low-income with the intention of improving power reliability, for safety reasons in the event of faults or surges on the main grid, and for the purpose of maximizing opportunities for additional revenue generation for the mini grid operator. The major- Mini grids are electric power generation and ity of mini grids in low-income countries are considered, at distribution systems that supply electricity present, to be totally isolated (for example, electronically to local communities, covering domestic, commer- disconnected) from the main grid. Figure O.1 illustrates a cial, and industrial demand. Mini grids come in all common setup for a solar hybrid mini grid. sizes, from systems that provide electricity to just In various contexts, the term mini grid is often replaced with a few customers in a remote settlement to systems or juxtaposed with the term microgrid. For instance, micro- that bring power to hundreds of thousands of cus- grids are often defined as mini grids with generation capac- tomers (usually groups of households, businesses, ity below 10 kilowatts (kW) (alternatively often referred to and public institutions) in a town or city. Mini grids as pico-grids), with mini grids described as having genera- can be fully isolated from the main grid or con- tion capacities from 1 kW up to 10 megawatts (MW) (IRENA nected to it. Those that are connected to the main 2016). The two terms are also often used differently in grid generally have the ability to intentionally iso- high-income (primarily the Organisation for Economic late, or “island,” themselves from it. Co-operation and Development [OECD]) and low-income Figure O.1 • Example of a common solar hybrid mini grid setup AC load in village AC appliances Poletop hardware Smart meter Generator Distribution line AC bus Residential Smart meter AC/DC inverter Service drop Pole PV array Charge Battery block Commercial controller Smart Solar-hybrid generation system Distribution system meters Efficient productive loads AC = alternating current; DC = direct current; PV = photovoltaic. MINI GRIDS FOR HALF A BILLION PEOPLE    17 countries, what is referred to as mini grids of similar gener- was coupled with, and amplified by, the coevolution of sup- ation capacities are generally built in areas not connected ply, demand, disruptive technology, and policy. Gradually, to the main grid and at some distance from it, and by devel- and as electricity systems became more complex, physical opers who often strive to achieve modularity, replicability, expansion and interconnection came as a natural conse- and economies of scale over time. quence, leading to today’s power systems. For the purposes of this report, the term mini grid will be Many dynamics affected these systems’ development, used to refer to all forms of mini grids, regardless of their some of which (technical advancements, innovation, entre- generation capacity or location, and regardless of whether preneurial drive, and decisions) were endogenous, while they are interconnected with the main grid, as long as they others (economic principles, legislative constraints and have the capacity to operate in an “island” mode. support, institutional structures, historical contingencies, and geography) were exogenous (Hughes 1983). Histor- Mini grids may be owned and managed by communities, ically, areas with robust socioeconomic activity were the local governments, utilities, private companies, or some earliest adopters. The first modern electric utility was the combination of the above. The delivery mechanism nec- Pearl Street Station in Manhattan, New York. Fired by coal, essary to finance, develop, operate, and maintain a mini this thermal power plant initially served electricity for lamp grid depends on characteristics like ownership structure, lighting in 1882 to about 80 customers via a direct current size, and technology (or a combination of technologies). (DC) distribution system (Hughes 1983). It was thus, by For example, diesel-only mini grids require lower up-front definition, an isolated mini grid. costs as compared with solar, solar hybrid, or hydropower mini grids, but are expected to have much higher and less From New York City and Chicago in the United States to predictable operations and maintenance (O&M) costs London and Berlin in Europe and Kimberly in South Africa, (Greacen, Nsom, and Rysankova 2015), often resulting in mini grids started to emerge and operate autonomously limitations or restrictions to the electricity supply service. in cities throughout that period. Other similar systems While government or utility-run mini grids often charge developed to provide electricity to industrial loads or to subsidized tariffs, mini grids that are owned and operated serve particular populations, such as rural US agricultural by private companies require rates of return sufficient producers. not only to cover O&M costs but also to turn a profit. This Various factors supported the early deployment of decen- means that, in the absence of a government subsidy, they tralized electricity systems in areas of high demand den- must charge cost-reflective tariffs, which are often higher sity (urban areas and industrial facilities) or low-cost than the average national electricity tariff. supply (such as hydro sites). First, DC systems and early THE HISTORIC ROLE OF MINI GRIDS IN low-voltage AC systems had physical limits that kept NATIONAL ELECTRIFICATION EFFORTS distribution local; technology in the late nineteenth and early twentieth centuries did not allow for larger systems Mini grids are not a new phenomenon: all current central- covering long distances. Second, electricity demand was ized power grid systems started with small, isolated power initially limited to a few services, such as public lighting. systems and mini grids. These systems were the initiat- Third, the capital intensity of electric power systems ing “spark” of electricity uptake some 130 years ago, and meant that cost recovery required the maximization of were pivotal to the early development and industrialization electricity output and sales. These early power systems of most modern economies, such as Spain, Sweden, the therefore sought to improve the load factor and economic United Kingdom, and the United States. Although these performance. systems were initially few and scattered, their development While government or utility-run mini grids Mini grids are not a new phenomenon: all often charge subsidized tariffs, mini grids current centralized power grid systems that are owned and operated by private companies started with small, isolated power systems and mini require rates of return sufficient to not only cover grids, which gradually interconnected. These sys- O&M costs but also turn a profit. This means that, tems were the initiating “spark” of electricity uptake in the absence of a government subsidy, they must some 130 years ago, and were pivotal to the early charge cost-reflective tariffs, which are often higher development and industrialization of most modern than the average national electricity tariff. economies. 18   MINI GRIDS FOR HALF A BILLION PEOPLE As technologies improved, demand increased, and the public systems, rural cooperatives, and large federally policy and regulatory regimes stabilized, larger generators owned power generation corporations—and supported could be built and electricity could be transmitted over through public and nonprofit entities, such as rural longer distances. These factors resulted in the emergence electrification agencies and, in the United States, the of centralized utilities (either privately or publicly owned). National Rural Electric Cooperative Association. Mini grids either became integrated with one another, • Local participation and ownership appear to be attri- forming the nucleus of a larger centralized system, or were butes of many public and cooperative efforts, par- absorbed by a larger grid system as it expanded. ticularly in small communities and rural areas. Rural The process was not always smooth. In Bolivia, for exam- communities were eager to access electricity. In most ple, lack of technical coordination meant that different mini cases, the local population was actively involved in the grids used different frequencies, making their integration in process. Community engagement and political com- a central grid challenging. In the United Kingdom, compet- mitment through financial and regulatory support ing business and institutional interests resulted in aggres- were crucial. sive competition and stranded assets. Over time, however, • Interconnecting neighboring mini grids was a way to the increasing variety of sources, loads, and control nodes cope with load variation and to increase system flexibil- created the extensive and complex grid network many ity. Increasing generating capacities (per unit) was a way countries have today. to lower costs through economies of scale. These historical systems can be described as the first gen- • Choosing between centralized grid versus mini grid elec- eration of mini grids, which faced many of the same policy, trification was a lengthy process that depended upon regulatory, and operational challenges experienced by mini technological advances, geographic factors, resource grids in developing countries in Sub-Saharan Africa and availability (for example, hydropower), sociodemo- Asia today. A retrospective overview of a number of these graphic factors (for example, demand density), and systems is available online (www.esmap.org/mini_grids_ policy. With the exception of resources and geography, for_half_a_billion_people), highlighting the origin stories of the other factors shifted and changed over time, with modern grids from isolated mini grids in Bolivia, Cambodia, accompanying changes in how electricity demand was China, India, Ireland, Spain, Sweden, the United Kingdom, met, both technically and institutionally. These factors and the United States. That brief historical review provides pushed the industry to ever-increasing interconnection, a number of insights: standardization, and centralization. • Rapid industrialization and the socioeconomic shifts Second-generation mini grids it spurred created demand for new, low-cost forms of Unlike the first generation, what has often been referred energy. to as the second generation of mini grids can be found • The electric power sector soon became a strong new in modern low-income countries (Tenenbaum, Greacen, business opportunity, attracting substantial entre- and Vaghela 2018). These systems are typically small and preneurial and investment activity. The competitive isolated, and generally built by local communities or local environment in the electric power industry promoted entrepreneurs to provide access to electricity in zones technological innovation, leading to new technical sys- with low population densities and low demand, primarily tems that were quickly adopted by utilities. in rural areas that have not yet been reached by the main • Regardless of their type or size, the earliest power sys- grid or where it would be too prohibitively expensive to tems were designed to be successful in terms of eco- extend it. Typically, such second-generation systems are nomics as well as engineering, contributing to their built to supply electricity to single villages. Tens of thou- profitability and competitiveness. • Early deployment of isolated stations and urban mini grids (and later peri-urban systems) was driven primar- Mini grid systems that came about in the ily by the growing demand for electricity. In rural areas, late nineteenth and early twentieth centu- system expansion was largely a function of an explicit ries can be described as the “first generation” of social welfare policy aimed at bridging the gap between mini grids, which faced many of the same policy, urban and rural areas. regulatory, and operational challenges as those • Private power companies would not or could not serve experienced by mini grids in developing countries all of the population and provide power at large scales. in Asia and Sub-Saharan Africa today. The unelectrified areas were filled with small municipal MINI GRIDS FOR HALF A BILLION PEOPLE    19 sands of these systems were built, starting in the 1980s to Haiti, Zambia, Rwanda, and elsewhere—to explicitly and ramping up through the 1990s and early 2000s. and preemptively provide economic options for mini grid developers when the main grid arrives. The second generation of mini grids provided important lessons about technical design, the importance of produc- Third-generation mini grids tive uses for financial viability, and economies of scale to In the past decade, a new, third generation of mini grid drive down costs. The developers of second-generation technologies and business models has emerged. These mini grids, whether public or private, were motivated by the third-generation mini grids differ from the earlier genera- overriding need to supply rural communities with a higher tions in several important ways. level of electricity service as soon as possible. Developers of such second-generation mini grids almost always relied New technologies. Technological developments have on standard existing technologies—such as diesel or mini allowed third-generation projects to use more modular hydro generation—and mini grids were built as one-off proj- technologies—especially solar photovoltaic (PV) genera- ects instead of as part of a larger portfolio. Second-gener- tion backed up with diesel, batteries, or both—and state-of- ation mini grids typically used basic meters, on-site meter the-art hydropower. In most of Africa and parts of Asia (for reading, and in-person bill collection, which was expensive example, Bangladesh, Myanmar, and India), the dominant and did not permit innovative pricing schemes that could emerging technology is solar hybrid mini grids. These new promote productive uses of electricity during the day. They systems are usually combined with sophisticated pay-as- also often charged flat monthly tariffs or postpaid fees cal- you-go (PAYG) billing, smart metering, mobile payment culated and collected at the end of each month based on options, and real-time internet-based monitoring systems, the customers’ power consumption for that month (Tenen- enabled by cellular data, allowing company engineers to baum, Greacen, and Vaghela 2018). spot problems as they start to emerge and make adjust- ments or repairs before small problems snowball into Second-generation mini grids also provided important larger ones. Some third-generation mini grid developers lessons about regulatory frameworks, particularly to use sophisticated load dispatch technologies to ensure reduce the risk of stranded assets once the main grid that priority loads always get electricity by automatically arrives (for a detailed discussion of what happened when shifting low-priority loads to times of energy surplus. the main grid arrived in Cambodia, Indonesia, and Sri Lanka, see Tenenbaum, Greacen, and Vaghela [2018]). New players. In addition to local entrepreneurs and com- When these mini grids were developed, little thought was munity organizations, new national and international pri- given to the possibility of later interconnecting with the vate companies are building or proposing to build these main grid, and many of them were simply abandoned third-generation projects. They seem to be motivated by when the main grid arrived (Tenenbaum, Greacen, and the possibility of using the modular (often proprietary) Vaghela 2018). If they did not go out of existence, these technologies that can be scaled up quickly to serve differ- mini grids often chose to become small power producers ent-sized villages and towns, providing opportunities for (particularly if using a more affordable renewable energy cost-reducing economies of scale that were not available generation source rather than diesel); or small power to second-generation developers. The very early evidence distributors, converting to buying all of their electricity suggests that this will be accomplished through joint ven- supply wholesale from the main grid and selling it to the tures with local firms. Large multinational corporations that local customers at retail prices. These options for what have previously not operated in the mini grid market—such happened when the main grid arrived in the service area as Caterpillar, Tesla, Siemens, General Electric, and ABB— of second-generation mini grids are now being codified have publicly announced their intentions to enter it. Unlike in new mini grid regulations—from Tanzania to Nigeria, second-generation local private entrepreneurs, these new third-generation companies have better access to national and international financial markets. Public-private partnerships. In Kenya, Sierra Leone, and “Second-generation” mini grids are com- elsewhere, governments have proposed public-private mon in low-income countries today. These partnerships to build and operate mini grids. This is an systems are typically small and isolated, and built alternative to pure publicly owned or pure privately owned by local communities or entrepreneurs to provide mini grid systems that have been used in second-gen- access to electricity in zones with low population eration mini grids. These new partnerships appear to be densities and low demand, primarily in rural areas motivated, in part, by the reality that it is politically easier that have not yet been reached by the main grid. to channel a subsidy through a government entity in a joint 20   MINI GRIDS FOR HALF A BILLION PEOPLE venture than to openly give the same or even a smaller sub- Access to new geospatial tools. In the last few years, low- sidy to a private company. cost geospatial planning tools have become more widely available to those planning to develop mini grids. These Not necessarily isolated. Mini grids are no longer being new tools use satellite imagery data that allow potential built only in isolated rural villages at a distance from the developers to obtain important market intelligence on the main grid. For example, in the Indian state of Uttar Pradesh, physical characteristics, likely initial customer base, and one private mini grid operator (OMC Power) has built many probable daily electricity demand profiles of individual mini grids in villages that are already served by a govern- villages. The cost of acquiring the data is rapidly coming ment-owned distribution utility, because the distribution down. Several years ago, one donor organization paid $1 utility has not been able to provide reliable service, espe- million to gather this information on 25 villages in Nigeria cially during peak evening hours (Rockefeller Foundation without the use of geospatial tools. More recently, simi- 2018). These mini grids are not currently interconnected lar information was obtained for 300 villages in Nigeria at with the main grid but have been built to be grid compati- roughly the same total cost with the application of latest ble in the future. A similar arrangement has been proposed geospatial analysis and planning applications. in the mini grid regulations recently issued by the Nigerian electricity regulator. FIGURE O.2 • The first, second, and third generations of mini grids Intake weir and setting basin Forebay tank Penstock Power house containing turbine and generator 1st generation 2nd generation 3rd generation Source: Upper left: International Magazine Co. 1925; upper right: World Bank design; bottom left: World Bank photo. MINI GRIDS FOR HALF A BILLION PEOPLE    21 reflective tariff for 39 utilities across Sub-Saharan Africa is $0.27/kilowatt-hour (kWh); 25 percent of utilities In the past decade, a new “third generation” require a cost-reflective tariff of more than $0.40/kWh, of mini grids has emerged, characterized about half require a tariff of $0.20–$0.40/kWh, and 25 by new technologies, new business models, new percent require less than $0.20/kWh. Only 2 of the 39 players, new types of partnerships, new tools, and utilities (Seychelles and Uganda) charged tariffs that tailored policy and regulatory systems. enabled them to recover their costs (Trimble and others 2016; Kojima and Trimble 2016). Mini grids are therefore often the least-cost, best solution to connect communi- ties where the cost of extending the main grid is simply Targeted regulatory systems. Until recently, developers too expensive. were “flying blind” on government policies and regulations that would apply to mini grid projects. This, too, is chang- Meanwhile, the penetration of off-grid solar—including ing. Mini grid regulatory systems have been developed by solar lanterns, pico PV systems, and solar home systems— governments in India, Kenya, Myanmar, Nigeria, Rwanda, grew rapidly over the last two decades, with more than 100 Sierra Leone, and Tanzania, among several others. These million systems sold in Africa alone. This market growth has systems reduce regulatory uncertainty for mini grid devel- been the result of increasing consumer demand for elec- opers, though there is always the remaining uncertainty tricity services in homes, as well as the pace of innovations as to whether the regulatory rules will be implemented as in telecommunications, which enabled the rise of the PAYG written. model for electricity access. Significant consumer data that emerged from the mobile money and PAYG revolution THE ROLE OF SOLAR MINI GRIDS IN provided lenders and investors with more confidence with UNIVERSAL ELECTRIFICATION regard to the credit risk of the end users, enabling them to Electrification programs have traditionally focused on raise more capital and consequently expand their services. extending the national grid, primarily through power Today, such solar home systems, depending on their size, generated from fossil fuels. Experience in electricity-ac- can typically cost $30–$200 and provide electricity ser- cess-deficit countries over the past five decades, however, vice at Tiers 1 and 2. Some larger, component-based sys- has shown that the main grid is typically unreliable. Across tems are also in use (GOGLA 2019). Sub-Saharan Africa, more than half of households con- WHERE DO SOLAR MINI GRIDS FIT IN? nected to the main grid reported receiving electricity less than half of the time (Blimpo and Cosgrove-Davies 2019). Mini grids have characteristics of both utilities and solar In most electricity-access-deficit countries, the main grid home system companies, creating both challenges and usually provides only Tier 3 or Tier 4 electricity.4 The main opportunities for their large-scale deployment. Like the reasons for this unreliability are the challenges with the main grid, mini grids have sunk cost assets, are subject national transmission and distribution networks, rather to regulatory oversight, and have the possibility of provid- than with the generation systems. Given the region’s size ing 24/7 electricity and supporting productive loads. Mini and frequently very low population densities, the vast dis- grids also have features of the solar home system industry, tances between rural economic hubs in many countries with the possibility for very rapid expansion when the value prove to be prohibitively expensive to connect to central- proposition is right for the market. ized systems. Both utilities and solar home system companies are enter- In addition, research has shown that most utilities in Africa ing the mini grid space for economic reasons, in ways that are not financially solvent. Most national utilities in Sub-Sa- mirror their respective business models, with utility mini haran Africa sell electricity at a loss, as the full cost of con- grids operating as rural distribution networks and solar necting residential customers (typically $800–$2,000 home system companies interconnecting individual stand- but often much higher for rural areas) is too expensive alone systems. This trend would lead to modest growth in for most households (Trimble and others 2016), and this the deployments of mini grids, as the two sectors develop cost is frequently subsidized by the national government. mini grids at the margins of their current target markets. If, In addition, the amounts that the rural, remote, and poor- however, the unique position of mini grids can build on the est groups of the population are able to pay for electric- strengths of both sectors—24/7 electricity from the utility ity generally do not reach the cost-recovery threshold for sector and agility and customer service from solar home national utility companies, and the tariffs charged to these system companies—mini grids will be able to bring afford- customer segments are often cross-subsidized across able access to high-quality electricity to millions of people the utilities’ large customer bases. The average fully cost- at an accelerated pace. 22   MINI GRIDS FOR HALF A BILLION PEOPLE Indeed, as a result of the declining levelized cost of energy “win-win-win” economic outcomes for the three key parties. (LCOE), increasing income-generating uses of electricity, The arrangement can eliminate or reduce financial losses and mainstreaming of geospatial planning, solar mini grids for distribution companies (DISCOs) that are forced to sell are on track to provide power at lower cost than many util- electricity at non-cost-recovering retail tariffs. Interconnec- ities by 2030. At $0.40/kWh, mini grid LCOE would be less tions also allow DISCOs to earn new revenues through bulk than the LCOE of national utilities in 7 out of 39 countries power sales to the mini grid as well as rental revenues from in Africa. At $0.20/kWh, mini grid LCOE would be less than the leasing of some or all of the DISCOs’ existing distribu- the LCOE of national utilities in 24 African countries (Trim- tion systems to the mini grid. For the mini grid operator, a ble and others 2016). This would make mini grids the least- physical connection to the contiguous DISCO offers the cost solution for grid-quality electricity for more than 60 possibility of purchasing bulk power, whether on a firm or percent of the population in Africa in a scenario assuming an “as available” basis from the interconnected DISCO or that national utilities do not dramatically change their oper- an upstream supply source. This can lead to lower operat- ations—with major implications for the allocation of both ing and capital costs (that is, lower LCOE) for the intercon- public and private investment funds. nected mini grid than if it operates in a pure stand-alone mode. And for the mini grid’s customers, this should lead to However, scaling up mini grids does not mean scaling back lower tariffs than would be possible if the mini grid operated the main grid. On the contrary, solar mini grids enhance the in a totally isolated mode. Finally, it is well documented that economic viability of expanding the main grid. By design- mini grids, whether interconnected or isolated, routinely ing the system from the beginning to interconnect with the achieve high levels of reliability for their customers than main grid and by promoting income-generating uses of DISCOs do for theirs (Tenenbaum, Greacen, and Shrestha electricity through effective community engagement and 2022 forthcoming). training, third-generation mini grids can provide early eco- nomic growth, so that significant load already exists by the Product cost: LCOE, portfolio development, CAPEX, time the main grid arrives, and customers have a greater OPEX including major replacements ability to pay. New regulatory frameworks give developers The plummeting cost of mini grid electricity is on pace to viable options for what happens when the main grid arrives, achieve $0.20/kWh by 2030. Indeed, the LCOE of “best- and reductions in the cost of components enable develop- in-class” mini grids already dropped from $0.55/kWh in ers to build grid-interconnection-ready systems while still 2018 to $0.38/kWh in 2021. Up-front investment costs per keeping tariffs affordable. customer have also fell dramatically, from around $2,000 Supporting solar mini grids therefore goes hand in hand per connection just a few years ago to $700–$800 per cus- with strengthening the power sector. Interconnecting tomer in 2021 (AMDA 2021). third-generation mini grids with the main grid can increase These cost declines are the result of decreases in the cost the resource diversity and overall resilience and efficiency of major components, and increasing economies of scale of the power system. However, this presents a couple as mini grids are built as part of ever-larger portfolios of of operational challenges that are better addressed in a projects by private-sector developers and national utilities. comprehensive strategy for developing the sector, for As table O.1 shows, prices for major components declined example, for governments through their electrification 60–90 percent between 2010 and 2020, and are projected strategies to allow for utilities, mini grid, and off-grid to decrease even further through 2030. companies to deliver services in the country, as well as for utilities to be able to introduce the practical technical In addition, economies of scale can further drive down functions to support power system operations and plan- costs for mini grids. Companies like Tata Power Renewable ning with multiple mini grids connected to the distribution Microgrids, Husk Power, Engie PowerCorner, OMC, and grid, such as short- and long-term forecasting and other others, are planning hundreds and thousands of mini grids procedures. over the next several years. At this scale, the unit costs of distribution infrastructure, batteries, solar PV modules, and First experiences with interconnected mini grid collabora- power electronics drop dramatically, as we discuss in more tions are emerging, for example, in Nigeria and India, and detail in chapter 1. are providing valuable lessons. These interconnected mini grids are built to serve different market segments: rural and With these cost declines, mini grid electricity is on pace to peri-urban towns and villages, large urban marketplaces, achieve an unsubsidized cost whereby $10 buys 50 kWh of commercial and industrial (C and I) installations, and sepa- energy each month— transformative consumption levels rate urban residential communities. Early evidence seems for hundreds of millions of people and millions of commu- to indicate that these interconnected mini grids can create nities worldwide. MINI GRIDS FOR HALF A BILLION PEOPLE    23 TABLE O.1 • Benchmarks and price projections, mini grid component costs, 2010–30 Mainstream Mainstream Best in Class industry industry 2030 LCOE Best in Class Mainstream benchmark estimate by modeling Percent Median cost 2020 LCOE industry in 2020 (% 2030 (% assumption of total in ESMAP modeling benchmark change from change from (% change Component Unit capital cost survey assumption in 2010 2010) 2020) from 2020) PV module US$/kWp 9.7 441 596 1,589 198 (–88) 114 (–42) 343 (–42) PV inverter US$/kWp * * * 320 80 (–75) 70 (–12.5) * Battery (Li-ion) US$/kWh 14.9 314 297 1,160 126 (–89) 58 (–54) 137 (–54) Battery inverter US$/kVA 8.6 † 415 303 565 113 (–63) 99 (–12.5) 265 (–12.5) Smart meters US$/ ‡ ‡ ‡ 106 40 (–62) 35 (12.5) ‡ customer Source: Bloomberg New Energy Finance Solar Spot Price Index; ESMAP analysis; Feldman and others 2021; Kairies 2017; National Renewable Energy Laboratory US Solar Photovoltaic System Cost Benchmark: Q1 2020. * PV inverter is included with PV module cost. † Battery inverter is grouped with EMS and monitoring equipment. ‡ Smart meters are included in distribution cost. Average, median, minimum, and maximum costs are all expressed in inflation-adjusted dollars. ESMAP = Energy Sector Management Assistance Program; kVA = kilowatt-ampere; kWp = kilowatt-peak; LCOE = levelized cost of energy; Li-ion = lithium-ion; PV = photovoltaic.  Addressable market and demand: Number of load mobile phones) show that these new solutions need to be centers, current expenditure, income-generating not a little, but much better than the current alternative: appliances why else would consumers take the risk to change their Meanwhile, as costs for mini grid electricity continue to fall, behavior? For these clusters of clients, the service provided the addressable market for its services remains immense by the solar mini grids should be a reliable source for their and continues to grow along with the population. Indeed, consumptive activities (lighting, charging, radio/TV) as well taking into account population growth, ESMAP’s Global as provide for life-changing productive activities within the Electrification Platform estimates that mini grids are the current expenditure of $5–$20 per month. From the end least-cost option to provide first-time access to electricity user’s perspective, a $5–$20 expenditure should not only to 430 million people. This represents around 86 million cover the cost of the reliable electricity provided but also mini grid connections and an estimated up-front invest- cover the cost of transitioning into electric appliances as ment cost of $100 billion. well as the purchase of appliances themselves. When cal- culating this over the lifetime of the technology, this would For Sub-Saharan Africa, nearly 291,000 load centers have mean that roughly $3–$15 per month covers the cost of the profile that favors the deployment of solar mini grids. electricity and about $2–$5 per month, the appliances. The analysis based on digitalized rooftops for the region This equation shows the tall order that needs to be met by shows that over 177,000 clusters consist of 100 to 500 peo- the mini grid industry to fulfil its full market potential. ple each, matching smaller solar mini grid of up to 20 kW, almost 96,000 clusters with 500 to 2,500 people matching This addressable market represents a major business medium-sized solar mini grids of up to 80 kW, more than opportunity not only for mini grid developers but also for 15,000 clusters with 2,500 to 10,000 people that would be suppliers and financiers of income-generating appliances best serviced with larger solar mini grids of up to 200 kW, and machines. Households in low- and middle-income and nearly 3,000 population centers of 10,000 to 100,000 countries receiving electricity for the first time have an people where customization rather than standard sizing of available budget for energy services of between $5 and mini grids would likely be appropriate based on GRID3 data $20 per month. If we use the 86 million connections esti- (CIESIN 2020). mate from the Global Electrification Platform, and a $10 per month expenditure in 2030, the resulting global mar- The end users in these load centers spend on average ket potential for mini grid electricity services is $10 billion an estimated $5–$20 per month on alternative forms per year by 2030. Furthermore, ESMAP’s research has of energy such as candles, kerosene, dry-cell batteries, identified more than 130 income-generating machines car-batteries, and petrol and diesel fuel for stand-alone and appliances with a payback period of less than a year, gensets. Lessons from the introduction of innovative tech- and a median up-front cost of $1,200. If we apply this nologies in the marketplace (like solar home systems or 24   MINI GRIDS FOR HALF A BILLION PEOPLE cost globally and assume 15 appliances per mini grid and properly planned and executed as part of a national elec- 200,000 new mini grids by 2030, the market potential trification strategy, it can increase the resource diversity for income-generating appliances connected to mini grid and overall resilience and efficiency of the power system. electricity is at least $3.6 billion. But this presents operational challenges such as those described earlier. This means that mini grid development— The private sector has taken note of this market opportu- as a viable strategy for helping deliver universal access to nity, and is already going after it, by treating communities electricity—also entails a greatly strengthened utility sec- as valued customers and partners rather than beneficia- tor able to accommodate interconnecting mini grids with ries. Leading developers today across Africa and Asia are the main grid. Many electricity-access-deficit countries expanding their portfolios, raising capital to deploy doz- lack clear procedures for integrating mini grids into the util- ens of new mini grids in the next two years, and hundreds ity’s system planning and operations. So national electrifi- of mini grids in the next five years, on a per-developer, cation plans will need to accommodate scenarios in which per-portfolio basis. mini grids are isolated from the main grid or connected only to other mini grids. A transformational end-user value proposition In addition to declining costs and a large, addressable mar- At the same time, mini grids developed today are challeng- ket, mini grids today are providing high-quality electricity ing the existing centralized approach to electricity service services to their customers. A recent benchmarking report delivery. The cost of mini grid electricity is expected to plum- by the Africa Minigrid Developers Association (AMDA) met over the next decade to levels that make it competitive found that modern solar hybrid mini grids in Africa had with main grid electricity in a large number of electrici- 99 percent uptimes on average, with only seven mini grids ty-access-deficit countries (more discussion on this point reporting uptimes below 95 percent. is provided in chapter 1). In addition, modern mini grids provide higher-quality service—in terms of reliability, avail- Third-generation mini grids are true engines of economic ability, and customer service—than many national utilities development, especially when taken as a package. Their in low-income countries. As mini grid developers establish declining costs are on pace to deliver 50 kWh of electric- strong reputations in their respective countries of operation, ity for $10 by 2030, while a large, addressable market is demand for their services in urban and peri-urban areas is already attracting private sector investment and superior likely to increase, incentivizing developers to target these service. They offer a powerful value proposition for end customers as well. This will put pressure on national utility users, communities, and governments. companies to evolve and improve their service offering. Strengthening the power sector: Win-win with utilities Older diesel-powered mini grids were expensive, ineffi- cient, polluting, and dangerous. Nor were they managed HOW TO SCALE SOLAR MINI GRID as businesses. The fact that they existed at all proved DEPLOYMENT TO SERVE HALF A that customers were willing to pay for electricity and suggested that demand would develop once the main BILLION PEOPLE grid arrived. Most customers used little electricity. But FIVE DRIVERS: COST, QUALITY, PACE, FINANCE, that meant the main grid would sustain ongoing financial AND ENABLING ENVIRONMENT losses when it reached areas served by these older mini grids. Modern mini grids are flipping this narrative. By Through a collaborative, iterative process, ESMAP and mini designing the system from the beginning to interconnect grid industry leaders—including AMDA5 and development with the main grid and by promoting productive uses of partners—have jointly identified five market drivers for the electricity through community engagement and training, sector to achieve its SDG 7 targets: mini grids can provide early economic growth. So by the • A more rapid deployment of mini grids through a port- time the main grid arrives, substantial load already exists folio approach; and customers are able to pay. In parallel, new regulatory frameworks give developers viable options for what hap- • Better service; pens when the main grid arrives, and lower-cost compo- • Crowding in private-sector and government finance; nents enable developers to build interconnection-ready • Creating an enabling business environment for mini grid systems while keeping tariffs affordable. grids in access-deficit countries; and To restate the important point above, supporting mini grids • Reducing the cost of solar hybrid mini grids—which the therefore goes hand in hand with supporting utilities. When other four market drivers will also support. interconnection of modern mini grids with the main grid is MINI GRIDS FOR HALF A BILLION PEOPLE    25 With support from ESMAP and the World Bank, the mini the overall value proposition of mini grids as an electrifica- grid industry can work toward clear and measurable tar- tion strategy. Mini grids can be deployed faster than main gets for these market drivers. These drivers will enable the grid extensions, often at a lower cost per connection; they sector to connect 490 million people by 2030. We summa- tend to provide better-quality electricity and customer ser- rize these targets in table O.2. vice than utility companies; they support productive uses, unlike solar home systems; and they can attract both pri- In addition to helping the mini grid sector achieve magni- vate- and public-sector finance (AMDA 2019). tude changes in scale, these five market drivers support TABLE O.2 • SDG 7 and mini grid industry targets, 2020–30 Target Objective/indicator What is measured 2018 Baseline 2020 2025 2030 1. Increase pace of mini grid development Time from purchase order to Cohort of leading private- 6–12 7 6 5 commissioning (weeks) sector developers Time from goods arriving on site to Cohort of leading private- 6–12 5 4 3 commissioning (weeks) sector developers Mini grids per key access-deficit country Portfolios from rural 20–75 150 500 2,000 per year electrification agencies, utilities, private developers, or industry associations 2. Provide superior-quality service Industrywide standard for minimum Industry associations Under Developed for Developed for Developed for technical specifications preparation solar hybrid mini solar hybrid all renewable grids and hydro mini energy mini grids grids Industrywide standard for reliability of Representative sample of 90–97 percent 97 percent 97 percent >97 percent electricity supply mini grid developers uptime uptime during uptime for 24/7 uptime for 24/7 promised electricity electricity availability times Customer satisfaction (percent) Representative sample of 82–84 85 88 90 mini grid customers Average load factor across the industry Representative sample of 22 25 35 45 (percent) mini grid developers 3. Establish enabling mini grid business environment in key access-deficit countries Average RISE score for mini grids Top 20 electricity-access- 59 60 70 80 framework in top 20 electricity-access- deficit countries deficit countries 4. Crowd in government and private-sector funding Cumulative government and development Estimated from a cohort 8 10 18 32 partner funding committed to mini grids in of leading development key access-deficit countries (US$, billions) partners Cumulative private sector debt and equity Global estimates from 5 10 27 73 invested in mini grids in key access-deficit market research countries (US$, billions) Total cumulative investment in mini grids for Sum of all funding for mini 13 20 45 105 energy access (US$, billions) grids in key energy-access- deficit countries 5. Reduce cost of solar hybrid energy Levelized cost of energy (US$/kWh) Average across a cohort 0.55 0.40 0.25 0.20 of leading mini grid developers Source: ESMAP analysis. Note: See the discussion of the underlying analysis for each target is presented in the overview. kWh = kilowatt-hour; RISE = Regulatory Indicators for Sustainable Energy. 26   MINI GRIDS FOR HALF A BILLION PEOPLE TEN BUILDING BLOCKS • Regulations and policies: Enacting regulations and pol- The World Bank’s experience over the past decade working icies that empower mini grid companies and customers. with mini grid developers, electricity regulators, investors, • Doing business: Cutting red tape for a dynamic busi- policy makers, ministries, rural electrification agencies, ness environment. experts, and donor partners has helped it identify a set of The market drivers focus mainly on PV and solar-die- 10 building blocks that need to be in place to achieve coun- sel hybrid mini grids, as these two types of mini grids are try-level scale-up in mini grid development. These 10 build- likely to be the most prevalent technologies for scaling ing blocks are as follows. up mini grid deployments in key electricity-access-defi- • Solar mini grid costs and technology: Reducing costs cit countries. A similar focus on solar PV and solar-diesel and optimizing design and innovation for solar mini hybrid mini grids is evident in the chapters on the 10 build- grids. ing blocks. The building blocks themselves, however, are conceptualized to support vibrant renewable energy mini • Geospatial planning: Planning national strategies and grid sectors at the national level, regardless of renewable developer portfolios with geospatial analysis and digital energy technology. platforms. Each building block contributes to different market drivers • Productive uses: Transforming productive livelihoods presented above. Collectively, they represent the founda- and improving business viability. tion of successful national mini grid programs. How each • Community engagement: Engaging communities as of the building blocks supports various market drivers is valued customers. shown in figure O.3. • Companies and utilities: Delivering services through As the matrix illustrates, there is a logic to the order in local and international companies and utilities. which we present these building blocks. The first six build- • Access to finance: Financing solar mini grid portfolios ing blocks, read from left to right in figure O.3, primarily and end-user appliances. support market drivers at the project and portfolio levels, while the remaining four building blocks primarily support • Skills and training: Attracting exceptional talent and the country-level market driver of establishing enabling scaling skills development. environments in key electricity-access-deficit countries. • Institutions and delivery models: Supporting institu- tions, delivery models, and champions to create oppor- The next 10 chapters of this handbook present these 10 tunities. building blocks. Each chapter presents the frontier of knowledge in its topic area and speaks directly to public- FIGURE O.3 • Matrix of market drivers and building blocks to support them Building blocks to support Solar Mini Grid Costs & Technology mini grid development at scale Institutions and Delivery Models Community Engagement Regulations and Policies Companies and Utilities Geospatial Planning Skills and Training Access to Finance Productive Uses Doing Business Market drivers of magnitude changes in scale Reducing costs Increasing the pace of deployment Providing superior quality of service Crowding in government and private-sector finance Establishing enabling environments in key countries Source: ESMAP analysis. Note: The darker the shading in the figure, the more direct the impact a building block is expected to have on a driver. MINI GRIDS FOR HALF A BILLION PEOPLE    27 and private-sector decision-makers working on mini grids Building block 4: Community engagement by trying to answer the “how” question. This building block entails engaging with local communi- ties at every stage of the mini grid development process to Building block 1: Solar mini grid technology ensure community buy-in, promote productive uses, and The objective of this building block is to benchmark and support gender equality, thus increasing the likelihood that analyze mini grid component costs and technologies to the mini grid will operate successfully over the long term. identify and promote opportunities to reduce costs and Chapter 4 presents examples and innovations from lead- improve the quality of mini grid services. Innovations pre- ing mini grid developers on their community engagement sented in the chapter include benchmarking mini grid strategies, and identifies community engagement tactics component costs based on in-depth analysis of more than at every stage of a mini grid project’s life cycle. This build- 400 operational mini grids in Africa and Asia, identifying ing block directly supports the superior quality of the ser- trends in costs and technologies, modeling the LCOE of vice market driver, and indirectly supports the pace of the solar and solar-diesel hybrid mini grids, and understand- mini grid deployment market driver by tackling an inherent ing the impacts of productive uses and subsidies on LCOE. paradox—community engagement is typically time and This building block directly supports the reduction of the resource intensive but the sector needs to scale quickly— mini grid costs market driver, and indirectly supports the by presenting innovative processes and strategies to con- pace of the mini grid deployment market driver by identi- duct community engagement at scale. fying technologies that decrease the time it takes to build a mini grid. Building block 5: Companies and utilities This building block aims to accelerate private-sector partic- Building block 2: Geospatial planning ipation in the deployment of mini grids, while strengthening The goal of deploying the latest geospatial analysis tools national-utility-led approaches to mini grid development in and techniques is to support national-scale least-cost elec- countries where this approach has a proven track record. trification planning and portfolio-scale mini grid design. Engaging the private sector also means facilitating deal Chapter 2 presents frontier knowledge on how geospatial making and collaboration between local and interna- analysis is making national electrification planning more tional industry players in the mini grid market. Chapter 5 precise and credible, thereby giving it more weight for pol- describes the industry’s value chain and companies tak- icy makers and other decision-making stakeholders, and ing part in various stages of the value chain, calculates the presents an innovative process to use geospatial analysis profit potential along the value chain, provides examples of alongside other portfolio planning and auctioning tools to effective deal making and collaboration between local and quickly and efficiently (in terms of cost per site) develop international players, and presents the results from nation- large portfolios of mini grids. Geospatial planning directly ally representative surveys of mini grid developers in three supports the pace of mini grid deployment by enabling a countries, giving readers a sense of what second-genera- portfolio approach to mini grid development, and indirectly tion mini grids look like. This building block directly sup- supports the reduction of the mini grid cost market driver ports the crowding in of the private-sector finance market by helping developers design mini grids that are appropri- driver by supporting collaborations between local and ate for their respective sites. international companies, and indirectly supports the pace of the mini grid deployment market driver by supporting Building block 3: Productive uses the upstream and downstream elements of the mini grid Given the vital importance of productive loads for the cost industry value chain. efficiency and sustainability of mini grids, ensuring that productive uses are a priority for the design and implemen- Building block 6: Access to finance tation of mini grid programs and facilitating their availabil- The objective of this building block is to develop financial ity to mini grid customers are the objectives of this building packages that complement private-sector debt and equity block. Chapter 3 answers the questions of how to increase and crowd in large private-sector investors. Chapter 6 pres- a mini grid’s load factor through productive uses of elec- ents frontier innovations in grants, equity, and debt that tricity, what impacts this can have on a mini grid’s profit- can address the different barriers that mini grid developers ability and long-term sustainability, and how to increase face when trying to finance their projects and portfolios, productive-use appliance uptake by mini grid customers. provides financing options for utility-owned mini grids and This building block directly supports the quality of the ser- public-private partnerships, and details the World Bank’s vice market driver, and indirectly supports the reduction in mini grid portfolio. In addition to directly supporting the the mini grid costs market driver because of the strong link crowding in of the private-sector and government invest- between increasing load factor and decreasing LCOE, as ment market driver, this building block indirectly supports discussed earlier. the reduction of the mini grid costs market driver by help- 28   MINI GRIDS FOR HALF A BILLION PEOPLE ing finance large portfolios of mini grids that lead to econ- presents innovative solutions to reduce red tape in two key omies of scale. areas: the long and complicated processes for environmen- tal and social approvals and for interactions with the main Building block 7: Skills and training grid, and the high costs of connecting with suppliers and The objective of this building block is to support training for investors and participating in tenders. In addition to directly key stakeholders in the mini grid sector, including develop- supporting the enabling environments market driver, this ers, financiers, technicians, regulators, and policy makers. building block indirectly supports the crowding in of the pri- Chapter 7 identifies existing training programs and training vate-sector finance market driver by making national mini program gaps for key mini grid sector stakeholders, and grid markets more attractive to private-sector investors. highlights innovative technologies and methods for large- scale training programs—an essential element of scaling up in mini grid deployment in key electricity-access-deficit GLOBAL MARKET SNAPSHOT, countries. This building block directly supports the enabling OUTLOOK 2030, AND CALL TO ACTION environments market driver by increasing human capacity across the mini grid ecosystem, and indirectly supports the Where we are, where we are headed, and where we pace of the mini grid deployment market driver by increas- need to go ing the availability of high-skilled, knowledgeable stake- The mini grid market globally is undergoing seismic trans- holders who can support a portfolio approach to mini grid formations, particularly as a source of electricity in populous development. countries with little current access. Many electricity-ac- cess-deficit countries are pursuing holistic approaches to Building block 8: Institutions and delivery models electrification, including main grid extension, mini grids, This building block aims to ensure that the agencies and solar home systems. Market analysis from Bloomberg responsible for implementing a mini grid program have New Energy Finance (BNEF 2018) predicts that in the next the required mandate and capacity and that collaboration five to seven years, decentralized renewables—both mini among agencies happens in the most effective way possible. grids and solar home systems—will bring electricity to Chapter 8 identifies the institutions both within and outside more people every year than extensions of the main grid. the energy sector with which mini grid developers interact, This section examines some key elements of the global as well as the relationships among these institutions, and mini grid sector. In particular, it presents data on installed presents country-level models that can support mini grid and planned mini grids to understand three important development at scale and the institutional arrangements questions for mini grid development that motivate the therein. This building block directly supports the establish- remainder of this report: ment of enabling environments for the private-sector mini grids market driver and, as a result, indirectly supports the • Where are we today, in terms of number of mini grids, crowding in of the private-sector finance market driver. number of connections, investment, and capacity? • Where are we headed if we keep up the current pace of Building block 9: Regulations and policies mini grid development? The goal of this building block is to ensure that mini grid • How big is the gap between where we are headed and regulations are clear, light-handed, and conducive to where we need to go to achieve the mini grid portion of private-sector participation in national-level mini grid SDG 7 (“universal access to modern energy services”) markets. Chapter 9 presents the five key decisions for reg- by 2030?6 ulators: market entry, tariffs, service standards, technical specifications, and main grid arrival. Recognizing that there Table O.3 provides a global overview of the installed and is no one-size-fits-all solution in these decision areas, the planned mini grid projects around the world.7 For the pur- chapter presents a decision tree approach as well as a way poses of our database, mini grids were defined as electricity for regulations to evolve as the sector evolves. This building systems that have both electricity generation and distribu- block directly supports the establishment of enabling envi- tion infrastructure, either isolated from the main grid or, ronments for the private-sector mini grids market driver if connected to the main grid, capable of operating as an and in doing so indirectly supports the crowding in of the electrically separate, or “islanded,” mini grid. Both alternat- private-sector finance market driver. ing current and direct current mini grids are included. To be included in our database, a mini grid had to serve multiple Building block 10: Doing business customers. Installed mini grids are those we know to have This building block focuses on reducing red tape and been built. Planned mini grids are those that developers, increasing the ease with which mini grid developers can do governments, and other organizations have said they plan business in each country where they operate. Chapter 10 to build over the next several years. MINI GRIDS FOR HALF A BILLION PEOPLE    29 TABLE O.3 • Installed and planned mini grid projects worldwide: A summary Number of Number of connections Number of Average capital Total capacity Total investment Totals calculated mini grids (millions) people (millions) cost (US$/kW) (MW) (US$, millions) Global totals: installed 21,557 10.3 47.9 3,955 7,224 28,571 Global totals: planned 29,353 8.0 35.4 3,501 2,657 9,304 Grand total 50,910 18.3 83.2 3,833 9,881 37,874 Source: ESMAP research and analysis; proprietary and in-house databases of mini grid projects from Bloomberg New Energy Finance, CLUB-ER, Guide- house, Infinergia, and Sustainable Energy for All; and, unpublished World Bank surveys. This book’s website provides a full list of sources. Note: A cascading process was followed to fill gaps in the data. When a country has many projects with usable data for a given metric (for example, connections per mini grid), that country’s median value for that metric was used to fill in gaps for the remaining projects in that country. When only few projects in a country had usable data for a given metric, or no data at all, that region’s median value for that metric was used to fill data gaps. Finally, in the rare cases when no data existed at the country or regional level for a particular metric, gaps were filled using the global median value for that metric. Mini grids smaller than 1 kW were excluded from the database, but no strict maximum capacity was used to exclude large mini grids. That said, of the more than 50,000 mini grids in our database, only 73 had an installed capacity of greater than 15 MW, and we used median values in our analyses in- stead of averages to minimize the risk that outliers skewed results. The data may be skewed toward mini grids that have a renewable energy component, as data are more abundant for renewable and hybrid mini grids. As a result, the data may underestimate the total number of mini grids globally, and may overestimate the capital costs. The data are incomplete for a number of countries where there are likely to be large numbers of mini grids, particularly countries in Eastern Europe, North Africa, and Asia. kW = kilowatt; MW = megawatt. Leveraging proprietary data from three leading market research firms—Guidehouse, BNEF, and Infinergia—as well as a large database compiled by SEforALL and BNEF, ESMAP identified 21,557 mini grids installed World Bank surveys of mini grid operators, and extensive in 131 countries and territories around the desk research, ESMAP identified 21,557 installed and world, providing electricity to 48 million people, and 29,353 planned mini grids in 138 countries and territo- an additional 29,353 mini grids being planned in 77 ries.8 countries that are expected to provide electricity to almost 35 million people. Mini grids currently provide electricity to about 48 mil- lion people worldwide. Mini grids that are currently being planned are expected to bring electricity to an additional 35 and deceleration in fossil fuels. And, almost 99 percent of million people. To put this number in perspective, the com- all planned mini grids are solar or solar hybrid. bined total number of people connected to, or expected to be connected to, a mini grid—about 83 million—is approxi- The mini grid market currently represents almost $29 mately equal to the entire population of Germany. billion of cumulative investment in capital costs, with an additional $9 billion capital cost investment expected for Installed mini grids have a combined power capacity of mini grids currently being planned. The total $38 billion more than 7 gigawatts (GW); however, the total opera- represents an average investment cost of around $2,100 tional capacity is almost certainly lower, since many mini per connection, though this is skewed higher by more grids—particularly small hydro—do not operate at their costly systems serving fewer customers in high-income full capacity.9 A further 2.7 GW of installed capacity is countries. Planned mini grids are expected to be much expected from mini grids currently being planned. less costly, with an average expected investment globally In terms of generation technology, the trend toward solar is of around $1,200 per connection, though this figure varies already clear today, and is accelerating. Approximately 51 by region. percent of installed mini grids are solar or solar hybrid, fol- Table O.4 summarizes the installed mini grid projects lowed by those powered only by hydro (35 percent), fossil regionally and table O.5 shows the breakdown of installed fuel (10 percent), and other generation technologies such and planned mini grids by region. as wind or fuel cells (5 percent). The trend is also accelerat- ing: more than 10 times as many solar mini grids were built Asia has the most mini grids installed and planned, with every year from 2016 to 2020 than fossil fuel mini grids. a combined total of 16,819 installed mini grids and 19,824 Meanwhile, from 2010 to 2014, by comparison, about three planned mini grids across South Asia and East Asia and times as many solar mini grids were built every year than Pacific. With 3,174 mini grids installed and 9,006 mini grids fossil fuel mini grids. This is a major acceleration in solar planned, Africa has many more mini grids than the com- 30   MINI GRIDS FOR HALF A BILLION PEOPLE The trend toward solar is clear already and Asia—including South Asia and East Asia is accelerating. Approximately half of all and Pacific—has a combined total of 16,819 installed mini grids to date are powered by solar, installed mini grids and 19,824 planned mini grids, but nearly 99 percent of all planned mini grids will which is 78 percent of all installed mini grids and be solar or solar hybrid. 