New methods and applications to explore the dynamics of least-cost technologies in geospatial electrification modelling

Sammanfattning: Access to modern energy services is a pre-requisite for sustainable development. As such, Sustainable Development Goal (SDG) 7 aims to ensure access to affordable, reliable, sustainable and modern energy for all. However, as of 2021, 675 million people lack access to electricity, and 2.3 billion people lack access to clean cooking fuels. Electricity in particular can bring benefits to many sectors of society, including households, health facilities, educational facilities, agricultural activities and businesses. Providing such access in currently underserved areas at the lowest cost requires an integrated approach, utilizing a combination of extension of the centralized grid networks, deployment of mini-grids and stand-alone technologies.Given the need for an integrated approach to increase access to electricity, geospatial electrification tools have been developed and used to inform policy- and decision-making. These tools are a category of energy system tools that draws on geospatial information to understand which technology to use where, depending on the local characteristics of each settlement in a country or region as well as the existing energy infrastructure. The number of geospatial electrification tools and analyses have seen a steep increase since the start of the millennia, particularly since the 2010’s. Some of these tools and analyses use simpler analytical expressions to estimate the least-cost technology in each location, whereas others provide detailed design of transmission, distribution and generation systems.  Geospatial electrification tools and analyses are increasingly