68 percent of all planned mini grids in the ESMAP database. Meanwhile, Africa has more mini grids installed (3,174) and planned (9,006) than all other regions outside of Asia combined. bined total of all other regions outside of Asia. It is import- ant to note that the number of mini grids we identified as being planned does not equal the total market potential for mini grids. Planned mini grids have already secured or been ($6 billion), and Europe and Central Asia ($6 billion). The allocated funding. high investment figures for these regions are explained by several factors, including relatively high up-front capital Mini grids provide electricity to about 18 million people in costs (for example, Africa and Europe and Central Asia), Asia, 27 million people in Africa, and 2 million people in Latin relatively large mini grids (for example, United States and America. A further 14 million people in Asia and 20 million Canada), and a large number of mini grids (for example, people in Africa are expected to receive electricity from mini East Asia and Pacific). South Asia leads the market share grids currently being planned. In Asia and Africa, this rep- for planned mini grids, with approximately $2.8 billion, fol- resents a small but significant percentage of the region’s lowed by Africa ($2.4 billion) and the United States and total population, using World Bank population data: the Canada ($1.6 billion). combined total of installed and planned mini grids in Africa would connect less than 3 percent of the region’s current Table O.6 presents top-10 lists of countries and companies population; in Asia, installed and planned mini grids would across a set of key mini grid metrics, focusing on installed connect less than 1 percent of the region’s population. mini grids. With almost 2 GW of installed mini grid capacity, Africa Six countries in the database have more than 1,000 has the most installed capacity of any region, followed by installed mini grids: Afghanistan, Myanmar, India, Nepal, the United States and Canada (1.8 GW) and East Asia and China, and Indonesia. Afghanistan has the most mini grids Pacific (1.5 GW). South Asia leads the world in planned mini of any country in the database, with more than 4,700 grid capacity (0.87 GW), followed by Africa (0.66 GW) and installed mini grids. The top 10 countries account for 84 the United States and Canada (0.50 GW). percent of all installed mini grids. Total cumulative investment in mini grids is spread out Installed mini grids in Afghanistan, the Democratic Repub- evenly among the top four regions: Africa ($7 billion), lic of Congo, and Madagascar serve electricity to about 19 United States and Canada ($6 billion), East Asia and Pacific million people, which represents about 40 percent of all TABLE O.4 • Summary of installed mini grid projects by region Number of Number of Number of connections people Total capacity Total investment Region mini grids (millions) (millions) (MW) (US$, millions) South Asia 9,592 2 12 407 1,555 East Asia and Pacific 7,227 2 6 1,530 6,271 Africa 3,174 6 27 1,960 7,238 Europe and Central Asia 624 <1 1 1,110 6,092 United States and Canada 615 <1 1 1,783 6,447 Latin America and the Caribbean 286 <1 2 390 810 Middle East and North Africa 39 <1 <1 46 158 Source: ESMAP analysis. Note: Data remain scarce for the Europe and Central Asia, Latin America and the Caribbean, and Middle East and North Africa regions, where there are likely to be many more mini grids than this table has captured. kW = kilowatt; MW = megawatt. MINI GRIDS FOR HALF A BILLION PEOPLE    31 TABLE O.5 • Number of installed and planned mini people served by mini grids today. Collectively, the top 10 grids by region countries in terms of people served by mini grids account for about 69 percent of all people served by the mini grids Installed Planned in the database. South Asia 9,592 19,035 The United States has the highest total capacity of installed East Asia and Pacific 7,227 789 mini grids of any country for which data are available, at 1.4 Africa 3,174 9,006 GW. This is a result of the relatively large number of mini grids identified in this country, and each mini grid tends Europe and Central Asia 624 226 to have a relatively large capacity compared with mini United States and Canada 615 198 grids in other countries. With 671 MW of installed capacity Latin America and the Caribbean 286 88 for installed mini grids, Russia is second in the top-10 list. Nearly all of this capacity is from 500 diesel-powered mini Middle East and North Africa 39 11 grids in remote parts of Russia operated by a regional utility Source: ESMAP analysis. company, RAO Energy. TABLE O.6 • Top-10 lists for key mini grid indicators for installed mini grids Number of people Total Total Utility portfolios Private-sector Number of (millions, % of capacity investment (country, installed portfolios (country, minigrids population) (MW) (US$, billions) mini grids) installed mini grids) 1 Afghanistan Afghanistan United States United States RAO Energy BRAC (4,712) (7, 19%) (1,424) (4.9) (Russia, 500) (Afghanistan, 356) 2 Myanmar Congo, Dem. Rep. Russian Russian NPC-SPUG Husk Power (4,016) (7, 8%) Federation Federation (Philippines, 278) (India, 300+) (671) (3.7) 3 India Madagascar China China NIGELEC OMC (3,192) (5, 15%) (529) (1.9) (Niger, 115) (India, 280) 4 Nepal Tanzania (3, 5%) Congo, Dem. Philippines JIRAMA Tata Power (1,541) Rep. (363) (1.8) (Madagascar, 110) Renewable Microgrids (India, 163) 5 China Kenya Canada Canada Eskom MeshPower (1,236) (3, 5%) (359) (1.6) (South Africa, 100) (Rwanda, 85) 6 Indonesia Burkina Faso Philippines Congo, Dem. CREDA Optimal Power (1,190) (2, 10%) (338) Rep. (1.3) (India, 32) Solutions (India, 59) 7 Senegal India Angola Angola EEU NS RESIF (677) (2, <1%) (333) (1.2) (Ethiopia, 32) (Senegal, 53) 8 Russian Philippines Madagascar India KPLC Sud Solar Federation (1, 1%) (253) (0.9) (Kenya, 32) (Senegal, 50) (501) 9 United States Nepal Kenya Madagascar Energie du Mali Jumeme (478) (1, 5%) (239) (0.9) (Mali, 30) (Tanzania, 42) 10 Philippines Myanmar Australia Australia Alaska Village Yoma Micro Power (455) (1, 2%) (217) (0.9) Electric Coop. (Myanmar, 42) (United States, 25) Total 17,998 33 4,724 $19 1,256 1,132 (% global total) (84%) (69%) (65%) (67%) (6%) (5%) Source: ESMAP analysis. kW = kilowatts; MW = megawatts; n.a. = not applicable; NPC-SPUG = National Power Corporation Small Power Utility Group. 32   MINI GRIDS FOR HALF A BILLION PEOPLE For mini grids built as part of a developer’s portfolio, the average size of the portfolio is 33 mini grids. We defined a Six countries have more than 1,000 installed portfolio as a collection of more than two mini grids built mini grids each: Afghanistan, Myanmar, India, by the same entity. Only 258 portfolios of mini grids were Nepal, China, and Indonesia. Mini grids serve more identified. Portfolios of planned mini grids, however, were than 2 million people in seven countries: Afghanistan, an order of magnitude larger than portfolios of installed the Democratic Republic of Congo, Madagascar, Tan- mini grids. zania, Kenya, Burkina Faso, and India. Planned mini grids are expected to be larger than installed mini grids. The median installed mini grid serves 137 con- Seven countries have seen more than $1 billion of cumula- nections, while the median planned mini grid serves 386 tive investment in mini grids, led by the United States, for connections. Similarly, the median capacity of installed a combined investment of $16 billion. These seven coun- mini grids is 123 watts (W) per connection, compared to tries account for around 57 percent of the market, with the 245 W per connection for planned mini grids. To be clear, United States alone accounting for 17 percent. the capacity per connection numbers here do not reflect the capacity that every customer is guaranteed at any The utility company with the largest portfolio of installed given time. Nor are they meant to represent the tier of ser- mini grids is RAO Energy, which provides power to remote vice provided by the mini grid. Instead, they are calculated areas of the country. The second largest, National Power as total installed capacity divided by total number of con- Corporation Small Power Utility Group (NPC-SPUG), is nections. That said, the capacity per connection of installed the national utility in the Philippines, a country with more mini grids varies by two orders of magnitude across than 7,600 islands. Notably, 6 of the 10 largest utilities regions—the capacity per connection of installed mini grids by number of installed mini grids are in Africa: NIGELEC, in the United States and Canada is more than 100 times JIRAMA, Eskom, EEU, KPLC, and Energie du Mali. The larg- higher than in South Asia—likely as a result of differences est private-sector developer is BRAC in Afghanistan, but across regions in household income, and therefore ability four of the top 10 private-sector developers are in India: to pay for electricity. Tata Power Renewable Microgrids, OMC, Optimal Power Solutions, and Husk Power. One area that our data are likely to underestimate is the size of the diesel mini grid market. The primary databases Though not shown in the table, the largest portfolios of and sources used to compile the data set focus principally planned mini grids are all vertically integrated private-sec- (but not exclusively) on mini grids that contain a renew- tor developers, led by Tata Power Renewable Microgrids able energy generation source. However, thousands of (more than 9,800 planned in India), Husk Power (5,000 diesel-fired and other nonrenewable mini grids are likely in mini grids across India and Africa), OMC Power (5,000 mini operation today, for which no data are available. grids in India), Engie PowerCorner (2,000 mini grids across Africa), and Renewvia (700 across Africa). One way to estimate the number of diesel-fired mini grids Table O.7 provides a snapshot of the characteristics of the is to use global estimates for diesel generator shipments installed and planned mini grid projects around the world. using trade statistics tracked by the United Nations. In TABLE O.7 • Characteristics of installed and planned mini grids Mini grids per People per mini Connections per Capacity per Capacity per Totals calculated portfolio* grid mini grid connection (watts) mini grid (kW) Global totals: installed Median 6 1,040 137 123 20 Average 33 1,524 291 371 540 Number of observations (N) 258 portfolios 7,489 mini grids 9,601 mini grids 8,659 mini grids 19,670 mini grids Global totals: planned Median 20 780 386 245 147 Average 544 1,836 853 405 1,304 N 45 portfolios 23,827 mini grids 26,189 mini grids 24,471 mini grids 26,758 mini grids Source: ESMAP analysis. *A portfolio is defined as a collection of more than two mini grids built by the same developer. kW = kilowatt MINI GRIDS FOR HALF A BILLION PEOPLE    33 its analysis of these data, BNEF found that in the 10-year able to make updates meaningful, but is frequent enough period 2008–17, 92.49 GW of diesel generator capac- to capture trends as they happen. The database developed ity—from generators with a capacity of less than 375 for this chapter will facilitate this effort: based in Microsoft kilovolt-amperes, or 300 kW, assuming a power factor of Excel, aggregate country-level data can be shared to iden- 0.8—was shipped around the world. If we assume that 5 tify areas for improving the accuracy and completeness of percent of this capacity is used to power diesel-only mini the data. grids (4.6 GW), and that the average diesel capacity per mini grid is 150 kW, then around 30,800 diesel-only mini PROJECTIONS grids may be currently installed today. Meanwhile, our This section explains the gap between where we are database identified only about 1,400 diesel-only mini grids. headed now, in a business-as-usual (BAU) case, and where Hybridizing existing diesel-powered mini grids by adding we need to go to achieve universal access to electricity by solar PV and battery capacity represents a market oppor- 2030. ESMAP estimates that under the right conditions, tunity of $7–$18 billion. In the database, diesel-only mini mini grids have the potential to be the least-cost way to grids had a total combined capacity of 1.8 GW. As men- provide electricity to almost half a billion people by 2030 tioned above, this is likely to underestimate the global total. (figure O.4). Therefore, it can be assumed that the total installed capac- ESMAP developed four scenarios for mini grid deployment ity for diesel-only mini grids is around 1.8–4.6 GW (using between now and 2030. Each is described in turn. the BNEF analysis of diesel genset trade data). Using an estimate of $4,000/kW for the investment costs of hybrid- ESMAP Mini Grid Outlook Scenario izing diesel-fired mini grids, the total market opportunity Under this scenario, mini grids are the least-cost option for for hybridizing diesel mini grids is $7–$18 billion. 430 million people to receive electricity for the first time, A market snapshot like the one presented above should and an additional 60 million people will be serviced through be conducted every three years. This allows for enough an interconnected network with mini grids due to reliabil- preparation time and sufficient new data to become avail- ity issues on the main grid or to increase resilience in the FIGURE O.4 • Number of people connected to mini grids under business-as-usual and universal access scenarios, 2020–30 500 450 400 350 Millions of People 300 250 200 150 100 50 0 2020 2021 2022 2023 2024 2025 2026 2027 2028 2029 2030 ESMAP Business As Usual SDG7 Tracking Stated Policies SDG7 Tracking Universal Access ESMAP Mini Grid Outlook (Mini Grids for Access) ESMAP Mini Grid Outlook (Mini Grids for Resilience) Source: ESMAP analysis. ESMAP = Energy Sector Management Assistance Program; SDG 7 = Sustainable Development Goal 7. 34   MINI GRIDS FOR HALF A BILLION PEOPLE face of climate shocks/severe weather. The resulting total together with relevant policy proposals not yet enacted. is 490 million people connected to more than 217,000 mini This scenario sees 260 million people gaining access to grids at a cumulative investment cost of approximately electricity between 2022 and 2030 (IEA 2021). We then $127 billion. These projections are based on the following conservatively estimate that 31 percent of these people will considerations. be served by mini grids—the same proportion used by the 2021 SDG 7 Tracking Report under the Sustainable Devel- • ESMAP ran country-specific scenarios for the 58 coun- opment Scenario (IEA and others, 2021, p. 160), resulting tries with data in the Global Electrification Platform and in 81 million people served by about 45,300 mini grids at a found that, under enabling circumstances, 430 million cumulative investment cost of $36 billion. people can be best served at least cost by mini grids by 2030, at a cumulative investment cost of about $105 The SDG 7 Tracking Sustainable Development billion.10 This assumes modest cost declines in key Scenario components such as batteries and solar PV and the This scenario is based on the IEA’s Sustainable Develop- main grid expanding at a rate of 2.5 percent of the pop- ment Scenario in World Energy Outlook 2021 (IEA 2021), ulation per year. which takes universal access to electricity as the point of • In addition, as more cities, islands, and utility compa- departure and thus sees 930 million people receiving elec- nies consider risks of extreme weather and invest in tricity access between 2022 and 2030, after taking into more resilient infrastructure, we expect to see more and account population growth. This scenario identifies mini more transitions toward interconnected mini grids that grids as the least-cost electrification pathway for 31 percent can isolate from the network in “island mode” if needed. of new connections (IEA and others 2021, p. 160), which This is complemented by grid-connected towns and ESMAP’s analysis and estimates indicate results in 288 mil- communities investing in grid-connected mini grids in lion people connected to approximately162,000 mini grids order to increase the fraction of renewable energy sup- at a cumulative investment cost of around $93 billion. plying their electricity. The team estimates that these Table O.8 presents a regional breakdown of the ESMAP resilience- and renewable-motivated mini grids could mini grid outlook scenario. serve about 2–3 million new connections (about 6–7 million people) per year globally, equivalent to about Upon analysis of the regional breakdown of the global 10–15 cities or small regional utilities per year deciding ESMAP Mini Grid Outlook Scenario, several important to strengthen their power systems by developing inter- points stand out. First, the largest number of mini grids, connected micro/mini/metro grids. Using costs from and associated investment, will be needed in Africa. microgrids in high-income countries, the cumulative Indeed, almost 80 percent of all access-related invest- expected investment by 2030 for these additional mini ment for mini grids between now and 2030 will need to grids is about $22 billion. go to Africa to achieve SDG 7 by 2030. This means con- necting 380 million people to 160,000 mini grids at a The ESMAP Business-As-Usual Scenario cost of about $91 billion. By contrast, the total number of The BAU scenario assumes that development in 2021– mini grids and investment required in Latin America and 30 follows the same linear growth trajectory that was the Caribbean is small relative to other regions because observed in the 2010–20 data in the ESMAP database, of this region’s current high energy access rates, with the for number of people served by mini grids, and uses exception of Haiti. Finally, mini grids for resilience and actual data from planned mini grids to estimate the total increased penetration of renewable energy represent a number of mini grids and total cumulative investment by major market opportunity, accounting for 12 percent of 2030. The 2021 baseline starting points for the scenario people connected to mini grids and 18 percent of cumu- are the totals from the database: 21,557 mini grids, 48 mil- lative investment, by 2030. lion people served by mini grids, and $29 billion of invest- ment. The results for 2030 are 80 million people served The gap between even the most optimistic BAU and univer- by almost 44,800 mini grids at a cumulative investment sal access scenarios is still vast. The gap for energy access cost of approximately $37 billion. alone (that is, counting only mini grids needed for providing first-time access to electricity, and not counting additional The SDG 7 Tracking Stated Policies Scenario mini grids built for resilience), using only ESMAP’s scenar- The basis for this scenario comes from the IEA’s World ios, is 382 million people, $76 billion, and 183,000 mini grids. Energy Outlook 2021, which developed a “Stated Policies The purpose of the remainder of this book is to identify Scenario” that accounts for policies and initiatives adopted concrete ways to bridge this gap. as of mid-2021 that have an impact on energy access, MINI GRIDS FOR HALF A BILLION PEOPLE    35 TABLE O.8 • ESMAP mini grid outlook scenario: A regional breakdown Population connected to Cumulative investment in Total number of mini mini grids (millions) mini grids (US$, billions) grids installed Region 2021 2030 2021 2030 2021 2030 Africa 27 380 7 91 3,100 160,000 South Asia 12 24 2 3 9,600 27,000 East Asia & Pacific 6 19 6 9 7,200 15,000 Latin America & Caribbean 2 6 <1 1 300 1,800 Rest of World including new mini grids for 2 60 13 22 1,400 12,300 resilience and renewable energy penetration Total 48 490 29 127 21,500 217,000 Source: ESMAP analysis. is made even more impressive given that Indonesia is an archipelago of 6,000 inhabited islands characterized by If the current pace of mini grid development jungles, mountainous terrain, and limited transport and continues, about 44,800 mini grids will be communication infrastructure. installed by 2030, serving around 80 million people. However, achieving universal access to clean and 2020 progress. In 2019–20, Tata Power Renewable Micro- reliable electricity by 2030 will require more than grids was able to achieve a similar pace of development 217,000 mini grids serving 490 million people, at a in India as the Indonesia program, constructing more cost of around $127 billion. For energy access alone, than 150 mini grids in a little more than a year. The rapid not counting new mini grids for resilience, this rep- pace of development for these projects was greatly aided resents an expected shortfall by 2030 of 382 million by the choice to use a standardized design for mini grids. people, $76 billion, and 183,000 mini grids. Standardization improved the efficiencies of carrying out tenders and enabled developers to bid on multiple sites knowing they would be installing the same type of equip- ment across all sites. Note that this does not mean that STATUS OF THE FIVE DRIVERS AND TEN every mini grid should be the same size but, instead, that BUILDING BLOCKS standardization across components facilitates modular ESMAP has begun tracking the mini grid industry’s prog- mini grid design. ress against the 5 drivers and 10 building blocks, to gauge the pace of development against the 2020 targets and Indicators embedded in mini grid development show how assess whether the sector is on track to meet the 2030 tar- long a build will take. From our conversations with AMDA, gets. More comprehensive stocktaking will be needed over we know that some are able to develop mini grids in around the next few years in order to gain a more accurate under- six weeks once initial site identification and assessment standing of the whole industry’s progress. work are completed. But the length of time from placing a purchase order to commissioning for the typical solar-die- Increasing the pace of deployment sel hybrid mini grid is usually measured in months.11 In The targets for the pace of mini grid development are general, components arrive on site at different times, while derived from what would be needed to achieve SDG 7 in delays in customs can set projects back days or weeks. each of the top 20 countries lacking access to electricity. Only a handful of the largest developers have systematized The pace of development grows from around 150 mini grids construction and installation in ways that allows quick and per country per year in 2020 to around 2,000 per country efficient deployment. The pace of deployment must speed per year by 2030 (figure O.5). up if we are to proceed from building tens of mini grids a year to building thousands by 2030. 2018 benchmark. A program in Indonesia supported by the German Society for International Cooperation (GIZ) set the One key innovation that has already been deployed to benchmark for the pace of development of mini grid port- reduce the setup time for individual mini grids is contain- folios. This Indonesia Solar Mini-Grid Programme installed erization and the associated standardization of mini grid 236 mini grids in just more than two years, from 2012 to components—the upstream integration of standardized 2014, setting the pace at approximately 100 mini grids per major mini grid components into one or two shipping con- year (Schultz, Suryani, and Puspa 2014). This achievement tainers,12 which are then delivered, unpacked, and installed 36   MINI GRIDS FOR HALF A BILLION PEOPLE FIGURE O.5 • Mini grids installed annually in each of the top 20 electricity-access-deficit countries, 2018–30 2,000 1,500 Number of Mini Grids 1,000 500 0 2018 2019 2020 2021 2022 2023 2024 2025 2026 2027 2028 2029 2030 target actual target trendline actual trendline Source: ESMAP analysis. at the mini grid site. Indeed, the 2030 target for solar-die- is the average load divided by the peak load in a specified sel hybrid mini grids—working toward a five-week time- time period. The more productive uses of electricity a mini frame from purchase order to commissioning—is already grid serves during the day, the higher the mini grid’s load possible for some containerized mini grids. Both large factor, and the more economically viable the mini grid. multinational companies like General Electric and ABB, as Boosting productive uses of electricity also contributes to well as smaller, more specialized companies like Redavia, the economic development of the communities in which BoxPower, and Nayo Tropical Technology, already have off- mini grids operate and helps entrepreneurs and small busi- the-shelf containerized hybrid mini grids at the tens to hun- ness customers get the most value for their money from dreds of kilowatts scale. While not a silver bullet, this type of their connection to the mini grid. For the mini grid industry, containerization—combined with standardization of mini increasing productive uses of electricity means increasing grid components and improved efficiencies in construct- demand for mini grid electricity—a necessary component ing the distribution network—will have to transition from of growing at scale. breakthrough technology to industry norm. 2018 benchmark. For the 2018 benchmark, we use Increasing the pace of deployment for portfolios of mini HOMER’s default load factor of 22 percent. grids will also require systematized construction and project management processes. The same practices and 2020 progress. While data remain scarce on load factor for processes in use today by large construction firms that a sizeable cohort of mini grids, recent analysis of new solar manage portfolios of hundreds of small- and medium-size hybrid mini grids in Haiti offers a reference point for 2020 projects are translatable to both private-sector developers progress. These mini grids were able to achieve a load fac- and public utility companies as they seek to scale up from tor of 30 percent, due in part to the fact that they served tens to hundreds of mini grids a year by 2030. large towns with significant daytime economic activity. This is ahead of pace toward the 2030 target of 45 percent, as Providing superior service figure O.6 shows, although the 2030 target is the industry LOAD FACTOR average. As shown in chapters 1 and 3, the viability of a mini grid Achieving the load factor targets will require integrating depends on productive-use customers, those who use the imperative to promote productive uses from the outset electricity at off-peak, typically daytime, hours. This makes of every mini grid project’s development process. In addi- intuitive sense: if the mini grid is only able to sell electricity tion, it will require developers to address the appliances’ during the evening peak hours, it is earning revenue only up-front costs by partnering with local financial institutions during this limited time period. One way to determine how such as microfinance institutions or selling customers well a mini grid is performing in terms of selling electricity the appliances on credit paid back through on-bill financ- during off-peak times is a metric called load factor, which ing. To achieve the productive-use targets, developers will MINI GRIDS FOR HALF A BILLION PEOPLE    37 FIGURE O.6 • Average mini grid load factor, 2018–30 50 45 (Average Demand / Peak Demand) 40 35 Load Factor 30 25 20 15 10 5 0 2018 2019 2020 2021 2022 2023 2024 2025 2026 2027 2028 2029 2030 target actual target trendline actual trendline Source: ESMAP analysis. also need to work directly with appliance suppliers to seek could agree with national governments and regulators on attractive commercial arrangements in terms of both price minimum performance requirements for mini grid batter- and after-sales service. Finally, developers must earn the ies, the mini grid market would become more attractive to trust of productive-use customers. Only if customers trust manufacturers of batteries that meet those specifications. that electricity will be supplied to them reliably and in the Third, regulatory and rural electrification agencies tasked long run will they invest in a productive-use appliance or with overseeing a national mini grid program must develop machine powered by the mini grid’s electricity. Meanwhile, their own mini grid technical specifications, which takes governments can also incentivize mini grid developers to time and resources away from other important activities. encourage productive uses of electricity in their mini grids. 2020 progress. Efforts are underway to develop an indus- In Tanzania, for example, Rule 43 of the 2018 mini grid reg- trywide standard for mini grid technical specifications. ulations in Tanzania allow developers to factor in the “asso- ESMAP has developed a set of minimum standards for ciated administrative costs” of promoting productive uses technical specifications that take a light-handed approach of electricity in their retail tariffs.13 and prioritize safety, with the goal of providing an off-the- TECHNICAL SPECIFICATIONS shelf product that countries can adopt for their mini grid programs. The ESMAP specifications are in use in Haiti and 2018 benchmark. Mini grid technical specifications typi- Rwanda. cally define the minimum power, safety, engineering, and other technical specifications for mini grid components MINIMUM QUALITY OF SERVICE STANDARDS: UPTIME and installations to which developers must adhere. Recent 2018 benchmark. Several years ago, a handful of leading examples of mini-grid-specific technical specifications can mini grid developers in AMDA’s membership were able to be found in Annex 7 of the Nigeria mini grid regulations,14 achieve around 97 percent uptime on average. This set the as part of a mini grid tender in Kenya,15 and as part of the benchmark, leading to the targets of achieving 97 percent new draft regulatory framework for mini grids in Zambia,16 uptime as the industry standard by 2025, and increasing among many other countries. Technical specifications this level of reliability through 2030.17 tend to differ in each country, if they exist at all as part of a national mini grid program, which creates at least three 2020 progress. In 2020, the average uptime for mini barriers to growth. First, developers who want to build port- grids in AMDA’s membership was already 99 percent for folios of mini grids in different countries cannot aggregate 24/7 electricity, surpassing the 2020 target of 97 percent their component orders if their portfolio spans jurisdictions uptime during promised service hours (AMDA 2021). This with different technical specifications. Second, different sets a high, but attainable, standard for new entrants to the technical specifications restrict the size of the potential market, and helps ensure that mini grids retain their good mini grid market for component manufacturers. Consider reputation as providers of reliable electricity. batteries as an example. If the global mini grid industry 38   MINI GRIDS FOR HALF A BILLION PEOPLE CUSTOMER SATISFACTION tion of development partner and government funding will Successful mini grid developers provide high-quality elec- decrease, from 60 percent or higher today to 30 percent or tricity service, delivering reliable and predictable power lower in 2030. As figure O.7 shows, total cumulative invest- while maintaining close relationships with their customers. ment in mini grids is not on track to reach the 2030 target that is necessary if SDG 7 is to be achieved. 2018 benchmark. In a study from Smart Power India (SPI 2019), mini grids scored 84 out of 100 for small business 2018 benchmark. By 2018, ESMAP data indicate that the customer satisfaction and 82 out of 100 for household cus- total cumulative investment in mini grids in Africa, South tomers. For comparison, the main grid scored 41 out of 100 Asia, East Asia and Pacific, and Latin American and the and 34 out of 100 for these two customer groups, respec- Caribbean stood at around $13 billion. We estimate that at tively (SPI 2019). least 60 percent of this funding came from governments and development partners, equal to roughly $8 billion. 2020 progress. We do not have more recent survey data to provide a quantitative update on 2020 progress for this 2020 progress. By 2020, the total cumulative investment indicator. To grow at scale, the mini grid industry will need in mini grids in these same regions was about $16 billion, to develop and sustain a reputation for high-quality cus- according to data from ESMAP’s database of mini grid tomer service, to ensure that customers feel that they are projects around the world. While we do not have good data receiving value for their money. This will require surveying on the global breakdown of public vs. private sector financ- mini grid customers on a regular basis to ascertain their ing, we estimate that still about 60 percent of this funding satisfaction with the services they receive. We know that came from governments and development partners, based many developers already invest in a variety of activities to on results-based grant programs across the World Bank’s increase customer satisfaction, including call centers for mini grid portfolio. This is behind pace toward 2020 targets, customer support and rapid response to customer com- both in terms of total cumulative investment (about $4 bil- plaints, as well as continued close engagement with the lion short of the $20 billion target for 2020), and in terms of communities they serve. the fraction of funding coming from the private sector (50 percent of funding from private sector sources by 2020). Cumulative investment in mini grids Achieving growth of two orders of magnitude in the global Establish enabling mini grid business environments mini grid industry by 2030 will require an unprecedented in key access-deficit countries level of investment from governments and their develop- Developing mini grids at scale will require major improve- ment partners. Achieving the SDG 7 targets will require ments in the regulatory environment, making it easier for total cumulative investment of around $105 billion in mini mini grids to operate as companies. A light-handed and grids for energy access by 2030, shared between the predictable business climate would address the needs of public and private sectors—but not in equal proportions. mini grids be conducive to private-sector participation in As the mini grid industry grows and matures, the propor- mini grid development. FIGURE O.7 • Total cumulative investment in mini grids for energy access, 2018–30 110 100 90 80 70 US$, b illions 60 50 40 30 20 10 0 2018 2019 2020 2021 2022 2023 2024 2025 2026 2027 2028 2029 2030 target actual target trendline actual trendline Source: ESMAP analysis. MINI GRIDS FOR HALF A BILLION PEOPLE    39 The World Bank tracks these elements of an enabling envi- TABLE O.9 • Top 20 countries with energy access ronment for mini grids in its Regulatory Indicators for Sus- deficits: Doing Business and RISE scores, 2020 tainable Energy (RISE) index (World Bank 2019b), which electricity access (%) covers the regulatory environment for mini grids in more electricity access in Population without population without than 50 countries. To achieve the SDG 7 objective, countries mini grids in 2020 2020 (millions) with energy access deficits must improve their enabling Share of global RISE score for (out of 100) environments, as measured by their RISE scores. Efforts to raise these scores should focus on the 20 countries that account for almost 80 percent of the global population that does not currently have access to electricity (figure O.8). Country Nigeria 85 11 100 2018 benchmark. In 2018, the average RISE score for these Congo, Dem. Rep. 68 9 62 countries was just 59 out of 100. India 64 8 78 2020 progress. By 2020, the average RISE score for the Pakistan 61 8 60 top 20 electricity-access-deficit countries rose to 64 out Ethiopia 60 8 70 of 100, ahead of the 2020 target of 60/100. Table O.9 pro- vides the RISE scores for each of these 20 countries. Tanzania 36 5 100 Uganda 25 3 73 Reducing the cost of solar hybrid mini grids Bangladesh 24 3 80 One chief driver of growth for the mini grid market is the Mozambique 20 3 45 cost to build and operate a mini grid. The typical metric Madagascar 19 2 52 used to combine and quantify these costs is the LCOE, Niger 18 2 73 which also serves as a proxy for the average tariff at which the mini grid must sell its electricity to break even over its Myanmar 18 2 73 lifetime. For mini grid deployment to scale up rapidly, the Angola 17 2 60 LCOE of mini-grid-based electricity will need to plummet Burkina Faso 17 2 42 by 2030. Much lower LCOE would raise market demand Sudan 17 2 37 and speed deployment in low-income areas, where ability Malawi 15 2 77 to pay often limits the potential for mini grids as a solution Chad 14 2 35 for electrification, since mini grid developers often set their tariffs at or below what customers pay per month for alter- Korea, Dem. People’s Rep. 13 2 No Data natives, such as kerosene, car battery- and phone-charging Kenya 13 2 82 services, and diesel gensets. These traditional energy ser- Yemen 11 1 20 vices can be expensive: households and small businesses Total 617 78 Average 64 FIGURE O.8 • Average RISE score in top 20 electricity-access-deficit countries 100 90 80 70 60 (out of 100) RISE Score 50 40 30 20 10 0 2018 2019 2020 2021 2022 2023 2024 2025 2026 2027 2028 2029 2030 target actual target trendline actual trendline Source: ESMAP analysis. 40   MINI GRIDS FOR HALF A BILLION PEOPLE FIGURE O.9 • LCOE of best-in-class solar hybrid mini grids 0.60 0.50 0.40 USD/kWh 0.30 0.20 0.10 0.00 2018 2019 2020 2021 2022 2023 2024 2025 2026 2027 2028 2029 2030 target actual target trendline actual trendline Source: ESMAP analysis. routinely pay the equivalent of $1/kWh or more for lighting Mini grid industry progress across all 10 building and charging services (Tenenbaum and others 2014). blocks Table O.10 presents an overview of the global mini grid The LCOE targets that the mini grid industry can work industry’s progress across all 10 building blocks, based toward—$0.40/kWh by 2020, $0.25/kWh by 2025, and on ESMAP’s internal assessment of energy access deficit $0.20/kWh by 2030 (figure O.9)—are therefore ambi- countries. Note that each country has made its own unique tious but not impossible. They are indicative targets that progress; the table below, meanwhile, gives an overall a cohort of leading mini grid developers can realistically sense of which building blocks have seen more progress achieve, thus setting the pace for other developers in and which generally require more work. the sector. Achieving these targets would bring mini grid electricity close to universal affordability levels by 2030. Key energy-access-deficit countries have made notable In addition, they make the value proposition of mini grids progress across all 10 building blocks over the past decade. that provide reliable power 24 hours a day every day com- Arguably the most progress, particularly in just the past petitive with—if not more attractive than—other options three to four years, has been made in geospatial planning such as backup diesel generation. Achieving these LCOE and in the development of regulatory frameworks and levels would greatly limit, if not eliminate, the need for policies specific to mini grids. Advances in technology, subsidies in areas where customer ability to pay aligns combined with widespread adoption as part of technical with these targets. assistance support to client governments, have main- streamed geospatial analysis at the national level to iden- 2018 benchmark. According to detailed costing analysis tify least-cost options for electrification over the short, that ESMAP conducted in 2019, the LCOE of a best-in-class medium, and long terms. Furthermore, whereas site-spe- solar hybrid mini grid was $0.55/kWh in 2018 (ESMAP cific geospatial analysis was typically used just for feasibil- 2019). ity studies, today’s technologies and service providers are 2020 progress. As we discuss in greater detail in chapter able to identify and analyze thousands of high-potential 1, the LCOE for a best-in-class solar hybrid mini grid was mini grid sites, complete with distribution and generation $0.38/kWh in 2020. This was ahead of the 2020 target and system sizing and costing, and demand estimation that puts the industry on pace to achieve $0.20/kWh by 2030. includes productive uses and public institutions. Countries have also made strong progress on developing mini-grid- On the one hand, hitting all the targets across the five mar- specific regulatory frameworks, whether embedded into ket drivers does not guarantee the 2030 scale of deploy- licenses (for example, Rwanda) or concession contracts ment required to realize the mini grid portion of the SDG 7 (many francophone countries), developed as stand-alone objective. On the other hand, it would make reaching the regulations (for example, Nigeria), or government direc- goals much easier. tives (for example, Ethiopia). We note, of course, that there is still a long way to go toward mainstreaming regulatory MINI GRIDS FOR HALF A BILLION PEOPLE    41 TABLE O.10 • The global mini grid sector and its progress across the 10 building blocks Building block Progress Notable achievements Solar mini grid costs and technologies Large portfolios of solar hybrid mini grids in India Geospatial planning National and portfolio-level planning in Pakistan, Nigeria, Ethiopia Productive uses Productive uses embedded in mini grid planning in Ethiopia Community engagement Signed community agreements in Nigeria, Haiti, Myanmar Companies and utilities AMDA’s growing membership; utility hybridization in Niger, Ethiopia Access to finance Comprehensive financial package in the Democratic Republic of Congo Skills and training Training programs in Mali and Nigeria Institutions and delivery models Strong institutions and private-sector approach in Bangladesh Regulations and policies High RISE scores for mini grids in Nigeria, Kenya, Bangladesh Doing business e-Government initiatives in Ghana and India KEY: Source: ESMAP analysis. Significant work needed AMDA = Africa Minigrid Developers Association; RISE = Regulatory Indicators for Sustainable Energy. Some progress made Significant progress made best practices, particularly in how regulations are imple- • Development partners to work with government coun- mented and the processes underpinning them. terparts and the private sector to create enabling environments for mini grids through investments in When considering the mini grid market globally, more sup- actual portfolios of projects and technical assistance port is needed for productive uses of electricity, community for developing workable regulations and strengthening engagement, skills and training, and the overall business institutions. environment than for the other building blocks. Of these, we see the most positive momentum in productive uses, • Regulators to adopt an evolving, light-handed approach with dedicated organizations like CLASP and EnerGrow, for a maturing mini grid sector, providing at each stage as well as the importance of (and funding for) income-gen- of development clear guidance on market entry, retail erating uses of electricity embedded in a growing number tariffs, service standards, technical standards, and of national mini grid programs. However, progress toward arrival of the main grid. large-scale initiatives to engage with communities and build • The mini grid industry and its associations to work skills, as well as innovative approaches to reducing red tape, toward increasing the pace of deployment, retaining lags behind the progress in other building blocks globally. superior-quality service delivery of third-generation mini We present some ideas for how to accelerate progress in grids, and reducing the cost of these systems through each of these areas in chapters 4, 7, and 10, respectively. innovation to reach a value proposition that is affordable to the end users. CALL TO ACTION • National utilities to adopt an openness to partnerships Connecting half a billion people to mini grids by 2030 is with the third-generation mini grid industry on the basis a monumental task that requires unprecedented levels of that the systems are grid-integration ready, which can investment, innovation, and commitment from develop- provide for more financially viable grid expansion pro- ment partners, governments, and the mini grid industry. grams for the utility in the long run. This book calls for action by stakeholders across the mini grid value chain. Key recommendations are for: Lastly, there is a clear need for accurate, up-to-date, and widely available data to inform any type of initiative that • Policy makers to leverage the latest geospatial analysis supports mini grids. To this end, we strongly recommend technology to develop national electrification plans that the development of a global tracking tool to monitor and can guide investment in mini grids, main grid extension, measure the global mini grid industry’s progress against and solar home systems, as well as develop initiatives the 10 building blocks and 5 market drivers outlined above. that promote productive uses of electricity and build human capital. 42   MINI GRIDS FOR HALF A BILLION PEOPLE REFERENCES Kojima, M., and C. Trimble. 2016. Making Power Affordable for Africa and Viable for Its Utilities. Washington, DC: World Bank. https://open- AMDA (Africa Minigrid Developers Association). 2019. “Untitled Pow- knowledge.worldbank.org/handle/10986/25091. erPoint presentation to the World Bank EEX Week.” Nairobi, Kenya. Rockefeller Foundation. 2018. “Smart Power for Rural Development: AMDA. 2021. Benchmarking Africa’s Mini Grids. Nairobi, Kenya: AMDA. Accelerating Energy Access to Economically Empower and Trans- form Lives.” New York. https://www.rockefellerfoundation.org/initia- Blimpo, M., and M. Cosgrove-Davies. 2019. Electricity Access in Sub-Sa- tive/smart-power-for-rural-development/. haran Africa: Uptake, Reliability, and Complementary Factors for Eco- nomic Impact. Africa Development Forum Series. Washington, DC: Schultz, R., A. Suryani, and A. F. Puspa. 2014. “Executive Overview: Indo- World Bank. https://openknowledge.worldbank.org/bitstream/han- nesia Solar Mini-Grid Programme (PVVP/PLTS Terpusat).” Energis- dle/10986/31333/9781464813610.pdf?sequence=6&isAllowed=y. ing Development Indonesia (EnDev Indonesia), Jakarta, Indonesia. https:/ /energypedia.info/images/1/1e/Indonesia_Solar_Mini-grid_ BNEF (Bloomberg New Energy Finance). 2018. Powering the Last Bil- Programme_EnDev_Executive_Overview_2014.pdf. lion: The Outlook for Energy Access. New York: BNEF. SEforALL (Sustainable Energy for All) and CPI (Climate Policy Ini- BNEF. 2020. “BNEF Solar Spot Price Index.” Proprietary online database. tiative). 2021. Energizing Finance: Understanding the Landscape. CIESIN (Center for International Earth Science Information Network), Washington, DC: SEforALL. https:/ /www.seforall.org/publications/ Columbia University, and Novel-T. 2020. “GRID3 Central African energizing-finance-understanding-the-landscape-2021. Republic Settlement Extents Version 01, Alpha.” Palisades, NY: SPI (Smart Power India). 2019. Rural Electrification in India: Customer Geo-Referenced Infrastructure and Demographic Data for Devel- Behaviour and Demand. Gurgaon, India: Smart Power India and opment (GRID3). Source of building Footprints ‘Ecopia Vector The Rockefeller Foundation. https:/ /www.rockefellerfoundation.org/ Maps Powered by Maxar Satellite Imagery’ .” Accessed June 1, 2022. report/rural-electrification-india-customer-behaviour-demand/. https://doi.org/10.7916/d8-y2ax-p859. Tenenbaum, Bernard, Chris Greacen, and Dipti Vaghela. 2018. Mini Grids Feldman, David, Vignesh Ramasamy, Ran Fu, Ashwin Ramdas, Jal and the Arrival of the Main Grid: Lessons from Cambodia, Sri Lanka, Desai, and Robert Margolis. 2021. U.S. Solar Photovoltaic System and Indonesia. Energy Sector Management Assistance Program Cost Benchmark: Q1 2020. Technical Report NREL/TP-6A20-68925. (ESMAP) Technical Report 013/18. Washington, DC: World Bank. Golden, CO: National Renewable Energy Laboratory. https:/ /www. https://openknowledge.worldbank.org/handle/10986/29018. nrel.gov/docs/fy21osti/77324.pdf. Tenenbaum, B., C. Greacen, and A. Shrestha. Forthcoming 2022. Under- GOGLA. 2019. Global Off-Grid Solar Market Report: Semi-Annual Sales grid Mini Grids in Nigeria and India: Interconnected and Non-Inter- and Impact Data. July–December public report. Utrecht, The Nether- connected. lands: GOGLA. https:/ /www.gogla.org/global-off-grid-solar-market- report. Tenenbaum Bernard, Chris Greacen, Tilak Slyambalapitiya, and James Knuckles. 2014. From the Bottom Up: How Small Power Producers Greacen, Chris, Stephanie Nsom, and Dana Rysankova. 2015. “Scaling and Mini-Grids Can Deliver Electrification and Renewable Energy in Up Access to Electricity: Emerging Best Practices for Mini-Grid Regu- Africa. Directions in Development Series. Washington, DC: World lation.” LiveWire 2015/51, World Bank, Washington, DC. Bank. doi: 10.1596/978-1-4648-0093-1. Hughes, T. P. 1983. “Introduction.” In Networks of Power: Electrification Trimble, C., M. Kojima, I. Perez Arroyo, and F. Mohammadzadeh. 2016. in Western Society, 1880–1930. Baltimore and London: The John “Financial Viability of Electricity Sectors in Sub-Saharan Africa: Qua- Hopkins University Press. si-Fiscal Deficits and Hidden Costs.” Policy Research Working Paper IEA. 2021. World Energy Outlook 2021. Paris: IEA. 7788, World Bank, Washington, DC. http:/ /documents.worldbank. IEA (International Energy Agency), IRENA (International Renewable org/curated/en/182071470748085038/pdf/WPS7788.pdf. Energy Agency), UNSD (United Nations Statistics Division), WB UN (United Nations) 2015. “Transforming Our World: The 2030 Agenda (World Bank), and WHO (World Health Organization). 2021. Track- for Sustainable Development.” General Assembly Resolution 70/1, ing SDG 7: The Energy Progress Report 2021. Washington, DC: World A/RES/70/1 (September 25). https:/ /www.un.org/ga/search/view_ Bank. doc.asp?symbol=A/RES/70/1&Lang=E. IEA (International Energy Agency), IRENA (International Renewable World Bank. 2019b. “Regulatory Indicators for Sustainable Energy Energy Agency), UNSD (United Nations Statistics Division), World (RISE).” http://rise.worldbank.org/scores. Bank, and WHO (World Health Organization). 2022. Tracking SDG 7: World Bank and IMF (International Monetary Fund). 2017. “Maximiz- The Energy Progress Report 2022. World Bank. Washington, DC. ing Finance for Development: Leveraging the Private Sector for https://www.iea.org/reports/tracking-sdg7-the-energy-progress-re- Growth and Sustainable Development.” https:/ /www.devcommit- port-2022. tee.org/sites/dc/files/download/Documentation/DC2017-0009_ International Magazine Co. 1925. “The Location of Places Served with Maximizing_8-19.pdf. Electricity.” 119 West 40th St., New York. IRENA (International Renewable Energy Agency). 2016. Policies and Regulations for Private Sector Renewable Energy Mini-Grids. NOTES Abu Dhabi: IRENA. https://www.irena.org/publications/2016/Sep/ Policies-and-regulations-for-private-sector-renewable-energy- 1. This report defines access to electricity in accordance with the mini-grids. Multi-Tier Framework (MTF), which is elaborated further in this Kairies,Kai-Phillip.2017.“Battery StorageTechnology Improvements and section. Cost Reductions to 2030: A Deep Dive.”PowerPoint presentation at the 2. The 20 countries are Angola, Bangladesh, Burkina Faso, Chad, the International Renewable Energy Agency Workshop, March 17. https:/ / Democratic People’s Republic of Korea, the Democratic Republic of www.irena.org/-/media/Files/IRENA/Agency/Events/2017/ Congo, Ethiopia, India, Kenya, Madagascar, Malawi, Mali, Mozam- Mar/15/2017_Kairies_Battery_Cost_and_Performance_01.pdf? bique, Myanmar, Niger, Nigeria, Pakistan, Sudan, Uganda, and Tan- la=en&hash=773552B364273E0C3DB588912F234E02679CD0C2. zania (data available: https://data.worldbank.org/). MINI GRIDS FOR HALF A BILLION PEOPLE    43 3. Notable exceptions include the solar DC mini grids built, owned, a similar issue: many generation plants are not operating at full and operated by Mera Gao India in the Indian state of Uttar Pradesh, capacity, and some countries are planning to export electricity and by Devergy in Tanzania. because of poor-quality in-country distribution infrastructure. 4. The World Bank’s MTF defines electricity access by five tiers of ser- 10. This is inclusive of ESMAP estimates for countries not covered by vice provision. The tiers rank seven attributes of electricity service: the GEP. capacity, service hours, reliability, quality or voltage fluctuations, 11. Before placing a purchase order and having the goods arrive on site, affordability, legality, and safety. The MTF then assigns any given mini grid developers and their partners will have already completed household to one of the five tiers, from no meaningful access at Tier a number of time-consuming activities, from site identification 0; basic lighting and charging at Tier 1; Tier 2 households can power and assessment to feasibility studies to community agreements. a few small appliances; Tier 3 households have formal grid connec- For the purposes of tracking the sector’s progress, however, clear tions with limited service; Tier 4 access supports refrigeration; and and measurable start and end dates are required as proxies. This Tier 5 is unrestricted continuous service. requirement motivated our decision to use the purchase order and 5. More information about AMDA is available on its website: https:// goods arriving on site as start dates as initial benchmarks for the africamda.org/. pace of mini grid development. 6. Sustainable Energy for All website: https://www.seforall.org/ 12. It is important to note that shipping containers, even if modified by 7. We have less confidence in the reliability of data for planned mini cutting doors and windows into them, do not always meet the min- grids than we do for installed mini grids, and fewer data points were imum standards for powerhouse facilities, particularly where these available for planned mini grids than for installed mini grids. For this standards specify rules for preventing overheating and insulation. reason, the main text and tables related to the mini grid market 13. The 2018 mini grid regulations in Tanzania are available at http:// today focus on installed mini grids. However, we will provide tables www.ewura.go.tz/wp-content/uploads/2018/06/The-Electrici- and analysis of planned mini grids on the website associated with ty-Development-of-Small-Power-Projects-Rules-2018.pdf. this book. For installed mini grids, we made every effort to deter- See the NERC Regulation for Mini Grids, annex 7, available at 14. mine their current operating status; when we found a mini grid was https://nerc.gov.ng/index.php/library/documents/Regulations/ no longer operational, we did not include it in the database. That NERC-Mini-Grid-Regulation. said, we cannot claim that every mini grid in our database is oper- 15. The mini grid technical specifications for the tender in Kenya are ational as of January 2022 owing to the sheer number of individual available at https://tenders.go.ke/website/tender/TenderDocu- projects in the database. ment/9034. 8. It is important to note that despite the scope and depth of the 16. Technical specifications for Zambia are available on the Energy database, it is almost certainly incomplete. For example, data Regulation Board’s website as a downloadable “zip” folder from are scarce for North African, Latin American, Eastern European, http://www.erb.org.zm/content.php?viewpage=mini. and Central Asian countries. It is therefore entirely possible that The Service Standards target for 2020 focuses on increasing reli- 17. the global mini grid market is much larger than what this chapter ability during the times of day when the developer has promised describes, and what the underlying data set supports. Neverthe- to provide electricity—for example, during evening hours. The tar- less, the quality of the available data is quite high, and the result get for 2025 focuses on maintaining high reliability and increasing of the data collection is the most comprehensive database of mini availability to 24/7 electricity—which brings the standards on par grids around the world to date. with (or above) those of the main grid. 9. We learned from conversations with AMDA and energy sector experts that many of the main grids in Sub-Saharan Africa have 44   MINI GRIDS FOR HALF A BILLION PEOPLE CHAPTER 1 REDUCING COSTS AND OPTIMIZING DESIGN AND INNOVATION FOR SOLAR MINI GRIDS CHAPTER OVERVIEW This chapter presents the results of the deepest and most extensive survey of the costs and technology innova- tions of solar mini grids and solar-diesel hybrid mini grids in developing countries conducted by any organization to date. Detailed data were collected from 411 solar and solar-diesel hybrid mini grids in Africa and Asia. According to our analysis, a 40 percent load factor plus expected decreases in component costs would lower the levelized cost of mini grid electricity to $0.20/kilowatt-hour (kWh) by 2030. After outlining the present status of mini grids’ capital and operating costs, the chapter concludes with an outlook for these costs through 2030. The solar mini grid industry is in the early stages of scale-up. panels and a diesel generator. Some run on just solar pan- Whether solar mini grids’ potential will be fulfilled depends els. They include battery storage and deliver alternating crucially on their cost. What are the key drivers of mini grid current (AC) electricity to customers.3 (Please see https:// costs? What do the data suggest are key opportunities for www.esmap.org/mini_grids_for_half_a_billion_people lowering mini grid costs without sacrificing quality and reli- for more information about the mini grids analyzed in this ability? What is the variation in costs among projects, and chapter.) what does it suggest about best practices that should be emphasized as the rollout of mini grids scales up? To help answer these questions with real-world data, the THE LEVELIZED COST OF MINI GRID Energy Sector Management Assistance Program (ESMAP) ELECTRICITY undertook the deepest and most extensive survey of the costs and technology innovations of solar1 mini grids and This chapter focuses on the capital expenditure (CAPEX) solar-diesel hybrid mini grids in developing countries con- required to build a mini grid, including for its preparation, ducted by any organization to date. This survey of 411 mini and the operational expenditure (OPEX) required to keep grids implemented by national electrification programs it going. We can combine these costs into a single cost per supported by the World Bank probes the technology unit of energy, called the levelized cost of energy (LCOE).4 design and costs of mini grids commissioned or contracted For mini grids, LCOE pertains to the cost of electricity on between 2012 and 2021.2 The portfolio of projects encom- a per kilowatt-hour (kWh) basis delivered to mini grid cus- passes 22 countries. The survey responses include detailed tomers over the lifetime of a mini grid. LCOE considers proj- data down to the component level (solar panels, batteries, ect development costs (engineering, obtaining permits, inverters and energy management systems, distribu- management), initial costs (for example, equipment and tion networks, land, logistics and transport, and so forth), installation), the costs of operations (for example, staff and including technical specifications. fuel), and equipment replacement over the lifetime of the project. As such, it is equivalent to the minimum average The mini grids analyzed in this chapter are isolated from tariff that electricity must be sold for in order to cover proj- the main grid. Most are powered by a combination of solar ect costs, including project financing. MINI GRIDS FOR HALF A BILLION PEOPLE    45 We calculate LCOE in two different ways: financial and eco- 221 mini grids in six countries6 and is restricted to mini nomic. The financial perspective on a project is from the grids that use lithium-ion (Li-ion) batteries, whereas case point of view of the developer, and incorporates all costs 5—“Global” (355 mini grids from 19 countries7)—calculates reported by developers in constructing and operating a an average LCOE for all mini grids, including those with mini grid, including import duties and taxes. The simpli- lead-acid batteries. The distinction between battery types fied economic perspective endeavors to remove the influ- is important because our data show a major shift from the ence of taxes, duties and subsidies, and thus represents use of lead-acid batteries (comprising about 97 percent of the cost to society at large, or equivalently, the cost of mini mini grids in our database up to year 2017) to Li-ion (69 grid electrification if a country were to impose no duties or percent of mini grids in our database installed between taxes or subsidies on mini grids. A private sector operator 2018 and 2021). Because Li-ion batteries have superior competing in the marketplace and paying import duties lifetimes and performance characteristics, they lower mini and taxes must consider the financial cost. Policy makers grids’ LCOE. For more on this, including the viable role of deciding among approaches to electrification, meanwhile, lead-acid batteries in mini grids, see the discussion on bat- are most concerned by the economic cost. A separate teries later in this chapter. analysis of how subsidies affect the affordability of mini Case 6 represents the best-in-class mini grid; its LCOE is grids for end users, and the viability of their development, based on component costs and load magnitude averaged is provided in chapter 6. from three high-performing mini grids in our database, one THE LEVELIZED COST OF ENERGY FROM each from Ethiopia, Myanmar, and Nigeria. MINI GRIDS: SEVEN ANALYTICAL CASES Finally, case 7 is a best-in-class 2030 mini grid, based on We developed financial and economic LCOE estimates for case 6 but with equipment cost reductions expected in seven mini grid cases, described below (and in detail in the 2030, as the mini grid industry ramps up and other asso- website accompanying this handbook). For each of these ciated industries achieve scale that drive cost reductions cases, we collected data from multiple mini grids, in some for important components such as solar panels (driven by cases hundreds of them, to determine the average unit global solar panel deployment in solar farms) and Li-ion cost for mini grid component categories. These component batteries (driven by global expansion of electric vehicles categories included the following: and utility-scale electricity storage). Drivers of cost changes are discussed at the end of this chapter. • Solar panels • Batteries The number of mini grids in each sample used to determine • Inverters representative mini grid unit costs, the average peak load • Energy management systems (in kilowatts, kW), and average number of customers per • Backup generators mini grid are shown in table  1.1. Mini grids varied consid- • Distribution networks erably across and within countries. For example, in Nigeria • Installation the average mini grid served an average of 916 customers, • Land but only had a peak load of 69 kW—about 75 watts per cus- • Management tomer; whereas Ethiopian mini grids on average served 228 Using case-specific costs, we used HOMER® Pro (Hybrid TABLE 1.1 • Representative mini grids from seven cases: Optimization of Multiple Energy Resources) software to An analysis of key characteristics optimize a solar hybrid mini grid for each case and to calcu- late the LCOE for each optimized mini grid over a 20-year Peak Number of lifespan. load (kW) of Customers Mini grids average for average The seven cases of LCOE analysis covered three different in sample mini grid mini grid levels: country, global, and best-in-class. 1. Nigeria 150 69 916 2. Myanmar 61 96 409 For the country level (cases 1–3), we picked three coun- tries deploying mini grids at scale: Ethiopia (10 mini grids), 3. Ethiopia 10 178 228 Myanmar (61 mini grids), and Nigeria (with unit costs based 4. Global Li-ion 221 79 716 on averages of 150 mini grids).5 5. Global 355 68 587 For the global average (cases 4 and 5), we pursued the  lobal best 6. G 3 141 793 same approach, but aggregated data that met internal in class data consistency check thresholds from countries around  lobal best in 7. G 3 141 793 the world. Case 4—“Global Li-ion”—is based on data from class 2030 46   MINI GRIDS FOR HALF A BILLION PEOPLE customers but with a peak load nearly three times higher mation about the modeling and assumptions is available (178 kW), averaging 780 watts per customer. at https://www.esmap.org/mini_grids_for_half_a_billion_ people. MODELING ASSUMPTIONS AND SCENARIOS Peak hours generally occur in the evening, when house- When calculating LCOE with HOMER Pro, we included a holds use the most electricity. So for each case, we looked number of assumptions based on prior research and expe- at five scenarios to analyze the impact of daylight and non- rience, including a 20-year lifetime for the mini grid,8 a dis- peak operation of local manufacturing on productive-use count rate of 9.6 percent,9 inflation of 3 percent,10 and diesel loads. The variable adjusted in each scenario is the load prices of $1/liter.11 For all cases except Myanmar, inadequate factor—a measure of the mini grid’s utilization rate, defined data were available on non-fuel OPEX and we assumed an as average load divided by peak load over a year. The first OPEX of $1,700 per staff person with three staff ($5,100 of these scenarios is a base case (22 percent load factor) per year) required for a mini grid with up to 500 customers, representing a typical rural residential load. The third and and four staff ($6,800 per year) for a mini grid with more fifth scenarios bring the load factor up to 40 percent (the than 500 customers. Individual mini grid components medium case) or 80 percent (the high case) (figure 1.1). such as generators, batteries, and photovoltaic (PV) pan- Higher load factor scenarios (40 percent and 80 percent) els had additional variable OPEX expenses.12 More infor- represent the addition of off-peak (primarily daytime) pro- FIGURE 1.1 • Load profiles for 22 percent load factor, 22 percent load factor (sun following), 40 percent load factor, 40 percent load factor (sun following), and 80 percent load factor 100 100 90 22% of load factor 90 22% of load factor, sun following 80 80 Percent of full load Percent of full load 70 70 60 60 50 50 40 40 30 30 20 20 10 10 0 0 0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 Time (hour of day) Time (hour of day) 100 100 40% of load 90 40% of load factor 90 factor, sun 80 80 following Percent of full load Percent of full load 70 70 60 60 50 50 40 40 30 30 20 20 10 10 0 0 0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 Time (hour of day) Time (hour of day) 100 80% of 90 load factor 80 Percent of full load 70 60 50 40 30 20 10 0 0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 Time (hour of day) MINI GRIDS FOR HALF A BILLION PEOPLE    47 ductive-use loads such as water pumping, agricultural pro- At the global level, mini grids with Li-ion batteries had cessing, cold storage with thermal inertia, and charging of similarly low LCOE. A representative mini grid based on electric vehicles. component costs from 221 mini grids from 6 countries (but dominated by the large number of Li-ion mini grids The second and fourth scenarios illustrate the benefits of in Nigeria and Myanmar) was calculated to have an LCOE lower LCOE that arise from using electricity during sunny of $0.46 per kWh. Adding in mini grids with lead-acid bat- hours, coincident with its production by solar panels, thus teries from 16 countries brought the average LCOE up to minimizing the costly battery storage and retrieval or the $0.53 per kWh for the 22 percent load factor case. It is burning of diesel fuel in backup generators. These two noteworthy that the mini grids with lead-acid batteries “sun-following” scenarios use the same amount of daily were often more expensive, not necessarily because of load in the 22 percent (base case) and 40 percent (medium the battery type, but because they tended to be older— case) load factor scenarios but modify the timing of the built as far back as 2012—and therefore had a variety of consumption to be concentrated during sunlight hours. higher-cost components. These sun-following load profiles should be seen as aspira- tional, and are included to provide a sense of the benefits of The economic LCOE from a representative best-in-class incentivizing daytime consumption. mini grid is $0.38 per kWh, reflecting a nearly 29 percent decrease from the LCOE of current global mini grids in our In calculating the financial LCOE presented in this chap- data set and a 17 percent decrease from current global ter, we have assumed zero grants. However, a second set Li-ion mini grids. At component prices expected in 2030, of scenarios considers the impact of performance-based this drops further to $0.29/kWh, a 22.5 percent drop from capital subsidies (grants), at levels of 40 and 60 percent current best-in-class LCOE. of initial CAPEX, on seven representative mini grid cases. Because these address financing, this set of subsidy sce- The financial LCOE follows the general trends of the eco- narios is discussed in detail in chapter 6. nomic LCOE but is higher by up to 13 percent. The amount of increase depends on the country, since duties and taxes MODELING RESULTS vary from country to country. The increase also depends Our economic and financial LCOE results are for Ethiopia, on the load factor and load curve since mini grids in the Myanmar, and Nigeria as well as representatives of global highest load factor cases rely heavily on batteries (both and best-in-class cases. The LCOE and renewable energy larger battery banks and also deeper cycling and thus fraction of each varies depending on the load curve, includ- more frequent need for replacement) to provide large ing the degree to which loads occur during sunlight hours amounts electricity in the hours of little or no sunlight, and (figure 1.1). batteries are generally assessed higher duties and taxes than solar panels. The next section discusses the portion of costs that each category of expense (for example, solar panels, batteries, LCOE analysis: Adding productive use loads and management, taxes, duties, and so on) involves, how these impact load shifting costs have been trending since 2012, and the projections The mini grids in our analysis had much lower LCOE when that justify our 2030 best-in-class LCOE projections. loads shifted from evening to daytime sunlight hours (see the “sun-following” cases in figure 1.1), or when produc- LCOE analysis: Base case load profile with tive uses that increase the profile’s load factor were added 22 percent load factor (resulting in a 40 percent or 80 percent load factor). Both Our economic analysis indicates that the economic cost shifting to daylight hours and increasing the load factor are of electricity delivered to households from representative likely to be accomplished through encouraging loads such mini grids in Nigeria and Myanmar is $0.43 to $0.46 per as agricultural milling, light manufacturing, water pumping, kWh for our base-case typical village residential load profile or cold storage, for which demand is greatest during day- with a 22 percent load factor (figure 1.2). In our modeling time hours. based on winning engineering, procurement, and construc- tion contract bids, Ethiopian mini grids on average cost A simple way to understand the benefits of daytime use is somewhat less, at $0.41 per kWh. This reflects that their to consider that the marginal cost of adding new genera- commissioning was relatively recent (in 2021) in our sam- tion capacity (solar panels, PV inverters) to meet solar-co- ple, and thus they benefited most from the declining global incident consumption costs around $0.10 per kWh at mini prices of solar panels and batteries, as well as economies of grid scales. Meanwhile, the levelized cost of new capacity to scale due to being relatively large in terms of peak load and cycle electricity into and out of a battery for later use adds average customer load (figure 1.2). at least twice that per kWh (taking into account losses of 4 8   MINI GRIDS FOR HALF A BILLION PEOPLE BOX 1.1 THE LEVELIZED COST OF ENERGY FOR BEST-IN-CLASS MINI GRIDS DROPPED NEARLY 31 PERCENT FROM 2018 Best-in-class mini grid costs have plum- TABLE B1.1.1 • Estimated and potential levelized cost of mini meted in the past few years. In 2018 grid energy, 2018 and 2021 ESMAP conducted a cost analysis of 53 LCOE ($/kWh) from best-in-class mini grid mini grids (ESMAP 2019). At that time, Percentage the best-in-class mini grid produced Load factor (%) 2018 2021 decrease (%) electricity with a levelized cost of energy of $0.55 per kilowatt-hour (kWh). By 22 $0.55 $0.38 31 2021, best-in-class costs had dropped 22 sun following — $0.30 nearly 31 percent to only $0.38 per kWh 40 $0.42 $0.28 35 in the unsubsidized 22 percent load fac- 40 sun following — $0.26 tor case, and more in cases with a higher 80 $0.35 $0.23 35 load factor. This drop in the levelized cost of energy Initial CAPEX $1,160,000 $847,000 in the past two years reflects dramatic Number of customers 1099 793 decreases in the costs of mini grid com- ponents, including solar panels (drop- Solar capacity (kWp) 228 286 ping 18 percent) and a shift in battery Battery type lead-acid OPzS Li-ion LiFePO4 type from lead-acid to lithium-ion, which Battery capacity 887 690 has similar upfront costs but superior (kWh) performance characteristics and thus Average daily load 890 758 lower life-cycle costs. While batteries (kWh) appear to be decreasing in capacity, the Firm power (kWfirm) 207 230 switch to lithium-ion batteries enables $/kWfirm $5,604 $3,659 35 deeper discharge, leading to higher Note: Levelized cost of energy data in 2018 is for the best mini grid in the ESMAP effective capacity. Cheaper storage and database at the time, representing a well-designed mini grid serving 1,100 customers PV generation, in turn, enables reduc- in Bangladesh; 2021 best-in-class data are from a representative mini grid synthesized from average costs and consumption levels in three mini grids in Myanmar, Nigeria, and tions in fuel usage: our 2021 best-in- Ethiopia commissioned in 2020 or 2021. Calculations assume that annual peak load class case used less than one-quarter of is 75% of installed battery inverter capacity and average daily load is calculated as the area under the daily load profile curve scaled to the peak load, accounting for 10 percent the diesel fuel consumption of the 2018 day-to-day and 20 percent hourly load volatility. best-in-class mini grid. These trends LCOE = levelized cost of energy; CAPEX = capital expenditure; kWh = kilowatt-hour; kWp also reflect decreased project develop- = kilowatt peak. ment and installation costs due to econ- omies of scale in deployment. electricity in the charge/discharge process). Benefits of LCOE drops by 4.5–10.6 cents per kWh (15–21 percent of solar-coincident consumption can be further maximized by igure 1.2). the total) in contemporary mini grids (f the use of dispatchable loads such as water pumping to a As the load factor is increased to 40 percent, reductions storage tank or non-time-sensitive agricultural processing in LCOE are even more substantial, shaving 8.8–16.5 US in which the activity needs to take place sometime, but can cents per kW (32–36 percent) from the base case 22 per- wait until there is an energy surplus. cent load factor scenario. If an 80 percent load factor can If the load curve remains at a 22 percent load factor but be achieved, reductions are 13.0–24.7 cents per kWh (39– demand is shifted to largely follow solar production (a 47 percent). 22 percent load factor in the sun-following case), the MINI GRIDS FOR HALF A BILLION PEOPLE    49 FIGURE 1.2 • Economic LCOE calculations for mini grids in 7 cases based on 0 percent subsidy and load profiles described in figure 1.1 Nigeria Myanmar $0.43 $0.46 $0.34 $0.37 $0.30 $0.32 $0.28 $0.29 $0.24 $0.26 22% LF 40% LF 80% LF 22% LF 40% LF 80% LF Ethiopia Global Li-ion $0.46 $0.41 $0.36 $0.33 $0.31 $0.29 $0.30 $0.27 $0.24 $0.25 22% LF 40% LF 80% LF 22% LF 40% LF 80% LF Global Global Best in Class $0.38 $0.53 $0.30 $0.42 $0.28 $0.36 $0.26 $0.34 $0.23 $0.28 22% LF 40% LF 80% LF 22% LF 40% LF 80% LF Global Best in Class 2030 $0.29 $0.25 $0.20 $0.19 $0.16 Normal daily load curve Sun-following 22% LF 40% LF 80% LF 50   MINI GRIDS FOR HALF A BILLION PEOPLE Shifting loads to daylight hours for a 22 Conservative ESMAP analysis indicates that percent load factor mini grid can decrease the combination of increased productive LCOE by up to 21 percent, while increasing the load uses and decreased component costs resulting factor of a solar-hybrid mini grid from 22 percent to from economies of scale and sector-wide technol- 40 percent decreases LCOE by up to 36 percent. ogy cost trends can bring best-in-class mini grid LCOE down to $0.20/kWh by 2030. EFFECT OF EXPECTED DECLINES IN CAPITAL AND THE SHARE OF RENEWABLE ENERGY OPERATING COSTS BY 2030 HOMER modeling calculated the optimum renewable The cost reductions seen over the past decade are expect- energy fraction for each case (table 1.2). Hybrid mini grids ed to continue until the end of the current decade for mini are largely powered by renewable energy but employ die- grid components, particularly solar panels and battery sel generators as backup during extended cloudy periods storage. In addition, ESMAP projects increased savings or for times of particularly high nighttime loads. HOMER in management, installation, and OPEX due to scaled-up modeling includes consideration of seasonal variations in deployment of clusters of mini grids. Together, these sav- sunlight as well as random day-to-day and hour-to-hour ings are expected to drive down the cost of electricity from variation in solar resources and electrical loads. Nearly mini grids. Our 2030 best-in-class case predicts mini grid all of the cases modeled have renewable energy fractions economic LCOE to reach $0.29 per kWh for a typical resi- exceeding 90 percent, meaning that over the course of the dential load curve with a 22 percent load factor. This is a 23 year less than 10 percent of the electrical energy is derived percent decrease from today’s best-in-class LCOE of $0.38 from operating the diesel generator.16 for the same load curve. When combined with a 40 percent The exception is for the global case at a very high load fac- load factor, best-in-class mini grids are expected to reach tor (80 percent), which indicates a renewable energy frac- an economic LCOE of $0.20/kWh by 2030. tion of 76 percent. The representative mini grid in the global For the best-in-class 2030 scenario, the following eco- case is derived from a large data set covering 355 mini grids, nomic cost assumptions were made: some installed as early as 2012. As such, it includes many mini grids built when solar panels were much more expen- • The costs of key mini grid components available to sive. Higher equipment costs in this global case combined developers decrease as follows:13 PV modules and PV with the high nighttime loads in the 80 percent load factor inverters (combined) cost $343 per kilowatt-peak case mean that there are more hours in the year in which (kWp), down from $596/kWp in the 2020 best-in-class the diesel generator is dispatched. This effect disappears representative mini grid;14 battery inverters cost $265 if the price of fuel is modeled at $1.50 per liter. For recent per kilovolt-ampere (kVA), down from $303/kVA; and mini grids and those in the future, high renewable energy Li-ion batteries cost $137/kWh, down from $297/kWh.15 fractions (above 90 percent) will continue to be expected. • Operation and maintenance (O&M) costs fall 50 per- cent, thanks to better bill collection through online pay- TABLE 1.2 • Optimum renewable energy share for mini as-you-go metering, and enhanced remote-monitoring grid cases considered technologies that streamline repairs and reduce staff Renewable energy share (%) costs through geographic clustering (Carlin and others 2018). 22% 40% 80% load 22% load 40% load • We conservatively assume that other CAPEX elements  Country factor sun factor sun factor and economies of scale remain constant, even though Nigeria 92 92 94 95 93 lower installation costs will be achieved by increas- Myanmar 93 95 94 95 93 ing portfolio sizes, among these, cost reductions from scalable plug-and-play building block components (the Ethiopia 93 90 94 93 91 “LEGO-fication” of mini grids), decreased management Global Li-ion 93 91 94 93 92 and engineering costs through economies of scale, and Global 87 92 91 91 76 better pricing of components through larger volumes of Best-in-class 92 90 93 93 90 purchases (Carlin and others 2018). Declines in this full range of costs are explored later in this chapter. Best-in-class 2030 94 94 95 95 94 MINI GRIDS FOR HALF A BILLION PEOPLE    51 MODELING RESULTS: LCOE OF OPTIMUM HYBRID IMPLICATIONS FOR NATIONAL UTILITIES VS. 0 PERCENT AND 100 PERCENT RENEWABLE OF IMPROVING THE QUALITY OF MINI GRID ENERGY SERVICES Powering these same loads using only a diesel generator Many of the mini grids that we analyzed in our study pro- is much more expensive. For example, using the best-in- vide 24/7 electricity and a level of service that consistently class case above, HOMER-calculated economic LCOEs are exceeds the level of service provided by the main grid. 55 percent to 126 percent higher for diesel only (table 1.3) Remote monitoring technologies and smart meters are compared to an optimized hybrid solar mini grid with bat- increasing the quality of customer service and the reliability tery storage and diesel backup. of mini grids. According to a 2022 benchmarking study by the Africa Minigrid Developers Association (AMDA), among But using a diesel generator to occasionally cover cloudy mini grid sites installed by their members in 2020, only 2 periods or periods of particularly high load lowers costs of 35 sites reported service uptime of less than 99 percent compared to the cost of a mini grid that is sized to meet (AMDA and ECA 2022). 100 percent of the load with renewable energy. The eco- nomic LCOE of an optimally sized 100 percent solar mini Across Sub-Saharan Africa, the main grid is much less reli- grid with battery storage is 24 to 39 percent higher than an able: households and small businesses typically experience optimally sized hybrid system. several hours a day of outage. In some countries—includ- ing Burundi, Ghana, Guinea, Liberia, Nigeria, and Zimba- With a renewable energy fraction of 90 percent, the deploy- bwe—more than half of households connected to the main ment of solar mini grids for half a billion people has vast grid reported receiving electricity less than half the time benefits for the environment. These systems are replacing (Blimpo and Cosgrove-Davies 2019). Disaggregated data diesel-fueled systems and/or kerosene-based appliances from the diagnostic survey reports carried out by ESMAP that on average emit 0.89 kilograms (kg) of carbon diox- in a range of countries based on the Multi-Tier Framework ide (CO2) per kWh. Assuming a rollout at scale covering the provide additional evidence of this lack of reliability, both addressable market of 217,000 systems by 2030, 1.2 billion in the Sub-Saharan region and beyond. The report from tonnes of CO2 emissions would be avoided. Rwanda indicates 97 percent of grid-connected house- IMPLICATIONS FOR NATIONAL UTILITIES OF holds experience more than four electricity disruptions a LOWER MINI GRID LCOE week (Koo and others 2018). The Ethiopia report shows that 57.6 percent of grid-connected households face 4 to 14 As a result of declining LCOE and increasing income- outages a week, and 2.8 percent face more than 14 outages generating uses of electricity, third-generation mini grids a week (Padam and others 2018). The report from Cam- can have transformational effects on power sectors. They bodia indicates that 69.3 percent of grid-connected house- are on track to provide power at costs lower than many holds face frequent, unpredictable power outages, and 9.9 utilities by 2030 (figure 1.3). At an LCOE of less than $0.30 percent of all grid-connected customers receive less than 4 per kWh, mini grids will become the least-cost solution for hours of service per day (Dave and others 2018). grid-quality electricity for more than 38 percent of African countries in a scenario in which national utilities do not dra- Utility information, while limited, corroborates this survey matically change their operations—with vast implications information. Only about a third of vertically integrated for the allocation of both public and private investment Sub-Saharan utilities reported figures for the average funds. At $0.20/kWh, electricity from mini grids is less duration and frequency of system interruptions in 2018, expensive to produce than electricity from the main grid in and only 5 of 21 distribution companies did so. Of those 24 out of 39 countries in Africa. that did, median reported duration and frequency of inter- ruptions in 2018 were 51.6 hours and 24.7, respectively. TABLE 1.3 • Economic LCOE of hybrid mini grid versus diesel only and renewables only Economic levelized cost of energy (% above hybrid)   22% load factor 22% sun 40% load factor 40% sun 80% load factor Optimized hybrid (91 to 94% 0.38 0.30 0.28 0.26 0.23 renewable energy) Diesel only (0% renewable energy) 0.85 (126%) 0.65 (115%) 0.55 (98%) 0.47 (85%) 0.35 (55%) No diesel (100% renewable energy) 0.47 (24%) 0.40 (30%) 0.37 (34%) 0.35 (38%) 0.32 (39%) Note: All equipment sizes optimized to meet load profiles based on best-in-class 2021 component pricing. 52   MINI GRIDS FOR HALF A BILLION PEOPLE FIGURE 1.3 • Comparison of levelized cost of energy of mini grids and utilities in Africa Liberia 5 of 39 At $0.50/kWh: At $0.40/kWh: 7of 39 At $0.30/kWh mini grid LCOE is less than 15 of 39 Mini grid LCOE at $0.20/kWh is less than the LCOE of 24 of 39 utilities in Africa Comoros Sierra Leone São Tomé and Príncipe Cape Verde Gambia, The Rwanda Guinea Senegal Mauritania Burkina Faso Togo Mali Madagascar Seychelles Benin Gabon Kenya Botswana Nigeria Côte d'Ivoire Mauritius Burundi Central African Republic Niger Swaziland Congo, Rep. Ethiopia Tanzania Malawi Cameroon Uganda Zimbabwe Sudan Ghana Mozambique South Africa Lesotho Zambia 0.0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 National utility LCOE ($/kWh) Source: Based on Trimble and others 2016. Note: Many customers in rural areas are charged tariffs that are much lower than the levelized costs shown above due to subsidies. kWh = kilowatt-hour; LCOE = levelized cost of energy. MINI GRIDS FOR HALF A BILLION PEOPLE    53 These are high by international standards. In order to cuted. But this presents an operational challenge, requiring receive any points under the scoring methodology used utilities to be able to introduce the practical technical func- by the World Bank’s Doing Business indicators, the maxi- tions to support power system operations and planning with mum SAIDI (System Average Interruption Duration Index) multiple mini grids connected to the distribution grid, such and SAIFI (System Average Interruption Frequency Index) as short-term and long-term forecasting and other complex is 12—equivalent to one hour-long outage each month procedures. This means that mini grid development—as a (Balabanyan and others 2021). Information on load factor viable strategy for delivering universal access to electricity— for these utilities was not available. entails a much stronger utility sector able to accommodate interconnecting mini grids with the main grid. WIN-WIN FOR MINI GRIDS AND NATIONAL UTILITIES How this can be achieved is laid out in several chapters that follow, including chapter 5, 8, and 9. However, scaling up mini grids does not mean scaling back the main grid. On the contrary, third-generation mini grids enhance the economic viability of expanding the main grid. By designing a system from the beginning to interconnect CURRENT STATUS OF SELECTED MINI with the main grid and by promoting income-generating GRID CAPITAL AND OPERATING COSTS uses of electricity through effective community engagement and training, third-generation mini grids can provide early COST PER UNIT OF FIRM POWER OUTPUT economic growth, so that significant load already exists by Of primary concern to developers and program adminis- the time the main grid arrives and customers have a greater trators alike is the total investment cost of the mini grid, by ability to pay. New regulatory frameworks give developers capacity. As a precursor to this discussion, it is important viable options for what happens when the main grid arrives, to define what we mean by mini grid capacity. How much and reductions in the cost of components enable develop- electric power (kW) can a mini grid reliably provide on an ers to build grid-interconnection-ready systems while still ongoing basis? For conventional dispatchable power plants, keeping tariffs affordable. New grid-connected mini grid this is called firm capacity, commonly understood as the business models are also providing win-win arrangements intended, sustained output of the facility at times of full load. in which utilities lease distribution assets and sell backup power to mini grid operators that take on customers that There is no consensus in the mini grid industry on how utilities have found unprofitable to serve directly, or that to define a metric comparable to firm capacity, based on require higher levels of service than the utility is able to pro- component specifications. The amount of power available vide (Tenenbaum, Greacen, and Shrestha 2022). at any given moment from a mini grid is shaped by diverse factors: the capacity of the solar array, the amount of sun- As a result, supporting third-generation mini grids goes hand light, the capacity of the inverter, the storage capacity of in hand with strengthening the utility sector. Interconnecting the battery bank, the capacity of the diesel generator, and, third-generation mini grids with the main grid as part of a as a practical matter, the availability of diesel fuel. Some of national electrification strategy can increase the resource these factors are at times limiting, and others offer alter- diversity and overall resilience and efficiency of the power native pathways to provide electricity. Some have specific system when interconnection is properly planned and exe- time durations or depend on factors such as the weather. In the absence of an industry metric, we offer an imperfect, As the cost of mini grid electricity continues but we believe useful back-of-the-envelope definition of to fall while its quality continues to rise, mini mini grid firm power. We define the firm power output of a grids will become competitive with the main grid in mini grid as the generator capacity (kW)17 plus 25 percent more and more countries. When the levelized cost of the solar array output rated peak (direct current, DC) of mini grid electricity hits $0.20/kWh, it will be less power output (kWp). In much of the world, about six hours expensive to produce than main grid electricity in of midday sunlight is the daily allotment—roughly the sun- 24 of 39 African countries. However, scaling up mini light intensity at which solar panels produce their rated grids does not mean scaling back the main grid. On output.18 Six hours is 25 percent of a full day. the contrary, third-generation mini grids enhance kWfirm = kWgen + 0.25 kWpPV the economic viability of expanding the main grid, and scaling up mini grids requires much stronger This definition assumes that the battery is sized large utilities, able to accommodate interconnection over enough to store sufficient solar electricity to redistribute time as the main grid expands. it over periods with inadequate sunlight, accounting for inefficiencies. But in some areas, especially those with a 54   MINI GRIDS FOR HALF A BILLION PEOPLE prolonged rainy season, this will not be possible. The defi- more of annual energy, solar panels or solar plus storage nition also assumes that diesel supply is not constrained. are sufficient. The low marginal cost of a diesel genset’s Thus, mini grids with large diesel generators seem to have installed capacity makes it affordable for it to carry the greater capacity than mini grids with strong solar invest- entire load. However, given high fuel costs, there is strong ment, despite the fact that operating a diesel generator for incentive to operate generator only when there is abso- anything other than backup generation is not cost effective. lutely no other choice. Within this context, the firm power The definition also ignores the effects of temperature on metric delivers valuable information on the ability of a sys- PV power output, as well as power lost through efficiencies tem to power a load for days or even weeks. Outside these in energy storage and conversion. On the other hand, the contexts, the kWfirm metric should be treated with caution. definition underestimates power available during sunlight Of 356 mini grids, there are wild cost variations per hours, which could reach as high as the sum of the genera- kilowatt of firm power (kWfirm) output (figure 1.4). The tor output, the PV array output (technically, the AC output median economic cost was $5,084 per kWfirm, while the from the PV inverter), and the battery inverter capacity. 25th and 75th percentile economic costs were $3,760 Despite these shortcomings, we find the metric useful and $6,953 per kWfirm, respectively. Most mini grids because it is easily calculated with available data, and gives below 200 kWfirm have costs around or below $5,000 an indication of the rough magnitude of a constant load per kWfirm. The economic cost of a best-in-class mini that could be powered by the mini grid for many days, if not grid in 2021 was $3,659 per kWfirm. indefinitely, if adequate diesel supply were available. The As expected, in general mini grids display economies of premise is that the provider of the electricity service can scale in generation, with smaller mini grids costing more guarantee electricity delivery upfront for any time the con- per kWfirm than larger mini grids. Most of the highest cost sumer wants it. The definition is useful for contemporary per kWfirm projects are built as individual projects and often solar mini grids that are basically designed as solar-storage lacked backup diesel generators. Many of these mini grids systems that have a diesel generator backup for less than represent early efforts to understand the marketplace and 10 percent of annual energy. For the other 90 percent or experiment with new technologies with less focus on cost FIGURE 1.4 • Total economic cost of mini grids per kWfirm as a function of firm power output 40,000 35,000 30,000 25,000 Cost per rm kW 20,000 15,000 10,000 5,000 0 0 100 200 300 400 500 600 700 800 Firm power output (kW) kW = kilowatt. MINI GRIDS FOR HALF A BILLION PEOPLE    55 reduction. Data on financing sources were not collected, INVESTMENT COSTS PER CUSTOMER but it is likely that these costlier projects had larger shares The median cost per customer for village mini grids was of grant funding. While generally only small projects (under $846, with 25th and 75th percentile costs of $468 and 50 kWfirm) had exceptionally high costs per kWfirm, many $1,413, respectively. As figure 1.5 shows, however, there are small projects also had low costs in this metric. We should notable and unsurprising outliers—many from 2020 and note that the lowest-cost mini grids from an LCOE per- 2021 mini grids reporting only a few customers—among spective are not necessarily the lowest cost in terms of firm newly commissioned mini grids. Others were relatively capacity. If a low LCOE is desired, high solar utilization is costly pilot projects, sometimes one of a kind, built with required, whereas mini grids with large diesel generators less emphasis on cost reduction. tend to score low on the $/kWfirm metric because diesel generators provide cheap capacity, albeit expensive to fuel. The total costs per customer reflect economies of scale, With this in mind, it is worth emphasizing that the $/kWfirm as mini grids serving more customers have lower costs metric is not the design parameter for optimization—but per customer on average. If only those mini grids with this indicator is useful when estimating investment. per customer costs below the median are included, every additional 100 customers lowers the per customer cost by Some high $/kWfirm projects appeared to follow a deliberate about $9. strategy of overbuilding their distribution network (in terms of both quality and scale), to easily accommodate upgrades Of the 411 mini grids in the database, a majority (217) had in generating capacity as the load grows. Other factors between 200 and 600 customers; 82 mini grids had fewer explaining the wide variation may be the amounts and ways than 200 customers, and 112 had more than 500 custom- in which project development costs are internalized into a ers. The preference for mini grids under 500 customers project or absorbed by a company and not reported as a may be because mini grids beyond this scale, both in terms mini grid development cost, the cost of doing business in of distance and cumulative consumption, require trans- the country in question, and a lack of competitive tendering. formers and medium-voltage lines to distribute power. Lack of regulatory certainty regarding grid arrival may also be a factor: in the absence of regulation, project developers may be choosing small sites farther from the main grid that Mini grid costs vary across projects and are less attractive for potential grid expansion. When a ver- countries, with a median cost per kilowatt of sion of this study was conducted in 2018 with 53 mini grids, firm power output of about $5,000. A low cost rel- the most popular size (28 mini grids) served under 200 ative to firm capacity points to the likelihood of the customers, likely reflecting relatively early preferences by mini grid having a large diesel generator relative to developers to limit risk by testing the waters with smaller its solar array, which could, in turn, raise the level- communities. ized cost of energy if it is dispatched often. FIGURE 1.5 • Mini grid economic costs per customer (left) and costs per customer for mini grids below median cost (right) 5,000 900 Cost per customer (USD) 800 Cost per customer (USD) 4,000 700 600 3,000 500 400 2,000 300 1,000 200 100 y = –0.0865x + 572.24 0 0 0 2,000 4,000 6,000 8,000 0 2,000 4,000 6,000 8,000 Number of customers Number of customers Source: ESMAP analysis. 56   MINI GRIDS FOR HALF A BILLION PEOPLE The data suggest that, on average, for every The largest cost components of the mini additional 100 customers a mini grid serves, grids in our data set were distribution (27 its per customer costs fall by about $9. Whereas in percent of total capital expenditure), batteries (15 2018 most mini grids had fewer than 200 custom- percent), import duties and taxes (12 percent), and ers, today most serve between 200 and 600 cus- installation (13 percent). tomers. cult for anyone to tell, especially early on in a project’s life COST OF INDIVIDUAL COMPONENTS cycle, how profitable the project will be and thus what the ultimate profit margin will be. For mini grids that were con- Solar mini grid components include solar panels, batteries, structed as engineering, procurement, and construction generators, inverters, other electronics, the distribution contracts, our data are derived from bid responses and in network, powerhouse, shipping and logistics, and installa- this case the profit margin is not explicitly stated, but rather tion. Mini grid total costs also include soft (but very real) is blended into line items (equipment costs, management, business and project development costs, such as site iden- installation, and so on). tification, demand assessment, design, and the process of obtaining necessary approvals. Table 1.4 provides a sum- It is important to note that the portion of reported costs that mary of the costs and characteristics of individual mini grid each component accounts for ranges widely across mini components from 351 mini grids in our database. The sec- grids (table 1.5). PV solar panels in a mini grid in Nepal, for tions below unpack the details of these components. example, were reported to cost nearly twice what they did in neighboring India. Li-ion batteries cost more than double In addition to the summary data reported in the table on a per kWh basis in Indonesia in 2017 than what they cost above, sufficient data were available from 294 mini grids to in Ethiopia for a 2021 mini grid. Distribution costs per cus- determine the average share of mini grid cost attributable tomer were more than five times higher on a per customer to each component (figure 1.6). The largest cost compo- basis in a 2016 Côte d’Ivoire project than reported in a 2021 nents were the distribution grid (26.6 percent), batteries project in India. (14.9 percent), installation (11.3 percent), PV modules (9.7 percent), and taxes and import duties (11.5 percent).19 There are likely multiple reasons for these differences. Taxes and import duties were back-calculated based on Most pronounced is that costs have come down over time tax and import duty rates provided by developers for the for major components, yet the data in the table above do countries they build and operate mini grids in, and vary not distinguish the year of project commissioning. This considerably from country to country and from compo- remarkable change in cost over time is discussed below. nent to component. The profit margin (reported at only 0.3 Another factor is that developers in different countries fold percent of economic costs in our data set) merits unpack- the profit margin into component costs in various ways. ing and further research. On the one hand, most mini grid Other reasons may include differences in interpretation projects earn revenues on electricity sales and it is diffi- by mini grid developer respondents to the surveys and TABLE 1.4 • Mini grid components: A summary of costs and characteristics Technical characteristics Costs Number of mini grids with 25th 75th 25th 75th Component available data percentile Median percentile percentile Median percentile Solar panels 351 45 kWp 76 kWp 125 kWp $388/kWp $441/kWp $599/kWp (including PV inverter) Battery Lead-acid: 133 144 kWh 288 kWh 432 kWh $154/kWh $193/kWh $224/kWh Lithium ion: 217 102 kWh 180 kWh 312 kWh $271/kWh $314/kWh $414/kWh Inverter + EMS 313 30 kW 63 kW 118 kW $325/kW $415/kW $716/kW Distribution and meters 317 N/A N/A N/A $163/kWp $250/cust $331/cust Customers 350 238 404 644 $480/cust $836/cust $1290/cust Source: ESMAP analysis. Note: percentile technical characteristics and unit costs are for each component separately. The rows should not be read together and interpreted in aggregate to represent the component capacities of a “25th percentile” or “median” mini grid. MINI GRIDS FOR HALF A BILLION PEOPLE    57 FIGURE 1.6 • Average share of component economic costs in total capital costs of mini grids 11.5% 0.3% 100% 90% 11.3% 80% 4.6% 26.6% 0.6% 70% Percent of cost 60% 50% 6.1% 8.6% 40% 14.9% 30% 20% 9.7% 10% 5.9% 0% t es y S S n nd s n s n en ic tie er io tio gi EM BO ul La st ar ut tt em du la od gi Ba M ib & al ag Lo m & tr s st er is s an PV In xe D rt M ve Ta In “Management + SG&A” (selling, general and administrative Distribution and smart meters expenses)—including engineering, planning, permits, approvals, Land licenses, and community engagement Logistics and installation Generation, including PV modules, battery, inverters, and Taxes and duties balance of system (BOS), which includes a diesel backup system Pro t margin Note: For costing purposes, PV inverters and PV controllers are grouped together. BOS = balance of system; EMS = energy management system; PV = photovoltaic. TABLE 1.5 • Average economic costs of key mini grid hardware components, by country Solar panels Lead-acid Lithium-ion Battery inverter Distribution $/kWp battery $/kWh battery $/kWh $/kW $/customer Bangladesh 622 185 — 1,242 355 Ethiopia 504 — 285 — 385 Ghana 798 143 — 1,011 520 Guinea Bissau 801 129 — 1,612 283 India 445 115 — 1,225 155 Indonesia 601 — 625 1,017 316 Ivory Coast 688 106 — 707 933 Kenya 834 142 — 928 307 Myanmar 497 231 422 467 321 Nepal 865 152 — 870 392 Nigeria 477 180 331 206 Palestine 656 158 — 1,181 303 Tanzania 585 159 614 1,431 496 Vanuatu 464 141 — 641 704 Minimum 445 106 285 467 155 Average 631 153 455 1,028 405 Maximum 865 231 625 1,612 933 Delta 94 119 119 245 504 Source: ESMAP analysis. — no data available; kVA = kilovolt-ampere; kWh = kilowatt-hour; kWp = kilowatt-peak. 58   MINI GRIDS FOR HALF A BILLION PEOPLE queries that solicited this information. In the distribution case, some projects may have needed only upgrades to existing distribution networks, whereas most others built Mini grids benefit from decreasing global new distribution networks, and there are considerable cost solar module prices, reflected in cost differences in distribution networks that are underground declines in the solar portion of mini grids of around vs. above ground, and for different sizes of poles and con- $32 per kilowatt-peak per year. ductors. The following sections explore in more detail some of the components that account for large fractions of the total Batteries investment cost. Batteries are a huge story in mini grids in the past several years. Li-ion batteries are rapidly becoming the dominant Solar panels (including racking and PV inverter) choice for new mini grids, driven by lower costs enabled Based on data from 351 projects, solar panel economic by their increasing use in consumer appliances, electrical costs for mini grids have been decreasing by about $32/ vehicles, and utility power storage. Of 211 mini grids under kWp per year on average since 2012 (figure 1.7). Based on construction or commissioned in 2020 and 2021, 145 (69 data from 278 projects built between 2019 and 2021, these percent) used Li-ion batteries while 66 (31 percent) used more recent installations had a median cost of $413 per lead-acid batteries. kWp, with a 25th percentile cost of $354 per kWp and a The common metric for battery pricing is $/kWh of battery 75th percentile cost of $599 per kWp. storage capacity. Figure 1.8 indicates economic cost trends In our data groupings, solar panel costs include not only for mini grid Li-ion (blue) and lead-acid (red) batteries. The the solar panels, but also the PV inverters.20 Both are con- graph shows lead-acid battery costs slightly increasing sistent with, yet not strictly comparable with, the mod- from our earliest project data in 2012 but holding roughly ule-level pricing that we discuss among global PV module steady at about $200 per kWh. Increasing lead-acid bat- cost trends later in this chapter. tery costs are consistent with global trends driven by the FIGURE 1.7 • Costs of solar panels (including PV inverters) for mini grids, by year, 2012–21 1,000 900 800 700 600 PV $/kWp y = –31.73x + 64568 500 R = 0.0829 400 300 200 100 0 2011 2012 2013 2014 2015 2016 2017 2018 2019 2020 2021 2022 Source: ESMAP analysis. Note: The relative capacity (kWp) of the solar array is indicated by the relative size of the dot. kWp = kilowatt-peak. MINI GRIDS FOR HALF A BILLION PEOPLE    59 FIGURE 1.8 • Economic cost trends for the storage capacity ($/kWh) of lithium-ion and lead-acid batteries used in mini grids between 2012 and 2021 700 600 500 Cost of storage capacity ($/kWh) Lithium ion: y = –36.9x + 74962 R = 0.126 400 300 Lead Acid: y = 1.7x–3238.3 R = 0.0034 200 100 0 2011 2013 2015 2017 2019 2021 Lead $/kWh Lithium $/kWh Linear (Lead $/kWh) Linear (Lithium $/kWh) Source: ESMAP analysis. Note: The nameplate capacity (kWh) of the battery is indicated by the relative size of the dot. kWh = kilowatt-hour. increase in the commodity price of lead, especially since TABLE 1.6 • Performance characteristics of lead- mid-2015 (Trading Economics 2022). acid and lithium-ion batteries as modeled in HOMER levelized cost of energy calculations Li-ion batteries appeared first in 2016 among the mini grids we tracked, and by 2018 were in 28 new projects. In 2019 Lithium- Unit Lead-acid ion they were included in 60 new projects and this number has grown every year since. Costs have been declining substan- Cycle life kWh throughput 800 3,000 (throughput) before failure tially for Li-ion batteries and their battery management systems at a rate of nearly $37 per kWh per year. Maximum depth % 60 80 of discharge A casual glance at figure 1.8 would suggest that while Li-ion Roundtrip % 80 90 batteries are decreasing in price, they are still more costly efficiency overall than lead-acid. But this would be an incorrect inter- pretation of the data, as the nameplate kWh capacities of the HOMER LCOE calculations in the first half of this chap- lead-acid and Li-ion batteries are not comparable.21 For ter are shown in table 1.6. a given kilowatt-hour of nameplate capacity, Li-ion bat- teries can be more deeply discharged22 and thus have a The implication of these performance differences is that larger usable kilowatt-hour capacity. Moreover, Li-ion bat- a single 1 kWh lead-acid battery will, over the course of its teries have superior cycle lifetimes (the quantity of kilo- lifetime, be able to cycle 800 times to 60 percent depth of watt-hour of electricity that can be charged and discharged discharge at 80 percent efficiency, storing and releasing into the battery before failure), higher efficiencies, as well 800 x 0.6 x 0.8 = 384 kWh of electricity before it must be as decreased temperature-related degradation, which is replaced. A single 1 kWh Li-ion battery, on the other hand, problematic for lead-acid batteries in tropical countries.23 will cycle 3,000 x 0.8 x 0.9 = 2,160 kWh of electricity, over Differences in lead-acid and lithium batteries as modeled in five times more. 60   MINI GRIDS FOR HALF A BILLION PEOPLE FIGURE 1.9 • Net present value of storage capacity for lithium-ion and lead-acid batteries, 2012–21 $2,500 $2,000 Lead acid NPV of storage capacity ($/kWh) y = 8.0591x – 15345 R = 0.0034 $1,500 $1,000 $500 Li-ion y = –62.74x + 127322 R = 0.126 $0 2011 2013 2015 2017 2019 2021 Corrected Lead $/kWh Corrected Li/kWh Linear (Corrected Lead $/kWh) Linear (Corrected Li/kWh) Source: ESMAP analysis. Note: The relative nameplate capacity (kWh) of the battery is indicated by the relative size of the dot. kWh = kilowatt-hour. To account for these differences, figure 1.9 compares the net present value (NPV) per kWh of energy storage capac- ity among these mini grid projects. This calculation takes Despite a higher sticker price, lithium-ion into account the factors listed in table 1.6 in a discounted batteries have replaced lead-acid batter- cashflow calculation that accommodates the battery’s ies due to their superior longevity, efficiency, and replacement schedules as predicted by the HOMER project deeper discharge capabilities. Lithium-ion battery modeling for mini grid projects serving the same load profile. costs are falling while the cost of lead-acid bat- teries is slowly increasing over time, in line with The revised figures (“Corrected Lead” and “Corrected Li” in global increases in the price of lead. Mini grids with figure 1.9) show that Li-ion batteries, despite their higher lead-acid batteries remain competitive, however, sticker price, have proven to be cost-competitive with lead- especially where strong supply chain relations can acid batteries since at least 2018. procure quality lead-acid batteries at high-volume The shift to Li-ion batteries is remarkable considering that pricing, and where discount rates are high. most subsidies for mini grids are for capital, per connection (performance-based grants), and therefore mini grid devel- opers must shoulder the higher upfront cost of Li-ion bat- Although our data set indicates that Li-ion batteries are teries at a time in the project cycle when revenue is not yet now the battery of choice in most mini grid projects, their generated. Moreover, lead-acid batteries were the incum- dominance is not complete. Our data set includes develop- bent technology and benefit from over a hundred years of ers, particularly in Nigeria and India, who are building very tried and tested operation; whereas Li-ion batteries are a competitive mini grids using lead-acid batteries at scale. new arrival accompanied by unknown technical risk as well Developers who have established lead-acid battery supply as the need to develop new supply chains. chains and low pricing through large volumes of orders will likely find it competitive to continue using lead-acid bat- MINI GRIDS FOR HALF A BILLION PEOPLE    61 teries, at least in the short term. Moreover, high discount costs are shown here on a per kilowatt-peak basis because rates reflecting high capital costs will, all things being many (solar support structure, fencing, civil works) are equal, favor lead-acid batteries since their upfront capital proportional to the size of the solar array. Somewhat low costs are lower. The country of battery manufacture is also costs in 2012 and 2014 may reflect shortcomings in data a consideration. Currently China dominates Li-ion battery collection in these categories. Likely contributors to declin- manufacture, whereas countries where lead-acid batteries ing costs are larger economies of scale through larger mini remain popular for mini grids (India and Bangladesh, for grids and clustering. example) have well-established and historically competi- Data collected in the survey circulated for the 2018 ver- tive lead-acid battery industries. sion of this analysis distinguished between conventional masonry powerhouses and powerhouses made from Battery inverters, energy management systems, and shipping containers. When a shipping container is repur- monitoring posed as a powerhouse, typically equipment arrives on site Battery inverters, energy management systems, and mon- prewired in the shipping container, which is also used to itoring compose on average 8.6 percent of project costs. transport the PV modules and racking materials to the site. Based on data from 327 mini grids, over time, these costs Measured in absolute costs, shipping containers as power- have been trending downward on a per kW basis (figure houses were, on average, the lowest cost, with an average 1.10), dropping from an average of $1,204 per kW in 2014 cost of $6,922 and a median cost of $7,235; powerhouses to $524 per kW in 2021, a decrease of nearly $100 per kW constructed on site averaged $29,700, with a median cost each year. This reflects the global decreases in the cost of of $26,253. We have anecdotal evidence, however, to the power electronics, as well as economies of scale both from contrary. Some developers have found working with local larger mini grid sizes over time as well as bulk purchases masons to be cost-efficient, particularly in areas where through expanded deployment. roads are poor, increasing shipping costs and challenges. Balance of system Shipping containers as powerhouses (figure 1.11) were also Balance of system (BOS) costs compose a catch-all cate- the lowest cost on a per kilowatt basis, accounting in the gory for the remainder of generation costs not captured in 2018 version of the study for the five mini grids with lowest the main categories of PV panels, batteries, and inverters. powerhouse cost, while conventional buildings accounted The BOS comprises the diesel generator, solar support for the most expensive five. For mini grids with shipping structures, fencing, foundations, lighting, civil works, pow- containers, the average powerhouse cost/kWfirm was $153, erhouse, and air conditioning system for the batteries, if whereas buildings constructed on site averaged $494/ installed. Based on data from 349 mini grids, average BOS kWfirm. costs are broadly trending downward (figure 1.10). BOS A recent innovation is to use weatherproof cabinets for the battery and energy management system enclosure, placed FIGURE 1.10 • Unit costs for inverters, energy manage- under a PV-paneled roof without the need for any addi- ment systems, and monitoring (blue), and balance of tional structure for a powerhouse. Even though the CAPEX system (orange) of the system is only slightly lower than its alternatives, the 1,400 major savings occur with the transportation and installa- 1,200 tion costs. These cabinets can be transported in lower-cost and more agile pickup trucks (figure 1.11, photo 6). 1,000 800 Distribution 600 For 349 mini grids with comparable data, the distribution costs of recent mini grids tend to cluster between $100 and 400 $500 per customer, with wide variation. Broadly, the distri- 200 bution costs appear to be trending downward slightly year 0 2012 2013 2014 2015 2016 2017 2018 2019 2020 2021 Average of inverter $/kW Average of BOS $/kW Shipping containers as powerhouses were Source: ESMAP analysis. Note: Balance of system (BOS) comprises the diesel genset, solar support the lowest cost on a per kilowatt-hour basis, structure, fencing, foundation, lighting, civil works, powerhouse, and accounting for the five mini grids with the lowest cooling. powerhouse costs. kW = battery inverter rated kilowatt capacity. kWp = kilowatt peak of the solar PV array. 62   MINI GRIDS FOR HALF A BILLION PEOPLE FIGURE 1.11 • Powerhouse innovations can lower costs and expedite deployment Power house: Remote Power Unit (RPU) 40-foot 1.  PV and battery inverters Inside the RPU 2.  shipping container under PV array Location: Bunjako Island, Uganda Location: Bunjako Island, Uganda Developer: Winch Energy Developer: Winch Energy  Photo credit: © Winch Energy. Used with permission by  Photo credit: © Winch Energy. Used with permission by Winch Energy. Further permission required for reuse. Winch Energy. Further permission required for reuse. Power house: 20-foot shipping container kiosk 3.  Brick power house 4.  Location: Katiko, Turkana North, Kenya Location: Kangitan Kori, Kenya. Developer: Renewvia Developer: Renewvia  Photo credit: © Jon Exel. Used with permission by Jon Exel.  Photo credit: © Jon Exel. Used with permission by Jon Exel. Further permission required for reuse. Further permission required for reuse. Micro-grid in a box (MIB) is the taller structure on the 5.  6. Power equipment in outdoor rated cabinets right. The diesel generator stands alone outside on a Location: Danchitagi, Niger state, Nigeria. platform (left). Elevating equipment protects against Developer: PowerGen flooding, increases natural cooling, and reduces risk of  Photo credit: © PowerGen. Used with permission by damage from dust, insects and animals. PowerGen. Further permission required for reuse. Location: rural India Developer: TPRMG Photo credit: © TPRMG. Used with permission by TPRMG. Further permission required for reuse. MINI GRIDS FOR HALF A BILLION PEOPLE    63 by year, with larger systems in recent years appearing to nection charge and then small monthly installments added have lower such costs per customer. to their customers’ bills for the first several months of ser- vice. While this strategy helps customers overcome what Distribution costs include poles, conductors, service drops, would otherwise be a prohibitively expensive one-time and meters, and customer wiring (or prewired “ready connection charge, implementing it adds an administrative boards” that contain a couple of light switches and one or burden—and cost—to the mini grid developer. From the two outlets). Included in this list are smart meters that can customer’s perspective, though, this pricing model is often send and receive data to and from the internet, and gener- familiar because a similar pricing model is used for most ally incorporate pay-as-you-go features by which custom- smartphones, where customers do not pay the full price of ers prepay for electricity (similar to prepaid minutes on a the phone up front but instead a portion of their monthly cell phone). Smart meters can help substantially reduce bill goes toward the cost of the phone. ongoing costs and increase revenues by lowering electric- ity theft; remove the costs of meter reading and postpay Intuition would suggest that increasing the number of cus- billing and collections; and, in some cases, provide data to tomers served would lead to decreases in costs per cus- mini grid operators on vital mini grid technical parameters tomer. Each increase of 100 customers per mini grid lowers that help operators and engineers identify and address costs by about $3 per customer, but the data suggest only problems before they become larger and more expensive. a weak correlation (figure 1.13).24 For mini grids with low consumption needs, a DC mesh can offer lower costs per Variations in cost and the technical sophistication of meter- customer (box 1.2). ing explain some of the wide variation in distribution costs per customer (figures 1.12 and 1.13). Other variation may be attributable to the fact that some mini grids provide inhouse wiring while others do not. Though not tracked in Each additional increment of 100 custom- the survey, it is nevertheless worth noting that the connec- ers correlates with declines in distribution tion fees charged to customers do not necessarily have costs per customer of about $3. But the data only a one-to-one relationship with the connection costs per weakly support this relationship. customer. Indeed, many mini grid developers choose to recoup the connection costs through a small upfront con- FIGURE 1.12 • Distribution costs per customer, 2012 to 2021 1,000 900 800 Distirbution cost per customer 700 600 500 400 y = –24.098x + 48934 300 R = 0.0379 200 100 0 2011 2012 2013 2014 2015 2016 2017 2018 2019 2020 2021 2022 Source: ESMAP analysis. Note: The size of the installation (number of customers) is indicated by the relative size of the dot. 64   MINI GRIDS FOR HALF A BILLION PEOPLE FIGURE 1.13 • Distribution costs per customer as a function of customers served 900 800 700 600 Distribution cost per customer 500 400 300 y = –0.0338x + 290.45 R² = 0.019 200 100 0 0 1000 2000 3000 4000 5000 6000 7000 Number of customers Source: ESMAP analysis. BOX 1.2 DIRECT CURRENT MESH GRIDS Although mesh grids are not included in our analysis, appliances, but they can upgrade to AC appliances they nevertheless hold promise for some communities by requesting an inverter. Larger productive use AC and states affected by fragility, conflict, and violence, appliances and higher-consumption households can with lower electricity needs. also be accommodated through networks of inter- connected customers. Approximately 90 percent of Mesh grids—or “skinny grids”—distribute DC electric- Alina’s customers are interconnected with at least one ity for lighting, electronics, and small appliances like other customer; only the remotest 10 percent of cus- fans and even efficient refrigerators or electric rick- tomers are served with isolated systems. shaws. They take the form of clusters of solar home systems made up of solar panels affixed to customers’ Alina encourages productive uses of electricity. The premises and connected in a mesh network. Special- company partners with a local appliance supplier and ized controllers allow surpluses to be shared. Examples conducts multiple community visits and workshops include Okra Solar, with installations in Cambodia, the prior to the arrival of the mesh grid arrival and as it Philippines, and Haiti; and SOLshare in Bangladesh. expands. In the rural Haitian province of Artibonite, Alina Enèji The mesh grids in Haiti typically have a capital cost has built direct current (DC) mesh grids that electrify about $800 per connection, about half of the cost for 300 households and small businesses using Okra conventional AC mini grids in Haiti. The modularity of Solar’s platform and equipment. Households start with the systems makes for quick installation and capacity systems that provide electricity for small, efficient DC upgrades as needed. a. Okra Solar, https://okrasolar.com/. b. SOLshare, https://me-solshare.com/ MINI GRIDS FOR HALF A BILLION PEOPLE    65 Land In particular, our data show how building a portfolio of mini Land comprised only a small portion (0.7 percent) of the grids can help lower costs by bundling approval processes mini grid sample’s average economic costs. Of 356 mini and exploiting economies of scale in project management, grids that provided plausible cost data, only 193 reported shipping, equipment procurement, and installation. land costs, and only 103 projects reported land costs over Project developers reported management costs (includ- $5,000. We are not sure to what extent this reflects lim- ing project development, general administration, planning, itations in data reporting. Land is often provided gratis by engineering, partnership, public relations, permits, approv- communities or local governments as part of agreements als, licenses, community engagement) for 309 sites. Logis- at project inception, although we have anecdotal evidence tics (transportation) and installation costs were reported that obtaining rights to suitable land is often a challenge. A for 327 and 297 sites, respectively. 100 kWp solar array requires about half an acre of land, or a square about 20 meters on a side. Though mini grids built separate from a portfolio tended to have fewer customers (single mini grids average 405 cus- Figure 1.14 shows two examples of solar arrays for mini tomers, while clustered projects averaged 657 customers), grids, illustrating the relatively small amount of land they had substantially higher average soft costs ($208,900, required. The array on the left is a 30 kWp mini grid devel- compared with $127,400 for portfolio projects). oped by Mandalay Yoma in Myanmar, with the powerhouse and diesel generator under the green roof at upper right. Governance requirements and developers’ internalized pre- The array on the right is a 40 kWp mini grid developed by feasibility assessments, technical standards, and account- Winch Energy in Lamwo District in northern Uganda. The ing and reporting requirements may give rise to widely solar array is built over the powerhouse to reduce land varying soft costs across similar projects. requirements, though with additional racking costs. REPLACEMENT COSTS Sales, general and administrative expenses, senior Replacement costs are for repairing worn-out or broken management, logistics, and installation equipment as the mini grid ages. These costs were not The data collected for this chapter illustrate how soft costs explicitly reported in the data that underlie this study, but (project development, logistics, and installation) might be an are nevertheless essential. For long-term sustainability, it is area where economies of scale can lower investment costs. critical to ensure that sufficient funds are available to cover replacement costs. Battery replacements in particular are problematic because it is generally necessary to replace the entire pack in order to ensure that new batteries are not Mini grids built as part of a portfolio saved electrically compromised by older batteries to which they $81,000 in soft costs on average, compared are electrically connected. In this regard, the transition with mini grids built as one-off projects. from lead-acid to Li-on batteries is important. As discussed above, although Li-on batteries have much higher upfront FIGURE 1.14 • A 30 kWp Mandalay Yoma mini grid in Myanmar (left) and a 40 kWp Winch Energy mini grid in Uganda (right) Photo credits: Left © Mandalay Yoma; used with permission; further permission required for reuse. Right: © Winch Energy; used with permission; further permission required for reuse. 66   MINI GRIDS FOR HALF A BILLION PEOPLE costs, they last much longer than lead-acid batteries, to understand whether revenues are sufficient to cover delaying the need for, and ultimately reducing, a project’s OPEX and other costs, such as debt service and equipment long-run replacement costs. replacement. While many projects are too young to report out their replacement costs, they are built into HOMER optimiza- tion and LCOE modeling. The replacement of many com- THE OUTLOOK FOR MINI GRID ponents (PV panels, inverters, monitoring equipment) is CAPITAL AND OPERATING COSTS based on years in service, diesel generators’ replacement is based on hours operated, and battery replacement is As discussed in the sections above, mini grid costs have dictated by the total throughput in kilowatt-hours based on declined substantially on an LCOE basis, including a 31 battery type. percent decline of LCOE in best-in-class mini grids since 2018. In addition, as discussed above, our data show that OPERATING COSTS the costs of key components (especially PV modules, bat- A mini grid’s operational expenses are important, espe- teries, and electronic components) reported by mini grid cially from the perspective of long-term sustainability. developers have been steadily declining. OPEX includes all costs associated with operating mini grid This section draws on research on costs in related global equipment, including fuel costs, maintenance, repairs,25 industries such as solar panels and batteries to better payment collection, and security. understand what levels component costs for mini grids OPEX was reported for 137 systems (113 in Myanmar, 4 in may reach by 2030. Industry trends suggest that compo- other Asian countries, and 20 in Africa). Reported OPEX nent costs in most key areas of generation, storage, meter- per customer varied widely from a low of $2 a year in Myan- ing, and power conversion can be expected to continue to mar to $267 a year in Kenya. decline thanks to their increasing deployment and spillover effects from technological development in much larger sis- Among the 19 mini grids (12 of them in Bangladesh) that ter industries. Table 1.7 and the sections in this chapter that reported a breakdown of staff, fuel, and other O&M costs, follow provide details of the expected cost declines for key fuel on average accounted for 30 percent of O&M, staff mini grid components. accounted for 49 percent, and other O&M accounted for 21 percent. Within this data set, there were considerable The best-in-class 2030 component price assumptions variations. In some, fuel or staff accounted for 0 percent; in used in the HOMER LCOE modeling discussed in the begin- others, 100 percent. ning of this chapter used the following approach: we started with the costs for each component in the best-in-class mini Some of the large variation in reported OPEX may reflect dif- grid from 2021, and then applied the same percentage ferences in staffing needs. Did the sale of electricity require drop to that component that is expected industry-wide. For staff, or was it accomplished automatically through a cell- example, PV costs in the best-in-class, representative mini phone-based prepayment system? Does the site require grid were $596 per kWp in 2021. Global industry PV, with security guards? How is the O&M of the mini grid plant a 2020 benchmarked cost of $198 per kWp, is expected accomplished? Are some staff responsibilities conducted to drop another 42 percent to $114 per kWp by 2030. For on an unpaid basis? Did the mini grid initially not work prop- the 2030 mini grid cost estimate, the same 42 reduction erly and therefore require more intense support until the is applied to the 2021 best-in-class price, yielding a 2030 system was operating robustly? How was OPEX allocated estimate of $343 per kWp (including PV inverters). This on a component basis? The data set does not provide suffi- is still several times higher than the industry benchmark ciently detailed information to answer these questions. price, reflecting the realistic cost multipliers that translate Developers face choices between CAPEX-intensive and an industry spot market price into the cost at a remote mini low OPEX installations (for example, a contemporary solar grid site far from a factory. hybrid system) versus those involving low CAPEX and high PV MODULE TRENDS OPEX (for example, diesel-fueled mini grids). With the availability of subsidies to help cover CAPEX for renewable Mini grids benefit from decreasing solar module prices, energy mini grids, effort is often made to set affordable tar- driven mostly by large grid-connected installations. PV iffs that cover OPEX and replacement costs. prices have fallen faster and lower than nearly any forecast. As of April 2021, global spot prices averaged $198 per kWp Further research to revisit the OPEX costs of the ana- for poly-crystalline modules (Energy Trend 2021).26 By the lyzed mini grids would be useful to understand how OPEX time PV modules arrive on the project site and are included changes over time, and how staff, fuel, and other OPEX with PV inverters they cost considerably more than global components evolve. Further research is also necessary MINI GRIDS FOR HALF A BILLION PEOPLE    67 TABLE 1.7 • Mini grid component cost benchmarks and price projections Component Unit Share Median Best-in-class Mainstream Mainstream Mainstream Best-in-class of total cost in 2021 LCOE industry industry industry 2030 LCOE capital ESMAP modeling benchmark benchmark estimate by modeling cost survey assumption in 2010 in 2020 (% 2030 (% assumption (%) change from change from (% change 2010) 2020) from 2020) PV module $/kWp 9.7 $441 $596 $1,589 $198 (–88) $114 (–42) $343 (–42) PV inverter $/kWp * * * $320 $80 (–75) $70 (–12.5) * Battery (Li-ion) $/kWh 14.9 $314 $297 $1,160 $126 (–89) $58 (–54) $137 (–54) Battery inverter $/kVA 8.6 $415 $303 $565 $113 (–63) $99 (–12.5 $265 (–12.5) Smart meters $/ †‡ ‡ ‡ $106 $40 (–62) $35 (12.5) ‡ customer Sources: ESMAP analysis; Bloomberg New Energy Finance Solar Spot Price Index; National Renewable Energy Laboratory U.S. Solar Photovoltaic System Cost Benchmark: Q1 2020; Feldman and others 2021; Kairies 2017. * PV inverter is included with PV module cost. † Battery inverter is grouped with EMS and monitoring equipment. ‡ Smart meters are included in distribution cost. Average, median, minimum, and maximum costs are all expressed in inflation-adjusted dollars. kVA = kilowatt-ampere; kWh = kilowatt-hour; kWp = kilowatt-peak; Li-ion = lithium-ion; PV = photovoltaic. spot prices. The median cost of PV (with PV inverters) for all From a materials perspective, new cell technologies like mini grids in our database was $441 per kWp. Just counting perovskite cells promise to radically reduce the amount of those mini grids commissioned between 2019 and 2021 highly pure silicon material required in solar cells, as well gives us a median cost of $413 per kWp, with 25th and 75th as improving efficiencies, paving the way to lower produc- percentile costs of $354 and $599 per kWp, respectively. tion costs (US DOE n.d.). Currently about 41 percent of the world’s supply of high grade polysilicon for solar panels Module prices have been roughly following Wright’s Law,27 comes from Xinjiang, China, a region where human rights falling 18–22 percent for every doubling of installed capac- groups and numerous governments have reported ongoing ity (Yu 2018). With growth rates averaging about 40 per- forced labor violations, specifically against the Uyghur eth- cent a year through 2017,28 production doubled about every nic minority (Jenkins 2022). 1.8 years. In recent years, growth slowed to about 24–27 percent, reflecting a doubling every 2.8 years. PV INVERTER TRENDS At the end of 2021, a cumulative total of 843 GWp of PV had PV inverters used in mini grids are similar (or in many been deployed, with 133 GWp commissioned in 2021 alone cases identical) to those used in residential and commer- (IRENA 2022). Bloomberg New Energy Finance (BNEF) cial grid-connected installations, which are projected to projects that solar PV prices will drop to $114 per kWp by increase by about 200 GW in China alone between 2021 2030 (BNEF 2020a), with a cumulative 2.4 terrawatts-peak and 2026 (IEA 2021). PV inverters used in mini grids are (TWp) of PV installed by that year. This reflects a compound also smaller cousins to the grid-tie PV inverters used in average 13 percent annual growth rate for solar PV, a con- utility-scale PV installations that account for the lion’s siderable decrease from contemporary growth levels. share of the 204 to 252 GW of PV expected in 2022 (BNEF 2022). These large volumes, together with cost declines There are wide variations in estimates of total PV that will associated with rapid expansion of other power electronic be added by 2030. As of 2020, new builds of solar PV farms markets, such as motor drives for electric vehicles, will con- are competitive with the marginal cost of existing conven- tinue to drive down costs for PV inverters and controllers. tional generation such as coal, nuclear, and combined cycle natural gas (Lazard 2020). With decreasing PV costs and A study by the National Renewable Energy Laboratory increasing electrification of transportation, heating, and (Feldman and others 2021) identifies a benchmark price industry, some scientists are envisioning that PV’s current of $80 per kWAC for three-phase string inverters (including annual percentage growth will be maintained for the next the cost of monitoring equipment, in 2019 US dollars) in decade, hitting 10 TWp of PV by 2030. To accommodate this the first quarter (Q1) of 2020 for commercial scale PV (100 level of PV would require considerable utility-level storage kW to 2 MW). Inverters are less than a third what they cost a and expanded ability to dispatch load (Haegel and others decade ago: in Q1 of 2010, three-phase inverters cost about 2019). If Wright’s Law continues to hold, expansion to 10 $270 per kWAC (also in 2019 US dollars). TWp is consistent with a price drop to below $90 per kWp. 68   MINI GRIDS FOR HALF A BILLION PEOPLE Assuming a conservative slowdown in the rate of decrease BATTERY INVERTER TRENDS in costs between 2021 and 2030, ESMAP estimates that PV Decreasing battery inverter costs are consistent with inverter costs could reach $70 per kilowatt-peak by 2030. broader trends in power electronics, driven by synergies with PV inverters and electric vehicle motor drives. While BATTERY TRENDS broader industry data were not available for battery invert- Costs for Li-ion batteries have declined dramatically since ers, using PV inverter costs as a proxy, ESMAP estimates 2010 and are expected to continue to decrease substan- that battery inverter costs will reach $99/kVA by 2030, tially. BNEF reported in December 2020 that Li-ion (pack assuming the same 12.5 percent decline by 2030 expected level) battery benchmark costs were at $126 per kWh on for PV inverters. a volume-weighted average basis, down from $1,100 per kWh in 2010. Even with rising commodity prices in the wake SMART METER TRENDS of COVID, BNEF predicts an average cost below $100/kWh The global market for smart meters has seen rapid growth for batteries by 2024 (BNEF 2021). It predicts an average in recent years, driven by strong policy support in China cost of $58/kWh in 2030 (BNEF 2020b), reflecting a cost and Europe. European utilities are projected to install 182 reduction of 54 percent from 2021 benchmark prices. million smart meters between 2016 and 2020, totaling Many Li-ion batteries used in high-end electric cars use nearly $38 billion in investment. Likewise, in Japan, 55 mil- cobalt in their cathodes to increase their energy density to lion meters costing $16.6 billion were installed between provide greater ranger or power output. Lower-end electric 2016 and 2020. Globally, the smart meter industry is a $20 vehicles (such as those increasingly sold in Chinese mar- billion a year market, expected to reach $30 billion by 2026 kets) and stationary power applications such as mini grids (Smart Energy International 2018; Global Industry Analysts or utility storage use lithium iron phosphate (LFP) batter- 2022). Smart meters used in mini grids are in some cases ies, which are less expensive, and less exposed to the risk identical to those deployed in large numbers by utilities. In of supply shortages for cobalt. Bloomberg found that LFP other cases, they are built specifically for the mini grid mar- cells were almost 30 percent cheaper than batteries with ket, with functionalities such as load dispatching, which are cobalt (BNEF 2021). Cobalt is a rare metal and the Dem- included to optimize mini grid load factor by reducing peak ocratic Republic of Congo accounted for 70 percent of demand but still benefit from technology improvements global production in 2019, and substantial concerns have and component cost reductions driven by the larger smart been raised concerning child labor and other human rights meter industry. abuses in cobalt mines in that country (Sanderson 2019). Competition among prepayment metering manufactur- The industry benchmark cost for lead-acid batteries is ers and increasing scale will allow development costs to $143 to $147/kWh (Kairies 2017; Wagman 2020). Lead-acid be spread over a larger product base. BNEF tracks smart technology is largely developed, but the industry makes meter installations and investments globally and found that improvements every year. For example, carbon added to the global average cost per smart meter in 2010 was about the negative electrode will reduce sulfation and increase $106 per unit; 2021 benchmark costs per smart meter in charge rates. ESMAP was unable to find cost projections low-income countries were around $40. ESMAP expects for lead-acid batteries by 2030, but efforts to increase their costs to continue to decline at a rate equal to inverters, life cycles beyond 800 and into the thousands would have reaching unit costs of $35 in 2030. the effect of reducing lead-acid battery storage costs even if TRENDS IN OTHER CAPITAL COSTS nominal costs remain the same. Even so, lead-acid batteries are unlikely to reclaim the mantle of battery of choice from One cost-saving trend in low-voltage distribution is the Li-ion based on expected cost decreases in Li-ion batteries. increase in local factories that build hollow reinforced con- crete poles.29 These poles are manufactured through a Other battery chemistries promise long-term energy stor- process in which concrete is poured into a mold together age that may allow solar mini grids to remove diesel gen- with reinforcing steel. The mold is spun, while centrifugal erators entirely and yet maintain high reliability. Recent force compacts the concrete into a smooth hollow cylin- innovations in iron air batteries have led to price targets der (Taizhou Amity Care 2013). Spun poles have higher for this technology at less than $20/kWh for 100 hours strength per unit weight than solid poles and require much of storage. Because of their lower-current, higher internal less concrete, lowering transportation costs by 25 percent impedance, and longer-duration chemistry, iron air batter- or more compared with solid concrete poles. Developers ies would not replace but could complement Li-ion batter- from Myanmar report that depending on local availability ies, allowing mini grids to weather a week or more of cloudy of sand and gravel, on-site construction of concrete poles weather (Plautz 2021). may also lower costs (Zaw Min 2019). MINI GRIDS FOR HALF A BILLION PEOPLE    69 For powerhouses, though the data on shipping container • The availability of big data that provide geotagged powerhouses appear promising, more research is needed points of interest that can be used to prepare a detailed to understand whether the construction cost savings of demand assessment of prospective load centers shipping containers outweigh the thermal management • Affordable high-resolution satellite imagery issues that arise from their use in hot, sunny environ- ments, and the other engineering issues (for example, • Easy-to-use but sophisticated software that can be stackable batteries) required to repurpose these contain- used to design hybrid generation systems together ers. Another option, still in its infancy, is building mini grids with the design of the distribution network that need no powerhouse, in which components are shel- • Data-driven web-based platforms that compile large tered under the solar array. The Rockefeller Foundation– amounts of geotagged market intelligence that can be supported Smart Power India program has partnered configured in different ways to be useful for mini grid with the Institute for Transformative Technologies in an developers, financiers, and government agencies approach that combines a 10 kW PV array with all neces- The introduction of geospatial and other digital technolo- sary electronics into modular units that can be scaled up, gies has decreased the cost of preparation and planning depending on the situation. In India, Tata Power Renew- by an order of magnitude (see chapter 2 for more details able Microgrids has targeted the installation of 10,000 on geospatial planning). In the past, the unit cost per microgrids using standardized equipment packages built site was more or less the same, irrespective of the num- around a mass-produced “micro-grid in a box” (figure 1.11). ber of sites—about $30,000 per site—because each site Increased factory integration of components in a “utility required a high level of on-site analysis. Today, portfolios in a box” model will lower on-site assembly requirements. of mini grids can be prepared to the point where they Through these economies of scale, the Rocky Mountain are ready for full feasibility assessment and community Institute expects these other CAPEX components to drop engagement at a cost of about $2,300 per site, based on by 15 percent (Carlin and others 2018). the World Bank’s recent experience in Nigeria. Meanwhile, the next generation of diesel generator incor- The socioeconomic surveys and energy audits (looking at porates power electronics in ways that allow engines to demand and willingness/ability to pay) make up 58 per- operate at variable speed as needed, increasing energy cent of per-site costs, which are largely linear since human efficiency (AP News 2018). Variable-speed generators, resources are the primary drivers. The time required together with a dump load and short-term battery storage for a household survey will not change with the scale of buffer, can accommodate up to 100 percent renewable the exercise, although streamlined travel logistics might energy penetration (Innovus 2015). One approach uses a produce savings. Nevertheless, technology can expedite fast-acting clutch that can disengage the motor from the these labor-intensive tasks—for example, through the use alternator when the renewables can fully support the load. of drones to map out a village and sequence household The alternator remains spinning, providing reactive power visits by enumerators. Tablet-based software can swiftly and voltage and frequency regulation (Danvest Energy and accurately capture survey data. Partnerships with 2019). cell-phone-based electronic payment companies can Local production of relatively low-tech items like PV racks obtain market data from targeted rural customers on has the advantage of low labor costs, low shipping costs, appliance purchases or other spending patterns. and local economic development. The mass produc- tion of these items in large factories can, however, take TRENDS IN OPERATING COSTS advantage of economies of scale. As mini grids scale up The introduction of remote-controlled, prepay smart and competition intensifies, mini grid developers in each meters has slashed labor costs. Reaching delayed or non- country can be expected to find context-specific solutions paying customers can now be done remotely. Consump- that optimize the costs of these components. tion patterns can also be tracked and analyzed remotely. In addition, preparation and planning costs have declined. Smart meters and cell-phone carrier-based, real-time data In the past, multidisciplinary teams prepared electrifi- collection enable detailed monitoring of system parame- cation plans, scoped sites, and conducted prefeasibility ters. When parameters exceed programmable thresholds, studies, at considerable cost. Today, most of this work,30 alarms alert technicians of problems that are much easier to address before they grow and cascade into expensive all the way up to feasibility-level analysis—including com- equipment failures and prolonged downtime. In addition, piling bills of quantity and bid documents or purchase smart meters enable developers to easily collect and ana- orders—can be done from behind a desk, thanks to the lyze their performance data, which can be aggregated and following factors: anonymized to share with development partners, indus- 70   MINI GRIDS FOR HALF A BILLION PEOPLE try associations, investors, and other stakeholders. Data grids themselves benefit from economies of scale due to uploaded to the cloud can be analyzed by machine learn- increasing portfolio size and from industry scaling at the ing algorithms, and allow early identification of problem- country level. As mini grid developers scale their portfolios atic patterns. Some companies are planning to respond from 10 to 100 and then to 1,000 or 10,000 mini grids, fixed to customer inquiries with artificial intelligence systems. costs like administration and management are spread over more units of production; sometimes a company can nego- Replacement costs have also fallen. Projects installed with tiate lower per-unit costs enabled by bulk purchases. lead-acid batteries that last three to six years can, if the bat- tery inverters are compatible, be replaced with Li-ion bat- To explore this effect, we analyzed mini grid cost data to teries with a life of ten or more years. Developers building discern changes across categories arising from portfolio mini grids would be wise to choose battery inverters (and and in-country market sizes. Categories included hardware battery chargers in the case of DC-coupled systems) com- (PV modules, batteries, inverters, and so on), manage- patible with Li-ion batteries. Replacement costs for elec- ment, logistics, and installation. We also made estimates of tronics, such as PV inverters and battery inverters, are also the net present value of the ongoing costs of O&M, major falling as they are manufactured at larger and larger scales. equipment replacements, and engaging with customers. To align portfolio-scale projections with declines in equipment Other costs are incurred in dealing with bureaucratic spot prices (see above), we assumed a representative port- processes, such as obtaining licenses, approvals, and folio size of 100 mini grids and doubled this at each time permits. These costs depend on a country’s enabling interval to 200, 400, 800, and 1,600 mini grids per portfo- environment. Several governments have incorporated lio. We also had the portfolio grow by orders of magnitude to mini grids as part of their energy policy, giving them and stress-test the boundaries of the different cost categories. the industry a place in the energy sector. Some countries have adopted mini grid regulations that allow for a light- The results show that with economies of scale, significant handed approach. In some countries, e-government has shifts are taking place in the three cost categories (table streamlined the process for obtaining location and build- 1.8). Overall, the portfolio development and management ing permits. Even though these costs are important, they cost category remains small, with less than 5 percent of are not expected to change much over the next decade the cost over the lifetime of the portfolio. This indicates in countries with high energy deficits, not unless enabling that additional cost reductions will have limited impact environments are introduced in these countries. on the overall LCOE of the portfolio. What is not incorpo- rated in the calculation is the cost of delay in processing for permits, licenses, approvals, and other red tape. More THE IMPACT OF ECONOMIES OF surprising, perhaps, is the minimal difference between the extended CAPEX and OPEX. On average the CAPEX con- SCALE tributes a little more than half the cost of the LCOE, while In addition to the benefits of decreasing spot market prices the OPEX is close to 45 percent, suggesting that the LCOE from the deployment of PV panels and batteries in large is sensitive to the makeup and design of the cost structure global industries like solar farms and electric vehicles, mini of O&M and major repairs over the lifetime of a project. In TABLE 1.8 • Net present value broken down by category with economies of scale Procurement, construction, Operation, maintenance, Portfolio size (number Portfolio development installation, and customer major replacements, and of mini grids) and management (%) engagement (%) customer engagement (%) 100 4.8 53.0 42.2 200 4.0 53.3 42.7 400 4.0 53.1 43.0 800 3.7 52.9 43.4 1,600 3.5 52.6 44.0 100 4.8 53.0 42.2 1,000 3.6 52.5 43.9 10,000 3.4 56.8 39.8 100,000 2.9 48.3 48.8 Source: ESMAP calculations and analysis using costing data described in this chapter. MINI GRIDS FOR HALF A BILLION PEOPLE    7 1 part due to difficulties in obtaining OPEX data, this topic REASONING FROM FIRST has not received the same level of attention in this hand- book and deserves more scrutiny in future work. PRINCIPLES Closer scrutiny of unit costs reveal important changes To further break down the complex setup of a solar mini (table 1.9). When doubling the size of the portfolio stepwise, grid, we tried to reason from first principles and analyze the from 100 to 1,600 mini grids per portfolio, unit costs plunge system’s basic elements. This analysis is a first attempt to across categories. Also, the LCOE falls from $0.36/kWh determine the cost asymptote for the hardware of a solar with a load factor of 22 percent for a portfolio of 100 mini mini grid. It is also an invitation to interested experts, stu- grids to $0.21/kWh for a portfolio with 1,600 mini grids. A dents, and professionals to elaborate further. We took the load factor of 40 percent produces a similar trend. typical system (see box 1.1) that consists of a 285 kWp solar system, a 690 kWh Li-ion (LiFePO4) battery, and a The analysis suggests that all component costs of mini 285 KVA back-up generator set. grids will see declines, but as imported equipment costs (PV modules, batteries, electronics) tumble downward As the generator set is expected to phase out over time through spot markets, the remaining components will due to economic forces, and optimization of this system assume more of the share of overall costs. The NPV of has been ongoing for more than a century, we have used ongoing major replacements such as batteries benefits a specific cost for the full system of $100/kW. For the solar from the future size of portfolios: they will be larger and unit and battery systems we looked into the composition of the costs lower. For example, the batteries installed at year 7 basic elements and found on the commodity market the to replace a failing pack will be part of a scaled-up battery estimated cost for each material. purchase to build 5,000 mini grids. A typical solar mini grid system needs an estimated 20 tons When moving from 100 to 100,000 mini grids in a portfolio, of glass, 16 tons of steel, 13 tons of concrete, 5 tons of alu- the marginal gain diminishes in terms of percentages. The minum, 2 tons of silicon, 2 tons of copper, 2 tons of plastic; largest gain is made from 100 to 1,000 and from 1,000 to the Li-ion batteries require an estimated 650 kg of alu- 10,000 systems per portfolio. Growing beyond this scale minum parts, 450 kg of graphite, 400 kg of copper parts, might call for closer scrutiny; perhaps multiple, smaller 250 kg of iron, and about 50 kg of lithium. Adding value to portfolios (several of 10,000 mini grids) might be optimum. these raw materials resulted in a total cost of $157k for a Additional research will need to be conducted to obtain solar-battery-genset power plant. This is a 53 percent cost more specific insights for the industry. reduction from what is reported in box 1.1 ($333,000 for the generation system). TABLE 1.9 • Change in unit costs with economies of scale, by cost category Portfolio development Procurement, construction, and management installation, and customer Total NPV per LCOE with 22 LCOE with 40 Portfolio size per mini grid (US$, engagement per mini grid mini grid (US$, percent load percent load (# of mini grids) thousands) (US$, thousands) thousands) factor ($/kWh) factor ($/kWh) 100 23 251 473 0.36 0.20 200 17 220 412 0.31 0.17 400 14 192 363 0.27 0.15 800 18 168 319 0.24 0.13 1,600 10 148 281 0.21 0.12 100 23 251 473 0.36 0.20 1,000 11 163 310 0.23 0.13 10,000 6 106 187 0.14 0.08 100,000 4 69 143 0.11 0.06 Source: ESMAP calculations and analysis using costing data described in this chapter. NPV = net present value; LCOE = levelized cost of energy; kWh = kilowatt-hour. 7 2   MINI GRIDS FOR HALF A BILLION PEOPLE When maintaining the rest of the upfront cost (manage- and commercial clients, can reduce their LCOE by up to ment, distribution system, land and logistics, installation 30 percent. When combined with the expected declines cost, taxes, and duties), the cost reduction is 21 percent in CAPEX and OPEX, the cost of electricity from a best- of the total upfront cost for the power plant. Carrying this in-class third-generation system will be $0.20 per kWh forward into the LCOE calculation, assuming a 75 percent by 2030. This is for mini grids with productive applica- CAPEX, 20 percent OPEX and 5 percent for preparation tions that enable a 40 percent load factor. costs, the power plant cost savings lowers the LCOE by • Expected decreases in component costs can reduce 15 percent, from $0.38/kWh to $0.32/kWh. If we use upfront investment costs to less than $2,500/kWfirm the breakdown as found in a database of 440 projects by 2030. In improving the design of a race car, a designer (CAPEX, 64 percent; OPEX, 31 percent; and 5 percent for might find it impossible to shave 1 kg off in a single loca- preparation costs), the reduction of the LCOE is less, and tion but could identify 20 places in the car where she when we use the breakdown as calculated in the “econ- could reduce 50 grams. Cost reductions in mini grids omies of scale” analysis (CAPEX, 52 percent; OPEX, 44 work the same way, with cost reductions in many dif- percent; and 4 percent for preparation costs), the impact ferent components adding up to a substantial overall of the power system’s cost reduction on the LCOE falls to cost reduction. If the prices that mini grid developers 11 percent. pay for the PV array, Li-ion batteries, and inverters and As also mentioned under the “economies of scale” analy- associated electronics decline by the same proportion sis, the overall reduction in power plant costs is essential as mainstream industry benchmarks between 2020 for an overall competitive product in the marketplace. and 2030, the upfront capital cost per kWfirm of a solar Equally important, and a topic that has not received the hybrid mini grid would fall by almost 25 percent. same level of attention in this handbook, is the innovation • Economies of scale will reduce the LCOE of mini grids necessary to also reduce the OPEX, including the cost of even further. As developers build portfolios of mini grids major replacements. instead of one-off projects, they benefit from increased economies of scale—primarily as a result of bulk pur- chases of components and increased efficiencies CONCLUSION through standardized processes and increased know- how. Analysis of the data collected in ESMAP’s survey Best-in-class mini grid costs have plummeted in the past of mini grids in Africa and Asia indicates that economies few years. In 2018 ESMAP conducted a cost analysis of 53 of scale can greatly reduce capital costs. As we describe mini grids published in the executive summary of Mini Grids in this chapter, for every additional 100 customers a for Half a Billion (ESMAP 2019). At that time, the “best-in- mini grid serves, its cost per customer falls on average class” mini grid produced electricity with a (financial) LCOE by about $9. Cost reductions from economies of scale of $0.55 per kWh. In the three years since this analysis, complement the downward effect on costs from greater best-in-class costs have dropped nearly 31 percent to only recourse to productive uses of electricity. $0.38 per kWh, thanks to decreases in the cost of solar • Using geospatial and other digital tools to develop panels, batteries, inverters, and efficiencies through econ- portfolios of mini grids will also reduce costs. Geo- omies of scale. spatial analysis allows developers to assess mini grid SUMMARY OF POTENTIAL COST REDUCTIONS sites at a fraction of the cost of traditional site assess- ment activities—from around $30,000 per site with- The trends in CAPEX and OPEX highlighted above will lead out using geospatial analysis, to approximately $2,300 to major cost reductions in four areas for third-generation per site using geospatial analysis. A number of estab- mini grids through 2030: lished mini grid developers in Sub-Saharan Africa use • Increasing income-generating uses of electricity can geospatial and other analytical software to plan their decrease the LCOE by 25 percent or more and, when portfolios remotely. They prioritize sites for mini grid combined with the expected cost declines described development and use technology-enabled processes below, will bring the economic cost of mini grid elec- to estimate demand, allowing them to optimize sys- tricity to almost $0.20/kWh by 2030. The baseline load tem design across their portfolios. Where government factor for mini grids of 22 percent reflects low levels of or donor entities are conducting the portfolio-level income-generating uses of electricity. Mini grids that analysis, the data can be analyzed and disseminated can increase their load factors to 40 percent through to developers on a web-based platform like Odyssey significant daytime consumption by local businesses Energy Solutions. MINI GRIDS FOR HALF A BILLION PEOPLE    7 3 GOVERNMENT’S ROLE IN REDUCING MINI GRID costing data for use in understanding detailed mini grid COSTS AND CATALYZING INNOVATION costs as they evolve in different markets. Governments can help keep the path open for mini grid More data are also needed on the standards to which component technology innovation and cost decreases by mini grids are built. For example, are poles and wires built designing and implementing regulatory frameworks and to standards that a utility would use? Or are cheaper, mini grid programs that provide light-handed regulation untreated wooden poles used to save costs? What is and exempting mini grid components from import taxes the expected life cycle of the battery? Mini grids built to (see chapter 9 for a detailed discussion of mini grid regu- different standards will naturally report different costs, lations). It is important to design standards that leave open reflecting these different standards. Without improved opportunities for innovation and not to assume (and thus knowledge of the underlying standards for each mini lock in) a particular technology or configuration. grid, variations that currently appear to be noise in data could more meaningfully and accurately reflect the real- Rural electrification agencies can harness these cost sav- ities on the ground and help identify areas where action ings by designing programs that provide opportunities for is warranted to reduce mini grid costs, improve quality, or capable developers to develop multiple nearby sites as part both. This chapter and the underlying database should be of a larger, comprehensive program. Doing so allows for viewed as living documents, which will benefit from better, economies of scale in project identification (especially har- and more, data over time. nessing geospatial information), engineering and design, site assessment and community negotiations, equipment procurement and installation, O&M, and tariff collection. REFERENCES Ensuring a competitive marketplace for mini grids will be important to promoting innovation and continued cost AMDA (Africa Minigrid Developers Association) and ECA (Economic Consulting Associates). 2022. Benchmarking Africa’s Minigrids. declines. The data presented in this chapter show sizable Nairobi, Kenya: African Minigrids Developers Association. https:// cost variations, implying in part the ability of mini grid devel- africamda.org/. opers to procure equipment at internationally competitive AP News. 2018. “Global Diesel Generator Market Size, Share & prices.31 In cases where costs are on the high end of the Trends Analysis Report, 2013–2022.” ResearchAndMarkets.Com, mini grids we analyzed, the systems were clearly overbuilt, October 19, 2018. https://www.businesswire.com/news/home/201 designed to meet a load that may not materialize for years. 81019005370/en/Global-Diesel-Generator-Market-Size-Share- Trends-Analysis-Report-2013-2022---ResearchAndMarkets.com Some subsidy programs, particularly those that subsidize a portion of renewable energy generation investments, Balabanyan, Ani,Yadviga Semikolenova, Arun Singh, and Min A Lee. 2021. “Utility Performance and Behavior in Africa Today.”World Bank, Wash- incentivize oversizing mini grids. Costs reported at the low ington, DC. https://openknowledge.worldbank.org/handle/10986/ end in this study indicate the best possible practice at the 36178. frontiers in a competitive market, keeping in mind the need Blimpo, M., and Malcolm Cosgrove-Davies. 2019. Electricity Access in to specify minimum customer-service levels and not stint Sub-Saharan Africa: Uptake, Reliability, and Complementary Factors on quality. As mini grids are deployed in larger quantities for Economic Impact. Africa Development Forum Series. Washing- and markets become more competitive, costs will trend ton, DC: World Bank. downward toward, and beyond, the best-in-class cost and BNEF (Bloomberg New Energy Finance). 2020a. “Deep Dive into Utili- performance benchmarks revealed in this study. ty-Scale PV System Cost.” BNEF. 2020b. “Battery Pack Prices Cited below $100/KWh for the THE IMPORTANCE OF COORDINATED First Time in 2020, While Market Average Sits at $137/KWh.” Bloom- COLLECTION OF DATA ON MINI GRID COSTS bergNEF (blog), December 16, 2020. https:/ /about.bnef.com/blog/ battery-pack-prices-cited-below-100-kwh-for-the-first-time-in- Data collection on mini grids is at an early stage. Better and 2020-while-market-average-sits-at-137-kwh/. more uniform data will produce more useful results and BNEF. 2021. “Battery Pack Prices Fall to an Average of $132/KWh, observations. More effort should be spent on standardizing but Rising Commodity Prices Start to Bite.” BloombergNEF (blog), data collection and integrating data collection into report- November 30, 2021. ing requirements into mini grid programs. One branch of BNEF. 2022.“Solar – 10 Predictions for 2022.” January 26, 2022. https:// about.bnef.com/blog/solar-10-predictions-for-2022/. this effort could take the form of a plug-in into a standard- ized mini grid bidding and accounting software package Carlin, Kelly, Josh Agenbroad, Stephen Doig, and Kendall Ernst. 2018. Minigrids in the Money: Six Ways to Reduce Minigrid Costs by 60% that provides developers front-end geospatial information for Rural Electrification. Boulder, CO: Rocky Mountain Institute. on prospective villages and markets, optimizes mini grid https://www.rmi.org/insight/minigrids-money/rmi-seeds-mini- system design, helps link developers with equipment sup- grid-report/. pliers and financiers, helps keep track of key milestones in Danvest Energy. 2019. “Danvest Energy A/S—Wind Diesel Systems, project development, and provides suitably anonymized Solar Diesel Systems.” https://www.danvest.com/how-does-it-work 74   MINI GRIDS FOR HALF A BILLION PEOPLE Dave, Rutu, Sandra Keller, Bryan Bonsuk Koo, Gina Fleurantin, Elisa Por- Framework. Washington, DC: World Bank. http://hdl.handle.net/ tale, and Dana Rysankova. 2018. Cambodia—Beyond Connections: 10986/30102. Energy Access Diagnostic Report Based on the Multi-Tier Framework. Plautz, Jason. 2021. “Form Energy’s $20/KWh, 100-Hour Iron-Air Bat- Washington, DC: World Bank. http:/ /hdl.handle.net/10986/29512 tery Could Be a ‘Substantial Breakthrough.’” Utility Dive, July 26, Energy Trend. 2021. “Solar Price | EnergyTrend.” April 6, 2021. https:// 2021. https://www.utilitydive.com/news/form-energys-20kwh-100- www.energytrend.com/solar-price.html. hour-iron-air-battery-could-be-a-substantial-br/603877/. Energy Trend. 2022. “Solar Price | EnergyTrend.” April 19, 2022. https:// Sanderson, Henry. 2019. “Congo, Child Labour and Your Electric Car.” www.energytrend.com/solar-price.html. Financial Times,July 7, 2019. https://www.ft.com/content/c6909812- Feldman, David, Vignesh Ramasamy, Ran Fu, Ashwin Ramdas, Jal 9ce4-11e9-9c06-a4640c9feebb. Desai, and Robert Margolis. 2021. U.S. Solar Photovoltaic System Smart Energy International. 2018. “Global Trends in Smart Meter- Cost Benchmark: Q1 2020. Technical Report NREL/TP-6A20-68925. ing.” Smart Energy International, December 31, 2018. https://www. Golden, CO: National Renewable Energy Laboratory. https:/ /www. smart-energy.com/magazine-article/global-trends-in-smart-me- nrel.gov/docs/fy21osti/77324.pdf. tering/ Global Industry Analysts. 2022. Global Smart Meters Industry. February Solarpower Europe. 2018. “Global Solar Market Grows over 29% in 2022. 2017 with Even More to Come in 2018.” PV Magazine, March 14, Haegel, Nancy M., Harry Atwater, Teresa Barnes, Christian Breyer, 2018. https://www.pv-magazine.com/press-releases/global-solar- Anthony Burrell, Yet-Ming Chiang, Stefaan De Wolf, et al. 2019. “Ter- market-grows-over-29-in-2017-with-even-more-to-come-in-2018/. awatt-Scale Photovoltaics: Transform Global Energy.” Science 364 Stevens, Pippa. 2021. “More than Half of 2022’s Solar Projects Threat- (6443): 836–38. https:/ /doi.org/10.1126/science.aaw1845.IEA ened by Spiking Costs, New Report Finds.” CNBC, October 26, 2021. (International Energy Agency). 2021. Renewables 2021. Paris: IEA. https://www.cnbc.com/2021/10/26/more-than-half-of-2022s-so- https://iea.blob.core.windows.net/assets/5ae32253-7409-4f9a- lar-projects-threatened-by-spiking-costs-new-report-finds.html. a91d-1493ffb9777a/Renewables2021-Analysisandforecastto2026. Taizhou Amity Care International Co., Ltd. 2013. “Prestressed Concrete pdf. Spun Pile Pole Centrifugal Spinning Machine.” https://www.youtube. Innovus. 2015. “Innovus Power Microgrid Platforms: Delivering the com/watch?v=fY4hirhg3c4. Highest Penetration with the Lowest Levelized Cost of Energy.” Tenenbaum, Bernard, Chris Greacen, and Ashish Shrestha. 2022. Inter- http://www.innovus-power.com/wp-content/uploads/2015/08/ connected and Non-Interconnected Mini Grids in Undergrid Areas of innovus-power_hybrid_v082015.pdf. Nigeria and India. ESMAP Technical Report. Washington, DC: World IRENA (International Renewable Energy Agency). 2022. “World Energy Bank. Transitions Outlook 2022: 1.5°C Pathway.” Abu Dhabi: IRENA. Trading Economics. 2022. “Lead - 2022 Data - 1993-2021 Historical - Jäger-Waldau, A. 2017. PV Status Report 2017. EUR 28817 EN. Luxem- 2023 Forecast - Price - Quote - Chart.” bourg: Publications Office of the European Union. https://publica- Trimble, Chris, Masami Kojima, Ines Perez Arroyo, Farah Mohammad- tions.jrc.ec.europa.eu/repository/handle/JRC108105. zadeh. 2016. Financial Viability of Electricity Sectors in Sub-Saha- Jenkins, Lisa Martine. 2022. “Chinese Solar Panels Seized at US Bor- ran Africa : Quasi-Fiscal Deficits and Hidden Costs. Policy Research der over Possible Human Rights Abuses.” Protocol, August 9, 2022. Working Paper;No. 7788. World Bank, Washington, DC. https:/ / https://www.protocol.com/bulletins/xinjiang-solar-panels-uy- openknowledge.worldbank.org/handle/10986/24869 US DOE (US ghur-enforcement. Department of Energy). N.d. “Perovskite Solar Cells.” Energy.Gov. Kairies, Kai-Phillip. 2017. “Battery Storage Technology Improvements Accessed May 1, 2021. https:/ /www.energy.gov/eere/solar/per- and Cost Reductions to 2030: A Deep Dive.” PowerPoint presen- ovskite-solar-cells. tation at the International Renewable Energy Agency Workshop, Wagman, David. 2020. “US Energy Storage Strategy Includes Tech Cost March 17. https:/ /www.irena.org/-/media/Files/IRENA/Agency/ Estimates.” PV Magazine International, December 23, 2020. https:// Events/2017/Mar/15/2017_Kairies_Battery_Cost_and_Perfor- www.pv-magazine.com/2020/12/23/us-energy-storage-strate- mance_01.pdf?la=en&hash=773552B364273E0C3DB588912F- gy-includes-tech-cost-estimates/. 234E02679CD0C2. Yu, Hyun Jin Julie. 2018. “A Prospective Economic Assessment of Koo, Bryan Bonsuk, Dana Rysankova, Elisa Portale, Niki Angelou, Residential PV Self-Consumption with Batteries and Its Systemic Sandra Keller, and Gouthami Padam. 2018. Rwanda—Beyond Con- Effects: The French Case in 2030.” Energy Policy 113 (February): nections: Energy Access Diagnostic Report Based on the Multi-Tier 673–87. https://doi.org/10.1016/j.enpol.2017.11.005. Framework. Washington, DC: World Bank. https:/ /openknowledge. Zaw Min. 2019. Interview. Kyaw Soe Win (KSW) Hydropower Company. worldbank.org/handle/10986/30101. March 18. Lazard. 2018. “Lazard’s Levelized Cost of Storage Analysis—Version 4.0.” https:/ /www.lazard.com/perspective/levelized-cost-of-ener- gy-and-levelized-cost-of-storage-2018/. NOTES Lazard. 2020.“Lazard’s Levelized Cost of Energy Analysis—Version 14.0.” https://www.lazard.com/perspective/levelized-cost-of-energy-and- 1. This chapter uses the terms solar and photovoltaic interchangeably levelized-cost-of-storage-2020/. to mean generation of electricity from sunlight. Lazard. 2021. “Lazard’s Levelized Cost of Energy Analysis—Version 2. For some 2021 mini grids, contracted costs were used rather than 15.0.” https://www.lazard.com/perspective/levelized-cost-of-ener- post-commissioning costs. gy-levelized-cost-of-storage-and-levelized-cost-of-hydrogen/. 3. Not included for detailed analysis in this chapter, but nonethe- Padam, Gouthami, Dana Rysankova, Elisa Portale, Bryan Bonsuk Koo, less promising (especially for communities with needs for smaller Sandra Keller, and Gina Fleurantin. 2018. Ethiopia—Beyond Con- amounts of electricity) are lower-cost direct current (DC) “mesh nections: Energy Access Diagnostic Report Based on the Multi-Tier grids” or “skinny grids” that distribute DC electricity for lighting, MINI GRIDS FOR HALF A BILLION PEOPLE    75 electronics, and small appliances like fans and even efficient refrig- average PV module costs. For our 2030 calculations we used aver- erators or electric rickshaws. See box 1.2. age cost of PV modules in these Li-ion battery mini grids and then 4. In the world of grid-connected power plants, LCOE is used to com- applied industry-projected cost declines discussed in the “PV Mod- pare on an apples-to-apples basis the cost of energy delivered to ule Trends” section of this chapter. the grid network from generating assets that have different capital 15. Because of Li-ion batteries’ ability to discharge energy more deeply, costs, fuel costs, and lifetimes. LCOE is typically expressed in cur- they can have a nameplate capacity that is 25 percent smaller than rency per kilowatt-hour. if lead-acid batteries were used. 5. In these three countries we restricted our analysis to mini grids 16. Somewhat counterintuitively, the optimum renewable energy frac- with lithium-ion batteries because they are the most common type tion decreases slightly in some of the cases as the load curve shifts (accounting for 76 percent of battery types in mini grids from our from normal to sun-following. For example, for cases 3 and 4 (Ethi- data set in Nigeria, 50 percent in Myanmar, and 100 percent in Ethi- opia and global Li-ion, respectively), the renewable energy fraction opia), and also have lower LCOE, on average, than mini grids with falls from a 22 percent load factor to a 22 percent sun-following lead-acid batteries. scenario. Why would the renewable energy fraction decrease when 6. Bolivia, Ethiopia, Indonesia, Myanmar, Nigeria, and Tanzania. shifting to a more solar-coincident load? The answer lies in the component sizing of the optimal mini grid in each case. Moving 7. Bangladesh, Bolivia, Chad, Ethiopia, Ghana, Guinea Bissau, India, more solar-coincident loads allows the system to rely less on bat- Indonesia, Ivory Coast, Kenya, Liberia, Myanmar, Nepal, Nigeria, tery storage, with the consequence that the system becomes a bit Palestine, Sierra Leone, Tanzania, Vanuatu, and Vietnam. more reliant on backup diesel during occasional rainy periods. 8. In addition to 20-year project economic life, we have also modeled 17. Generators are often rated in apparent power (kVA). The genera- the impact of 15- and 25-year lifetime assumptions on LCOE. Add- tor’s real power output (kW) is the apparent power multiplied by ing five years decreases LCOE by about 2.2 US cents per kWh in the power factor, typically assumed to be 0.8 for design purposes. the 22 percent load factor / 0 percent subsidy case, with a smaller reduction in other cases. Subtracting five years increases LCOE by 18. Peak power output of a solar panel is the power output at a solar about 2.7 US cents per kWh for the same case, with lesser impact irradiance of 1,000 watts per square meter, 1.5 air mass, and a tem- in other cases. perature of 25º C. 9. The weighted average cost of capital of a capital structure compris- 19. In 2019 an early version of ESMAP’s Mini Grids for Half a Billion ing 40 percent equity at 12 percent return and 60 percent debt at 8 included a similar waterfall graph based on data from 36 mini grids percent interest is 9.6 percent. Assumptions consistent with Lazard commissioned between 2012 and 2018. A comparison of the 2019 (2021). graph with this 2022 version reveals that PV costs (including for PV inverters) plummeted, from 16 percent to about 10 percent of 10. The combination of a 9.6 percent nominal discount rate and 3 per- total project costs, reflecting lower PV costs in recent years. Bat- cent inflation yields a real discount rate of 6.41 percent. tery and battery inverter + EMS costs remained the same. Distribu- 11. The World Bank tracks pump prices for diesel from Sub-Saharan tion and meters as a portion of total project costs increased from Africa at https:/ /data.worldbank.org/indicator/EP.PMP.DESL.CD? 21.0 percent to 26.6 percent as other costs fell, as did installation, end=2016&locations=ZG&start=2010. Corrected for inflation, the which increased to 11.3 percent from 8.0 percent. On the other most recent (2016) pump price is $1.08 per liter. The fuel cost sen- hand, project development costs dropped from 9.0 to 5.9 percent, sitivity analysis investigated $0.75 and $1.50 per liter of diesel fuel which likely reflects benefits of clustering and perhaps also a trend in addition to the base case of $1.00 per liter. Because the mini grids in more recent projects to fold project development costs into have high penetrations of renewable energy, the cost of diesel had a reported equipment costs. relatively small effect on LCOE. In HOMER sensitivity runs with die- 20. PV inverters convert the direct current (DC) electricity produced sel fuel costs of $0.75 and $1.50 per liter, the variation in LCOE was by the solar array into alternating current (AC) power on the mini less than +6 percent of the $1.00 per liter base case. The project grid’s network. lifetime analysis considered project economic lifetimes of 15 and 25 years in addition to the base case of 20 years. The +5 year project 21. Battery nameplate capacity is typically indicated in ampere-hours lifetime assumptions affected LCOE by less than +7 percent, with (Ah). Battery nameplate kWh is calculated as the Ah multiplied by the strongest impacts in the 22 percent load factor case. the battery’s nominal voltage. 12. Diesel generators’ nonfuel OPEX is estimated at $0.03 per kWh; 22. Lead-acid batteries are typically not discharged more than 50–60 solar PV OPEX, at $10 per kW a year; and battery OPEX, $10 per kW percent, whereas lithium-ion batteries can be discharged to 80 per- a year. These variable O&M assumptions are held constant across cent depth of discharge. all HOMER modeling runs. 23. For an apples-to-apples comparison across different battery chem- 13. The prices that developers pay for individual components are typ- istries, it is useful to compare the levelized cost of storage (LCOS). ically higher than the wholesale price direct from the factory. Even LCOS is analogous to the levelized cost of energy (LCOE) but uses as third-generation mini grid developers build larger portfolios, the discounted cost of purchasing and operating the battery over all but the very biggest developers will still pay higher prices than the course of its lifetime (in lieu of the cost of generating and distrib- those available at the factory door. As a result, we conservatively uting electricity), divided by the discounted discharged electricity. assumed that by 2030, the typical third-generation mini grid devel- It is the levelized cost associated with storing and withdrawing one oper would be able to purchase components at their 2020 factory kWh of electricity. The data set does not provide sufficient data for spot prices. an LCOS calculation. Lazard (2018) finds that for US applications in 2018 at the scale of 40 kWh of storage capacity (“residential scale” 14. In our data set, the average cost of PV modules in mini grids built in their analysis, equivalent in storage capacity to the smallest mini with Li-ion batteries (mostly built in 2019 to 2021) was $534 per grids considered in this chapter), the LCOS for lithium-ion batteries kWp, reflecting the fact that our best-in-class mini grid, while best- was $0.476–$0.735/kWh. Lead-acid batteries have a comparable in-class overall and for other equipment costs, had higher than range of LCOS values at this scale, of $0.512–$0.707/kWh. For proj- 76   MINI GRIDS FOR HALF A BILLION PEOPLE ects at the scale of 2 megawatt-hours of energy storage (Lazard’s (Stevens 2021). As of April 2022, the global average spot price “Commercial & Industrial scale,” about twice as much storage as for PV was $230 per kWp for 330–335 W multi-crystalline mod- the largest village mini grids studied in this chapter), the LCOS for ules. Industry experts expects these bottlenecks to be transitory lithium-ion batteries was $0.315–$0.366/kWh, several cents lower (Energy Trend 2022). than the $0.382–$0.399/kWh LCOS for lead-acid batteries. The 27. Wright’s Law posits that every cumulative doubling in the cumula- analysis assumed a 20:80 percent debt to equity ratio, with debt at tive amount of a product produced leads to a consistent percent- 8 percent and the cost of equity at 12 percent (Lazard 2018). age cost decline. 24. Parameters not captured in the distribution network cost data are 28. PV deployment has been growing at an average of 40 percent in the standards to which the low-voltage distribution network is built. recent years (50 percent in 2016, 29 percent in 2017) (Solarpower Projects built to a high standard or a grid-ready standard will have Europe 2018) and a compound average growth rate of more than much higher costs per customer and per kilometer than those built 40 percent over the past 15 years (Jäger-Waldau 2017). to lower standards (for example, using untreated wooden poles, 29. See http://haiyuindustry.sell.everychina.com/p-101752826-concrete- low pole heights, and undersized conductors and hardware), as will spun-electric-pole-production-machine.html for photos and a distribution grids that are deliberately oversized to accommodate more detailed description of this process. future growth. Other factors influencing the wide variations in cost per kilometer and per customer that are not captured in the survey 30. Field visits at the preparation stage are needed to engage with likely include whether poles for distribution were constructed of local communities to discuss agreements, such as the terms of land pur- materials and not costed as part of the project, accounting practices chases or leases, and to verify the geospatial analysis data—and related to in-kind labor and materials supplied by local communities, they can be handled by a much leaner team. whether service is single phase or three phase, and whether public 31. A related issue more germane to the data is that some companies lighting costs were bundled into this category by a developer. appear to report project development and business development 25. Replacement of large assets such as batteries is not included in costs explicitly, others blend them into equipment costs in the these O&M costs. These costs are included, however, in LCOE cal- form of markups, and still others internalize these costs and do not culations earlier in this chapter. report them at all. The background study carried out for this chap- ter did not include data gathering on subsidy amount or in-kind 26. With high commodity prices for polysilicon, aluminum, and other accounting (for local materials and community contribution) or raw materials due to post-COVID supply bottlenecks prices were address competition in markets. pushed higher for PV modules in 2021 and the first part of 2022 MINI GRIDS FOR HALF A BILLION PEOPLE    7 7 CHAPTER 2 NATIONAL STRATEGIES AND DEVELOPER PORTFOLIOS: THE ROLES OF GEOSPATIAL ANALYSIS AND DIGITAL PLATFORMS CHAPTER OVERVIEW This chapter discusses geospatial analysis and other digital tools that can support electrification planning at both the national and portfolio levels. Drawing on real-world examples from Nigeria and Ethiopia, and leading mini grid developers, the chapter lays out how to use cutting-edge technologies like geospatial software and online platforms to develop large portfolios of mini grids. It also introduces some of the leading technology providers for such planning tools. Thanks to new geospatial analysis technologies, a port- household income, poverty, commercial activities, willing- folio approach to mini grid development is becoming ness to pay). Spatial modeling delivers a least-cost plan by mainstream in the industry and in national electrification identifying beforehand the technology best suited to local planning. This is occurring at the national level for least-cost circumstances—technically feasible and economically via- electrification planning and among mini grid companies ble. At the same time, geospatial plans can also identify themselves. Geographic information system (GIS) software communities requiring decentralized solutions (mini grids) and geospatial data are becoming key tools for planning as they wait for the grid. electrification at the national level and performing rapid site Geospatial plans are essential in siting mini grids and sig- assessments. Mainstream digital tools are expediting tech- naling the likelihood of grid arrival, information that cur- nological advances and cost reductions, including: tails asset stranding. The identification of communities • Satellite imagery and spatial products for which mini grids offer the optimal technology solution • Big data and cloud-based computing requires at least the following: • More sophisticated algorithms and analytical solutions • Electricity demand estimates, including for productive (for example, heuristics and machine learning) uses; • Global positioning system devices and the proliferation • The location of existing infrastructure and modeling of of web-based and mobile technologies grid rollout; and • Higher-quality open-source software • Estimation of local renewable generation potential. A geospatial approach ensures that national electrification Using geospatial analysis in planning mini grid portfolios is mapped cost-effectively on the existing grid network and could cut the time spent on deployment. its attributes digitalized. Demand and supply of electricity can be geolocated by overlaying demographic data (such Geospatial planning cannot replace field-based feasibility as population density and growth patterns) on social infra- studies, but it can determine mini grid potential. It does structure (for example, schools, health centers, admin- this by evaluating current and anticipated service needs istrative offices) and the economic landscape (such as (including productive use) and the time frame for grid 78   MINI GRIDS FOR HALF A BILLION PEOPLE arrival. This exercise prepares engineers and policy makers understand how mini grids could support a speedy roll- for planning electricity services and allocating public fund- out of electrification; how many people or households can ing, ensuring that public interventions (where and why) are mini grids serve with high-quality and sustainable elec- done with equity foremost in mind. System optimization, tricity over the long term? Investors and financiers, on the network design tools, and online platforms analyze data, other hand, are interested in the addressable market and develop project proposals, select developers, solicit financ- the economically viable potential of mini grids in Sub-Sa- ing, and monitor and verify implementation. haran Africa. This section briefly describes how new spa- tial data and analysis can help address these questions, This chapter assesses the market potential for mini grid providing qualitative and, to the extent possible, quanti- sites selected by geospatial data. It then looks at tools and tative data. analyses that support least-cost electrification and plan- ning exercises. We then assess how geospatial and other The single most critical data set required for this analysis digital tools are being used to save both time and costs is the settlement distribution—that is, the location of set- in mini grid project development, from site prospecting tlements or buildings over the area of interest. Over the and analyzing demand to right-sizing solutions, packag- past few years, several data sets have been developed in ing projects, and taking them to market. Examples from this regard, based on high-resolution satellite imagery and the “frontier” of these planning exercises, particularly the processing techniques (for example, machine learning). We World Bank’s Nigeria Electrification Project and its Ethio- describe and use some of them below. pia project, Access to Distributed Electricity and Lighting in Ethiopia, illustrate their practical application. SIMPLE EXPLORATORY SPATIAL DATA ANALYSIS FOR SUB-SAHARAN AFRICA USING GRID3 In the first example, we explore the settlement distribution ASSESSING THE MARKET layer for Sub-Saharan Africa provided by GRID3 (CIESIN POTENTIAL FOR MINI GRIDS 2020). The layer constitutes a comprehensive set of set- tlement polygons classified into built-up areas, small set- With programs ramping up worldwide as countries seek tlement areas, and hamlets; 326,000 settlements were to meet their Sustainable Development Goal (SDG) 7 tar- found to be more than 1 kilometer (km) from the existing gets, what roles do mini grid systems play in that process? grid1 (figure 2.1), including a preponderance of settlements Different stakeholders come at this question from various in the 100–1,000 population range. Extracting additional standpoints. Governments and policy makers want to information about the settlements to facilitate site selec- FIGURE 2.1 • Scatter plot of settlement population vs population density in Sub-Saharan Africa All 326,000 settlements more than 1 km from the grid 10k 8k Settlement density (people/km2) 6k 4k 2k 0 0 100 1k 10k 100k Source: ESMAP analysis of GRID3 data. Settlement population Note: Point size reflects the area of the settlement. MINI GRIDS FOR HALF A BILLION PEOPLE    7 9 tion, the following criteria were thus set to define a settle- 10,000 to 100,000 people, likely requiring mini grids at the ment suitable for mini grid electrification: 500 kW to 1 megawatt (MW) scale. • Number of people: more than 100 and less than 100,000 Settlement population distribution is presented by country in figure 2.2, with the estimated addressable market for • Distance from the existing grid: more than 1 km mini grids presented in absolute numbers atop the bars for • Population density: more than 1,000 people/km2 each country. The height of the bars represents the share of Settlements with populations of more than 100,000, or this segment as a percentage of the total population of the any settlements located less than 1 km from the grid, are country. This addressable market for mini grids is further considered either already electrified or candidates for grid disaggregated into the same settlement sizes as described electrification and thus excluded from this analysis. On a in table 2.1. similar note, any settlement with less than 100 people is Using the same selection criteria and settlement sizes, fig- considered a better candidate for solar home systems ure 2.3 maps and visualizes their population distribution. (SHSs) than mini grids. The density assumption was set as such to satisfy mini grid design criteria; that is, to avoid THE GLOBAL ELECTRIFICATION PLATFORM AND the selection of settlements that have sparse populations. LEAST-COST ELECTRIFICATION ANALYSIS FOR These criteria reflect the prevailing view of what constitutes SUB-SAHARAN AFRICA a good candidate site for a mini grid; they were based on The Global Electrification Platform (GEP) is an open-ac- past projects and experience on the ground. The results cess, interactive, online platform that models and visu- presented below are bound to these selection criteria and alizes pathways toward universal access, split into an should be interpreted with caution accordingly. intermediate strategy for 2025 and full electrification by As table 2.1 indicates, about 23.7 percent of Sub-Saharan 2030, for countries marked by severe access deficits. The Africa’s population (that is, 276.9 million of 1.17 billion peo- current version, GEP V.2.0, officially launched in April 2022, ple) lives in settlements that in theory could be markets for explores 96 unique scenarios. The set of results was mod- mini grids. Assuming an average size of 5 people/house- eled with a modified version of the Open Source Spatial hold, about 55.4 million households on the subcontinent, Electrification Tool (OnSSET). This is a flexible and mod- or about 291,000 clusters, could be served by mini grids. ular GIS-based energy modeling tool developed to support electrification planning and decision making by estimating, As an initial assessment of the potential for standardiza- analyzing, and visualizing the most cost-effective electrifi- tion in the rollout of mini grids in Sub-Saharan Africa, we cation strategy. In doing so, it takes into account spatially also explored whether we could find convergence around explicit characteristics related to energy, such as popula- certain sizes of mini grids to serve the clusters described tion density and distribution, proximity to transmission and above. The growing pipeline of mini grid projects under road network, night-time lights, and local renewable energy development under the World Bank’s Nigeria Electrification potential, among others. Project provided some data on the sizing of private-sec- tor-led mini grids vis-à-vis the customer base or settlement The GEP considers current and projected values of key size, indicating an average firm power allocation of roughly parameters such as population growth, demand level, 100 watts (W) per connection. We used this as a bench- technology costs, and other policy/planning limitations mark, while acknowledging variations from project to proj- in generating electrification scenarios. Results indicate ect, including in the ratios of commercial and productive the least-cost electrification technology per settlement end users to residential customers, and associated power (or cluster)2 across millions of clusters in 46 countries requirements. We anticipate that mini grids of 20 kilowatts of Sub-Saharan Africa. The addressable market for mini (kW), 80 kW, and 200 kW may be best suited to serve grids falls within a range that depends on the input param- the settlement sizes listed in table 2.1. Custom solutions eters and assumptions, with the key parameter being the will continue to be the preferred option for settlements of level of demand for unelectrified households. Usually, the Table 2.1 • Characteristics of Sub-Saharan African settlements suitable for electrification via mini grid Size (average pop.) 100–500 500–2,500 2,500–10,000 10,000–100,000 Total: Settlements 177,087 95,702 15,188 2,948 290,925 Population total 46,886,543 97,073,397 67,627,653 65,303,591 276,891,184 Share of total population (%) 4.01 8.31 5.79 5.59 23.71 Optimum mini grid sizing (kW) 20 80 200 Custom 80   MINI GRIDS FOR HALF A BILLION PEOPLE FIGURE 2.2 • Sub-Saharan Africa’s addressable market for mini grids 70% 5M 0. 60% Percent of total population 4M 5. M 4M M .3 50% .0 4. 23 24 11 M .5 26 M M 4M 40% 1.3 .8 M 9M 16 .4 2. .7M M 12 .1M 6. .5 M 42 7M 13 11 0M .2 .1M M 28 6. .6 0M 4. M 13 21 M 16 30% .5 2M .4 4. 16 M 6M 11 0M 4. M 1.2 M .0 4. 3. .0 2M 31 8M 33 0. M M 20% 8. .8 M .5 22 M .9 18 .5 27 M 36 .0 M 5M 9M 60 .5 M 5M M 0. 56 3. .4 .2 10% 4. 10 14 .7M 42 M 4M .2 18 2. 55 4M M M 1.9 1.1 2. 0 ep de lic ali C to d'I r rit ia ba ia Ug we So l Gu ire Za da ts ia Ke a h S ea Ga an Gu rra inea Su -Bis e n( u ia Si Gu ia So he) Rw alia er a Be n a Na nya Le ibia Es tho i Ni nin ria Sa a Co d Se go Bu gal Er di Dj rea Lib ti oz Gab ia biq n ue Pr ana ipe ina go M Ma o ag wi S A ar la tin ua te ige a on rn fric Ta ar n Ca and s da sa oo am o a u DR M nzan an Zi hiop Bo mb b er W out ngo n c ad la ub M Fa ut in wa ud n R Ver an Ch rk To n ria vo ge ibo b ne m ínc ru wa as it h so T m m nd Gh ine Le N A m Et au es h fri bo m l A Ca e te Eq Co Bu ca M éa om ra oT nt Sã Ce 100.0–500 500–2,500 2,500–10,000 10,000–100,000 Source: ESMAP analysis of GRID3 data. Note: Total country population shown in values atop each bar. Results indicate the percentage of people located in settlements (clusters) that fulfil the selection criteria for mini grid candidacy (that is, between 100 and 100,000 people, located more than 1 km from the main grid, with a distribution of more than 1,000 people/km2). Clusters derived from GRID3 (CIESIN 2020). FIGURE 2.3 • Sub-Saharan Africa’s addressable greater the targeted demand, the greater the share of market for mini grids, mapped by settlement mini grid potential. Table 2.2 presents how three different population levels of demand can dictate the share of mini grids in the least-cost mix. When demand is low, mini grids are the least-cost option for about 66.0 million people (or 13.2 million connections). Intermediate demand pushes mini grid potential to about 87.2 million people (17.5 million connections), while with high-demand scenarios, the potential is estimated at about 131 million people (26.2 million connections). Mini grids can also serve as pre-electrification solutions, which is to say, they could be least-cost options for settle- ments expecting the arrival of the main grid. Political, eco- nomic, and other considerations will ultimately determine if and when the grid reaches these communities. On the one hand, the newer mini grids mostly meet code and could connect to the main grid once it arrives. On the other hand, falling costs and decentralized renewable technologies tell us that not every community may need to connect to the Settlement population main grid. 100–500 2,500–10,000 Including pre-electrification, in the low demand scenario, 500–2,500 10,000–100,000 mini grids are cumulatively the least-cost option for 66 mil- lion people (or 13.2 million connections)—the same as the Source: ESMAP analysis of GRID3 data. count in 2030. In the bottom-up and high-demand scenar- Note: Results indicate the number of people located in settlements ios, we see huge increases in the population served at least- (clusters) that fulfil the selection criteria for mini grid candidacy (that is, cost by mini grids once we account for mini grids that are between 100 and 100,000 people, located more than 1 km from the main grid, with a distribution of more than 1,000 people/km2). Clusters derived eventually connected to the grid: 105 million people (or 21 from GRID3 (CIESIN 2020). million connections) and 325 million people (or 65 million MINI GRIDS FOR HALF A BILLION PEOPLE    81 TABLE 2.2 • Selected electrification results for 2030 retrieved from the Global Electrification Platform, aggregated for 46 countries in Sub-Saharan Africa Mini grid potential Mini grid potential Demand target a (including pre-electrification role) People Connections People Connections Low demand 66,004,359 13,200,872 66,004,359 13,200,872 Bottom-up demand 87,249,707 17,449,941 104,716,450 20,943,290 High demand 131,052,705 26,210,541 325,025,815 65,005,163 Source: ESMAP analysis of Global Electrification Platform results. Note: GEP V.2.0 released in April 2022. Low demand reflects targets equivalent to Tier 3-4 for urban households and Tier 1 for rural households. High demand indicates Tier 4-5 for urban a.  households while Tier 2-3 for rural. Tier values differ per country depending on the current electrification status and/or goals. The bottom-up value reflects an intermediate level of demand that is based on the combination of socio-economic indicators that vary spatially (poverty rate and GDP). TABLE 2.3 • Breakdown of electrification results from bottom-up demand scenario <10 kW 10–100 kW 0.1–1 MW 1–10 MW 10–100 MW >100 MW Total Settlements 36,914 136,465 19,313 1,056 94 3 193,845 Population 1,781,648 39,190,271 35,259,640 17,100,927 9,281,040 2,102,924 104,716,450 Households 356,330 7,838,054 7,051,928 3,420,185 1,856,208 420,585 20,943,290 Percentage of total new 0.19 4.16 3.75 1.82 0.99 0.22 11.13 connections in Sub-Saharan Africa Source: ESMAP analysis of OnSSET data and Global Electrification Platform results. connections) respectively. Note that all values are inclusive generation for grid electricity and the cost of solar pho- of population growth and reflect aggregated data for all tovoltaic (PV) systems, affect the least-cost option and modeling years between 2020 and 2030. together account for the 96 unique scenarios modeled in the GEP. The scenario most favorable for the deployment Looking more closely at the bottom-up demand scenario, of mini grids as the least-cost solution thus varies from we see that least-cost mini grids serve close to 200,000 country to country. settlements, numbers that correspond to the aforemen- tioned 105 million people (or 21 million households). Table The finding from this exercise is that 430 million people 2.3 displays the distribution of this population by the size can receive access at least cost via mini grids. This includes of the mini grids projected to serve them. One can see that 380 million people in Sub-Saharan Africa living in the 58 most settlements (and their mini grids) fall in the 10 to access-deficit countries covered by the GEP, which rep- 100 kW range, while mini grids in the 10 kW to 1 MW range resent nearly 40 percent of all new connections achieved serve about 7.9 percent of all newly electrified population in in these countries. See figure 2.4 for a breakdown of this Sub-Saharan Africa. population by region and country, and table 2.4 for the GEP scenario codes for readers interested in exploring country- Based on the estimated new capacity in settlements where specific scenarios. mini grids are the least-cost option for all or part of the expected population between 2020 and 2030 For those interested in electricity access or the mini grid industry, it will come as no surprise that Sub-Saharan MAXIMAL MINI GRID DEPLOYMENT MODELED IN Africa is by far the most important market for mini grid THE GLOBAL ELECTRIFICATION PLATFORM electrification. While three African countries—Ethiopia, the Rather than selecting a particular demand scenario (low, Democratic Republic of Congo, and Nigeria—stand out for bottom-up, or high) and modeling results for all coun- their massive mini grid potential (49.4 million, 47.9 million, tries for that scenario, we use the GEP to review which of and 42.9 million people, respectively), one can also see the 96 modeled scenarios for each country deploys the from figure 2.4 that many others have huge populations most mini grids. While high-demand scenarios tend to that could be served by mini grids. Elsewhere, the electri- favor mini grids as the least-cost solution, they also tend fication potential is vast: 20 million people in Pakistan, 15 to label grid densification or extension as the least-cost million in Myanmar, and, in Haiti, almost 5 million people option. Besides, other parameters, like those referring to could gain least-cost electricity by mini grid. 82   MINI GRIDS FOR HALF A BILLION PEOPLE FIGURE 2.4 • Distribution by country of 429.5 million people served at least cost by mini grids in 58 access- deficit countries Regional population totals • SSA: 380.4 million • SAR: 24 million • EAP: 19.5 million • LAC: 5.6 million SSA SAR Niger; 15.8 Chad; 15.0 Uganda; 14.3 Sudan; 16.6 Pakistan; 20.3 Mali; 8.1 Angola; 7.8 Kenya; 7.4 Bangladesh; 3.6 Ethiopia; 49.4 EAP Somalia; 14.0 Zimbabwe; 4.7 Senegal; 4.6 Burundi; 4.6 Guinea; 6.7 Sierra DRC; 42.9 Madagascar; 16.6 Leone; Rwanda; 2.9 2.7 Zambia; 4.2 South Sudan; 12.7 Myanmar; 15.0 Malawi; 5.8 Other Papua New Cameroon; EAP; Guinea; 3.3 1.1 3.7 LAC Côte Nigeria; 47.9 Tanzania; 20.7 Mozambique; 15.8 Burkina Faso; 10.4 d’Ivoire; 5.6 CAR; 3.3 Other SSA: 16.4 Haiti; 4.9 Other LAC; 0.7 n Sub-Saharan Africa (SSA) South Asia (SAR) East Asia & Paci c (EAP) Latin America & Caribbean (LAC) Source: ESMAP analysis of Global Electrification Platform results. Note: Under the scenario most favorable for mini grids. Data from the Global Electrification Platform. SSA: Sub-Saharan Africa; SAR: South Asia; EAP: East Asia & Pacific; LAC: Latin America & Caribbean. Other SSA: Ghana (2.5); Eritrea (2.3); Benin (2.2); Liberia (1.9); Congo (1.5); Mauritania (1.4); Togo (1.1); Guinea-Bissau (1.0); Gambia (0.9); Lesotho (0.5); Equatorial Guinea (0.3); Comoros (0.2); South Africa (0.2); Eswatini (0.2); Namibia (0.1); Djibouti (0.1); Gabon (0). Other LAC: Nicaragua (0.4); Honduras (0.3). Other EAP: Cambodia (0.6); Solomon Islands (0.4); Vanuatu (0.1). TABLE 2.4 • GEP scenario codes for each country’s maximum number of new mini grid connections by 2030 People newly connected to Region Country Name GEP Scenario mini grids (millions) SSA Ethiopia et-2-2_0_1_0_0_0 49.4 SSA Nigeria ng-2-2_1_1_0_1_0 47.9 SSA Democratic Republic of Congo cd-2-2_1_1_0_1_1 42.9 SSA United Republic of Tanzania tz-2-2_1_1_0_0_0 20.7 SSA Sudan sd-2-2_1_1_0_1_1 16.6 SSA Madagascar mg-2-2_1_1_0_1_0 16.6 SSA Mozambique mz-2-2_0_1_0_0_0 15.8 SSA Nigeria ne-2-2_1_1_0_0_0 15.8 SSA Chad td-2-2_0_1_0_0_0 15.0 SSA Uganda ug-2-2_1_1_0_1_1 14.3 SSA Somalia so-2-2_0_0 14.0 SSA South Sudan ss-2-2_0_1_0_0_0 12.7 SSA Burkina Faso bf-2-2_1_1_0_1_0 10.4 SSA Mali ml-2-2_1_1_0_1_0 8.2 SSA Angola ao-2-2_1_1_0_1_0 7.8 SSA Kenya ke-2-2_1_1_0_1_1 7.4 SSA Guinea gn-2-2_1_1_0_1_1 6.7 SSA Malawi mw-2-2_1_1_0_0_0 5.8 continued MINI GRIDS FOR HALF A BILLION PEOPLE    83 TABLE 2.4, continued People newly connected to Region Country Name GEP Scenario mini grids (millions) SSA Côte d’Ivoire ci-2-2_1_1_0_1_0 5.6 SSA Zimbabwe zw-2-2_1_1_0_1_1 4.7 SSA Senegal sn-2-2_1_1_0_1_0 4.6 SSA Burundi bi-2-2_1_1_0_0_0 4.6 SSA Zambia zm-2-2_1_1_0_1_1 4.2 SSA Cameroon cm-2-2_1_1_0_1_0 3.7 SSA Central African Republic cf-2-2_0_1_0_0_0 3.3 SSA Rwanda rw-2-2_0_1_0_1_0 2.9 SSA Sierra Leone sl-2-2_1_1_0_1_0 2.7 SSA Ghana gh-2-2_1_1_0_1_0 2.6 SSA Eritrea er-2-0_0_1_0_0_0 2.3 SSA Benin bj-2-2_1_1_0_1_0 2.2 SSA Liberia lr-2-2_1_1_0_0_0 1.9 SSA Congo cg-2-2_1_1_0_1_0 1.5 SSA Mauritania mr-2-0_1_1_0_1_0 1.4 SSA Togo tg-2-2_1_1_0_0_0 1.1 SSA Guinea Bissau gw-2-2_1_1_0_1_0 1.0 SSA The Gambia gm-2-2_1_1_0_1_1 0.9 SSA Lesotho ls-2-2_1_1_0_1_0 0.5 SSA Equatorial Guinea gq-2-2_1_1_0_1_0 0.3 SSA Comoros km-2-2_0_0_0_0_0 0.2 SSA South Africa za-2-2_0_0_0_1_1 0.2 SSA Eswatini sz-2-2_0_1_0_1_0 0.2 SSA Namibia na-2-0_1_1_0_1_0 0.1 SSA Djibouti dj-2-2_0_1_0_0_0 0.1 SSA Gabon ga-2-0_1_1_0_1_1 0.0 SSA Botswana bw-2-2_1_1_0_0_0 0.0 SSA Sao Tome and Principe st-2-0_0_0_0_0_0 0.0 SAR Pakistan pk-2-2_1_1_0_1_0 20.3 SAR Bangladesh bd-2-0_0_1_0_1_0 3.6 EAP Myanmar mm-2-2_1_1_0_1_1 15.0 EAP Papua New Guinea pg-2-0_1_1_0_1_1 3.3 EAP Cambodia kh-2-2_1_1_0_1_1 0.6 EAP Solomon Islands sb-2-0_0_0 0.4 EAP Vanuatu vu-2-0_0_0 0.1 EAP Timor-Leste tl-2-2_0_0_0_0_0 0.0 EAP Federated States of Micronesia fm-2-0_0_0 0.0 LAC Haiti ht-2-2_0_1_0_0_0 4.9 LAC Nicaragua ni-2-2_1_1_0_0_0 0.4 LAC Honduras hn-2-0_0_1_0_0_0 0.3 Source: ESMAP analysis of Global Electrification Platform results. GEP Scenario code definition: cc-: two letter country code 2-: default value indicating GEP V.2.0 1st value: [0: “Bottom up” , 1: “Top-down low”, 2: “Top-down high”] for “Electricity demand target” 2nd value: [0: “Social and productive uses demand included” , 1: “Residential demand only”] for “Productive uses inclusion” 3rd value: [0: “Estimated” , 1: “High”] for “Grid generation cost” 4th value: [0: “Estimated” , 1: “High”, 2: “Low”] for “PV cost” 5th value: [0: “No connections cap” , 1: “Capped connections in 2025”] for “Intermediate investment & Grid connection Cap” 6th value: [0: “Least-cost nationwide” , 1: “Only grid within 2 km”] for “Rollout Plan” 84   MINI GRIDS FOR HALF A BILLION PEOPLE What size mini grids could deliver electricity to these 430 potential for 8.6 percent and 11.7 percent of new mini grid million people? Figure 2.5 displays the population distri- connections to be served by systems in the 500 kW to 1 bution by applicable mini grid size. For example, systems MW and 1 MW+ range, respectively, which corresponds to of less than 20 kW are expected to serve 10.5 percent of more than 87 million people and aligns with discussions new mini grid connections. Mini grids roughly correspond- about so-called metro grids in some markets. ing to 20 kW, 80 kW, and 200 kW systems can serve about The GEP also estimates the investment required at about 52 percent of all new mini grid connections, which gener- $100 billion, or 66 gigawatts (GW) of installed capacity, ally accords with most mini grids built to date. The GEP almost all of it solar hybrid, with over 90% of both this analysis also suggests there is a great deal of scope for installed capacity and investment needed in Africa. The mini grids in the 200 to 500 kW range, which could serve GEP, however, defines3 the installed capacity of solar 17.5 percent of all new mini grid connections. There is also hybrid systems as the PV capacity plus diesel generator capacity, which results in a higher measure of installed FIGURE 2.5 • Distribution by mini grid size of 429.5 capacity than if measuring firm power as defined in chap- million people served at least-cost by mini grids in ter 1. Table 2.5 presents the number of settlements, elec- 58 countries with severe access deficits tricity connections (or households), population, installed 11.7% capacity in MWs, and the investment requirement to real- 429.5 million people newly connected to mini grids ize the delivery of electricity to 430 million people by mini 8.6% grid system size. 17.5% NATIONAL ELECTRIFICATION PLANNING 22.9% Geospatial plans represent a data-driven approach to planning for the efficient and effective deployment of lim- ited resources, particularly aimed at supporting countries 28.8% with low rates of electrification. Spatial modeling delivers a least-cost plan that identifies the optimal grid or off- grid technology tailored to local circumstances (including local cost parameters) and appropriate in its technical feasibility and economic viability. It also integrates social and economic planning objectives, like equity, which may 10.5% target universal service delivery or priority access for schools and clinics. The (local) costing associated with the deployment of different technology solutions (for ≤20 kW (20, 80] (80, 200] (200, 500] (500, 1000] >1000 kW example, grid, mini grid, or SHS) is triangulated and com- kW kW kW kW pared across various dimensions. The most important of Source: ESMAP analysis of Global Electrification Platform results. these are population (or institutional) density, distance, Note: Under the scenario most favorable for mini grids. Data from the and isolation from the main grid, in addition to current and Global Electrification Platform. ] means up to and including; ( means greater than but not equal to. forecasted demand. TABLE 2.5 • Distribution by mini grid system size of 429.5 million people served at least-cost by mini grids in 58 countries with access deficits <20 kW 20–80 kW 80–200 kW 200–500 kW 500–1,000 kW >1,000 kW Total Settlements 622,061 421,358 104,279 33,202 7,322 3,634 1,191,856 Connections 8,986,188 24,767,050 19,694,967 15,035,819 7,384,990 10,024,393 85,893,408 Population 44,930,942 123,835,249 98,474,836 75,179,094 36,924,951 50,121,965 429,467,039 Capacity (MW) 5,617.70 16,817.20 12,646.50 9,878.00 4,972.60 16,063.10 65,995 Investment (US$, millions) 14,513.80 28,311.50 20,408.40 15,261.70 7,296.80 14,412.80 100,205 Source: ESMAP analysis of Global Electrification Platform results. Note: Under the scenario most favorable for mini grids. Data from the Global Electrification Platform. MINI GRIDS FOR HALF A BILLION PEOPLE    85 FIGURE 2.6 • Geospatial least-cost rollout plans in Kenya and Rwanda a. Kenya b. Rwanda Sources: Kenya: World Bank 2007; Rwanda: World Bank 2009. GIS = geographic information system; km = kilometers; kW = kilowatt; mi = miles. Geospatial analysis surfaces the most efficient technol- therefore help rural electrification agencies and mini grid ogy solution by using not only location (where do pro- developers define the addressable market for mini grids spective beneficiaries reside?) but also over time. Hence both as interim and permanent electrification solutions in the focus on off- and mini grid programs. Even if the grid the country. may, in some cases, be the least-cost solution, decentral- Examples of geospatial planning exercises for least-cost ized solutions have a role in providing nationwide access electrification can be drawn from Kenya and Rwanda (fig- in the short term, making up for what could be halting ure 2.6) and Myanmar and Nigeria (figure 2.7). Their expe- progress in network extension, and providing backup riences led to the development of the GEP, also presented solutions. National least-cost electrification planning can below in more detail. 86   MINI GRIDS FOR HALF A BILLION PEOPLE FIGURE 2.7 • Geospatial least-cost electrification plans for Myanmar and Nigeria by 2030, by technology component Myanmar Nigeria (with a focus on 4 states) Source: Castalia 2014; World Bank 2015. HV = high voltage; MV = medium voltage. OPERATIONAL EXPERIENCE AND THE GLOBAL on-grid/off-grid distinction and indicating least-cost solu- ELECTRIFICATION PLATFORM tions between mini grids and SHSs. Figure 2.7 shows the least-cost access solutions for Myanmar and four states in Kenya and Rwanda were early adopters in the use of GIS Nigeria by 2030, broken down by technology component. tools for electrification planning. Electrification programs in Such analyses provide first-order estimates of potential both countries were informed in 2009 by investment pro- sites for mini grid projects or SHS programs. spectuses relying on the results of geospatial analysis. These early experiences with least-cost electrification planning Building on this momentum, the World Bank has engaged focused on the least-cost rollout for grid extension without in national least-cost electrification planning with its giving explicit insights about the size and space for off- Global Electrification Platform (GEP), a multiphase proj- grid solutions. In particular, transitional off-grid solutions ect.4 It will improve, standardize, and simplify the use of (whereby no distinction was made between mini grids geospatial tools in least-cost electrification planning. To and stand-alone solar) were assumed to be inversely pro- achieve this, it is designed and developed at two levels, portional in terms of required space and time to progress briefly described below. in grid expansion, while the long-term targets for off-grid electrification were presumed to lie in areas not expected The GEP Explorer to be connected to the grid even in the long term. The GEP Explorer (figure 2.8) is an open access, interac- tive, online platform that provides overviews of electrifica- Since then, several countries have undertaken geospatial tion investment scenarios for all countries with less than least-cost planning, with accurately sized components of 90 percent electrification, which now includes 58 countries electrification programs, which helps countries update worldwide. The GEP Explorer allows the user, in two steps, their existing plans or develop new ones. Initially, the loca- to navigate nearly 100 electrification scenarios to meet tion and sizing of decentralized electrification were based access goals in those countries. The first step is an outlook on short-term grid extension. For example, a five-year roll- for an intermediate investment strategy (up to 2025). The out plan for grid densification and extension (prospectuses second explores full electrification by 2030. The number, typically have a five-year overview) indicates the space for type, and parameters of investment scenarios, along with transitional off-grid solutions, whereas a long-term plan for their inherent assumptions, are presented in the form of six the rollout of connections indicates the space for long-term levers designed to reflect different socio-techno-economic off-grid solutions. assumptions about the country context. Gradually, least-cost geospatial plans have achieved further All scenarios indicate the least-cost option, investment, sophistication in geospatial planning by going beyond an and capacity required to achieve full electrification at MINI GRIDS FOR HALF A BILLION PEOPLE    87 FIGURE 2.8 • The GEP Explorer Source: https://electrifynow.energydata.info/. both settlement and national levels. Results include three users to discover the values for the levers outside the pre- types of technologies, namely grid extension, mini grids, scribed values in the GEP Explorer and to investigate the and stand-alone systems. The user can also apply filters many variables not exposed as levers. to narrow results as well as toggle on different base lay- Other code modules. A number of modules can be used ers (for example, distribution network, location of health to further customize GEP elements. Examples include facilities, MapBox satellite) that can help better assess the the backend code here, the code for estimating custom modeling results. demand for settlements here, the code for population The GEP Explorer targets high-level decision makers in cluster generation here, the code for GIS data extraction addition to policy and investment analysts that can use to those clusters here, and the code for a high-level result its output to assess geo-infographic electrification invest- analysis here. The list is expected to expand as the project ment for an area of interest. It does provide some flexibility evolves. through scenario selection, but all scenarios are pre-run Training/teaching hub (access here). Online videos, with no option to customize on the fly. presentations, short lectures, and training material (for example, exercises) support capacity-building activities The GEP Toolbox around the GEP. The GEP team has run four capacity- The backbone of the GEP initiative, the GEP Toolbox, offers building events and has established an annual training in a range of tools and material that support reproducibility, Trieste, Italy, every June. Recently, the training material replicability of the GEP Explorer, as well as capacity build- has been bundled into a self-paced, online, open course ing, dissemination, and inter-organizational collaboration.5 offered by Open University6 (access here). The course A few components are described below. seeks to introduce trainees to geospatial electrification The GEP-OnSSET code. The GEP Explorer displays results modeling and planning by providing lectures on theoret- developed in conjunction with the Royal Institute of Tech- ical concepts and practical exposure through hands-on nology (KTH), building on a special version of the Open exercises. Source Spatial Electrification Tool (OnSSET). This is called Finally, it is worth mentioning that the GEP is part of a GEP-OnSSET and is available on GitHub along with online continuous data and model discovery process. To ensure documentation that supports its installation, setup, and that new data and models can easily be integrated into the use. GEP ecosystem, guidelines for its form and description, The GEP Generator. This open access, user-friendly Jupy- as well as handling protocols, have been developed more ter notebook allows a user to reproduce and customize the here. Based on these, expect annual updates of the GEP electrification models behind GEP Explorer. The notebook to reflect advances in algorithms and models, better data requires little to no programming experience to operate; input, and more scenarios defined by increasingly relevant it hides coding complexities and presents only key input and available levers. decision parameters to the user. The GEP Generator allows 88   MINI GRIDS FOR HALF A BILLION PEOPLE INDICATIVE WORKFLOW FOR THE DEVELOPMENT geospatial analysis. These data may come from national OF A GIS-BASED NATIONAL ELECTRIFICATION agencies, such as the Census Bureau, public statistics, PLAN the survey department, and other departments/minis- tries; international agencies such as the World Bank, IEA, As described above, a spatial least-cost electrification UN, FAO, IRENA, EU JRC, WRI, and so forth; open access scheme could support planning undertaken by various databases such as ENERGYDATA.info, OpenStreetMap, stakeholders; it could help form policy and design around HOTOSM and so forth; and in some cases proprietary nationwide pathways for electrification. Such modeling sources (for example, satellite imagery, Maxar’s building activities—and the plans they might inform—should be footprint). based on rigorous models and analytics as well as good governance principles. The literature indicates some over- The type of data required depends on the modeling frame- arching principles to guide such initiatives—for example, work, but usually it covers infrastructure (for example, U4RIA (DeCarolis and others 2017; Howells and others the power network, roads, settlements, public facilities), 2021). natural resources (solar irradiation, wind speed, hydro resources, land cover, protected land) and socioeconomic ESMAP has adopted those principles and, based on its activity (night lights, population, travel time, gross domes- operational experience, converted them into a more practi- tic product, electricity demand, affordability). Note that cal, five-step workflow presented in figure 2.9. The activities non-GIS data are also collected at this stage in order to are often linear; however, in reality, the process depends on support model calibration. These may include population country-specific conditions, including feedback loops and growth, urbanization, and electrification rates; household reiterations. For example, in some cases, capacity building size; electrification targets; and so on. Finally, planners may take priority over analytical work. should collect the technical and costing parameters for the technologies used—namely, medium-voltage (MV) lines, Diagnostic and preliminary analysis mini grids, and SHSs—as well as cost curves/projections, The first step in the workflow includes a thorough investiga- and discount rates. tion of data availability and know-how over the area of inter- est. Any existing and/or past applications of electrification Once collected, cleared, and compiled, all data and infor- planning techniques should be reviewed. The public or pri- mation should be reported to and shared with stakehold- vate stakeholders involved in the project should be listed; ers. Wherever technically feasible with regard to security their capacity in the use of GIS-based analytics to support and privacy, data should be shared on an open-source data electrification planning should likewise be assessed. The repository such as ENERGYDATA.INFO. Any missing data, status of the assessment should be documented in great- like the locations of productive activities (potential and est possible detail (for example, data types, quality, meta- existing), energy expenditure, and ability to pay may be col- data, level of knowledge, etc.) as this will determine the lected by site visits, geolocated surveys, or top-down sec- level of effort required in the following steps. Therefore, the tor-based analysis working through regional government diagnostic and preliminary analysis should delineate any and private organizations and commercial associations. analytical gaps and guide the project structure. This is usu- ally presented in a short yet concise inception report that Development of least-cost electrification plan guides the project thereafter. Here, the information collected in the previous step(s) informs the analytical work. To bound the least-cost plan- Data collection, mapping, and database preparation ning exercise, planners must identify the constraining The next step involves the collection, review, and com- parameters. Then they need to explain how each parame- pilation of the best readily available data required for the ter is defined or measured. Such parameters may include FIGURE 2.9 • Typical least-cost electrification planning sequence (best practice) Diagnostic Data collection, Development Reporting, Capacity and mapping, and of least-cost data transfer, building and preliminary database electri cation and knowledge analysis preparation plan dissemination transfer Source: ESMAP analysis. MINI GRIDS FOR HALF A BILLION PEOPLE    89 definitions for (1) starts, plateaus, and endpoints; (2) elec- renewable energy. The output is a prospectus with details tricity demand targets and projections; (3) costs and ser- on the upgrades necessary to achieve on-grid targets and vice standards for networks and individual systems; (4) the associated financing requirements. availability of renewable energy resources; (5) ability of Analysis of mini grid and off-grid solutions. Alongside consumers to afford upfront investments (such as connec- this probe of on-grid solutions, an economic analysis of the tion charges) and recurring expenditures (such as monthly potential for mini grid and stand-alone systems (namely, tariffs); and (6) criteria for temporal and spatial prioritiza- SHS, diesel gensets, and so forth) is recommended as part tion. These parameters are just indicative and depend on of least-cost electrification planning. Aimed at securing the scope of the analysis. The combination of those param- sector-wide support, the study might look at representa- eters creates scenarios that can be used to assess the sen- tive samples of high-potential off-grid sites, using data and sitivity of results to different input values. Typical sensitivity comparisons from existing sites. This analysis can help analyses examine the impact of various electrification tar- articulate the most important considerations for both pri- gets (following the Multi-Tier Framework) or demand lev- vate and government stakeholders in pursuing the off-grid els (based on demand sensitivity); different commodity sector, such as the potential profitability of different busi- prices; economic forecasts; and other variables (such as ness models and technologies, the tariffs and subsidies grid supply cost, technology costs, and service standards). required to achieve profitability, and promising sites for The analytical work provides a basis for the systematic roll- public-private partnerships. Prospectuses could be devel- out of a least-cost national electrification program for both oped using the results of this analysis. urban and rural areas and aims to either maximize cover- age for a given investment level or minimize investment for Reporting results, including technical model, data the targeted coverage. The objective function depends on transfer, and dissemination the model used or the scope of the analysis. Key outputs The national least-cost electrification planning exercise include the following: usually produces the following output: • A technology mix (grid connections, mini grids, and • A well-structured and -informed database that includes stand-alone systems) that fulfils the objective function input and output data (both GIS and non-GIS) and is subject to parameters and/or constraints. • An electrification model built and customized for the • System components’ characteristics (for example, size, country (or area of interest) capacity, investment, service quality, and other opera- • Documentation related to the project; this might tional features) required to implement the least-cost include an inception report, an intermediate report or technology mix. a final report that describes the methodological frame- The results of the least-cost model can be overlayed with work, key assumptions, results of the analysis, lessons some of the input data (or other information) and provide learned, and recommendations. It may also include any a greater level of analysis as per need. The following para- user guide for data processing or model running. graphs show how this can further support on- and/or off- The output must comply with global best practices and grid rollout plans in particular. ensure the project’s long-term sustainability. The U4RIA Detailed analysis of on-grid solutions. Building on the framework is highly recommended. Its output and pro- least-cost electrification, planners could articulate the need cesses can be retrieved, repeated, and rebuilt. They can to expand generation capacity (or electricity trading with be audited and are interoperable. All stakeholders should neighboring countries) and upgrades to grid infrastructure be consulted before recommending any institutional and needed to support the stated targets. Doing so might require organizational arrangements. This will help to ensure that access to data, such as installed generation capacity, exist- the GIS database is maintained and regularly updated ing transmission network, geotagged on-grid demand and and that the GIS electrification planning exercise can be demand projection, potential generation capacity, reserve replicated. Determining the appropriate institutional and constraints and operational constraints, and interconnec- organizational arrangements involves identifying the orga- tions with neighboring countries. The sequencing of new nization responsible for hosting the national power sector connections (and related costing) and extensions, along GIS database and the arrangements by which stakehold- with other needed changes to the supply system (namely, ers will update their database. Furthermore, which organi- network reinforcement, increased generation, and trans- zation will house the electrification planning models? Who mission), can be elaborated at this stage. A power-flow anal- will be responsible for replicating the geospatial electrifica- ysis might also reveal needed grid infrastructure upgrades tion planning exercise in the future? These decisions will to support targets and potential integration of variable need to be made. 90   MINI GRIDS FOR HALF A BILLION PEOPLE Capacity building and knowledge transfer do not require distribution networks. But its recurring costs Finally, all output, as described above, should be trans- are relatively steep because of battery storage needs over ferred both to the government (or its designated counter- the long term. Mini grids typically offer an intermediate parts) and to the institution funding the geospatial work. option to serve demand levels that are too high for SHS but Those who are analyzing the least-cost electrification plan- not great enough (or too remote) to justify connection to ning should be asked to: the main grid. (See chapter 1 for more on mini grid costs.) • Train professional staff throughout the assignment; Unlike stand-alone solar systems, mini grids and grid extension both require the installation of an electrical dis- • Familiarize them with the capabilities of the models; tribution system throughout the village in addition to a • Teach them about the methodology and analysis frame- minimum density of customers to justify this installation. work for updating the geospatial high-level analysis in Table 2.6 shows the maximum distance justifying the cus- the future; and tomer-connection cost as a function of the level of ser- • Explain the key variables, such as technology costs, for vice that the customer requires. For example, for a mini future sensitivity analysis. grid or main grid distribution system to be cost-effective, a group of customers requiring Tier 1 service would have The consultant should list any licenses needed to ensure to be densely co-located (within approximately 3.3 meters the functionality of the GIS planning platform and provide of one another). By way of contrast, a group of customers estimated costs for acquiring them and also instructional requiring Tier 5 service can be about 1.7 kilometers (km) materials for ongoing capacity building and knowledge distant from the other group members for a distribution transfer efforts. system to make economic sense. ANALYTICAL INSIGHTS AND GENERIC In practice, communities that require only Tier 1 service can OBSERVATIONS almost never justify a distribution system, and communi- Although geospatial electrification plans are country and ties requiring only Tier 2 service will rarely justify a distribu- context specific, some insights with general application tion system unless one or more customers require Tier 4 or can be gleaned from experience. They are presented below. 5 service. Distribution systems, whether powered by mini grids or the main grid, are generally justified for areas that Estimated (or targeted) electricity demand of beneficia- require Tier 3 and higher levels of service. ries shapes the cost-effectiveness of various technologies. Varying demand also affects the type of system recom- Figure 2.10 presents the indicative results from a simula- mended by electrification modeling tools: household/cus- tion run using the Hybrid Optimization Model for Multiple tomers with strong demand typically favor grid extension Energy Resources (HOMER) planning tool. It indicates that if the load centers are close to the grid, and mini grids if large loads close to an existing grid are more cost-effec- they are farther away, whereas low demand favors off-grid/ tively served by a grid extension. Small loads far from an SHSs. existing grid are more cost-effectively served by a mini grid. For this exercise, the same level of service was assumed The different balances of initial and recurring technology from both approaches, and the same cost for the distribu- costs affect how economies of scale are leveraged. To illus- tion system and for operation and maintenance. trate: grid electrification has relatively high initial costs but lower recurring costs. By way of contrast, SHS has lower Both electricity demand and customer density thresholds initial costs, at least for small, remote communities, as they presented above refer to residential loads. The addition of Least-cost electrification planning at the Mini grids are rarely justified from an eco- national level using geospatial analysis tools nomic standpoint in areas with demand typically follows a five-step process: (1) diagnostic for electricity that correspond to Tiers 1 and 2. In and preliminary analysis; (2) data collection, map- contrast, distribution systems, whether powered ping, and preparation of the database; (3) develop- by mini grids or grid extensions, generally make ment of least-cost electrification planning, including sense from an economic standpoint for Tier 3 and detailed analyses of main grid, mini grid, and off- higher levels of electricity demand, all other things grid solutions; (4) reporting results; and (5) capac- being equal. ity building and knowledge transfer. MINI GRIDS FOR HALF A BILLION PEOPLE    91 TABLE 2.6 • Maximum cost-justified distance for connecting a customer as a function of the required level of service Service tier 1 2 3 4 5 PV size (Wp) 10 100 1,000 3,000 10,000 Energy requirement (kWh/month) 1 10 100 300 1,000 Generation and storage technology Solar Solar home PV, battery, PV, battery, PV, battery, inverter, Lantern system inverter inverter, backup backup generator generator Capital cost $50 $300 $3,000 $9,000 $25,000 Maximum distance between customers to 3.3 20 200 600 1,667 cost-justify a distribution system (meters) Source: HOMER Energy. kWh = kilowatt-hour; PV = photovoltaic; Wp = watts peak. FIGURE 2.10 • Distance as a function of load size: therefore crucial in efforts to boost the rate of load growth Break-even grid extension (IEG 2015). Chapter 3 discusses this topic in further detail. 250 The quality of services that beneficiaries receive is another Breakeven grid extension distance (km) useful parameter when comparing the technologies in the Mini grid is cheaper least-cost plans. 200 Stand-alone solar systems tend to have smaller capacity 150 (10–200 watts [W]) and provide on average 1–20 kilo- watt-hours (kWh) per month. Within the Multi-Tier Frame- work for measuring household electricity access, this is 100 Tier 1 or 2 access. Stand-alone systems generally don’t use an inverter or a backup generator. This means that they 50 Grid extension is cheaper provide direct current (DC) power whose availability on any given day may be determined by the weather. The supply of 0 electricity may be sufficient for households that need only 0 1,000 2,000 3,000 4,000 5,000 light-emitting diode (LED) lighting and cell phone charging Load size (kWh per day) or possibly some other small DC appliance, such as a radio. Source: HOMER Energy. By way of contrast, the national grid extension can supply km = kilometer; kWh = kilowatt-hour. 24-hour power. In practice, however, many of these grids, particularly those in low-income countries, cannot meet productive activities could change those dynamics. That is, this level of reliability, and their customers suffer frequent communities where households have low levels of power outages and load shedding. Third-generation mini grids, on demand but are close to productive loads might also be the other hand, can provide high-quality electricity service. good candidates for mini grids or grid extension. Therefore, Members of the Africa Minigrid Developers Association electrification technologies should be compared not only (AMDA) report an average of 97 percent system uptime. according to the number of households they serve but also Both the main grid and mini grids can also supply suffi- according to the productive uses and community services cient alternating current (AC) power for productive use (for they enable. Productive uses have only recently been incor- example, grain milling, water pumping, sewing, woodwork- porated into geospatial electrification modeling efforts; ing), businesses (for example, telecommunications towers; thus, their impact on national least-cost electrification local, small and mid-size enterprises), and public services plans is not yet directly quantifiable. like schools and hospitals. It should also be noted that estimating the required level Note that the cost of electrification can vary widely depend- of service of unelectrified beneficiaries can be difficult, as ing on local subsidies, but the true unsubsidized cost of demand for energy tends to grow once energy becomes power is the appropriate metric for comparing options. As available. But the rate of growth varies and depends on mentioned before, the reliability of grids in many low-in- whether and how productive uses are promoted. Outreach come countries can vary significantly. In most places, out- efforts that demonstrate productive appliances and com- ages are a common occurrence, and mini grids are often mercial opportunities enabled by reliable electricity are deployed in areas already connected to an unreliable main 92   MINI GRIDS FOR HALF A BILLION PEOPLE Because of the vast differences in the qual- Early experiences with least-cost electrifica- ity and reliability of the energy service pro- tion planning have demonstrated the impor- vided by mini grids, solar home systems, and the tance of engaging with the private sector during the main grid, it is not appropriate to compare them diagnostic and preliminary analysis phase of a proj- according to their respective costs for the provision ect. This ensures that realistic assumptions about of electricity to the same groups of customers. costs and demand growth over time, among other assumptions, are built into the least-cost model’s calculations. grid to ensure reliable electricity service. The mini grids operated by OMC Power in Uttar Pradesh offer one such by technology, location, and sizing of the different compo- example. They serve villages where a government-owned nents of electrification programs. In addition, a plethora distribution utility is already present, but with low service of spatial data is increasingly available and continues to reliability, particularly during peak evening hours. improve in quality, coverage, and availability. Amid wild variations in the quality and reliability of the Nevertheless, there is always room for improvement. For energy service provided by mini grids, SHSs, and the example, better MV line mapping and improved demand main grid, comparisons based solely on their respective estimation are essential in order to improve the sophistica- costs for energy provision are inappropriate. Any com- tion of planning and tailor services to beneficiaries’ needs. parison should—to the extent possible—internalize costs associated with the reliability of supply (for example, From a planning perspective, knowledge of existing elec- value of lost load). tricity infrastructure is fundamental to ensuring that the results of geospatial modeling tools reflect conditions on LESSONS LEARNED AND CHALLENGES AHEAD the ground. Knowing the reach of electricity infrastructure Drawing on new developments in geospatial analytics, is critical if developers are to, first, identify who already has many countries are updating their geospatial least-cost a connection and, second, to cost the investment neces- plans, taking stock of the results achieved so far with grid sary for access provision. This knowledge is based on the extension, and analyzing the off-grid space to inform their location of the beneficiaries and their distance from exist- electrification programs. These updated programs can ing infrastructure. In the absence of this information, plan- provide more guidance on the design of implementation ning tools may overestimate the number of beneficiaries. frameworks and modalities for scaling up off-grid solutions, Mapping of MV lines is not yet common in most developing as they will be more specific about location and sizing of countries. Analysts may infer the extent of the MV network off-grid and mini grid potential in the country; the location from other parameters, but this approach would result in and sizing of long-term off-grid beneficiaries; and the pre- more errors in determination of electrification status than liminary location and sizing of mini grid solutions (to be if reliable maps of MV networks are available. followed by feasibility studies on the ground), based on Demand forecasting is perhaps the single most critical population density and loads (including their forecasting), modeling parameter for electrification planning, from geo- and local renewable generation resources. spatial least-cost plans to power sector planning, although Previous national least-cost geospatial planning exercises the willingness and ability of customers to pay for electricity have taught us the need to engage the private sector during are also critical from the developer/investor point of view. the diagnostic and preliminary analysis phase of a project. Improved demand estimates are also crucial to support As early least-cost planning activities have demonstrated, existing economic centers (and maximize the economic projections of the costs of mini grid electricity, main grid returns of electricity access) through adequate access to expansion plans, and demand growth in areas not expected electricity services and to forecast locations for productive to be connected to the main grid in the near to medium uses (and future economic growth potential) that may be terms have underestimated—by far—the actual potential prioritized by electrification programs. for mini grids. Engaging with the private sector early in a Finally, the models themselves need to improve constantly national least-cost electrification plan can enable the inte- in order stay current with new data and policy/planning gration of more realistic assumptions about mini grids, needs. Better methods could help internalize the costs of SHSs, and the main grid. reliability (for example, the value of lost load) and other Most modeling frameworks available at the moment have policy mandates (like energy access equity or equality tar- evolved to provide an explicit analysis of electricity access iff and subsidy schemes). They could also better accom- MINI GRIDS FOR HALF A BILLION PEOPLE    93 Looking ahead, the next critical advances Geospatial analysis provides a broader pic- in geospatial planning are improvements to ture of communities’ locations and char- network mapping and demand estimation, which acteristics a portfolio can consider, a picture that will further increase the accuracy of national least- enables mini grid developers to exploit economies cost electrification plans. of scale and prepare quicker, more cost-effective rollout plans and plans for service and maintenance. modate new configurations like hybrid or biomass-based and reduces the time and resources spent on prospect- systems and climate aspects (for example, resilience to ing for such communities, and it can mitigate the risk of climate change or disasters). demand uncertainty by incorporating a larger number of customers in a single investment (the portfolio) as com- pared with a single mini grid. MINI GRID PORTFOLIO PLANNING Without geospatial tools, developers must often rely on anecdotal suggestions from local governments to identify OVERVIEW promising communities to visit and investigate further for Geospatial analysis can also be used as part of a portfolio consideration. While such human intelligence is still useful planning approach for mini grid development, to comple- and can complement or be used to validate the recom- ment a comprehensive national least-cost electrification mendations from geospatial portfolio planning, a broader planning framework and, in the absence of such a frame- picture of the locations and characteristics of communities work, where grid extension is expected to be limited or that can be considered at a portfolio level will enable mini unlikely because of political considerations, insolvency of grid developers to exploit economies of scale and prepare the distribution companies, and so forth. If national least- quicker, more cost-effective rollout plans and plans for ser- cost electrification planning exercises have carved out vice and maintenance. At a more micro level, geospatial areas that mini grids can serve as the least-cost solution, tools can be used for mini grid generation sizing and distri- mini grid developers and electrification agencies may bution network planning. wish to focus their time and resources on investigating the Technological advances and cost reductions in satellite potential for developing mini grids to serve communities in imagery and in machine learning, increased sophistication these areas. of algorithms and analytical approaches, and the prolifer- Developers would do well to remember that political and ation of web-based technologies have made available a other considerations may affect the likelihood of grid host of new digital tools to improve the efficiency of mini extension regardless of the underlying economics. The grid development. This section examines how some of grid may be extended to areas where it might make more these tools can expedite the process of identifying poten- sense to pursue decentralized solutions and, inversely, dis- tial sites for mini grids; collecting, estimating, and analyz- tribution companies may not be in a position to extend the ing customer data; optimizing mini grid system designs; grid to areas even when it may be the least-cost solution. and finding and selecting developers and investors, using Nevertheless, national least-cost electrification plans can innovations from the frontier as examples. These exam- serve as a guide and a starting point when prospecting for ples from the frontier include private developers, who are suitable sites for mini grid deployment. using geospatial and other digital technologies to improve preparation of portfolios of mini grid projects, as well as Geospatial portfolio planning, which is already being public-sector programs that are taking advantage of such used by a number of established mini grid companies in disruptive technologies to facilitate implementation of Sub-Saharan Africa, greatly reduces the pre-investment mini grid projects. cost associated with preparing sites for mini grid devel- opment compared with traditional approaches, which rely Figure 2.11 presents an indicating sequence of activities heavily on the deployment of full multidisciplinary teams (workflow) involved in geospatial portfolio planning for mini to villages to explore the scope for mini grid electrification. grids. The rest of this chapter describes each of these steps Geospatial portfolio planning does not eliminate the need and looks at how disruptive technologies are helping gov- to conduct feasibility studies or engage with beneficiary ernments and private developers prepare and implement communities, but it does provide guidance on where com- mini grid portfolios. The Access to Distributed Electricity munities suitable for mini grid electrification are located and Lighting in Ethiopia (ADELE) project and the Nigeria 94   MINI GRIDS FOR HALF A BILLION PEOPLE FIGURE 2.11 • Geospatial portfolio planning sequence Site identi cation, Analysis, sensitivity, Characterization System design Financial attribution, and visualization, and of load pro le and optimization modeling prioritization dissemination Source: ESMAP analysis. Electrification Project (NEP), a flagship initiative of the fed- The cluster data set improves when it is merged with OSM eral government of Nigeria, supported by the World Bank, land-use and mapped-buildings data. are two especially apt exemplars of geospatial portfolio This process identified nearly 200,000 clusters across planning. Lessons from studies of these two projects are the country—188,014 clusters, to be precise—occupying used interchangeably, hereafter, to illustrate the implemen- from 1.1 hectares to 48,500 hectares. Figure 2.13 visualizes tation of the suggested workflow in the real world. the distribution of these clusters across Nigeria, and rep- THE WORKFLOW PHASES FOR MINI GRID resents their distribution by size (in hectares). PORTFOLIO PLANNING: SPATIAL DATA AND More recently, alternative methods have been generating ANALYTICS population clusters with vector data (polygons or cen- troids) that show the distribution of buildings (or rooftops). PART 1. Site identification One example is the building data set developed by NRECA The initial phase of the workflow involves the collection and as part of the USAID Distribution Systems Strengthening processing of geospatial (and other) data and information Project. This visualizes digitized housing structures in Ethi- for the identification of sites with potential for mini grid opia based on satellite imagery (NRECA 2019). In addition, development. The process is split into three main activities: the Digitize Africa building footprint data set has been pro- namely, the generation of population clusters and the attri- duced for Sub-Saharan Africa by Ecopia.AI and Maxar with bution of those clusters and their prioritization based on a funding from the Bill and Melinda Gates Foundation (Eco- set of criteria. This activity produces a list of possible mini pia AI and Maxar Technologies 2021) (figure 2.14). The first grid sites. data set identified about 13.7 million buildings across Ethi- opia; the latter identified about 32.8 million rooftops in the CONVERTING RAW GIS DATA AND SATELLITE IMAGERY INTO country with a high estimated accuracy (>95 percent valid POPULATION CLUSTERS precision and recall). A key input for electrification planning and projects is to understand where people live. The location of settlements The new methods grouped rooftops into clusters using and their boundaries is therefore the baseline for any geo- density-based clustering (DBSCAN) and three parameters. spatial planning exercise. Rooftops in proximity are called “core,” whereas dispersed single rooftops are called “noise” and omitted from the In Nigeria, the administrative areas (from higher to lower cluster. This identifies more populous areas (clusters) with levels) are federal states, local governmental areas (LGAs), enough density to justify mini grid development. and wards. Population statistics and administrative bound- aries are well known at the LGA level, but not at the ward The DBSCAN algorithm is based on two parameters. The level. But having the exact population figure and boundar- first is the maximum distance between rooftops (or, eps), ies for wards is not sufficient for designating suitable sites and the second is either the density threshold (or, minPts) for mini grid electrification. That exercise would require or the minimum number of neighboring rooftops each having the exact locations of the buildings and settlements. building needs to have within the maximum distance in a So a cluster identification algorithm was developed for the potential cluster. The process is iterative; the clustering NEP to automatically identify the location and boundaries algorithm needs to be run for different parameter combi- of population settlements.7 nations in order to identify those yielding the most repre- sentative results. Based on the data sets described in box 2.1, it is possible to identify clusters with great accuracy following the pro- Note that there is a modified version of the DBSCAN cess illustrated in figure 2.12. First, the HRSL buildup raster algorithm (bounded DBSCAN suggested by Village Data is vectorized, buffered, and dissolved to define boundaries. Analytics, VIDA) that includes an optional parameter MINI GRIDS FOR HALF A BILLION PEOPLE    95 BOX 2.1 DATA SOURCES FOR CLUSTER DEFINITION IN THE NIGERIA ELECTRIFICATION PROJECT A population cluster is an area that could be supplied ments are located. Combining those estimates with by a single distribution network. All households that OSM building features presents a clearer picture of are “relatively” close to one another form one cluster. the settlement boundaries and building counts within A cluster, in principle, could be as small as a hamlet, or them. Note that some of these data sets are incom- as large as a city. Key data sets were used for cluster plete—for example, in Nigeria, Niger state is mapped identification: more precisely than its neighboring states. So addi- tional sources of data should complement the meth- • OpenStreetMap (OSM) data (contains vector layers odology, when and if they are available. with buildings, residential land use, roads, water- ways, and so forth) a For example, the WorldPop’s peanutButter web appli- cation, a more recent development, allows the cus- • A high-resolution settlement layer (HRSL) b tom generation of gridded population estimates at • Administrative boundaries c 100-meter spatial resolution. e The application builds A HRSL population raster estimates population distri- on the high-resolution building footprint data set and bution accurately. d Spatial processing of the HRSL can complements both population and building counts in thus help roughly delineate where population settle- the candidate sites. In the case of Nigeria, OSM data sets were retrieved from the open access Geofabric Server, available at https:/ a.  /download. geofabrik.de/africa/nigeria.html. Vector data list coordinates that define points, lines, or polygons. The HRSL layer was retrieved from the Facebook Connectivity Lab and Center for International Earth Science Information b.  Network (CIESIN), Columbia University, available at https://www.ciesin.columbia.edu/data/hrsl/. In the case of Nigeria, administrative boundaries have been retrieved from the Database of Global Administrative Areas c.  (GADM), available at https://gadm.org/download_country_v3.html. d. Raster data consist of pixels (or cells) where each pixel has an associated value. e. The WorldPop’s peanutButter web application is available as a beta version at https://apps.worldpop.org/peanutButter/ FIGURE 2.12 • Methodology for the generation of population clusters in Nigeria, using the HRSL and OSM data Input data Processing Result Built-up raster data Vectorizing, buffering, Settlement clusters Pixels with presence of and dissolving in order Outline of settlement structures built-up structures are to indicate precise result from merging the three input depicted in light colors settlement boundaries data types OSM land use Filtering, buffering and Land use types are colored dissolve individually (residential in Extracting residential land use red, industrial in blue) and reducing polygon count OSM buildings Clipping and clustering Open source mapped Finding additional buildings polygons . . . not covered in other data sets Source: Integration and Reiner Lemoine Institut 2016. Note: OSM = OpenStreetMap. 96   MINI GRIDS FOR HALF A BILLION PEOPLE FIGURE 2.13 • Nigeria’s population clusters: Spatial distribution (left) and size histogram, in hectares (right) Cluster size distribution 25 20 Cluster count (thousands) 15 10 5 0 Source: Integration and Reiner Lemoine Institut 2016. 1 10 100 1,000 10,000 ha = hectare. Cluster size [ha] FIGURE 2.14 • Sample outputs from the Digitize Africa building footprint data set  axar’s building footprint (blue polygons) and a. M  axar’s building centroids and HRSL’s population b. M NRECA’s building centroid (red dots) distribution (raster data set) Source: VIDA 2021. FIGURE 2.15 • Concept of the DBSCAN algorithm Border p and q are Noise p q density-reachable from o Core o Therefore p and q are density-connected Eps = 1 Eps = 1 minPts = 4 minPts = 4 Border Source: Hahsler, Piekenbrock, and Doran 2019. MINI GRIDS FOR HALF A BILLION PEOPLE    97 (minClusterSize) related to the minimum number of Once the set of parameters are selected and the rooftops grouped into clusters, then the boundaries are drawn to buildings a cluster requires in order to be considered. This parameter filters out settlements below a certain buildingidentify the exact contours of the potential village. This count and can be used in cases where this is needed (as is usually done by using an alpha shape generating algo- shown in figure 2.16). rithm, which estimates a village’s size, area, and density (by comparison, a simple convex hull-based village boundary always overestimates the village area and underestimates FIGURE 2.16 • Ethiopia’s rural population settlements and density). Figure 2.17 offers a comparison of convex hull and mini grid deployment: Bounded DBSCAN clustering alpha shapes. Number of o -grid settlements (with N>100) found across Ethiopia ATTRIBUTING POPULATION CLUSTERS 6,000 Once the clusters have been identified, attributes are Maximum distance between 50 730 393 192 72 23 8 added. These are used to rank the most suitable locations 5,000 for mini grid deployment. Common attributes in this rank- rooftops (meters) 2,231 1,496 1,114 883 713 578 ing exercise are shown below: 100 4,000 • Cluster name (if available, or its ID) 3,000 150 4,541 3,216 2,369 1,836 1,496 1,265 • Cluster size 2,000 • Administrative division(s) (municipality, district, region, country) 200 6,464 5,327 4,108 3222 2615 2,206 1,000 • Building count 5 10 15 20 25 30 • Population Minimum number of neighbors • Power situation and nighttime light intensity (electrified Source: VIDA 2021. or not, existing mini grids or SHS) Note: The y axis represents the eps parameter, maximum distance • Distance to infrastructure (grid network, roads, substa- between rooftops (in meters); the x axis, the minPts, or minimum number tion/transformer, etc.) of neighbors. The minimum minClusterSize is set to 100. Highlighted box indicates the reference pair of input parameters as derived from experience (eps: 150, minPts: 20, minClusterSize: 100). BOX 2.2 FINDING THE OPTIMAL INPUT PARAMETERS FOR DBSCAN The modeling exercise in Ethiopia has made it hard ematically good cluster will not necessarily capture a to identify a single set of input parameters that gen- physical feature such as a village. erate highly accurate results for the whole country. Yet another approach might validate results against What works well for one area might be suboptimal in real settlements via statistical validation. This would another owing to the differences in geography, topol- require a training data set and machine-learning- ogy, population densities, or even building structure based and -supervised validation methods. The latter and distribution. may contain boundaries and characteristics (such But a purely mathematical method might help identify as population and building counts) of actual villages. robust clusters. One can, for example, use a density The comparison could calibrate the input parameters based cluster validation (DBCV) approach (or similar, under a framework that reflects the situation on the for example, SC or AMI) and compare parameter val- ground more accurately. ues (Moulavi and others 2014). DBCV yields a valida- Finally, it should be noted that it is imperative to engage tion index ranging between -1 and +1, with higher values local counterparts in calibration and validation. indicating better clustering performance. But a math- The implementation code of the density-based cluster validation (DBCV) methodology is available at https:/ a.  /github.com/christopherjen- ness/DBCV; another implementation is available at HDBSCAN’s original GitHub repository at https://github.com/scikit-learn-contrib/hdb- scan/blob/master/hdbscan/validity.py. 98   MINI GRIDS FOR HALF A BILLION PEOPLE FIGURE 2.17 • Cluster contour delineation: Ministry of Power and the Nigerian Energy Support Pro- Convex hull (left) and alpha shapes (right) gramme have compiled and collected data from the field Convex hull Alpha shape on the extent of the medium voltage grid as well as other electricity infrastructure and made this information avail- able on a web-based mapping platform. Finally, some attributes will need to be calculated based on available data. For example, the area of each cluster can be estimated simply using spatial analysis. Or, as another example, determining the reach and extent of the main grid can help calculate its distance from a cluster. This kind of information is usually available in developed economies, Source: VIDA 2021. but in developing countries may be hard to find; one may need to estimate or infer it from other data, like nighttime • Number of public institutions (for example, health facil- lighting satellite imagery. See Figure 2.18 for an example. ities, education institutions, other services, and admin- Where possible, knowledge of the grid distribution and istrative offices) grid-connection status of population clusters can be • Commercial buildings, stores, and other anchor loads improved by better collaboration with distribution com- (mobile tower) panies or by site visits. In case data on the existing grid • Agriculture (crops, harvested area, production, yield, network is not available for the country or area of interest, and so forth) it can be simulated using other available GIS data sets as proxy (for example, night lights, road networks). In fact, • Post-harvest activity (milling, drying, cooling, storage, gridfinder.org12 has developed a methodology that predicts and so forth) the routing of the transmission network using the above • Resource availability (solar, wind, hydro, biomass etc.) data and a minimum spanning tree approach. Results are • Other socioeconomic characteristics (poverty rate, available globally and the model is open source (Arderne income level, household profile, and assets) and others 2020). A growing pool of databases has recently been providing VIDA has further refined the code in the Gridfinder repos- access to this information. Some examples worth men- itory. As a result, the predictions have improved, making tioning: the code deployable across the globe for any given time • The open data platform Energydata.info8 is a good place period. A sample VIDA GridLight is shown in figure 2.20. to start finding geospatial data for electricity and energy The GridLight algorithm has been extensively tested and access planning. utilized in several countries. In Nigeria the algorithm was tested against 300 on-ground survey data points with an • OpenStreetMap (OSM) and Humanitarian OpenStreet- accuracy level of 82 percent. In Ethiopia, we identified an Map (HOTOSM)9 are open-source projects with millions overlap of 80 percent with the latest utility data. of geospatial features pertinent to population distribu- tion, commercial and public buildings, infrastructure, Apart from a proxy for the existing grid network, nighttime and resources. lighting13 can be used, provisionally, to classify electrified population clusters in areas with high night light emis- • FAO’s GAEZ Data Portal10 and the International Food Pol- sions. These regularly visible light emissions are usually icy Research Institute’s MapSPAM & Harvest Choice11 based on the number of streetlamps or other permanent series provide a suite of data sets related to agricul- lights in villages or communities. Highly visible nighttime tural activity and productivity (for example, land, water, light activity suggests electricity sources. This logical soil, terrain and agro-climatic resources, protected assumption (of available electricity) should then be vali- areas, actual and potential production/yields as well as dated. In Nigeria, for example, data on education facilities selected socioeconomic and demographic data). from the Nigeria Millennium Development Goals Infor- On top of publicly available resources, one could also con- mation System, which includes information on whether sult with local counterparts who might share additional, a facility is grid connected, have been used to verify the proprietary information that might be useful later in the electrification status of the population cluster where the ranking process. Often, geospatial data do exist in differ- educational facility is located. ent governmental agencies, but their permission may be needed in order to use these datasets. Figure 2.19 illus- trates a great example from Nigeria, where the Federal MINI GRIDS FOR HALF A BILLION PEOPLE    99 FIGURE 2.18 • Health facilities and education facilities within 500 meters of village boundary Source: VIDA 2021. FIGURE 2.19 • Main grid coverage in Nigeria Source: Screenshot from https://nigeriase4all.gov.ng/map. 100   MINI GRIDS FOR HALF A BILLION PEOPLE FIGURE 2.20 • VIDA GridLight prediction for Ethiopia FIGURE 2.21 • Nighttime lighting in Nigeria (blue) compared to Ethiopian Electric Utility data (red) Source: Integration and Reiner Lemoine Institut 2016. Note: Nighttime lighting shown in white. Electrified consumer clusters appear in purple. Source: VIDA 2021. Figure 2.21 presents an example of the application of night FIGURE 2.22 • Nigerian night lights and 20 km buffer lights’ data, validated using data from other sources, to pre- zones dict the electrification status of population clusters. Electri- fied clusters can then be interconnected with an automatic grid-extension algorithm, which takes several factors into account—including topography, roads, and water bodies— to derive the most realistic grid connections. The informa- tion on population or population density and electrification status and distance from the grid may be supplemented with additional socioeconomic data to allow for more cri- teria when prioritizing the population clusters for mini grid electrification. LONG AND SHORT LISTS OF POTENTIAL MINI GRID SITES After the identified clusters are attributed, they can be prioritized according to certain criteria to generate long lists and short lists stating the project’s requirements and needs. The scope of this step is to eliminate clusters that don’t meet certain criteria. Elimination saves compu- Source: Integration and Reiner Lemoine Institut 2016. tational time and effort in the workflow steps that follow. Note: Purple areas show 20 km buffer zones. Night lights are shown in yellow with population clusters in black. Below are some examples. Once the locations of electrified clusters are known, a buf- fer zone can be applied around the clusters assumed to be depends on the local context. In an analysis by the Nige- electrified via grid connection. This represents a method- rian Energy Support Programme, for example, a 20 km ological shortcut to a least-cost electrification modeling radius around electrified clusters was taken to be suitable approach: clusters within the buffer zone are considered for electrification via grid connection in the base scenario. likely to be subject to grid extension and therefore unlikely Meanwhile, a 10 km buffer zone was thought appropri- to attract investment from private mini grid developers. ate in a low-grid electrification scenario (Integration and Meanwhile, clusters outside the buffer zone are unlikely to Reiner Lemoine Institut 2016). Figure 2.22 illustrates the be served by the grid within the time horizon considered 20 km buffer zones around electrified areas (based on and may thus be good candidates for mini grid electrifi- night lights), where grid extension would likely be the pre- cation. The decision on the size of a suitable buffer zone ferred option. MINI GRIDS FOR HALF A BILLION PEOPLE    101 Another approach would be to take the known grid cov- Environmental issues might also impose additional con- erage map, such as that in figure 2.22, and apply buffer straints in the prioritization process. For example, poten- zones directly around the network. This was done for the tial complications to implementing mini grid projects in NEP, and given the poor financial health of the distribution environmentally sensitive areas, and the time and cost of companies in Nigeria, a more aggressive scenario for mini obtaining necessary environmental permits and approvals, grids—where grid extensions are stalled—was consid- can be avoided by determining as early as possible whether ered, and the buffer zone was reduced to 5 km. Clusters a population cluster being considered lies in such an area, outside these buffer zones were considered suitable for so that an informed decision can be made on whether to decentralized electrification, but further screening based pursue a mini grid project at that site. on population was conducted. The economic viability of Key biodiversity areas (KBAs) are designated by the KBA mini grids was deemed unlikely for population clusters of Partnership,14 a consortium of wildlife conservation groups less than 1,000, which would be more effectively served including the World Wildlife Fund, while the International via SHS. Union for Conservation of Nature (IUCN) maintains a com- After sites within a specified distance from the grid (5 km prehensive database of legally protected areas around the for Nigeria) and with a population below a certain thresh- world. Protected areas are classified into different catego- old (1,000 people) are excluded from consideration, the ries, such as IUCN I-VI, Ramsar, and World Heritage sites, remaining clusters can be ranked based on scored criteria. which are all legally protected areas, while KBAs are not Recall the additional data collected on the clusters (pres- necessarily legally protected (though some protected ence and location of schools, clinics, and telecommuni- areas coincide with KBAs). cations towers). These can now be used to prioritize the The boundaries of population clusters can be compared clusters for mini grid electrification. with those of legally protected areas such as forest reserves The prioritization categories for the NEP were population, and national parks, as well as KBAs, using the Integrated density, distance to grid, and presence of telecommuni- Biodiversity Assessment Tool.15 This online geospatial tool cations towers, schools, and health facilities. Normalized hosts and maintains the three key global biodiversity data values for the prioritization categories were created and sets: the IUCN Red List of Threatened Species, the World summed up with weighting factors, and the clusters were Database on Protected Areas, and the World Database of then evaluated according to the prioritization criteria and Key Biodiversity Areas. ranked by state. Figure 2.23 exhibits the top 100 clusters The NEP decided to avoid the additional scrutiny and com- projected to be suitable for mini grid electrification for each pliance costs that come with implementing mini grid proj- federal state. FIGURE 2.23 • Results of prioritization of clusters for mini grid electrification in Nigeria Source: Integration and Reiner Lemoine Institut 2016. 102   MINI GRIDS FOR HALF A BILLION PEOPLE ects in protected areas and KBAs (if they are permitted at sible, to verify whether the community is served by the grid, all); therefore, any population clusters that intersected with and deploying agents (for example, on motorbikes where them were excluded from further consideration. Figure this is a swift and safe option) to quickly check on the elec- 2.24 shows an instance of population clusters within pro- trification status of all the population clusters shortlisted tected areas or KBAs being flagged for exclusion. for potential mini grid projects. FIELD VERIFICATION Part 2. Load profile characterization While the site identification and screening methodology The first part of the analysis identified, attributed, and vali- described in this section uses available geospatial data to dated—to the extent possible—the candidate sites for mini propose communities where mini grid electrification may grid deployment. The following step focuses on estimating be suitable, in the absence of reliable data on the reach the load profile for each candidate site. Estimating the load of the main grid, it does so partly by making assumptions profile is usually subject to requirements, as listed below about the electrification status of the population clusters. and illustrated in figure 2.25: Before deploying multidisciplinary survey teams to collect • Identify the composition of potential customers data on these communities, wasting resources on false positives (communities thought to be off grid that are actu- • Estimate the daily consumption of each customer seg- ally on grid) may be avoided with validation exercises. Such ment exercises include calling someone in the community, if pos- • Estimate the load profile in the cluster • Incorporate seasonality of demand over the year • Estimate growth rate (forecasting) The site-screening phase of a geospatial plan for portfolio development includes There are two paths. The first is to utilize all collected data identifying, characterizing, and ranking popula- and information, measure them against past experience, tion clusters, as well as undertaking environmental and estimate the load (or simulate it with machine learn- screening and field verification. Throughout this ing). The second is to survey all the candidate sites—if phase, the criteria used for all these tasks need to they’ve been shortlisted—or, failing that, conduct a ground be carefully considered. survey of sample sites on the load in those locations. The selection depends on the scope of the project as well as the available resources. FIGURE 2.24 • Population clusters falling in protected areas or KBAs flagged for exclusion in Nigeria Source: Screenshot from https:/ /worldbank.africageoportal.com/. IUCN = International Union for Conservation of Nature. MINI GRIDS FOR HALF A BILLION PEOPLE    103 FIGURE 2.25 • Requirements for estimating load profile 1. Composition of custmers (% split and quantity) Additional demand 2. Daily consumption A daily kWh value for each of the customer segments 3. Load pro le Residential Commercial Public Productive Irrigation Averaged daily load pro le 12 10 Low 8 Load kW 6 Medium 4 2 Agricultural 0 High equipment 2 4 6 8 10 12 14 16 18 20 22 24 Hour of day For each of these activities 4. Seasonality 5. Growth rate the secondary datasets will • Years to plan the location for be assessed and used • Presumed growth rate for those years where applicable Source: Integration 2021. kWh = kilowatt-hour. ROUGH ASSESSMENT BASED ON COLLECTED GIS DATA • Water pump (if existing) demand estimated at 5,000 (EASILY SCALABLE) kWh/day The rough assessment is based on general assumptions. • Telecommunications tower (if existing) demand esti- Some of them have been used (for example, in the case of mated at 84 kWh/day Ethiopia) to provide a quick, high-level estimate of the total demand in the candidate sites. These assumptions are MORE ACCURATE ASSESSMENT FOCUSED ON A FEW PILOT listed below: SITES (SURVEY DATA COLLECTION, MANUAL INTERVENTION) In the second approach—and for the shortlisted sites—it • Percentage of households connected to the mini grid is recommended that survey teams be deployed to col- (58 percent) lect data. Sending these teams only to those communities • Residential vs commercial customers likely to host a viable mini grid project will save time and – residential = (small rooftops and 50 percent medi- resources, and sending teams after the building mapping is um-sized rooftops) x 0.58 (connection rate) completed will enable them to conduct their surveys more – commercial = (large rooftops and 50 percent medi- efficiently. um-sized houses) x 0.58 (connection rate) Under the NEP in Nigeria, survey teams were deployed to • Residential customer demand, estimated at ~0.22– collect data on the community, households, public institu- 0.32 kWh/day tions, and commercial and productive users of energy in the prioritized unelectrified communities. Each survey cov- • Commercial customer demand, estimated at ~1.1 kWh/ ered community, commercial, household, and environmen- day tal-social data. The survey teams geotagged and surveyed • Public institution demand (if existing) estimated at: nonresidential structures in each community to identify – 2.97 kWh/day for primary schools the key loads (large, daytime, productive and commercial) – 11.23 kWh/day for health clinics that might be critical to mini grid viability. They recorded the count and wattage of large appliances, light bulbs, and • Flour mill (if existing) demand estimated at about 43.77 fans for each geotagged building. They also geotagged a kWh/day subset of households, categorizing each as large, medium, or small. 104   MINI GRIDS FOR HALF A BILLION PEOPLE Watts Power (kW) Watts Watts 10 5 0 0 0 15 10 50 70 25 35 20 30 40 60 80 90 20 30 200 400 600 800 0 10 20 30 1,000 0: 00 – 0:00–1:00 0:00–1:00 0:00–1:00 1:0 1:0 0– 0 1:00–2:00 1:00–2:00 1:00–2:00 2: 2:0 2:00–3:00 2:00–3:00 2:00–3:00 Source: VIA 2021. 00 0 – 3:00–4:00 3:00–4:00 3:00–4:00 Source: VIDA 2021. 3: 3:0 00 0 4:00–5:00 4:00–5:00 4:00–5:00 –4 5:00–6:00 5:00–6:00 5:00–6:00 Commercial 4: :0 00 0 6:00–7:00 6:00–7:00 6:00–7:00 –5 7:00–8:00 7:00–8:00 7:00–8:00 Shops load pro le 5: :0 00 0 8:00–9:00 8:00–9:00 8:00–9:00 –6 6: :0 9:00–10:00 9:00–10:00 9:00–10:00 Health clinics load pro le 00 0 10:00–11:00 10:00–11:00 10:00–11:00 –7 7: : 11:00–12:00 11:00–12:00 11:00–12:00 00 00 Households –8 12:00–13:00 12:00–13:00 12:00–13:00 Average residential load pro le :0 Time of day Time of day 8: 00 0 13:00–14:00 13:00–14:00 13:00–14:00 Time of day 9: –9 14:00–15:00 14:00–15:00 14:00–15:00 00 :00 – 15:00–16:00 15:00–16:00 15:00–16:00 10 10: :0 0 16:00–17:00 16:00–17:00 16:00–17:00 0– 0 17:00–18:00 17:00–18:00 17:00–18:00 11 11 :0 :0 18:00–19:00 18:00–19:00 18:00–19:00 0– 0 12 12 19:00–20:00 19:00–20:00 19:00–20:00 :0 :0 0– 0 20:00–21:00 20:00–21:00 20:00–21:00 13 13 21:00–22:00 21:00–22:00 21:00–22:00 :0 :0 Time 0– 0 22:00–23:00 22:00–23:00 22:00–23:00 14 14: :0 0 23:00–0:00 23:00–0:00 23:00–0:00 0– 0 15 15 :0 :0 0– 0 16 16: Watts :0 0 Watts 0– 0 Watts 17 17 :0 :0 0 500 1,000 1,500 2,000 2,500 3,000 3,500 4,000 4,500 0 50 0 100 200 300 400 500 150 100 250 200 0– 0 18 18 :0 :0 0 0 0:00–1:00 0:00–1:00 0:00–1:00 19 –19 :0 :0 1:00–2:00 1:00–2:00 1:00–2:00 0– 0 2:00–3:00 2:00–3:00 2:00–3:00 20 20: 3:00–4:00 3:00–4:00 3:00–4:00 :0 0 0– 0 21 4:00–5:00 4:00–5:00 4:00–5:00 21 :0 :0 5:00–6:00 5:00–6:00 5:00–6:00 0– 0 6:00–7:00 6:00–7:00 6:00–7:00 FIGURE 2.27 • Demand curve for a randomly selected candidate site in Ethiopia 22 22 :0 :0 0 0 7:00–8:00 7:00–8:00 7:00–8:00 23 –23 8:00–9:00 8:00–9:00 8:00–9:00 Flour mill load pro le :0 :0 0– 0 9:00–10:00 9:00–10:00 9:00–10:00 24 :0 10:00–11:00 10:00–11:00 10:00–11:00 0 11:00–12:00 11:00–12:00 11:00–12:00 12:00–13:00 12:00–13:00 12:00–13:00 Drinking water pump load pro le 13:00–14:00 13:00–14:00 13:00–14:00 Time of day Time of day 14:00–15:00 14:00–15:00 14:00–15:00 15:00–16:00 15:00–16:00 15:00–16:00 Elementary school (Yr 1-8) load pro le 16:00–17:00 16:00–17:00 16:00–17:00 17:00–18:00 17:00–18:00 17:00–18:00 FIGURE 2.26 • Indicative load profiles for various customer segments in potential mini grid locations 18:00–19:00 18:00–19:00 18:00–19:00 MINI GRIDS FOR HALF A BILLION PEOPLE   19:00–20:00 19:00–20:00 19:00–20:00 20:00–21:00 20:00–21:00 20:00–21:00 21:00–22:00 21:00–22:00 21:00–22:00 22:00–23:00 22:00–23:00 22:00–23:00  105 23:00–0:00 23:00–0:00 23:00–0:00 FIGURE 2.28 • Sample load profile for a village 120 100 80 Power (kW) 60 40 20 0 0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 Hour of the day Source: ESMAP analysis. kW = kilowatt. The surveys capture crucial information on electricity con- sumption, including: During the survey collection and load model- • Community willingness and ability to pay ing phase, survey teams should be sent only • Current expenditure on electricity generation using pet- to those sites that have been prioritized as having rol or diesel generators great potential for a viable mini grid, because of the costs of conducting on-site surveys. Survey teams • Expenditures on candles, kerosene, and dry cell batteries typically collect data that can inform the estimation • Other data pertinent to demand estimates of potential demand from households, commercial We recommend that, with smartphones and tablets so and productive loads, and public and community ubiquitous, personal interviews be conducted with com- institutions such as churches and hospitals. puter assistance. This practice ensures that data are col- lected with optimal efficiency and quality, minimizing the time and effort spent on data cleaning. user, recording the wattage and count of lightbulbs and Survey data, along with data from the literature, should fans at each location. The data were cleaned to adjust for inform the development of load profiles for different cus- outliers and to remedy survey errors. Equipment-use pro- tomer segments. The load profile is a view of estimated files emerged from applying utilization factors to each hour electrical demand at a given hour or day. This information of the day (varying by “high” and “low” months), and each will be vital to designing the mini grid. Figure 2.28 presents piece of large equipment was assigned a unique use profile a sample load profile. by state and community size (small communities and large communities use commercial/productive equipment dif- Load profiles for the NEP were developed by aggregating the ferently), and customer-type load profiles were generated expected demand from households, commercial and pro- by combining the equipment profiles with the survey data. ductive loads, and public infrastructure, as described below. It was assumed that all existing commercial lights would be Residential sector loads. Residential load profiles were replaced by 18W LED bulbs and residential lights by 6 W. All developed with Rural Electrification Agency (REA) classi- geotagged commercial and productive loads were included fications for small, medium, and large households. These in each site’s load profile. were also based on assumptions about the number of Public sector loads. The locations of public institutions appliances per household type and the use profiles for such as schools, health facilities, and religious centers each appliance (total watts and utilization factors for each in each community were determined from a geospatial hour of the day, considering seasonal variations). database of such points of interest16 and the REA geotag Commercial and other productive loads. In the geo- survey. Generic load profiles were written up for each type tag survey, surveyors assigned commercial and produc- of public institution, based on consultations with a devel- tive loads to 13 different business categories. Then they oper experienced with sizing and installing photovoltaic recorded the count and wattage of the high-load equip- systems for public institutions,17 and applied to the public ment associated with each commercial and productive end loads at each site. 106   MINI GRIDS FOR HALF A BILLION PEOPLE Part 3. System design and optimization tive function to design the generation as well as network assets for rural electrification. REM can help project Once the power load is estimated, then follows the system developers and investors with initial technical design and design and/or optimization. There are multiple options cost estimations, including both capital and operational for system optimization. Based on the load profiles devel- expenditures. The modeling tool seeks to aid developers oped for each community, modeling tools, each running on in making viable decisions regarding mini grid design by unique algorithms, calculate optimal solutions—least-cost providing the analytics needed to conduct technical and mini grid system designs that meet predefined parame- financial evaluations. REM offers a single package capable ters—for each community. of computing generation investment, operational perfor- For example, for mini grids to be included in the tender mance, and detailed design of the network starting from for the Nigerian NEP, a renewable fraction threshold of 60 the building level. percent was prescribed (that is, the mini grid design would Renewable Energy Integration and Optimization (REopt). need to produce at least 60 percent of its annual energy NREL has developed a tool called REopt™, which it uses output from renewable sources). The NEP also has min- to provide decision support, analyzing and optimizing imum technical requirements specific to several distinct mini grid designs for different systems.20 REopt is a criti- system architectures. Depending on the system archi- cal tool for understanding the technoeconomic trade-offs tecture selected for a particular community, the mini grid in the mini grid sector, which can lead to more sustainable would need to meet the minimum technical requirements business models and promote universal energy access. for that particular system architecture. These technical Formulated as a mixed-integer linear program, it solves a requirements cover both quality and sizing of components deterministic optimization problem to establish the opti- as well as quality of service, such as the number of hours mal selection, sizing, and dispatch strategy of technologies the mini grid can be offline for scheduled or unscheduled chosen from a candidate pool, such that electrical, thermal, maintenance. These design constraints are incorporated and or water loads are met at every step in the minimum into the optimization exercise. life-cycle cost. REopt is a time series model that looks at OPTIMIZATION MODELS a full-year energy balance to determine multiyear cash flows by applying appropriate discount and cost-escalation For the NEP in Nigeria, three optimization models (HOMER, rates. As opposed to algorithmic dispatch strategies, REopt REM, and REopt) were used to propose optimal mini grid finds the global optimum by anticipating load and resource designs for each community, given the load profile of that changes over the full analysis period. In the mini grid con- community and some design constraints, such as those text, this allows REopt to dispatch batteries to maximize described above. Each model takes a different approach renewable energy utilization and minimize generator run to mini grid optimization. All three models provide gener- time, maximizing economic efficiency. ation system sizing, but REM also generates a customized distribution design for each mini grid. Each of these three OTHER APPROACHES optimization models is briefly described below. Additional Other approaches, such as those described below, can also details are available at the companion website to this book: be implemented in the manner just presented. www.esmap.org/mini_grids_for_half_a_billion_people. Village Infrastructure Angels (VIA) employs another meth- Hybrid Optimization Model for Multiple Energy Resources odology to propose least-cost distribution network designs (HOMER). Originally developed at the U.S. National Renew- for mini grids, which excludes households that might be able Energy Laboratory (NREL), and enhanced and dis- better candidates for SHSs. VIA has designed electrical tributed by HOMER® Energy, HOMER software nests three power distribution networks for mini grids in the Philip- powerful tools—simulation, optimization, and sensitivity pines and Haiti (Craine, S) using a minimum spanning tree analysis—in one software product, so that engineering algorithm, which solves for the shortest network of lines to and economics work side by side.18 HOMER is the industry connect a given set of points. VIA conducts load-flow analy- standard for optimizing mini grid design in all sectors, from ses, assuming an average load per building and a maximum village power and island utilities to grid-connected cam- 10 percent voltage drop between the power source and the puses and military bases. end user, to determine what size wire (conductor) is nec- The Reference Electrification Model (REM). Developed essary to distribute power to households, the distribution by the MIT-Comillas Universal Energy Access Labora- losses involved, and the cost. A Critical Distance analysis, tory, REM is a computational modeling tool designed to based on the prevailing cost of SHSs, suggests the length help plan detailed medium- and low-voltage distribution of the mini grid or grid extension line that is viable before networks, with an implementation focus on developing a SHS makes much more sense. This guides which lines countries.19 REM uses cost minimization as the objec- of the distribution network should be kept or discarded in MINI GRIDS FOR HALF A BILLION PEOPLE    107 favor of SHSs. For the mini grids that remain, VIA lists wire special consideration because they require higher peak sizes, poles, and any transformers (if required) and pub- power than a standard user. The power asset systems will lishes them on a GIS platform. This gives aspiring develop- then be sized according to HOMER demand analysis. The ers, perhaps considering a project at the site, a useful bill scope of the study employs a standardized simulation to of materials. scale up the modeling to more sites. Integration combines manual designs and HOMER opti- Village Data Analytics (VIDA) conducts a hyperlocal den- mization. Based upon the analysis of building footprints sity analysis to identify village cores, outskirts, and outly- and satellite imagery, a grid design is created remotely, and ing areas. Within the core area, VIDA’s algorithm identifies potential locations for the power asset suggested. The grid high-value mini grid customers. using building category, design can be made to assume that all or a portion of the the location of buildings with respect to roads, and the den- buildings are connected within a certain distance from the sity of the built-up area (figure 2.30a). Special importance central village cluster. From this data, a voltage drop model is given to anchor/institutional loads. Once demand is iden- is generated to approximate the cable sizes required (fig- tified for the different customer segments, an algorithm ure 2.29). Specific load centers or anchor loads receive then generates a distribution layout that connects the high-value mini grid customers. First, a minimum spanning tree is generated to get an upper limit for the connection FIGURE 2.29 • Output of the voltage drop model density. Next, the trunk lines are generated (figure 2.30b). These are either single or three-phase and typically follow the main roads through the village. The poles are located There are multiple options for optimizing mini grid system design at each site. Based on the load profiles developed for each community, modeling tools, each running on unique algorithms, calculate the least-cost mini grid system designs that meet predefined community parameters. The most common system optimization tools are those from HOMER, MIT’s REM, and NREL’s REopt. Source: Integration 2021. FIGURE 2.30 • Sample outputs of hyperlocal density analysis from Village Data Analytics  igh-value mini grid customers a. H  ini grid distribution layout with trunk line, poles, and b. M (colored buildings) dropdown lines connecting high-value mini grid customers Source: VIDA 2021. 108   MINI GRIDS FOR HALF A BILLION PEOPLE at equal distances and then dropdown lines are generated would provide a first indication of sites where affordable so they connect every high-value mini grid customer to the tariffs can be expected. nearest poles. The bill of materials for distribution is then • Viability-gap calculations. Here, reverse logic can be estimated, which leads to estimations of distribution cost, applied whereby an agreed tariff is used as input to including the length of the predicted trunk line, the number determine the viability gap of the projects and grant of poles, and the length of the dropdown lines. The algo- share needed in CAPEX to reach the required financial rithm uses average cost per meter of wiring (of different returns. This differentiates sites according to grant lev- sizes) and the average cost of poles to generate this infor- els needed for sustainable mini grid operation. mation on a village-by-village basis. Part 5. Result visualization and dissemination via an Part 4. Sensitivity analysis and financial modeling online platform More scenarios are generated during the final part of the workflow, and these assess how sensitive the optimal solu- Geospatially planned portfolios of mini grids contain a tion is to input parameters and financial assessments. They wealth of information for developers. When produced as might also assess sensitivity to demand levels, quality of ser- part of a mini grid program in partnership with a govern- vice, level of renewable generation, to name a few. The sensi- ment or a developer, these portfolios should be shared with tivity analysis goes hand in hand with the financial modeling, mini grid developers, presenting relevant information in an which can be developed for site-specific characteristics accurate and transparent way. allowing for outputs of tariff, capital expenditure (CAPEX), One example is the VIDA software, a tool for site identifi- viability-gap analysis, and other relevant outputs in view cation, selection, prioritization, visualization, and collabo- of project needs (figure 2.31). Besides the usual financial ration (figure 2.32). The software visualizes the mini grid model parameters, some context-specific aspects include: portfolio analysis on an interactive platform that offers both • Scenarios with and without PUE. The definition of project a high-level overview of the modeling exercise (national and CAPEX could omit investments in productive activities regional levels) but also granular descriptions of all candi- and run the model under both scenarios. Such an anal- date sites. VIDA provides access to mini grid viability indi- ysis could, for example, identify viable sites (1) without cators and distribution layout characteristics. Users can depending on productive loads, (2) only if a minimum download results, upload data, share information, and col- basic productive load is ensured, (3) only if extensive laborate on the software platform. investment is undertaken to promote large-scale pro- Another example of such a dissemination tool is Odyssey ductive loads. Energy Solutions, which provides a web-based data plat- • Tariff-based calculations. In this mode the financial form that facilitates deployment of mini grids in emerging model can use predetermined CAPEX structure (equity, markets for developers, financiers, vendors, and govern- debt, and grant) and operating expenditure as inputs ments and donors. It developed a customized version of its to drive the tariff needed to meet the required finan- online platform for Nigeria’s REA to manage data analysis cial returns. Furthermore, time-of-use tariffs and tar- and the bidding process for the sites included in the mini iffs based on customer type can also be modeled. This grid tender of the NEP. FIGURE 2.31 • Indicative flow of financial modeling process of mini grids • Finacial paramaters (discount • Excel based financial model • Income statement rate, interest rate) • Cashflow statement • Income statement • Financing structure • Balance sheet • Cashflow statement • Generation system assets • Balance sheet • Key performance indicators • Distribution system assets – LCOE • Key performance indicators • Consumer assets ? – Profit margin • Depreciations – NPV • Energy sales per category – IRR • Tariff structure – Payback Source: Integration 2021. MINI GRIDS FOR HALF A BILLION PEOPLE    109 FIGURE 2.32 • VIDA interactive platform Source: VIDA 2021. The user interface organizes data into modules that con- tain the technical and financial dimensions of each mini Geospatially planned mini grid portfolios grid site: location, load forecast, generation system and offer a wealth of information for developers. distribution design, costing, tariffs, and financial model. When produced as part of a government or devel- The online platform has detailed, site-specific informa- opment partner mini grid program, this information tion on the customers and loads as well as suggested should present the relevant information to devel- system sizing. opers in an accurate and transparent fashion. The Once officially registered, program bidders gain access Village Data Analytics platform and Odyssey both to information posted at the tender sites, where they offer excellent ways to visualize and disseminate can register to participate in the tender. Registered bid- the modeled results. ders may access key program information within the 110   MINI GRIDS FOR HALF A BILLION PEOPLE FIGURE 2.33 • Mini grid tender preparation in Odyssey Source: Screenshot of Nigeria Electrification Project portal on https://www.odysseyenergysolutions.com/. NEP = Nigeria Electrification Project. platform, including all tender documents, deadlines, and instructions for submitting a bid. They may also view a list Geospatial analysis provides a broader pic- of all sites and run analytics within the software to view ture of the locations and characteristics of aggregate statistics about the sites (grouped by state, communities that can be considered at a for example). Bidders can copy all site data into their own portfolio level, which enables mini grid developers accounts in the platform, allowing them to use the soft- to exploit economies of scale and prepare quicker, ware to assemble a technical and financial proposal for more cost-effective rollout plans and plans for ser- each lot using Odyssey’s software tool suite and struc- vice and maintenance. Governments can also use tured workflow (figure 2.33). geospatial and other digital tools to catalyze deploy- Once the technical analysis for each site is complete, sites ment of mini grids led by the private sector to sup- are bundled into portfolios for each of the tender lots. Bid- ply electricity to off-grid communities. For example, ders must provide the business plan, financing approach, Nigeria’s Rural Electrification Agency is holding and required documents uploaded to the data room. minimum-subsidy tenders for portfolios of promis- Incomplete proposals cannot be submitted; a final check- ing mini grid sites that it has identified and for which list ensures all sections are complete and all required files it has collected market intelligence. uploaded to the data room. Only final portfolios for the tender lot are submitted to the evaluation committee. The VIDA and Odyssey platforms are compatible and can form a strong toolkit for governments, REA, development sey could navigate to VIDA to view and access granular financial institutions, or private organizations. VIDA’s soft- geospatial and on-ground data of the villages being ten- ware can host village-level data, geospatial, and others in dered. Similarly, a user in VIDA software could push the an interactive user interface that can then be accessed by village-level data to the Odyssey platform to tender and Odyssey for tendering and deployment. A user in Odys- deploy mini grids. MINI GRIDS FOR HALF A BILLION PEOPLE    111 LESSONS LEARNED AND NEXT STEPS REFERENCES The introduction of geospatial and other digital tech- Arderne, C., C. Zorn, C. Nicolas, and E. E. Koks. 2020. “Predictive nologies has lowered preparation and planning costs by Mapping of the Global Power System Using Open Data.” Scientific an order of magnitude: from about $30,000 per site— Data  7: Article  19. https://doi.org/10.1038/s41597-019-0347-4. because each site required high-level on-site analysis—to https://www.nature.com/articles/s41597-019-0347-4 around $2,300 per site, based on the World Bank’s recent Castalia. 2014. Myanmar National Electrification Program (NEP) Road- experience in Nigeria. Furthermore, by 2024, high-resolu- map and Investment Prospectus. Washington, DC: Castalia. https:/ / www.seforall.org/sites/default/files/Myanmar_IP_EN_Released. tion satellite imagery is expected to fall by nearly 60 per- pdf. cent from 2014 levels (Selding 2015). This too will drive CIESEN (Center for International Earth Science Information Net- down the cost of ever more accurate geospatially planned work), Columbia University and Novel-T. 2020. “GRID3 Central portfolios. Meanwhile, taking a portfolio approach to mini African Republic Settlement Extents Version 01, Alpha. Palisades, grid development, instead of building mini grids as one- NY: Geo-Referenced Infrastructure and Demographic Data for off projects, can slash upfront capital costs by around Development (GRID3). Source of building Footprints ‘Ecopia Vector Maps Powered by Maxar Satellite Imagery’.” Accessed June 1, 2022. $100/kW, according to analysis of the ESMAP’s data- https://doi.org/10.7916/d8-y2ax-p859. base of installed and planned mini grids presented in the Craine, S. 2019. Haiti Geospatial Exercise for the Global Facility on Mini overview to this handbook. In addition, geospatial analy- Grids of the Energy Sector Management Assistance Program: Incep- sis can help identify potential productive-use customers, tion Report. Washington, DC: World Bank. Unpublished. thereby shaping developers’ community engagement DeCarolis, Joseph, Hannah Daly, Paul Dodds, Ilkka Keppo, Francis Li, strategies to promote income-generating uses of mini Will McDowall, Steve Pye, Neil Strachan, Evelina Trutnevyte, Will grid electricity. Usher, Matthew Winning, Sonia Yeh, and Marianne Zeyringer. 2017. “Formalizing Best Practice for Energy System Optimization Model- As developers achieve economies of scale by developing ling.” Applied Energy 194 (May): 184–98. https://www.sciencedirect. economically viable mini grid portfolios that support pro- com/science/article/pii/S0306261917302192. ductive uses of electricity, and as mini grid component Ecopia AI and Maxar Technologies. 2021. “Digitize Africa.” https:// costs plummet over the next decade, as discussed in chap- blog.maxar.com/earth-intelligence/2018/gis-ready-building-foot- ter 1, the cost of mini grid electricity is on pace to reach print-shapefiles-for-accelerated-analysis. Internal resource. $0.20/kWh by 2030. As mini grid electricity approaches Hahsler, Michael, Matthew Piekenbrock, and Derek Doran. 2019. “dbscan: Fast Density-Based Clustering with R.” Journal of Statistical this cost threshold, mini grids become the least-cost option Software 91 (1): 1–30. DOI: 10.18637/jss.v091.i01. for more and more people. This means national least-cost HOMER Energy. N.d. “Hybrid Optimization Model for Multiple Energy electrification plans will need to weigh expected cost Resources (HOMER).” https://www.homerenergy.com/. declines in mini grid electricity as they anticipate main grid Howells, Mark, Jairo Quiros-Tortos, Robbie Morrison, Holder Rogner, expansions, more mini grids, and SHSs. Taco Niet, Luca Petrafulo, Will Usher, William Blyth, Guico Godinez, Luis F. Victor, Jam Angulo, Franziska Bock, Eunice Ramos, Fran- To catalyze deployment of private-sector-led mini grids to cesco Gardumi, Ludwig Hulk, Patrick Van-Hove, Estathios Peteves, supply off-grid communities, Nigeria’s REA, the implement- Felipe de Leon, Andrea Meza, Thomas Alfstad, Constantinos Talio- ing agency for this project, is holding minimum subsidy tis, George Partasides, Nicolina Lindblad, Benjamin Stewart, Ash- tenders21 for portfolios of promising mini grid sites it has ish Shrestha, Dana Rysankova, Adrien Vogt-Schilb, Chris Bataille, identified. The World Bank and the REA have developed an Henri Waisman, Asami Miketa, Pablo Carvajal, Daniel Russo, Morgan Bazilian, Andrii Gritsevskyi, Mario Tot, and Adrian Tompkins. 2021. innovative protocol for mini grid site identification, screen- “Energy System Analytics and Good Governance—U4RIA Goals of ing, and analysis using geospatial tools, including a geospa- Energy Modelling for Policy Support.” Unpublished preprint version. tial portfolio planning methodology to assess and select https://www.researchsquare.com/article/rs-311311/v1. the communities to be included in the minimum subsidy IEG (Independent Evaluation Group). 2015. World Bank Group Sup- tenders. The protocol enables governments, development port to Electricity Access, FY 2000–14. Washington, DC: World Bank. partners, or other public institutions to prepare portfolios https://ieg.worldbankgroup.org/sites/default/files/Data/reports/ Electricity_Access.pdf. of mini grid projects and “crowd in” private sector cofinanc- ing. We hope this may offer useful guidance to those seek- Integration. 2021. “Pilot Location Modelling Results”, (Internal report). ing to develop mini grid projects at scale elsewhere. For Integration and Reiner Lemoine Institut. 2016. “Preliminary Modelling of Off-grid PV Capacities for the Whole of Nigeria.” Nigerian Energy Sup- example, governments could replicate these steps in other port Program and the Federal Ministry of Power, (Internal report). countries interested in competitive tenders to kickstart or Khavari, B., A. Korkovelos, A. Sahlberg, M. Howells, and F. F. Nerini. scale up the market for mini grids. 2021. “Population Cluster Data to Assess the Urban-Rural Split and Electrification in Sub-Saharan Africa.” Scientific Data 8: Article 117. https://www.nature.com/articles/s41597-021-00897-9. 112   MINI GRIDS FOR HALF A BILLION PEOPLE Moulavi, Davoud, Pablo A. Jaskowiak, Ricardo Campello, Arthur Zimek, 5. This material is organized and shared through the GEP’s GitHub and Joerg Sander. 2014. “Density-Based Clustering Validation.” workspace. https://epubs.siam.org/doi/pdf/10.1137/1.9781611973440.96. 6. The development of open access training material has been under- NRECA International. 2019. “National Electrification Investment Plan. taken by the Climate Compatible Growth (CCG) program (https:/ / USAID Ethiopia—Distribution Systems Strengthening Project” , climatecompatiblegrowth.com/). (Internal report). 7. In the case of Nigeria, this task was carried out by Integration and Odyssey Energy Solutions. N.d. “Finance, Build and Operate Distrib- the Reiner Lemoine Institut. A similar approach has been employed uted Infrastructure at Scale.” https://www.odysseyenergysolutions. at the Global Electrification Platform (GEP) following a methodol- com/. ogy suggested by Khavari and others (2021) (https:/ /www.nature. Selding, Peter. 2015. “Established Imagery Providers Face Changing com/articles/s41597-021-00897-9). Competitive Landscape.” SpaceNews, September 24. https:/ /space- 8. Energydata.info is an open data platform providing access to data news.com/established-imagery-providers-face-changing-competi- sets and data analytics that are relevant to the energy sector, avail- tive-landscape/. able at https://energydata.info/. VIA (Village Infrastructure Angels). 2021. Geospatial Mapping of Least 9. HOTOSM is an international team dedicated to humanitarian action Cost Mini Grid Potential in Ethiopia. Project Completion Report for and community development through open mapping; the database the Energy Sector Management Assistance Program (ESMAP)/ is open and available at https://www.hotosm.org/tools-and-data. World Bank. Internal report. 10. The Food and Agriculture Organization of the United Nations (FAO) VIDA (Village Data Analytics). 2021. Geospatial Mapping of Least Cost Agro-Ecological Zones (AEZ) database is available at http://www. Mini Grid Potential in Ethiopia. Project Completion Report for the fao.org/nr/gaez/en/#. Energy Sector Management Assistance Program (ESMAP)/World 11. The International Food Policy Research Institute’s HarvestChoice Bank, (Internal report). products are available at https://dataverse.harvard.edu/dataverse/ World Bank. 2007. “Geospatial Plan for Kenya (National Electrification harvestchoice. Coverage Planning). Investment Costing Estimation Model.” Unpub- 12. The Gridfinder.org visualization can be found at https:/ /gridfinder. lished paper, World Bank, Washington, DC. org/; the modeled results for all countries are available at Zenodo World Bank. 2009. Project Appraisal Document on a Proposed Credit (2020). https://zenodo.org/record/3628142#.YIgO85BKhPY. in the Amount of SDR 45.1 Million (US$70 Million Equivalent) to the In the case of Nigeria, the night lights’ data set was retrieved from the 13. Republic of Rwanda for a Rwanda Electricity Access Scale-Up and NOAA Earth Observation Group (https:/ /ngdc.noaa.gov/eog/). The Sector-Wide Approach (SWAP) Development Project. Washington, World Bank’s Light Every Night initiative provides open access to DC: World Bank. all nightly imagery and data from the Visible Infrared Imaging Radi- World Bank. 2015. Achieving Universal Access in the Kaduna Electric ometer Suite Day-Night Band (VIIRS DNB) from 2012 to 2020 and Service Area. Energy Sector Management Assistance Program. the Defense Meteorological Satellite Program Operational Linescan Washington, DC: World Bank. https:/ /documents1.worldbank.org/ System (DMSP-OLS) from 1992 to 2013. You may find more infor- curated/en/782491487138851242/pdf/112802-Report-Kadu- mation at https:/ /registry.opendata.aws/wb-light-every-night/. na-Electric.pdf. More information about the source is available at http:/ 14. /www.key- Zenodo. 2020. “Data from: Predictive Mapping of the Global Power Sys- biodiversityareas.org/assets/8f1535aed3316ae2b720364019f- tem Using Open Data.” Zenodo, January 16, 2020. https:/ /zenodo. 8cb1c. org/record/3628142#.YIgO85BKhPY. 15. Tool available at https://ibat-alliance.org/. 16. The NGO ehealth Nigeria gave the authors access to this informa- tion from its database. NOTES Em-One designed and installed solar solutions in public institutions 17. such as schools and health centers in Lagos, Kaduna state, and 1. The electricity distribution grid has not been mapped or the data northeastern Nigeria. are not available in many countries. Gridfinder, an open-source tool 18. HOMER® Energy’s software suite consists of two desktop prod- for predicting the location of electricity network lines, using night- ucts—HOMER Pro and HOMER Grid—and application program- time lights satellite imagery and OpenStreetMap data, was used ming interfaces for building web-based tools, such as HOMER to determine the location of the grid for countries where the data QuickStart and QuickGrid. weren’t available. See https:/ /gridfinder.org/ and https:/ /github. com/carderne/gridfinder. 19. More information about MIT’s REM model and application is avail- able at http://universalaccess.mit.edu/#/main. 2. The GEP processes HRSL population data to transform them into population clusters (similar to GRID3) based on proximity and den- 20. More information about NREL’s REopt model and application is sity. You may read more about this approach at https://www.nature. available at https://reopt.nrel.gov/. com/articles/s41597-021-00897-9 21. REA has grouped these potential mini grid sites into lots by state 3. Firm power output of a mini grid is defined in chapter 1 as the gen- and will invite private developers to build, own, and operate these erator capacity (kW) plus 25 percent of the solar array output rated portfolios of mini grid projects. Through a competitive process, peak (DC) power output (kWp). REA will award grants per connections to the private developers selected to implement these projects. Their bids will be evaluated 4. Consortium members include KTH Royal Institute of Technology, based on the quality of their technical proposal and on their subsidy Development Seed & Derilinx, World Resources Institute (WRI), and requirement. Cambridge University. MINI GRIDS FOR HALF A BILLION PEOPLE    113 CHAPTER 3 PRODUCTIVE LIVELIHOODS AND BUSINESS VIABILITY CHAPTER OVERVIEW This chapter highlights why productive uses of electricity can be a game changer for both mini grid developers and socioeconomic development. It presents an everyone-wins scenario for developers, local entrepreneurs, commu- nities, and national utilities. Using real-world examples, the chapter outlines a six-step approach to implementing productive-use interventions and discusses who can organize such interventions. THE MULTIPLE BENEFITS OF dred times more than the energy consumed by a nearby mini grid customer. Grain mills have power ratings in the CONNECTING INCOME-GENERATING 1,000–10,000 watt (W) range, about ten times more than MACHINES AND APPLIANCES TO commercial appliances like refrigerators and a hundred MINI GRIDS times more than household appliances (CrossBoundary 2020). Figure 3.1 shows the load profiles for mini grids Increasing productive uses1 of mini grid electricity cre- operating with 22 percent and 40 percent load factors. ates an “everyone-wins” scenario for mini grid developers, Greater demand for electricity generates additional reve- rural entrepreneurs, communities, and national utilities nues for mini grid operators while improving the utilization over time. It reduces the levelized cost of energy, which capacity of their systems, which reduces the unit cost of increases the mini grid developer’s margins and therefore electricity (per kilowatt-hour) and ensures efficient use of financial viability. Entrepreneurs and small businesses ben- the mini grid’s assets. Although increased demand requires efit from switching from expensive diesel generators to more capital investment, it can optimize the use of sys- affordable mini grid electricity. Communities benefit from tems, especially during daytime, when the residential load the new jobs that mini grids create and the increased eco- is small and systems are underused. Daytime use is critical nomic activity. The growth of rural economies also benefits for solar-based mini grids, which produce electricity at min- national utilities once interconnection to the main grid is imal marginal cost during the day. considered, because it increases customers’ demand for Furthermore, an Energy Sector Management Assistance high-quality electricity and their ability to pay for it. Program (ESMAP) analysis of 1,028 mini grids in Cambo- Boosting productive uses benefits mini grids and their dia, Myanmar, and Nepal indicates that every additional 1 operational efficiency and financial viability (see also percent of nonhousehold customers (for example, micro- chapter 1). When income-generating machines and appli- entrepreneurs and small businesses) served by a mini grid ances boost demand for mini grid electricity, a mini grid’s adds 20 percent to the mini grid’s total average monthly load factor gets a corresponding boost too. Meanwhile, a consumption in terms of kilowatt-hours sold. So, if a mini higher load factor (jumping from 22 percent to 40 percent) grid serves 1,000 connections, going from 10 nonhouse- cuts the cost of electricity by 27 percent (see chapter 1). hold customers (1 percent of total customers) to 50 non- On a site analyzed by CrossBoundary Lab, a grain mill oper- household customers (5 percent of total customers) would ator consumes 300 kilowatt-hours (kWh)/month, a hun- increase the mini grid’s average monthly consumption 114   MINI GRIDS FOR HALF A BILLION PEOPLE FIGURE 3.1 • The impact of productive electricity uses on the daily load profile and levelized cost of energy 100 LCOE = $0.23/kWh 80 Percentage of peak load LCOE = $0.28/kWh 60 40 20 LCOE = $0.38/kWh 0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 22% load factor 40% load factor 80% load factor Source: ESMAP analysis. kWh = kilowatt-hour; LCOE = levelized cost of energy; PUE = productive use of electricity. by 80 percent (5 percent minus 1 percent, times 20). This In addition, more than 130 income-generating appliances additionality seems to be fairly consistent until 15 percent have a payback period of less than 12 months, according of a mini grid’s customers are nonhouseholds, after which to an analysis that ESMAP conducted in Ethiopia. The point, the effect begins to dissipate. It is worth noting here up-front investment costs, power consumption, pay- that most of the mini grids (about 60 percent) surveyed had back periods, and revenue potential for some of these a customer base that consisted of 1–5 percent nonhouse- income-generating machines and other appliances are hold customers, indicating substantial room to increase the presented in table 3.4. Up-front investment costs typ- share of nonhousehold customers in the mini grids’ cus- ically ranged from $500 to $1,500, with an average of tomer base. about $1,200. These appliances and machines generate between $50 and $500 of revenue per month after the The combination of savings and reliability brought by conclusion of the payback period, with an average of the mini grid makes business sense for local entrepre- $300 per month across the range of appliances identified. neurs. Reliable electricity will reduce costs for businesses as power outages and unreliable supply detract from reve- Finally, mini grid electricity enables entrepreneurs to earn nues. In Sub-Saharan Africa, outages inflict sales losses of additional profits by extending the shelf life of goods, mak- 30 percent on businesses. In some of the region’s largest ing productive processes more efficient, increasing output, economies—Angola, Ghana, and Nigeria—more than 25 and improving access to information and markets (IEG percent of businesses lose more than 10 percent in sales 2008). In Tanzania’s Ludewa District, access to the 300 because of power outages, with individual firms reporting kilowatt (kW) hydro mini grid operated by Lumama cut losing more than 70 percent. The firms with the greatest milling costs in half (USAID 2018). challenges average more than 200 hours a month without Stimulating demand for electricity from productive activities power, while even the companies getting highly reliable ser- can, in particular, assist women-run enterprises to boost vice still report more than 10 hours a month without elec- their earnings through the use of lighting, electrical equip- tricity (Ramachandran, Shah, and Moss 2018). ment for cottage industries, baking, ceramics, and so on. MINI GRIDS FOR HALF A BILLION PEOPLE    115 TABLE 3.1 • The Mwenga hydro mini grid: Estimated costs and benefits Productive uses of electricity can assist women in notable ways through their higher Present value of cost or benefit earnings achieved through better lighting and appli- Type of benefit (US$, millions) ances for cottage industries, baking, ceramics, and so on. Development subsidies received by project –7.1 Household cost savings a 6.4 Tea company savings from reduced 1.4 diesel backupb Households and communities also benefit from produc- Jobs created by electrifying villagesc 8.6 tive uses of energy, which bring socioeconomic gains in Economic net present value 9.3 addition to better opportunities to sell goods and services. Source: Banerjee and others 2017. If these opportunities existed prior to the mini grid, they a. Monthly savings of $14 for 5,600 households. were constrained by the need to rely on expensive diesel b. Diesel backup requirement of 10 percent of total power consumption. generation. Food and farm-related goods and services— Assuming that 65 percent of businesses create 1.5 jobs each, and that c.  for example, cooling, drying, processing, and so on—are each job created is valued at the average expected annual salary of two affected areas. Communications and connectivity are $1,500 a year. another (internet points), along with mechanical power Note: The estimated peak load averages about 700 kilowatts (kW) with a summer peak of 90 kW and a winter peak of 400 kW, and annual power (woodworking and metalworking machines), as well as consumption of 2,880 megawatt-hours. lighting and entertainment. These goods and services lead to newly available and improved outputs, including: backup, and job creation from new electrified businesses— • Household and community well-being; are estimated at about $9 million (table 3.1) (Banerjee and • Longer life of, and added value to, agricultural products; others 2017). • Higher productivity; Increasing the productive use of energy (PUE) also has important benefits for women in the communities served • Better-quality manufactured goods, particularly in car- by mini grids. The physical and time burdens of some pro- pentry, upholstery and tailoring, and metalworking; ductive activities mainly run by women can be alleviated by • Reduced costs; and ensuring that power goes to shared community facilities • Service availability after dark. such as mills. Women’s labor is dominated by the drudg- ery of preparing grain for household consumption (called These outputs in turn lead to more jobs, higher incomes, “agro-processing” in the literature), particularly in Sub- time saved, and improved well-being (GIZ and others 2013). Saharan Africa. For a family to eat over four to five days, a The developer Mlinda has installed almost 100 mini grids woman (and her daughters) will spend up to 13 hours to in India. Load analysis and grid design have enabled pound enough maize. Time-use estimates obtained for Mlinda to power single- and three-phase electrical devices Nigeria show that two to three hours are spent each day targeting productive and residential end users. The three- just to prepare grains for pounding—that is, threshing and phase loads go to income generation. To support pro- milling. Eighty-two woman-hours are spent processing one ductive uses within communities and identify business drum of oil palm fruits. It takes two hours to grate a basin opportunities, Mlinda set up a team to assist local busi- of cassava; a grating machine can process a basin in one ness development. An impact assessment of the first 24 minute (Kes and Swaminathan 2006). operating mini grids showed that microenterprise reve- But it takes more than the introduction of electricity to nues rose 28 percent; 115 new local jobs have been cre- boost enterprise generally and women-led businesses in ated (Mlinda 2021). particular. Outreach and capacity building are also needed. An economic impact assessment offers more evidence. A SolarAid study of the solar lighting market in East Africa in Tanzania’s Mwenga hydro mini grid, with a capacity of 4 2012–15 found that 38 percent of households interviewed megawatts, was commissioned in 2012 and operated by reinvested their energy savings into agricultural production Rift Valley Energy. Mufindi Tea Estates and Coffee Ltd. is the or to seed other small enterprises (ODI and others 2016). main client and anchor load, requiring electricity mainly to The World Bank’s work in Mali reveals some of the chal- process tea leaves and power large motors, fans, and sieves. lenges in applying a gender lens to foster productive uses Over a 20-year project life cycle, economic benefits—from of rural electrification. household energy cost savings, reduced reliance on diesel 116   MINI GRIDS FOR HALF A BILLION PEOPLE The benefits of increased productive uses of Increasing the productive uses of a mini energy are especially profound for women, grid’s electricity presents an opportunity for who spend a disproportionate amount of their time everyone to win. Mini grid developers can grow their on farm labor (agro-processing chores), particularly revenues and lower their costs. Local businesses and in Sub-Saharan Africa. Transitioning from manual entrepreneurs can transition from expensive on-site labor to machine-assisted processing can boost diesel generation to less costly mini grid electricity, or productivity and save many hours of drudgery each develop new businesses that use electricity services week. But evidence suggests that productivity and to generate revenue. Local communities benefit from time savings do not automatically follow from the the creation of new jobs and greater economic activ- introduction of electric machines and appliances. ity. National utilities benefit from the growth of rural Investments in outreach and capacity building are economies and demand for electricity once intercon- also needed. nection to the main grid is considered. Increased PUE in mini grids can also enhance the eco- nomic viability of expanding the main grid. In the absence The most effective interventions to foster of PUE, most mini grid customers in low-income areas use productive uses of energy acknowledge that little electricity. So the main grid would sustain ongoing men and women occupy different spheres in the financial losses when it reached the mini grid’s service area. productive economy. Men and women also benefit Mini grids that stimulate demand for electricity through from electricity in different ways. Equitable interven- income-generating appliances and machines flip this tions are designed to overcome the gender-based narrative. They can provide economic growth in rural and barriers around productive use. They need to deliver peri-urban areas when they are designed to connect with communitywide welfare improvements and grow the main grid and when PUE has been promoted through the productive customer base—helping electricity community engagement and training. By the time the grid suppliers become financially sustainable. arrives, a substantial load already exists and customers are better able to pay. ROLLING OUT PROGRAMS TO PROMOTE PRODUCTIVE USES returns, and profitability; and the setting of tariffs condu- cive to productive-use appliances. Finally, financial guar- AND STIMULATE DEMAND antees and up-front financing can mitigate default risks Electricity demand does not rise automatically with the (expanded from RMI [2018]). arrival of a mini grid. The barriers to demand are numer- In addition, interventions should acknowledge that men ous, among them limited markets, information, lack of and women occupy different spheres in the productive skills, up-front costs, inefficient appliances, and scant economy. Yet measures to increase productive uses of access to financing. But efforts to promote PUE will pay off. energy tend to be gender blind and assume that men In Indonesia, for example, local nongovernmental organi- and women benefit from electricity in the same way. For zations (NGOs) promoted productive use at the outset of instance, women are less likely to be employed than men, a rural electrification program, quadrupling annual elec- more likely to run informal enterprises from their homes, tricity use from the main grid (World Bank 2000). and are overrepresented in low-productivity businesses, The adoption of electricity-powered productive equipment while men tend to engage in mechanized, electricity-inten- depends on demand, competition, and other sources of sive sectors such as construction, welding, manufacturing, power—like diesel gensets and manual labor. What are and repair. So, acknowledging the gender-based barriers the up-front costs of equipment? The “Diffusion of Innova- to productive use must be part of equitable interventions tions” theory (Rogers 2003) highlights the importance of that deliver communitywide welfare improvements and peer-to-peer conversations and peer networks to stimu- expand the productive customer base. Ultimately these late productive use. These include end-user education and will improve the electricity suppliers’ financial sustainabil- vocational training; manufacturers’ road shows for poten- ity. Community-based institutions such as self-help, sav- tial entrepreneurs; more end-user engagement to raise ings, and farmers associations can become platforms for awareness about what has worked, expected investment productive-use discussions. MINI GRIDS FOR HALF A BILLION PEOPLE    117 TABLE 3.2 • Six steps to roll out a PUE program Description Outputs Step 1 Market/demand assessment with geospatial analysis Online data platform overlying mini grids, appliances, and end-use finance List of key stakeholders List of high-impact opportunity areas Step 2 Community engagement confirming and improving List of communities validated as areas of high-impact opportunity, data from Step 1 through survey(s) and workshops combined with community-specific market data List of appliances that are relevant for these communities based on local contexts List of potential PUE customers in these communities List of local providers of microfinance in or near these communities List of local suppliers of appliances that serve these communities List of community leaders and district-level government officials who are supportive of the PUE program Step 3 Demand analysis for mini grid design and market Detailed characteristics of an initial set of community-relevant potential for appliances and associated end-user appliances finance List of prioritized appliances Step 4 Preparation of road shows involving local government, Road show logistics finalized: who, what, where, when, and how community leaders, interested appliance providers Information and marketing campaign launched ahead of the road and end-user financiers, mini grid companies shows Step 5 Road shows to load centers explaining the value Road shows propositions to potential end users by mini grid devel- Customer sign-up for mini grid connections, appliances, and end- opers, appliance suppliers, and end-user financiers user finance based on current and aspiration lifestyles of the end users; document sign-ups by end users for mini grid connections, appliances, and end-user finance Step 6 Rollout of mini grid connections, sales of appliances, Customers connected and end-user finance Appliances sold and connected Financing secured PUE = productive use of energy. This next section presents six steps to roll out initiatives • Building footprints, supporting the uptake of income-generating appliances in • Road and electricity networks, towns served by mini grids. • Nightlight imagery, STEP 1 • Crop cover and yields, Assessing markets and demand: Geospatial • Socioeconomic and demographic data (like gender analysis superimposed over mini grids, appliances, composition and household income), and finance for end users • Cell-phone coverage, A good rollout can accelerate uptake of productive uses • Topography, and of energy, especially when developers understand the sector’s market potential and engage the stakeholders. In • Precipitation. 2022, leveraging geospatial analysis is the state-of-the-art Then, to identify market opportunities, the machine- way to assess the productive-use market and its demand learning algorithms highlight possible high-impact collab- potential. This data informs collaborations among mini grid orations for mini grid developers, appliance suppliers, and developers, appliance suppliers, and end-user financiers. local finance providers. Exemplifying these high-impact opportunities are town clusters that lack electricity but Machine-learning algorithms now analyze geospatial data are located in highly productive agricultural regions with sets and produce maps that highlight areas suitable for PUE good cell-phone coverage and good roads to major cities programs. This technology-enabled, data-driven approach or trade hubs. But high-impact areas are everywhere: in to market assessment first collects a range of geotagged arid regions, coastal settlements, agricultural towns, and in data, including: rural and peri-urban locales. 118   MINI GRIDS FOR HALF A BILLION PEOPLE Once uploaded to an online platform, this geospatial data tially enabled can become a platform for partnerships and (and their maps) can be presented to governments and collaborations likely to increase sales of PUE appliances their PUE development partners, who can then share the and machines. information with mini grid developers, local microfinance The outputs of Step 1 are: (1) an online platform that pres- providers, and appliance suppliers. Stakeholders can then ents the data in a user-friendly interface; (2) a long list of overlay the opportunity areas with their own operations— key stakeholders, including mini grid developers, local pro- for developers, their mini grid sites (existing and planned); viders of microfinance, and appliances suppliers; and (3) a for microfinance institutions (MFIs), their branch network; long list of high-impact opportunity areas. for appliance suppliers, their distribution networks and hubs. In addition, developers, suppliers, and MFIs can be Working with TFE Energy’s Village Data Analytics, ESMAP matched when mutually beneficial opportunities are iden- is deploying this approach in the Democratic Republic of tified. For example, a high-impact opportunity might fall Congo, Ethiopia, and Nigeria. Table 3.3 presents some of within the expansion plans of a mini grid developer, close the PUE program stakeholders—appliance suppliers and to a local MFI branch and an appliance supplier. In this local finance providers. These lists will grow alongside the way, a market assessment that is data driven and geospa- ESMAP project. TABLE 3.3 • Example of PUE program stakeholders identified in the Democratic Republic of Congo, Ethiopia, and Nigeria Stakeholders Congo, Dem. Rep. Ethiopia Nigeria Appliance Agrimont Group “Tsehay Roschli Industrial Adebash Manufacturing Hanigha Nigeria suppliers Association of Cocoa and and Agricultural Engineering Company Ibraham Onsachi Coffee Exporters of the (Selam Children Village)” Alanco & Son Steel Fabricator Kenny Construction Company DRD (ASSECAF)” Afesol Technology Plc Alaral Tech Engineering Design Kola Adekunku AVSI Foundation Amio Engineering Plc & Fabrication Koolboks CONAPAC Beza Industry Plc Alayan Metals Fabrication Nig Koolmill Fouani Bumzal Amadis Technical Company Lawod Metal Nig Lushebere Dairy and DYD Trading Plc Apexskill Works Magi Rches Limited Cheese Factory Arcadem Ethio-Mercantile MCAN Makita DRC Basicon Engineering Company General Mercantile Plc Muharib Machine Strategos Plantations Bennie Agro ltd. Kaleb Service Farmer’s Muhat Nigeria The Breeders’ Society of House PLC Besuga Global Investment Bandundu (SEBO) N.C. Gilbert Ind Dev Co Kalmeks Engineering Bifem Technologies Nigeria The Society of Livestock Nanyang Goodway Machinery & Manufacturing Limited Farms of Bas-Congo (GEL Equipment Co., Ltd Lacomelza Plc Blessed Silver Brothers Bas-Congo) Niji Lukas Nig Marast Bomik Adeyeera Engineering Nova Technologies Roda Business Group Plc Camco Nui-Lukas Ten Tools Chinige Technology Services Oladimeji Success Tsion Industrial Engineering LTD Olaleye Eliseri Coldbox PAF Metal Fabrication and Youth Coldhubs Development Confidence Technical Work Peak Products Enterprise Pentawork Technical Work Dangote Process Concepts & Technologies Deban Faith Agro Allied Ventures S. Adiss Engineering Works DEE Technica Sakilan Engineering Company Doing Segun Towoju E. K. Fabricating Engineering Sominie Nigeria Limited Eamak Technical Services Starron Ecotutu—Interview Sunday Omowaye Emeka & Sons Construction Talitha Fabrication Company Company Teekay Tronics ESE Engineering Service Tropical Development Engineering Ltd Fatoroy Steel Industry UNIC & Sons Gensaes Enterprises Weilai Machinery Zheng Zhou Sida continued MINI GRIDS FOR HALF A BILLION PEOPLE    119 TABLE 3.3 • continued Stakeholders Congo, Dem. Rep. Ethiopia Nigeria Finance Advans Bank Addis Credit & Saving Justice Development and Peace Commission (JDPC) microfinance / providers Altech Institution (ADCSI) Integrated Development Programme Business MFI Agar Microfinance Addosser Microfinance Bank Credit YA MPA Amhara Credit and Savings Afex Institution (ACSI) Babban Gona Finca Benshangul MFI Baobab Microfinance Bank Hekima/Goma Buusaa Gonofaa Barnawa Microfinance Bank IFOD Dedebit Credit and Savings Chase Microfinance Bank Kitumaini Institution (DECSI) Light in Business SA/ Corebank / CORESTEP MICROFINANCE BANK Dire Microfinance Butembo Development Exchange Center (DEC) Microfinance Metemamen Microfinance  Mam Tombuama GRASSROOTS DEVELOPMENT MICROFINANCE BANK Omo Microfinance MFI APE Ibile Microfinance Bank Oromiya Credit and Savings Micropop Life Above Poverty Organization (LAPO) microfinance Share Company (Ocssco) Paderu Nirsal Microfinance Bank Poverty Eradication and Paidek SA Community Empowerment Richway Microfinance Bank Procfin (PEACE) SMICO S.A./Goma Sidama Tout Pour le Genre Dans Somali Microfinance Le Developpement (TGD) Specialized Financial and Trust Investment Promotional Institution Development TID SA / (SFPI) Butembo Vision Fund Microfinance Tujenge Pamoja/Goma (VFMFI) Tujenge S.A. Wasasa Microfinance Vision Fund Yoasi Source: ESMAP and TFE-VIDA analysis. STEP 2 assessments of productive demand. The systematic Surveys and workshops build on Step 1 data approach maps all potential activities in a community that through community engagement could benefit from access to electricity. Every stage of the existing and potential value chains—from inputs and pro- Once the high-impact opportunity areas are identified, plan- cessing to outputs and end uses—is screened to capture ners need to “ground truth” by engaging with communities. actors, market dynamics, cycles, and seasonality. This This engagement entails multiple rounds of site visits to approach assesses the role electricity already plays and identify the context-specific potential for PUE. Each succes- could play in the prioritized sectors. sive round would home-in on top-priority locations, appli- ances, and stakeholders. In practice, this means sending The pragmatic approach, by way of contrast, does not look teams of surveyors and community engagement special- at the spectrum of productive uses. Instead, it takes advan- ists into the high-impact areas identified in Step 1. Equipped tage of existing opportunities to ease the interface between with smartphones and tablets, the team will collect data and mini grid developers and productive sectors. The process take photos and videos, obtaining a deeper understanding of identifying PUE is speeding up, so moving on to imple- of local socioeconomic dynamics, critical for effective inter- mentation takes less time. Although less comprehensive, ventions that promote productive uses and build a support- the pragmatic approach is faster and more affordable and ive ecosystem. If the mini grid developer leads this effort, builds on extant cross-sector ties. it could gain support from the rural electrification agency, After gathering information on appliance usage (current NGOs, and development partners in facilitated interaction and potential), teams could tell productive users they may with productive sectors and local businesses. sign up for mini grid electricity at a later site visit. Further- Two complementary methods—systematic and prag- more, survey teams could also approach local suppliers of matic—exist to identify opportunities for productive uses microfinance and appliances to gauge their interest in a at the community level (de Gouvello and Durix 2008). PUE program. This might require visits to MFI branches and Field investigation is, in both cases, a requirement to refine appliance or hardware shops in nearby towns to interview 120   MINI GRIDS FOR HALF A BILLION PEOPLE Identifying context-specific potential for At the community level, demand assess- productive uses should begin as early ment should also look into the gender gaps as possible in the mini grid development project in entrepreneurial activity. What is constraining the cycle. Two complementary methods—systematic ability of women-owned enterprises to thrive? and pragmatic—exist to identify opportunities for productive uses. The systematic approach maps activities that might benefit from access to electric- ity. The pragmatic approach appraises a sector (or STEP3 location) first, and then identifies economic activi- Demand analysis for mini grid design and market ties that are improvable through mini grid access. potential for appliances and associated end-user finance During the early stages of the PUE program, a team’s direct engagement with communities is important in identifying staff. Finally, the community engagement teams will need high-priority appliances and machines. Once these appli- to meet with community leaders and district-level govern- ances have been identified, their economic and technical ment officials to introduce them to, and garner their sup- attributes should inform mini grid design, so it may more port for, the PUE program. precisely ascertain the market potential for appliance sup- pliers and local providers of microfinance. Demand assessment should also assess the gender gaps that women and men face as they start up enterprises in The range of energy-efficient appliances on the market an effort to identify the chief constraints on women-owned today is continuing to expand. Table 3.4 presents indicative enterprises. For example, in the agriculture sector, men information on power requirements, costs, and payback often mediate women’s access to resources and commu- periods for a sample set of widely distributed income-gen- nity participation, and these men tend to be fathers or hus- erating appliances and machines. Of the more than 160 bands. So the agricultural contributions that women make appliances that ESMAP identified as being available on the often go unrecognized. market today, more than 130 have a payback period of less than 12 months. The typical up-front investment ranged Gender gaps are apparent everywhere, of course. But in from $50 to $1,500, with an average of about $1,200. At Nigeria, the gender productivity gap for agriculture stands the conclusion of the payback period, the appliances gen- at 18.6 percent. Another notable gap is found in Dominica, erate between $50 and $500 in monthly revenues, with an a Caribbean island state, where women work mostly in average of $300. the informal market in microenterprises and subsistence farming, which constrains their ability to advance (Mukasa Once the technical and economic information about pro- and Salami 2016; ILO 2014). Gender gaps are also driven ductive-use appliances is better understood, potential cus- by the cultivation of smaller land parcels, discriminatory tomers will need to decide if acquiring the appliance is right land laws, constrained access to financial products and for them. CrossBoundary’s Mini Grid Innovation Lab has services, snubbed by extension or business development shared some of the questions that customers can ask to services, and restricted to older technologies. help them make this decision. These include: Step 2 outputs are lists of: • What level of electricity service—primarily in terms of capacity and reliability—is required to produce the final • Communities validated as areas of high-impact oppor- product desired, and is the mini grid capable of provid- tunity, combined with community-specific market data; ing this level of service? • Appliances relevant in terms of the communities’ local • How will converting from diesel or using an energy-ef- context; ficient appliance affect the technical aspects of the • Potential PUE customers in these communities; machine? • Local providers of microfinance in or near these com- • What tariff will make converting to electric power from munities; diesel cost-effective? • Local suppliers of appliances that serve these commu- • What time of consumption (across the day or night) will nities; and yield the greatest profits? • Community leaders and district-level government offi- • Would it be more profitable for me to be located closer cials who are supportive of the PUE program. to the generation source? MINI GRIDS FOR HALF A BILLION PEOPLE    121 TABLE 3.4 • Power requirements, costs, and indicative payback periods of selected income-generating appliances Average monthly Power required revenue after (kW unless speci- Cost from Payback period payback period Sector Activities and appliances fied otherwise) supplier (US$) (months) (US$) Primary Crop dryer 6–1 1,000–5,000 12–18 236 industries Egg incubator 80–16 watts (W) 50–100 1–3 58 (agriculture, Hammer mills of various types for the 5–10 700–1,200 6–12 129 fishing) respective grains Electric sawmill 1.5–3.0 500–800 12–24 44 Threshing machine 7–9 500–1,000 3–6 208 Grinder for pulses and beans 5.2 1,500–4,000 6–12 396 Garlic/ginger paste machine 2–7 1,500–4,000 6–12 396 Tomato ketchup/paste–making machine 5 800–2,500 3–6 483 Water irrigation pump 3.7–22.4 200–1,000 3–6 183 Oil expeller 6 (600 W for 800–2,000 for 3–6 400 ordinary oil press) ordinary oil press Sterilizer (for dairy processing) 3–6 600–2,000 1–3 1,100 Packager 250 W–3 kW 500–1,000 6–12 104 Peanut roaster 1–10 1,000–5,000 6–12 458 Electric crusher for peanut paster 3 1,000–3,000 3–6 583 Automatic measurer/bagger 1 3,000–5,000 12–18 292 Light Electronic welding machines 3–7.5 200–300 6–12 33 manufacturing Angular grinder 2 100–200 6–12 21 Circular saw for wood 1–2 150–200 6–12 23 Electric smoothing plane 200–300 W 50–100 3–6 21 Jigsaw 400 W 100 3–6 25 Electric drilling machine 400 W 20–50 3–6 10 Ice maker 4–6 500–1,500 6–12 146 Socks knitter 4 2,000–5,000 6–12 500 Biscuit maker 2 3,500–6,500 6–12 688 C-brick maker 9 2,000–5,000 3–6 1,000 PET bottle maker 24 3,000–10,000 3–6 1,917 Popcorn maker 1.5–2.1 50 1–3 33 Commercial and Phone charger 5–10 W 10 1–3 7 retail activities Refrigerator for cold drinks in cafeterias 100 W 150–300 1–3 175 or medicines in dispensaries Computer 15–100 W 250–800 3–6 154 Printer/scanner for stationary 0.5–2 150–250 3–6 54 Appliances for hairdressers’ shops 1.5–2.5 (hair 40–60 (hair 1–3 19 (hair clipper, hair dryer) dryer) dryer) 10–20 W (hair 15–30 (hair clipper) clipper) Sewing machine 200 W 30–100 3–6 17 Television for local cinemas and bars 50–200 W 100–200 1–3 46 (including decoder) Hi-fi stereo system 20–500 W 20–200 A hi-fi system n.a. in itself does not provide a payback period Electric cookstove 3.5 50–100 Does not have a n.a. payback period due to high consumption Deep fryer 2 100–500 6–12 46 Sources: de Gouvello and Durix 2008; Alibaba 2022; ESMAP and INENSUS analysis. Note: Battery chargers come in a range of sizes. The one presented here would charge from 50 percent to 100 percent in about three hours. The ice maker can produce 1,000 kilograms of ice per day. kW = kilowatts; W = watts; n.a. = not applicable. 122   MINI GRIDS FOR HALF A BILLION PEOPLE is important to enable developers to effectively promote their uptake among customers, design their tariffs appro- Of the 160 income-generating machines and priately, provide financing options to end users, and design appliances on the market today, more than their mini grids. 130 have a payback period of less than 12 months. The typical up-front investment cost ranged from The different productive-use appliances come with a range $50 to $1,500, with an average of about $1,200. The of load profiles, and developers can design their mini grids monthly revenues generated by these appliances to prevent under- or oversizing the system to strengthen its after the payback period typically ranged from $50 economic viability. Undersizing could restrict revenues for to $500, with an average of $300 across the identi- the mini grid operator and push consumers toward alterna- fied appliances. tive sources of energy. Excess capacity raises investment costs above revenues, lengthening the payback period, hiking operational costs, and reducing overall efficiency. Because prediction of demand tends to be unreliable, over- For mini grid developers, adding productive users to their sizing is common for mini grids in Africa. customer base adds complexity to the project design, because they have to determine whether and how to The key elements of system design include: connect these loads, which differ in terms of time of use, • Peak power, magnitude of power and energy demand, and seasonality. As such, mini grid developers will also need answers to a • Reactive power, series of questions on how to include them in their busi- • Single- or three-phase distribution networks, ness models as early as possible in the project cycle. Cross- • Capacity utilization, and Boundary’s work with developers has helped identify some of these questions, including: • Incorporation of backup generators versus batteries. Three-phase distribution systems are able to power large • How much additional generating capacity is needed to productive-use loads across a wider range of machinery, support the load demand? especially equipment above 10 kW or 10 kilovolts-ampere, • What inverter size and distribution system will allow or rural clusters of smaller appliances in a productive facility multiple productive-use machines to operate simulta- at an isolated site. Productive users usually require a min- neously? imum service at Tier 3, or peak available capacity of more • From how far away can the grid support a large produc- than 200 W and eight-plus hours of energy supply, including tive load running on three-phase power? at least two hours in the evening (Bhatia and Angelou 2015). • How low can the tariff be while still proving sustainable? Mini grid developers generally have two ways to manage electricity demand from PUE appliances and machines: • What time of consumption will allow least-cost genera- tariff incentives and contractual obligations. tion? Tariff incentives can stimulate demand and greatly boost • How should the tariff structure be adjusted to account the use of income-generating appliances and machines. for seasonality? In regulatory environments where developers are free to Adding electric appliances to the systems is often not tech- charge cost-recovery tariffs (see chapter 9 for a detailed nically straightforward and requires demand-side manage- discussion on mini grid tariffs), the guiding principle of set- ment skills. Income-generating appliances typically require ting an appropriate tariff should be to keep it easily man- more power and energy to operate than consumer appli- ageable for the operator and easily understandable for end ances (fans or televisions). Some types of income-gener- users. The tariff structure should also allow a reasonable ating appliances are already in use in the community but return on investment while remaining attractive and afford- are powered by diesel engines and may not be the most able for productive end users. It should take into account efficient options to maximize limited load profiles. Others the ability and willingness to pay of productive clients as will be introduced into the community for the first time. well as the availability of alternative power sources or back- As a result, mini grid developers often lack information on ups, the predictability of supply, and the possibility of com- the power and energy requirements and load profiles of bining different sources of electricity. appliances and machines, which could lead to inaccurate demand forecasts and wrong-sizing the mini grid’s gener- Tariff structures with a range of applications exist, including ation and storage capacity. Accurate information on the tariffs that are: technical characteristics of income-generating appliances • Capacity-based, MINI GRIDS FOR HALF A BILLION PEOPLE    123 • Inverted block, • Per device, Demand and load management are essen- • Time-of-use, and tial for productive users, who depend on Tier 3+ service. Two strategies are available to man- • Seasonal. age demand: tariff incentives and contractual obli- When tariffs are tailored to end users, rates can be adjusted gations. Tariffs can incentivize productive uses. One by load location and size and by type of connection and common approach is to charge low daytime energy business. The mini grid operator can also decide to intro- tariffs but, during peak morning and evening hours, duce lower, curtailable load tariffs for customers who agree to charge higher energy tariffs. Other approaches to be curtailed. include allowing productive users to postpay for their electricity consumption, while waiving stand- In Tanzania, Jumeme and Energy 4 Impact designed a tariff ing charges for productive users with seasonal busi- schedule aligned with end users’ needs after collecting data ness activities. Operators can also encourage or on businesses’ energy consumption. They determined that, require a shift of some productive consumption to to be able to compete with diesel gensets, the mini grid tar- daytime or other nonpeak periods, through service iff to power a maize mill should be about $0.32/kWh. This agreement contracts. rate was much lower than what mini grid developers usu- ally charged in Sub-Saharan Africa ($0.50–$1.20/kWh). Based on an incentivized use pattern, daytime tariffs were set at about a quarter of evening tariffs. Ultimately, one of the key outputs of Step 3 is to iden- tify a set of high-priority appliances that appeal to both Payment terms should also be adjusted to reflect produc- mini grid developers and customers. For developers, tive users’ constraints. Some users may be reluctant to such appliances tend to be energy hungry at predictable pay a connection charge (or a flat standing charge) when daytime hours (or that run consistently over 24 hours). their activities are highly seasonal, with no electricity needs For customers, high-priority appliances have the short- during certain periods of the year. Prepayment can also be est payback periods while generating income over time. problematic for productive users with limited cash flows, Step 3 outputs also document community-relevant appli- especially early in their development. As a result, two ways ances—knowledge that informs the go-to market strate- to accommodate PUE customers are, (1) waiving stand- gies of mini grid developers, local microfinance providers, ing charges and (2) allowing productive-use customers to and appliance suppliers. postpay for their electricity. Operators can also use contracts to shift some productive STEP 4 consumption to daytime or other nonpeak periods. The Preparation of road shows involving local Tanzanian villages of Lupande, Mawengi, and Madunda are government, leadership communities, connected to a 300 kW hydropower mini grid. Corn milling communities, interested appliance providers and and welding are permitted only during business hours (9 end-user financiers, mini grid companies a.m. to 6 p.m.) so households have enough electricity at Steps 1–3 focus on gathering data and information that will night (USAID 2018). inform Steps 4–6, which are focused on implementing road shows and other community engagement activities and Contracts can make demand-side management more ultimately deploying income-generating appliances and efficient. They mitigate demand risk and anticipate the machines in partnership with mini grid developers, local effects both on capital expenditures and the levelized cost finance providers, and appliance suppliers. of electricity. In doing so, contracts define the tariff struc- ture, which determines what productive users should be Step 4 consists of preparing the logistics for road shows connected when, and at what tariff. Beyond collaborations and other community engagement events. Ever since elec- between mini grid operators and productive users—and tricity systems were first installed outside major cities—for in order to better secure the balance between demand example, in the 1920s and 1930s in the United States— and supply—developers can secure a guaranteed level of electricity providers, governments, and other community demand through contractual agreements with customers, development organizations have organized road shows, which increase load predictability and therefore revenue competitions, and other events that introduce communi- streams. Signing up the load before the final system design ties to electrical appliances and machines (Duke University improves the ability to right-size the system and gives the 2022). Demonstrations were key. How do electrical appli- end user the option of securing the amount and quality of ances and machines work? Productive users of electricity power it desires. share their experiences with prospective users, and explain 124   MINI GRIDS FOR HALF A BILLION PEOPLE the availability and terms of appliance financing, describing Information and marketing campaigns should preview the mini grid developer’s plans (timing, tariffs, and so on). the road shows, which are integral to Step 4. Notices on Finally, potential customers sign up for mini grid electricity, local radio, posters, leaflets, and other materials are vital appliance ownership, and consumer financing. Road shows, for community presentations. Centralized resource cen- supported by information and marketing campaigns, have ters or NGOs can handle this part, and they can gather proved their effectiveness for more than a century, increas- information on productive equipment and techniques, ing uptake of income-generating machines and appliances. advising on business capacity and productive processes. Analyzing these rural road shows in the United States Campaigns should disseminate high-quality information between 1938 and 1945, Duke University found that they materials, messages, and methods that focus on market increased consumption by 64 kWh per customer per year surveys and targeting. (Plutshak, Free, and Fetter 2020). The outputs of Step 4 are finalized logistics (who, what, But these events require careful planning and coordination where, when, and how) for road-show marketing and infor- across a diverse mix of stakeholders. In fact, six core stake- mation campaigns that target the high-impact opportu- holder groups that will need to come together are: nity communities identified in Steps 1 and 2. These steps demonstrate the first set of machines and appliances iden- • Current and potential users of income-generating tified and analyzed in Steps 2 and 3. appliances, • Mini grid developers, STEP 5 Road shows to load centers where mini grid • Local providers of consumer finance, developers, appliance suppliers, and end user • Appliance suppliers, financiers explain the value propositions to • Community leaders, and potential end users based on their current and aspirational living standards • District government leaders. With the logistics in place and the marketing and infor- All of these stakeholders should have been identified at mation campaigns implemented, the road shows can be Step 2—the task during Step 4 is to coordinate collec- deployed to the high-impact opportunity communities. tive action to plan the road shows and other community Road shows should be fun, informative, safe, and memo- engagement events. Additional stakeholders may also rable events. Indeed, the road shows implemented by the need to participate depending on the context, such as agri- Rural Electrification Administration of the United States culture agencies and local training institutes. were affectionately referred to by participants as the “Elec- The high-impact opportunity communities identified in tric Circus” (Duke University 2022). Step 1 and validated in Step 2 are the geographic targets for District-level government officials often serve as strong road shows. Events should take place in areas accessible voices of support to open the event. A successful road show to several different high-impact opportunity communities. will offer: The geospatial data collected in Step 1 should guide road- show activities, taking into account access, community • Highly interactive and participatory environments that clusters, and other relevant information. engage the mini grid developer, local authorities, local community organizations, financing agencies, vendors For road shows, good rules of thumb (de Gouvello and of electrical equipment, and potential customers; Durix 2008) are: • Events or demonstrations that target women. Because • Integrate initiatives from other sectors like agriculture to women are less likely to be employed than men, and avoid duplication; more likely than men to run informal businesses, they • Anticipate the effects of PUE on people, which is to say, are less familiar with mechanized work. They need to be aware of how increased use changes household and see the benefits of the productive uses of appliances. social behaviors; In addition to the road shows themselves, key outputs of • Favor local initiatives and equipment; Step 5 are concluding agreements with potential customers for the suite of products and services so they can acquire • Tailor activities to targeted sectors and regions and and use income-generating machines and appliances. Step adapt them to the literacy levels and habits of the tar- 5 should therefore conclude with signing customers up for geted audiences; mini grid electricity; for an appliance; and for consumer • Design all activities to include women, and develop finance (if necessary and desired by the customer). activities that specifically target women. MINI GRIDS FOR HALF A BILLION PEOPLE    125 TABLE 3.5 • Stakeholders that could be involved in road shows and their respective roles Stakeholders Roles Local distributors of targeted equipment, local Demonstrate equipment cottage industries Provide technical adaptation to local uses Train users Develop retail network to facilitate local purchase of appliances Energy service providers and their Build consumer awareness and support; help providers adhere to standards associations Educate and train customers Adapt energy infrastructure to local growth of energy demand Microfinance and credit unions Provide customized financing to potential users Conduct risk analysis Build sector awareness Educate customers about loans Agriculture and other institutes, enterprises, Provide sector knowledge, field presence, networking and outreach capacities, and promotion centers demonstration capacity, and vocational training Build user confidence Train users Training and vocational centers Develop vocational curriculum and train teachers Source: De Gouvello and Durix 2008. Clustering productive users in an area close to the source of generation is one way to increase the connectivity of Road shows, combined with information and businesses. There are different ways to encourage clus- marketing campaigns, are effective ways to tering, including multifunction platforms, solar kiosks, increase the uptake of income-generating appli- productive-use centers have been experimented. Beyond ances and machines. They require careful planning the technical and financial advantages to the developer and collective action by six core stakeholder groups: (higher-quality service, limited distribution investment), current and potential users of income-generating clustering facilitates day-to-day interactions, strengthens appliances, mini grid developers, local providers of innovation and business development, and increases tech- consumer finance, appliance suppliers, commu- nology and knowledge sharing within the productive com- nity leaders, and district government leaders. Road munity. The role of public and central authorities, assisted shows should be preceded by information and mar- by multilateral partners and donors, is key in clustering keting campaigns, should be highly interactive and small business activities by enabling land tenure, facilitat- participatory, and should actively target women, ing permitting, and providing infrastructure (roads). (See who are less likely to be employed than men, more chapter 8 for a discussion of the different institutions that likely to run informal businesses than men, and less are relevant for mini grids.) likely to engage in mechanized work than men. In addition, to unlock the entrepreneurial potential of communities, local authorities, NGOs, or the developers themselves need to provide gender-specific business STEP 6 development services and mentoring activities for targeted Rollout of mini grid connections, sales of local entrepreneurs. A study that covered Tanzania, Ghana, appliances, and end-user finance and Myanmar, highlighted differences found between Upon completion of a road show, the mini grid developers, enterprises run by men and women and the need for gen- appliance suppliers, and local finance organizations can der-specific support (IDS and GIZ 2019). Men own more begin to deploy their products and services to the custom- businesses and spend more on electricity than women, ers they signed up at Step 5. It is rarely easy, however, for who spend more on cooking fuels. Women operate in less microentrepreneurs and local small businesses to make electricity-intensive sectors that are mainly devoted to food the leap from sign-up to acquisition. Two strategies are preparation, hospitality, tailoring, hairdressing, and retail, emerging to directly support local entrepreneurial uptake while men are familiar with mechanized and electricity- of income-generating machines and appliances as they intensive work. This gap between women and men and arrive in the community: clustering, and advisory support. their respective patterns of productive energy use is eas- 126   MINI GRIDS FOR HALF A BILLION PEOPLE Two strategies to improve small business Most productive-use appliances and equip- uptake of mini grid electricity services are ment have relatively high up-front costs clustering productive users in an area close to the compared with the disposable-income levels of source of generation and providing gender-specific prospective entrepreneurs and small businesses, advisory support to entrepreneurs and small busi- but they provide clear opportunities to generate ness managers. or increase revenue. Financing the up-front costs of the appliance—whether from a mini grid opera- tor or a third party—will be necessary to increase productive uses of mini grid electricity. ESMAP estimates that approximately $3.6 billion in micro- Women often operate in jobs related to food finance for 3 million income-generating appliances preparation, hospitality, tailoring, hairdress- is needed under the scenario in which Sustainable ing, and retail, which tend to be less electricity inten- Development Goal 7 is achieved, in part through the sive. Meanwhile, men take jobs that tend to be more development of 200,000 new mini grids. mechanized and electricity intensive. They also enjoy better starting conditions in terms of capital, resources, and skills. By way of contrast, women 2018). Appliance financing schemes are therefore needed are constrained by household responsibilities and for end users to overcome up-front costs of appliances restrictive social norms. Nevertheless, access to through either the mini grid developer’s “own-managed” electricity can generate income for both men and financing facility or mechanisms managed by third par- women who own and run enterprises, highlighting ties, such as financing agencies, MFIs, or equipment sell- the need for gender-specific advisory support to ers (NREL and E4I 2018). boost the PUE. Financing by third parties Involving third parties as asset-financing companies lets ily explained: men start their working lives with more capi- developers remain focused on their primary activity—lever- tal, resources, and skills, while women take on demanding aging technology, competency, and capacity (Factor[e] household-care responsibilities, while social norms restrict Ventures 2020). Partnerships with pay-as-you-go com- their participation in certain occupations, such as fishing. panies for appliance financing services have been tested. Yet access to electricity led to increased profits for both Asset-financing companies such as Rent-to-Own in Zambia men and women, thus highlighting the need for targeted, and EnerGrow in Uganda are emerging with a specific focus gender-specific advisory support to increase overall PUE. on financing appliances for on- and off-grid customers. Consumer financing will need to arrive alongside mini Third parties, however, should tackle the issue of financial grid electricity and income-generating appliances and inclusion with end users, many of whom operate outside machines. The market potential is substantial. If mini grids the formal financial system and often lack collateral. They are to reach their full potential—serving half a billion people often face high costs and short repayment periods for loans by 2030 is a core solution for achieving universal access from the commercial banking sector in remote areas, espe- to electricity—then $3.6 billion in microfinance is needed cially when they want to develop new business activities. for 3 million income-generating appliances, assuming an Women are particularly burdened by the lack of collateral average of 15 productive-use appliances per mini grid for and are less likely to have a bank account.1 Many coun- 200,000 new mini grids at an average cost of $1,200 per tries continue to have laws that restrict women’s access appliance (see table 3.4). to inheritance and land titling, which hinders their ability to access assets that can be used as collateral when securing Expensive electrical appliances and high connection fees a loan (World Bank 2020). can impose a financial burden on entrepreneurs and small businesses, slowing or preventing the decision to invest Third parties that finance productive uses include MFIs, in productive-use appliances and equipment, especially community savings groups, rural electrification agencies, when no formal financing institution exists. A study from and ad hoc structures locally established by development the Rocky Mountain Institute showed that when end users partners, NGOs, and other local entities. As deposit-taking were offered 12-month loan terms to finance produc- institutions targeting low-income people and microenter- tive-use appliances, consumption almost doubled. After prises in developing countries, MFIs are well established 11 months, mini grid revenues grew by 18 percent (RMI and offer a variety of financing products. They also ben- MINI GRIDS FOR HALF A BILLION PEOPLE    127 Offering financing directly to productive users enables the developer to better monitor its strategy toward productive When trying to secure finance to purchase clients, including whom to connect and what appliances productive-use appliances and equipment, they choose. Controlling the type of appliances connected women are particularly burdened by the lack of col- to the mini grid and favoring high-quality equipment may lateral and are less likely to have a bank account. facilitate system management and result in better opera- Many countries continue to have laws that restrict tional efficiencies. As an example, a lease-to-own model (or women’s access to inheritance and land titling, on-bill financing) could enable end users to pay 30 percent which hinders their ability to access assets that can of the purchase price up front, paying the rest to the mini be used as collateral when securing a loan. grid operator month by month for a certain period through the electricity bill (NREL and E4I 2018). This mechanism helps end users, who pay less interest to the operator than efit from existing customer bases and loan distribution they would to a commercial lender. networks. Not yet seen are equity investments by impact In Tanzania, Jumeme, a private operator, has run a 90 kW investors in productive-use technologies. solar mini grid in Bwisya on Lake Victoria since 2016. Its Development partners can support access to credit for presence has allowed for the automation and expansion of equipment by increasing the awareness of financing agen- existing businesses (grain milling, carpentry, and bicycle cies or equipment sellers regarding nontraditional proxies repair) and supported the emergence of new businesses for creditworthiness and new approaches to assessing con- (egg incubation, ice block production, and metal welding). sumer risk (Cheney 2016). Lenders can use new sources of The company runs a shop selling appliances in the larg- data—such as records of timeliness of phone bill payments, est village. It has helped Jumeme avoid technical issues social network data, and mobile phone use—to determine and control consumption by ensuring the use of effective the ability and willingness to repay of unbanked users (Baer, and adapted equipment. Small and medium enterprises Tony, and Schiff 2013). Another option is to work through acquire appliances on credit (usually for about six months) local governance structures. provided by the mini grid operator. About 20 businesses have received appliances through in-house financing Financing by the mini grid developers (USAID 2018). In some cases, mini grid developers offer financing to their This model may impose a financial burden on the mini grid customers to support the acquisition of income-generating developers’ balance sheets, however, and divert the devel- appliances and machines. One approach is to offer custom- oper from its core business. It implies managing software ers a fee-for-service model. In one example of this model, systems for managing appliance loans and “pay-as-you-go” the mini grid company provides a productive hub and offers lockout systems in parallel with maintaining a billing sys- the use of income-generating appliances against the pay- tem (Factor[e] Ventures 2020). Mini grid developers do not ment of a fee. necessarily have the skills or expertise to handle in-house Another example comes from a deployment of 600 financing. solar mills in Indonesia, Vanuatu, and Papua New Guinea Once operational, mini grid developers should monitor funded by the United States Agency for International users for one to two years after their connections. Moni- Development and implemented by Village Infrastructure toring will enrich their knowledge of the demand dynamics Angels and Sumba Sustainable Solutions. Customers from productive-use appliances and help them address could trade goods and services (that is, share the produc- customer issues as—or even before—they arise. By mon- tion increase through sellable products made possible itoring demand, Vulcan Impact Investing (which owns by time savings) for the use of income-generating appli- about 10 mini grids in rural Kenya) found that the average ances. This approach has demonstrated positive results, revenue per user generated from the 10 percent of its cli- guiding end users through time savings gained from the ents that are small businesses was five times greater than mills and longer productive hours gained through night- the revenue generated from the other 90 percent. Even time lighting to make tradeable products. The approach though most customers consumed less than 250 watt- proved particularly efficient and robust during the COVID- hours a day, they were still critical to mitigating the risk of 19 lockdown, when cash in the communities was tight and losing larger clients (Blodgett and others 2017). village agents who disbursed noncash payments fared better than those dealing only in cash (Village Infrastruc- ture Angels 2019). 128   MINI GRIDS FOR HALF A BILLION PEOPLE other entities to take a coleading role in organizing the PUE program alongside the mini grid developers– notably, gov- Monitoring productive users for one to two ernments and local change agents. The following sections years after their connection to the mini grid describe how these two stakeholder groups can organize, can enrich a developer’s knowledge of the demand or help organize, PUE programs. dynamics from productive-use appliances and enable the developer to quickly address customer GOVERNMENT AGENCIES AND POLICY MAKERS issues as—or even before—they arise. Government agencies such as rural electrification author- ities or agricultural extension programs hold a strategic position in fostering productive use demand because of their national-level influence. They can serve as a coordinat- WHO ORGANIZES PUE PROGRAMS? ing agency to supervise and facilitate collaboration among stakeholders involved in promoting PUE (such as NGOs, MINI GRID DEVELOPERS developers, local communities, and equipment suppliers). As the providers of electricity to the end users of Operational support from rural electrification agencies income-generating machines and appliances, mini grid can include designing and implementing comprehensive developers are one of the main entities that can lead a PUE approaches that enhance the PUE in agricultural, indus- program. In particular, they can take a lead role in all six trial, and service sectors, for example, by enhancing the steps outlined above. In Step 1, they can inform the long knowledge and skills of small and microbusinesses on list of high-impact opportunity communities by directing how to use their newfound electrical and motive power analysis toward communities that they are planning to for profitable enterprise. Additional enabling interventions serve. Similarly, they can use the data collected in Step 1 could include, for example, strengthening the technical and to inform their expansion plans. In Step 2, mini grid devel- financial management capacity of women’s enterprises, opers will be engaging with communities already, as part expanding access to markets, creating linkages and access of their community outreach activities, so they can play a to financial products and services, enhancing extension lead role in identifying prospective PUE customers, local or business development services, and possibly address- community leaders, local finance providers, appliance sup- ing discriminatory land laws. As an interface between mini pliers, and even district-level government officials. Mini grid grid developers and other stakeholders, rural electrification developers also have the technical knowledge to assess the agencies can pursue these activities through the signature technical and economic impacts of connecting different of individual memoranda of understanding with developers appliances and machines to their networks, and thus are in to formalize their collaboration and precisely define roles a good position to lead on Step 3. This can also help ensure and responsibilities. A deep understanding of local socio- that the PUE program targets appliances that are attractive economic dynamics is critical, however, to ensure these to the mini grid developer. For Steps 4 and 5, the developer interventions succeed. can also take a leading position in organizing and imple- Partnering with other stakeholders enables rural elec- menting the road shows, although logistical and marketing trification agencies to improve such understanding and support from local, regional, or national governments can build their capacity on the topic of productive uses. Part- facilitate large-scale rollout of road shows. Lastly, the mini nerships could include engagement with in-country sec- grid developer is one of the three main actors in Step 6, and toral associations, microbusiness support entities, aid can coordinate the on-the-ground activities of appliance agencies and donors, governments, NGOs, private-sector suppliers and local finance entities after a road show. firms, and researchers. For example, a rural agency could Even if the mini grid developer takes the lead in organiz- contract directly with competitively selected local NGOs ing the PUE program, they will still need the support of the to assess the market and identify and promote activities other stakeholders—indeed, it requires collective action in close collaboration with developers. These contracts toward a common goal. Orchestrating a PUE program, par- can be structured in two phases. The first phase includes ticularly on a large scale, is a resource-intensive activity, conducting surveys to identify and assess productive requiring not just money but also capabilities, networks, potential and building the interface with other sectors and legwork. As a result, the overarching recommendation or programs. During the second phase, the NGO designs is that mini grid developers can be the driving force behind and launches marketing and promotional campaigns to a PUE program, but the program itself will need resources build capacity and awareness among entrepreneurs. The (time, money, and people) from a variety of stakeholder contract with the NGO would define targets, objectives, groups. And, in some cases, it might also make sense for and implementation frameworks to measure the NGO’s MINI GRIDS FOR HALF A BILLION PEOPLE    129 Government agencies like rural electrifi- Coordinated, concerted efforts to promote cation agencies and agricultural extension productive uses of electricity have been programs hold a strategic position with regard to quite successful, in one case (Bangladesh) increas- the productive use of energy, able to design and ing customer uptake by almost 500 percent. implement comprehensive, multistakeholder pro- ductive use in agricultural, industrial, and service sectors. These government-supported initiatives development programs, microfinance organizations, appli- can also impart knowledge and skills to micro and ance companies, energy service companies, municipali- small business, including, for example, how to use ties, regional/district officials, and local associations are all newfound electrical and motive power for profitable important stakeholders. Coordinating all of them requires enterprise, strengthen the management capacity of institutional support, which could be achieved through a women’s enterprises, improve access to markets, platform that facilitates dialogue. create linkages and access to financial products and services, and enhance services. TIMING PRODUCTIVE USE performance against indicators such as the amount of PROGRAMS FOR MAXIMUM EFFECT investment made in productive equipment, the amount of When should the programs launch? The answer is short electricity supplied to productive users, and the number and simple: Steps 1–5 should begin as soon as planning for of additional businesses connected. The contract should mini grid deployments starts. Step 6 should coincide with provide enough room for NGOs to tailor their strategies the arrival of the mini grid. and approaches to address local constraints and the Steps 1 through 6 can then be repeated throughout imple- specificities of each community. mentation not only to consider new data and learn from Two examples of concerted, coordinated, government-sup- previous iterations but also to maintain momentum and ported programs are offered in box 3.1, on the Infrastructure customer anticipation for expanded electricity access and Development Company Limited’s program in Bangladesh, its benefits. and box 3.2, on Ethiopia’s efforts to introduce productive Indeed, the activities presented in this chapter should be uses into its mini grid–based rural electrification program. sustained throughout the lifetime of a mini grid system, LOCAL CHANGE AGENTS with new geospatial analyses to be conducted every year or every other year, and Steps 2 through 6 undertaken at Local change agents such as village councils and commit- least once per year. tees, NGOs, and civil society organizations can also support productive uses and facilitate interactions among local businesses, mini grid operators, and equipment suppliers. WHAT’S NEXT? NGOs and civil society organizations have demonstrated Boosting productive use nearly ensures the chances of a that they can be successful partners for tasks that require mini grid’s success by offering an everyone-wins scenario substantial and continued support. Their diverse skills for developers, entrepreneurs, households and communi- (technical, social, and financial) combined with their local ties, and national utilities. More systematic and concerted presence allow them to work with entrepreneurs and coor- efforts are required, however, to promote, finance, and ulti- dinate with stakeholders in the field. Beyond identification mately boost the uptake of PUE. At the same time, devel- of productive uses and promotional activities, NGOs could opers and potential productive-use customers need to do business development, advising small enterprises on understand the economic and technical characteristics of business challenges. Field-based teams give NGOs the income-generating machines and appliances to be able to ability to analyze market opportunities and assist entre- make investment decisions accordingly. Successful exam- preneurs in preparing their business models and apply for ples of PUE programs, such as those described in boxes credit, while coordinating with mini grid developers on how 3.1, 3.2, and 3.3 below, show what is already possible, but a to provide adequate connections. dramatic scale-up of national mini grid markets will require MFIs, small business development centers, chambers of greater efforts from government agencies across sectors, commerce, and small business accelerators could also be developers, MFIs, NGOs, and other stakeholders along the mobilized. Implementation units of agriculture and rural income-generating appliance value chains. 130   MINI GRIDS FOR HALF A BILLION PEOPLE BOX 3.1 HOW IDCOL INCREASES PRODUCTIVE USES OF ENERGY IN SOLAR-HYBRID MINI GRIDS IN BANGLADESH The Infrastructure Development Company Limited Mini grids have clear, long-term development impact (IDCOL) is providing concessional project financing potential, but uptake of productive uses varies widely to enable mini grid developers to deliver improved and—despite careful and detailed consumer surveys energy services via solar and solar-hybrid mini grids in and expected load analysis—customer uptake was remote parts of Bangladesh. More than 20 mini grids lower than predicted (figure B3.1.1). After three years, are in operation, with a total capacity of almost 5 mega- the financials under the IDCOL package show that only watts-peak, and many more mini grids are planned for two out of seven mini grids reached their expected development. Sites are located primarily on islands not level of demand. Eleven were only recently commis- reachable by grid extension because of rivers often sev- sioned, while the remaining mini grids for which data eral kilometers wide in the monsoon season. are available, struggle to reach the expected level of demand. Mini grid sites have their own microeconomies, which are usually a mix of seasonal activities (such as fisher- Uptake lags were particularly striking among larger ies, agriculture, milling, husking, and oil pressing) and mini grids (>200 kilowatts-peak) at which productive wood production and sawmills. They also host a range energy use was expected to account for 40–65 per- of supporting businesses: carpentry and metal work- cent of demand. Daytime productive energy users ing shops; diesel engine repair shops; and many small were not connecting as planned, and in some cases, retailers of services, products, and foods in the local larger nighttime customers were saturating plant bazaars, which are open in the evenings. Almost all capacity more quickly than expected. Higher invest- productive energy use is provided by dedicated diesel ment in addition to low demand and underutilization engines with belt drives. Most households and shops of the plant exposed these mini grids to negative cash have had access to stand-alone home systems for sev- flows and risks. eral years, under IDCOL’s home system program. FIGURE B3.1.1 • Share of expected load achieved by selected mini grids in Bangladesh 3-month moving average of percent of expected load achieved 180% 160% 140% 120% 100% 80% 60% 40% 20% 0% 0 3 6 9 12 15 18 21 24 27 30 33 36 39 42 45 48 Months after commissioning Site A Site B Site C Site D Site E Site F Site G Source: IDCOL analysis. Note: Figures show three-month moving averages. Site names are not mentioned for confidentiality purposes. MINI GRIDS FOR HALF A BILLION PEOPLE    131 BOX 3.1, continued FIGURE B3.1.2 • Effect of extensive customer awareness campaigns on uptake 120 100 Number of customers 80 60 40 20 0 Site 1 Site 2 Site 3 Site 4 Site 5 Site 6 Site 7 Site 8 Site 9 Site 10 Site 11 Site 12 Site 13 Site 14 Mini grid sites Monthly asquisition before customer training (3 months average) Customer asquisition one month after training Source: IDCOL. Note: For purposes of confidentiality, sites are not named. To increase the uptake of productive uses of electricity, is under cultivation and has three cropping seasons. IDCOL launched intensive, three-day customer aware- Guidance was provided on irrigation services for farm- ness campaigns starting in October 2017, conducted ers at lower cost than their diesel-powered pumps. by international experts and trainers from major equip- IDCOL continues to broaden its training program to ment manufacturers. These campaigns combined cus- sensitize developers and operators on leveraging the tomer training with public shows featuring folk singing use of electricity to maximize socioeconomic benefit. and street theater, increasing productive-use custom- Training content is under development for the gen- ers and total customer acquisition by nearly 500 per- der dimension in mini grid interventions, highlighting cent (figure B3.1.2). the benefits of social facilities, streetlighting, schools IDCOL also arranged for training in management skills and clinics, rickshaw charging stations, and so forth. and development to strengthen understanding among This information will enable entrepreneurs to connect mini grid developers about business opportunities. It community benefits with their own sustainability as also required all funded mini grids to install remote, businesses. In addition, IDCOL plans to incentivize real-time monitoring capabilities for instantaneous daytime loads via time-of-use packages and financ- troubleshooting and tracking of historical trends. ing conversion packages ($120–$400, depending on Meanwhile, IDCOL provides assistance in mini grid– industry and load). Most of the larger sites need softer powered irrigation, which can greatly increase utiliza- loan finance terms (longer grace periods and terms) tion rates. In-house agriculturists identified irrigation to be viable for the sponsor. A sponsor will need to play potential for the grid area, since almost all arable land the lead microfinancing role. Source: Analysis by IDCOL and the World Bank RERED II team. 132   MINI GRIDS FOR HALF A BILLION PEOPLE BOX 3.2 RURAL, PRODUCTIVE USES OF ELECTRICITY: LESSONS FROM ETHIOPIA In 2019, the government of Ethiopia issued the updated ductive potential, to plan targeted mini grid develop- National Electrification Programme (NEP 2.0), which ments. This comprehensive approach has emerged formalized its ambitious goal of universal electrifica- from partnering with the World Bank, the Ethiopian tion by 2030. The NEP 2.0 detailed a large-scale rollout Agricultural Transformation Agency, the Rockefeller of mini grids both through the national electrical util- Foundation, AfDB SEFA, and others, putting together ity (the Ethiopian Electric Utility, or EEU) and through cutting-edge geospatial analyses and overlaying many engaging the local and international private sectors. layers of data and maps. Since then, the government, primarily through the To foster productive uses, ADELE is conducting ana- Ministry of Water and Energy, has been developing and lytics to get a better understanding of the potential deploying programs and projects with the support of for a new appliance result-based financing scheme. development partners like the World Bank, the African This scheme would complement the ongoing project Development Bank and its Sustainable Energy Fund activities and maximize the transformational eco- for Africa (AfDB SEFA), the Deutsche Gesellschaft für nomic impact in the communities targeted with mini Internationale Zusammenarbeit (GIZ), the Rockefeller grid rollouts. Under the utility-led modality of ADELE, Foundation, the Foreign, Commonwealth and Develop- the EEU intends to build in a number of activities to ment Office, the Ikea Foundation, the Rocky Mountain improve appliance availability and appliance financing Institute, and many others. for the communities expected to be electrified by EEU One important example is the $500 million Access with mini grids. Such activities are expected to range to Distributed Electricity and Lighting in Ethio- from community engagement and appliance demon- pia (ADELE) project, financed by the World Bank. strations to connecting communities to microfinance Approved in 2021, ADELE is a comprehensive, nation- institutions, nongovernmental organizations, appli- al-level rural electrification program powered through ance distributors, and so on. the grid, mini grids, and stand-alone solar solutions. The Ministry of Water and Energy, in partnership with So far, $270 million had been committed to the mini the Agricultural Transformation Agency and with grid program alone. The mini grid component under extensive support from the Rockefeller Foundation ADELE is being implemented by the EEU and covers and AfDB SEFA, is currently piloting nine mini grid sites two key modalities for rolling out greenfield mini grids. with major agricultural and irrigation potential, as part As a substantial scale-up of the utility-led model, the of the Rockefeller Foundation-supported Distributed EEU expects to deploy about 200 mini grids over the Renewable Energy Agriculture Modalities (DREAM) coming years through various engineering, procure- project. The pilot is intended to test a grant-supported, ment, and construction (EPC) and operation and private-sector-led modality, with results-based financ- maintenance modalities; it is also supported by the ing from Rockefeller and concessional financing launch of a large-scale, private-sector-led, perfor- expected from AfDB SEFA. mance-based grant program. Finally, the Rockefeller Foundation is preparing a Pro- Through ADELE and other programs, the Government ductive Use Appliance Financing Facility (integrating of Ethiopia has chosen to center all its mini grid and off- grid, mini grid, and off-grid electrification), with sup- grid rollout activities around income-generating uses port from the Collaborative Labeling and Appliance of electricity: identifying, prioritizing and stimulating Standards Program (CLASP) and Nithio. The facility them. In practical terms, this means that it has been should make income-generating appliances more assembling a holistic and up-to-date map of every site affordable and accessible. across the country with the most agricultural and pro- Source: ESMAP and the World Bank ADELE team. MINI GRIDS FOR HALF A BILLION PEOPLE    133 BOX 3.3 LESSONS FROM A UTILITY-NGO PARTNERSHIP IN INDONESIA Indonesia’s Rural Grid Electrification projects in the in leadership, proximity to larger markets, landholding 1990s, funded by the World Bank, also promoted pro- and crop patterns, and existing nonfarm income-gen- ductive uses. The lessons learned then are relevant erating activities, and the NGOs would need to craft for mini grid development today (Finucane, Besnard, their marketing to fit the specifics of the individual and Golumbeanu 2021). They show how partnerships businesses and their contexts. between energy providers (in Indonesia the national For the NGOs, the task was to market and sell an utility PLN) and local nongovernmental organizations already designed, deployed, and priced service (rural (NGOs) built an ecosystem that boosted rural electrifi- business services) in ways that would motivate pur- cation through jobs, income, and productivity. Fieldwork chase decisions in the different rural and business helped to tout the potential role of NGOs as channels contexts. The role of the utility, PLN, was to manage to promote PLN service. NGOs in rural Indonesia had the program, including: (1) NGO contracting, supervi- impressive operational capabilities, experience in the sion and payments; (2) village selections, which were marketing of changes, and experience with rural cot- expected to be recently electrified communities; (3) tage and small businesses and the poorest households. target setting (by village, NGO, and key metrics); (4) With the support of NGOs, project stakeholders held vetting of NGO marketing materials for accuracy; and consultations in order to better understand the char- (5) performance and impact monitoring and report- acteristics of current and potential customers and the ing. An effective outreach program, such as Rural reasons behind the slow take-up of grid services and Business Services, can improve load use and generate design-marketing campaigns. economic activity and employment. To run the campaigns, PLN contracted experienced Although the grid-based activity implies an exten- NGOs, skilled in outreach to families and community sive service area and varied clientele, the concept of groups in rural literacy, health, nutrition, and microen- customer-responsive service that makes good use of terprise development. The goal was to determine strat- capacity is transferable to other supply situations. The egies, marketing mixes, and communication methods success of this approach stemmed mostly from the suitable for the different village contexts, and conduct a holistic, opportunistic, context-specific design pro- series of time-limited marketing campaigns. The com- cess, a process that outsourced marketing to a local plexities of each village would be different, for instance, NGO as a possible market entry point. Source: ESMAP analysis of World Bank project documentation. 134   MINI GRIDS FOR HALF A BILLION PEOPLE REFERENCES IEG (Independent Evaluation Group). 2008. The Welfare Impact of Rural Electrification: A Reassessment of the Costs and Benefits. Washing- Alibaba. 2022. “Alibaba Online Store.” Accessed 2022. https://www. ton, DC: World Bank. alibaba.com/. ILO (International Labour Organization). 2014. “Evolution of Informal Baer, Tobias, Goland Tony, and Robert Schiff. 2013. “New Credit-Risk Employment in the Dominican Republic.” Notes on Formalization Models for the Unbanked.” McKinsey & Company. https:/ /www. Series, International Labour Organization, Geneva. https://www.ilo. mckinsey.com/business-functions/risk/our-insights/new-credit- org/americas/sala-de-prensa/WCMS_245893/lang--en/index.htm. risk-models-for-the-unbanked. Kes, Aslihan, and Hema Swaminathan. 2006. “Gender and Time Poverty Banerjee, Sudeshna Ghosh, Kabir Malik, Andrew Tipping, Juliette in Sub-Saharan Africa.” In Gender, Time Use, and Poverty in Sub-Sa- Besnard, and John Nash. 2017. Double Dividend: Power and Agri- haran Africa, edited by Mark C. Blackden and Quentin Wodon. World culture Nexus in Sub-Saharan Africa. Washington, DC: World Bank. Bank Discussion Paper Series No. 73. Washington, DC: World Bank. https://openknowledge.worldbank.org/handle/10986/26383. Mlinda. 2021. “Renewable Energy for Rural Communities India.” https:// Bhatia, Mikul, and Niki Angelou. 2015. Beyond Connections: Energy www.mlinda.org/democratising-energy-supply/. Access Redefined. Energy Sector Management Assistance Program Mukasa, A. N., and A. O. Salami. 2016. “Gender Productivity Differen- (ESMAP) Technical Report 008/15. Washington, DC: World Bank. tials among Smallholder Farmers in Africa: A Cross-Country Com- https://openknowledge.worldbank.org/handle/10986/24368. parison.” Working Paper Series No. 231, African Development Bank, Blodgett, Courtney, Peter Dauenhauer, Henry Louie, and Lauren Kick- Abidjan, Côte d’Ivoire. ham. 2017. “Accuracy of Energy-Use Surveys in Predicting Rural NREL (National Renewable Energy Laboratory) and E4I (Energy4Im- Mini-Grid User Consumption.” Energy for Sustainable Development pact). 2018. Productive Uses of Energy in African Micro-Grids: Tech- 41: (December): 88–105. https://www.sciencedirect.com/science/ nical and Business Considerations. Washington, DC: NREL. article/pii/S0973082617304350. ODI (Overseas Development Institute), GOGLA, Practical Action, and Cheney, Catherine. 2016. “How Alternative Credit Scoring Is Transform- Solar Aid. 2016. Accelerating Access to Electricity in Africa with Off- ing Lending in the Developing World.” Devex. https:/ /www.devex. Grid Solar: The Impact of Solar Household Solutions. London: ODI. com/news/how-alternative-credit-scoring-is-transforming-lend- https://cdn.odi.org/media/documents/10229.pdf. ing-in-the-developing-world-88487. Plutshack, V., J. Free, and R. Fetter. 2020. “Taking Electrification on CrossBoundary. 2020. “Study Design: Appliance Financing 3.0 Ener- the Road: Exploring the Impact of the Electric Farm Equipment gy-Efficient Productive Use.” https://www.crossboundary.com/wp- Roadshow.” Published online by the Rhodes Information Initiative content/uploads/2020/02/CrossBoundary-Innovation-Lab- at Duke University. Accessed June 12, 2022. https:/ /bigdata.duke. Study-Design-Appliance-Financing-3.0-Energy-Efficient- edu/projects/taking-electrification-road-exploring-impact-elec- PU-Anonymized-25-Feb-2020.pdf. tric-farm-equipment-roadshow. De Gouvello, Christophe, and Laurent Durix. 2008. Maximizing the Ramachandran, V., M. Shah, and T. Moss. 2018. “How Do African Firms Productive Uses of Electricity to Increase the Impact of Rural Elec- Respond to Unreliable Power? Exploring Firm Heterogeneity Using trification Programs: An Operational Methodology. ESMAP Formal K-Means Clustering.”CGD Working Paper 493, Center for Global Devel- Report 332/08, Energy Sector Management Assistance Program. opment,Washington,DC.https:/ /www.cgdev.org/sites/default/files/ Washington, DC: World Bank. https:/ /openknowledge.worldbank. how-do-african-firms-respond-unreliable-power-exploring-firm- org/handle/10986/17538. heterogeneity-using-k-means.pdf. Duke University. 2022. “Come Join the Electric Circus! What the US RMI (Rocky Mountain Institute). 2018. Closing the Circuit: Stimulating Rural Electrification Story Can Teach Us Today.” Website published End-Use Demand for Rural Electrification. Golden, CO: RMI. by the James E. Rogers Energy Access Project at Duke University. Accessed June 12, 2022. https:/ /energyaccess.duke.edu/come-join- Rogers, Everett. 2003. Diffusion of Innovations. 5th ed. New York: Simon the-electric-circus/. and Schuster. ISBN 978-0-7432-5823-4. Factor[e] Ventures. 2020. “Looking Beyond Appliances: Systemic USAID (U.S. Agency for International Development). 2018. “How Can Barriers to Minigrid Demand Stimulation.” https:/ /www.factore. Productive Uses Enhance the Economics of Mini-Grids?” USAID, com/systemic-barriers-to-stimulating-electricity-demand-in-afri- Washington, DC. https:/ /www.usaid.gov/energy/mini-grids/eco- can-off-grid-energy. nomics/productive-use/enhancement/. Finucane, James, Juliette Besnard, and Raluca Golumbeanu. 2021.  Village Infrastructure Angels. 2019. Fixed Amount Award (FAA): PAEGC “Raising Rural Productive Uses of Electricity: A Case Study of a Suc- Innovator Annual Report. Washington, DC: USAID. https:/ /pdf.usaid. cessful Utility-NGO Partnership in Indonesia.” Live Wire 2021/119,  gov/pdf_docs/PA00WGSS.pdf. World Bank, Washington, DC. https:/ /openknowledge.worldbank. World Bank. 2000. Implementation Completion Report on a Loan in org/handle/10986/36670. the Amount of US$398 Million to the Government of Indonesia for GIZ (Gesellschaft für Internationale Zusammenarbeit), BMZ (Federal the Second Rural Electrification Project. Washington, DC: World Ministry for Economic Cooperation and Development), ESMAP Bank. https://documents1.worldbank.org/curated/en/1208214680 (Energy Sector Management Assistance Program), and AEI (Africa 49833272/pdf/multi-page.pdf. Electrification Initiative). 2013. Productive Use of Energy—PRO- World Bank. 2020. Women, Business, and the Law. Washington, DC: DUSE: Measuring Impacts of Electrification on Small and Micro-En- World Bank. doi:10.1596/978-1-4648-1532-4; https://wbl.world- terprises in Sub-Saharan Africa. Eschborn and Bonn: GIZ. bank.org/. IDS (Institute of Development Studies) and GIZ (Gesellschaft für Internationale Zusammenarbeit). 2019. Unlocking the Benefits of Productive Uses of Energy for Women in Ghana, Tanzania and Myan- mar. Research report RA6. ENERGIA. https:/ /www.energia.org/ NOTES assets/2019/03/RA6-Unlocking-the-benefits-of-productive-us- es-of-energy.pdf. 1. Global Findex database: https://globalfindex.worldbank.org/. MINI GRIDS FOR HALF A BILLION PEOPLE    135 CHAPTER 4 ENGAGING COMMUNITIES AS VALUED CUSTOMERS CHAPTER OVERVIEW This chapter focuses on the role of community engagement in mini grid projects and the impact that inadequate or insufficient community engagement can have on their sustainability. It begins by underscoring the value of con- tinuous community engagement through every phase of the mini grid project—starting at the early design and planning stages, through financing, procurement, operation, and maintenance. It next delves into concrete steps that a mini grid developer can take at each project phase to engage the community. The chapter concludes with examples of innovative ways to scale up community engagement across multiple projects at the national and inter- national levels. WHY IS COMMUNITY ENGAGEMENT mann (2010) reveal that rural electrification programs ben- efit greatly from local participation. Involving communities IMPORTANT? from the start can help improve the design (Peru, Vietnam), As shown in previous chapters, mini grids offer a least-cost avoid disputes and gain local support (Bangladesh), mobi- option for providing reliable, affordable electricity to mil- lize contributions in cash or in kind (Nepal, Thailand), and lions of people now living without access, people who would increase local ownership, thus contributing to operational otherwise have to wait years for the main grid to arrive. sustainability. In line with this, a study covering mini grid projects of the Energy and Environment Partnership (EEP) However, bottlenecks must be cleared before mini grids Trust Fund in 13 countries in Sub-Saharan Africa concluded can take off on a large scale. Some, such as access to that building strong relationships in the community is crit- finance, workable regulations, and enabling business envi- ical to the financial sustainability of mini grids (EEP 2018). ronments, are systemic and best addressed at the national and international levels. Others are highly localized, such Community engagement is beneficial at all phases of a as the specific socioeconomic, cultural, and environmental mini grid project, as discussed below. characteristics of each community to be electrified with a DESIGN AND PLANNING: mini grid. A review of relevant literature and interviews with • As the primary customer of the mini grid, the local com- mini grid developers1 demonstrates that, to identify the munity should be the first to be consulted and engaged best technical solution for a given site and ensure its long- if the mini grid developer is to obtain the “social license” term sustainability, it is essential to continuously engage (that is, buy-in and acceptance from the community) to with the local community and determine the optimal fit operate the system. between the community and the new mini grid. • The local community has the best understanding of Existing evidence points to several reasons why community the surrounding conditions and resources (Mishra and engagement is essential for the successful implementation Sarangi 2016), which can help the mini grid developer of mini grid projects—and rural electrification more broadly. choose the optimal technology mix and operating Case studies reviewed by Crousillat, Hamilton, and Ant- model. 136   MINI GRIDS FOR HALF A BILLION PEOPLE Existing evidence highlights reasons why The central role of community engagement community engagement is essential for the in ensuring the long-term sustainability of successful implementation of mini grid projects: mini grids is further illustrated by mini grid failures improving the design (for example, in Peru, Viet- linked to insufficient local community involvement. nam); avoiding disputes and gaining local support Overlaying the common causes of mini grid failure (Bangladesh); mobilizing cash or in-kind contribu- with the typical stages of mini grid development tions (Nepal, Thailand); increasing local ownership indicates that (1) more than half of the causes of and operational sustainability; and ensuring finan- failure involve inadequate community engagement, cial sustainability (for example, in 13 Sub-Saharan and (2) community engagement plays a particularly African countries). decisive role in the stages that precede the com- mencement of operations. • During this early stage, community engagement will • In the case of a project backed by a public-private part- ease communication with local authorities, identifica- nership, transparency in the tender procedure will also tion of reliable and capable local (technical and sales) raise the community’s trust in the project. staff, and identification of prospective customers • In some instances, mini grids may also recruit commu- (including anchor clients and businesses). nity members to install the systems. • Early community engagement will facilitate land acqui- OPERATIONS AND MAINTENANCE: sition or right of use for the project, assist in obtaining good socioeconomic and cultural information from pro- • Community engagement will improve the communica- spective customers to underpin the business case of the tion between operator and customer, creating trust and project, and inform the community as well as the project mutual understanding about system repairs and other developer about expectations, requirements, roles, and matters. responsibilities. • Assuming that sufficient capacity has been developed • During this phase, a mini grid developer will use its con- within the community, the participation of locals in the tact with the community to accurately assess customer setup, maintenance, and repair of systems has been demand2 (including willingness and ability to pay), anecdotally shown to reduce the frequency of repairs compare the likely demand with prospective revenue (Fahey and others 2014) and increase the likelihood of streams (tariffs and subsidies), and reach conclusions more judicious use of systems by the community.3 about the project’s viability. • At this stage, the developer can also set up training • In addition, during this phase, community engage- sessions to stimulate demand and drive consumption ment will enable the developer to gauge residential among productive users. and business energy needs and expectations, and • In addition, mini grid developers often recruit local vil- to recognize the roles of men and women as energy lage agents to assist in sales and ensure rapid responses consumers and stakeholders. Raising awareness and to customer complaints and concerns. engaging in open dialogue will allow communities to The central role of community engagement in ensuring make well-informed decisions about their mini grid the long-term sustainability of mini grids is further illus- options, improving the quality of “service fit” (in terms trated by mini grid failures linked to insufficient local com- of situation, costs, payment modalities, business munity involvement. An extensive study of mini grids in Fiji case) and protecting the project’s viability by making found many of them failing as a result of various technical customer satisfaction more likely. It will also assist in and nontechnical issues (Dutt and Macgill 2013). Overlay- market segmentation. ing the causes of failure identified in the analysis with the FINANCING, PROCUREMENT, AND CONSTRUCTION: typical stages of mini grid development (as shown in fig- • The information obtained during the design and plan- ure 4.1) indicates that (1) more than half of the identified ning phase will yield data useful in developing the proj- causes of failure have a community engagement element ect’s tariff structure and improving its risk profile for (shown in the boxes outlined in red), and (2) community investors. engagement plays a particularly decisive role before oper- ation commences. MINI GRIDS FOR HALF A BILLION PEOPLE    137 In another comprehensive analysis, the main risks of mini Community engagement can thus be considered an essen- grid deployment were evaluated in terms of the potential tial component of the mini grid process during the design, impacts of the risk and the probability of the risk occur- implementation, operation, and maintenance phases. ring (Manetsgruber and others 2015). Proper community Based on existing evidence and interviews with mini grid engagement can significantly limit the exposure of mini developers, the next section will explore the principal steps grids to some of these risks (table 4.1). that a mini grid developer can take during each phase to engage the local community. FIGURE 4.1 • Typical issues hindering the mini grid development process Financial/ Implementation/ Operation Design Planning procurement construction Inadequate load Failure to consider local Inadequate and Inadequate transfer of skills to assessment needs and expectations inappropriate tariff local population (system upkeep, of the system structure productive use, payment, etc.) Inadequate risk analysis Inadequate technical evaluation of system performance Inadequate evaluation of project sustainability Component failure Inadequate evaluation of commercial and economic viability Inadequate technical evaluation of system performance Lack of consultation between suppliers and end users (services, land acquisition, etc.) Poor maintenance Inadequate evaluation Lack of budget to maintain the system of renewable source General lack of community participation Source: ESMAP analysis of Dutt and MacGill (2013). TABLE 4.1 • Potential for community engagement to limit mini grid risks Risk Community engagement aspect Impact Probability Nonpayment of electricity bills as a result of either Inadequate assessment of customers during High Medium inability or unwillingness to pay planning, limited social consensus regarding grid services, inadequate services Unpredictable electricity demand, negatively Poor feasibility assessment that fails to address Medium to Situational affecting project sizing and cost structure “A-B-C” considerations a; insufficient promotion of high services to productive users Insufficient social acceptance leading to poor Unfavorable public opinion, lack of transparency, Moderate Low to embedding of the project in the sociocultural context insufficient involvement and development of local medium (for example, social services, agricultural activities) capacity Theft (of materials for which there is a secondary Inability to create a “social compact”; insufficient Moderate Medium to market) and vandalism (particularly in cases of understanding of local power structures low conflict of interest among stakeholders) Defective operation resulting from Insufficient understanding of “A-B-C” expectations, Low to Medium miscommunication between business and needs, and limitations; insufficient understanding medium customer; conflicts of interest of local power structures Unfavorable mini grid regulations do not permit Insufficient understanding of “A-B-C” expectations, High High attractive tariffs (for either supplier or consumer) or needs, and limitations; insufficient consumer competition organization Policy and planning fail to sufficiently stimulate Insufficient market segmentation, limited High High productive and income-generating activities, productive-use awareness, and insufficient access undermining the viability of the investment to appliances for productive uses Source: ESMAP analysis of Manetsgruber and others (2015). In mini grid market segmentation, A-B-C refers to anchor, business, and community customers. a.  138   MINI GRIDS FOR HALF A BILLION PEOPLE COMMUNITY ENGAGEMENT DESIGN AND PLANNING PHASE THROUGHOUT THE MINI GRID During the design and planning phase, community engage- ment can PROJECT CYCLE • Establish a relationship between local authorities and Examples from the field indicate that successful com- the project developer munity engagement begins with raising awareness in the • Identify partners, local stakeholders and authorities, potentially connected community. It continues during and local staff adoption and productive operation, as satisfied custom- ers promote the technology to their neighbors and friends. • Improve community confidence in the project developer Each community requires a flexible approach, with a clear or service provider understanding of the local socioeconomic and cultural • Provide insight into local expectations of the technology characteristics, and a potential tailoring of the promotional and services tools, materials, and channels.4 The typical mini grid project • Improve the assessment of demand and load cycle, outlined in figure 4.2, maps these steps. • Create awareness of the technology and the potential uses of electricity, and develop a portfolio of prospective customers Successful community engagement begins • Assist with segmentation of the market into A-B-C cus- with raising awareness in the potentially tomers connected community. It continues during adop- • Profile (and segment) potential customers or arche- tion and productive operation, as satisfied cus- types and forecast their ability to pay (ATP) and willing- tomers promote the technology to their neighbors ness to pay (WTP). and friends. Each community requires a flexible In this initial phase, community engagement is preceded by approach, with a clear understanding of the local a preliminary market assessment, and public data are ana- socioeconomic and cultural characteristics, and a lyzed by the developer to identify potential mini grid sites. If potential tailoring of the promotional tools, materi- the developer is new to the country, this exercise is also likely als, and channels. to include an evaluation of national regulatory, policy, politi- cal, economic, financial, and environmental considerations. FIGURE 4.2 • Typical mini grid project cycle DESIGN AND 1 PLANNING OPERATION & PROMOTION / 5 MAINTENANCE 2 INFORMATION FINANCING / 4 IMPLEMENTATION / 3 PROCUREMENT CONSTRUCTION MINI GRIDS FOR HALF A BILLION PEOPLE    139 When a site or a group of potential sites has been identified, Customer profile the developer will typically conduct a site visit to assess Where possible, the information gathered during an initial conditions on the ground, accompanied by a brief survey survey of potential customers is supplemented with the of residents and contacts with the local government. This findings of a domestic energy baseline study and, in a later step—outreach to the local authorities and, in particu- stage, user surveys. The surveys inform a detailed customer lar, village chiefs—is emphasized by developers as key to profile that ensures prospective customers are effectively demonstrating respect and securing the favorable atten- served. In practice, these profiles are further differenti- tion of the village residents.5 Once communication with the ated—by type of farming, type of enterprise, and so on. local village representatives is underway, the developer can The customer profile prepared by the developer based go on to sign a memorandum of understanding with the on the initial data gathering can be used to inform a gen- local authorities, discuss land acquisition or rights of use, der-specific approach to connecting customers. In such and pursue any subnational government approvals that an approach, the cultural and socioeconomic nuances and may be required. potential barriers to women’s participation are taken into Following this, the developer will typically dispatch a full account. customer survey team to carry out household-size surveys using tablets and standardized survey instruments. For Segmenting the market into A-B-C customers larger developers that operate in several markets, these An additional approach to segmenting the market further— surveys are often adapted to the local context.6 The objec- into A-B-C customers—may enhance the robustness and tive is to gain a deep understanding of how to satisfy cus- viability of a mini grid. The A-B-C approach is a business tomers’ needs. Good surveys will help developers gain an approach rather than a community engagement method- understanding of six areas that ultimately affect the satis- ology, but it can offer added value in helping the mini grid faction of their customers. developer identify A, B, and C client segments through proper community engagement. Customer satisfaction is directly influenced by four con- siderations, all of which can be addressed in the survey The A-B-C business model (figure 4.3) was first developed questions: by companies such as OMC Power in India, which built mini grid companies around anchor clients (GIZ 2014). • Livelihood: The potential role of mini grid electricity in Under the model, the supply of electricity from the mini domestic or commercial activities and the effect the grid is prioritized for an anchor load customer, typically services may have on employment, entrepreneurship, a commercial or industrial user, followed by businesses and quality of life (shops, small enterprises, and so forth), and then house- • Influences on the mini grid’s activities, including sea- holds. The A-B-C model stipulates that the size of the sonal labor migration; national and regional media; mar- generation unit is typically determined primarily by the ket effects of nearby major towns (particularly if those demand needs of the anchor load (Bhati and Singh 2018), towns are connected to the national grid); and existing with the aim of securing stable and predictable revenues. marketing campaigns, including those developed by the OMC and other mini grid operators in India have used government telecommunications towers close to population centers • Aspirations of prospective consumers, including educa- to scale up mini grid systems and expand into surround- tion, careers, and communication • Consumption: Need-based or additional con- FIGURE 4.3 • The A-B-C model sumption, return on investment, purchasing behavior. Community: Community Low electricity demand, mostly for These considerations are in turn influenced by lighting, mobile phone charging and household appliances • Community: The role of the family, gender, and community institutions Business Businesses: Higher electricity demand for • Infrastructure: Energy and appliance mar- productive use kets; available technology, the existing infra- Anchor structure for power and information and Anchor: Financially sound, guarantees electricity communications technology, and the avail- purchase, secures commercial operation ability of finance and services (health care, education). Source: GIZ 2014. 140   MINI GRIDS FOR HALF A BILLION PEOPLE ing businesses and households. Agroprocessing and min- consume the lion’s share of the energy generated during ing activities have also frequently been used in Africa to the day, whereas household consumption tends to inten- provide reliable loads. sify during the hours before and after work. A study of mini grid projects in 13 countries in Sub-Saha- PROMOTION AND INFORMATION-SHARING ran Africa (EEP 2018) finds that “the most financially sus- PHASE tainable mini grids use an A-B-C strategy: first, identify and During the promotion and information-sharing phase, negotiate an agreement with an anchor load client (often in community engagement can agroprocessing); then identify, or help develop, small local businesses; and only last target domestic consumers.” Or, • Improve the information on mini grid services shared as Kennas and Barnett (2000) put it in their microhydro with the community assessment: “It is easier to make a profitable micro-hydro • Fine-tune customer segmentation plant socially beneficial than to make a socially beneficial plant profitable.” • Leverage women’s social and trust networks to spread information Growing evidence (discussed in detail in chapter 3) sug- gests the need to increase the prioritization of small and • Increase consumers’“energy education” on matters such medium-size enterprises—the B of the A-B-C model—as as understanding energy bills, performing maintenance, income streams for the mini grid. This is largely because handling grievances, and maximizing safety and health securing anchor clients can be a challenge, particularly • Select, record, and share early successful adoption in rural economies based on subsistence agriculture and stories. small-scale artisan networks. In addition, anchor clients The objective of this phase is to increase awareness of may have unrealistic expectations about tariffs, and oper- the technology, its services, and the facilities offered by ators may become overreliant or even dependent on them, the developer. Community engagement activities in this threatening their financial position. phase traditionally target fairs, radio, word of mouth, local Some developers have been combining multiple business extension services, construction activities, and—increas- clients to act as the anchor. In this approach, the mini ingly—social media. Marketing hubs are an additional grid’s generation assets are sited near a cluster of small engagement “tool,” as described below. In addition, through businesses that may be housed in the same building. billboards, social media, radio, and television, projects can This approach allows for mitigation of commercial risks target family members who migrated to urban centers, as and diversification of the mini grid customer base. A good they are often interested in helping parents gain access to example is the community engagement, load acquisition electricity. and micro-enterprise development approach (CELAMeD), To build up their presence in the community after the ini- which has been promoted by Smart Power India and imple- tial site visits and meetings, mini grid developers often mented by several energy service companies in India— cohost workshops for community members with mem- notably TARA Urja. CELAMeD is discussed in further detail bers of relevant cooperatives, nongovernmental organi- in the next section. zations (NGOs), microfinance institutions, and financial Including both anchor and business clients will enable the institution workshops for community members. These project developer to connect households at affordable workshops can be used to build awareness of (1) oppor- rates, because these clients (1) may cross-subsidize con- tunities to connect income-generating and household nection and consumption fees for households and (2) will appliances and machines to the mini grid; (2) agricultural applications of mini grid electricity; (3) financing available for the purchase of household, agricultural, and commer- Segmenting the customer base into anchor, cial appliances and machines; and (4) entrepreneurship business, and community customers can and business training. enhance the robustness and viability of a mini grid. The A-B-C approach is a business technique rather Establishing marketing hubs than a community engagement methodology, but it For the community engagement process to be success- can offer added value in helping the mini grid devel- ful, the underlying communication platform must have oper identify client segments through proper com- sufficient reach and trust within the community. Some munity engagement. established mini grid companies have developed commu- nity engagement practices as a core capability. However, new entrants to the market are likely to have insufficient MINI GRIDS FOR HALF A BILLION PEOPLE    141 resources and skills to properly develop a robust commu- Information should also be provided to local leaders, staff nity engagement process. of value chain organizations, and financial institutions, as they are often consulted about innovative technologies and Marketing hubs are one solution that can help mini grid practices. startups engage with their communities. These local plat- forms strengthen consumer confidence and allow for com- Engaging women as leaders of promotional and bined community engagement efforts between the mini information-sharing activities grid developer and organizations over the entire project Energy service companies are increasingly engaging cycle. Strong candidates for marketing hubs are entities women in promoting their service. Women are well aware that perform local or regional network functions, exhibit of the challenges that other women face, such as unique entrepreneurial drive, and have a natural affinity with mini time constraints, household responsibilities, and energy grid or off-grid electricity services. needs. Their access to other potential female customers Examples include savings and credit cooperatives, dairy may be significantly less constrained by social or cultural or agricultural cooperatives, agricultural input dealers, norms than male sales agents in certain cultural contexts. and rural development NGOs. The project developer may However, despite women’s potential to build robust distri- support the marketing hubs with capacity building, a ded- bution networks for energy solutions, especially in rural icated liaison or extension officer, and incentives for sales areas, they are still underrepresented in the sector. and extension activities. One successful example of engaging women in promot- The advantages of such hubs include lower operating costs ing energy access is Solar Sister, an African initiative that (promotion, training, extension), access to a more targeted recruits, trains, and supports female entrepreneurs to market, and knowledge sharing. In addition, hubs often serve as last-mile distributors of clean energy products, perform initial client eligibility screenings and may serve such as solar lights, mobile phone chargers, and clean as a physical location where microfinance institutions and cookstoves. By 2012, Solar Sister had empowered 2,000 other local financiers can meet their potential (or current) female entrepreneurs in Uganda, Nigeria, and Tanzania, customers. The hub-hosting entity, on the other hand, ben- who in turn provided solar and clean cooking solutions to efits from being able to offer an additional service to its more than 370,000 beneficiaries (Arc Finance 2012). members. FINANCING AND PROCUREMENT PHASE Experience with marketing hubs from other sectors During the financing phase, community engagement can, shows that a hub’s activities must fit into its regular on the demand side: agenda, and the hub should not be converted into a “sales machine” for mini grid developers, as this erodes • Assist in ATP/WTP assessments and the development trust within the local community. When set up properly, of a viable demand forecast marketing hubs can instead develop into “marketing • Organize in-kind contributions (for example, construc- beachheads” (Moore 1999), preparing the market for tion, at both individual and community levels) uptake by early adopters. • Improve consumer access to energy and credit for pro- Sharing information with prospective customers ductive-use appliances. After prospective customers have a basic awareness of the technology, they will require more detailed information to properly assess its usefulness to them. It is important to Energy service companies are relying ensure that the information provided is correct, unbiased, increasingly on women to promote their and comprehensive (addressing the merits, costs, and lim- service. Women are close to their female custom- itations of the technology). ers, can more easily tap into social networks, and are well aware of the challenges that other women The information will have to be provided by well-quali- face, such as unique time constraints, household fied project staff. At this stage, information can be shared responsibilities, and energy needs. Their access to either individually or in smaller functional groups (for other potential female customers may be signifi- example, hosted by local authorities, microfinance institu- cantly less constrained by social or cultural norms tions, NGOs, or other community stakeholders). In addi- than male sales agents in certain cultural contexts. tion, exchange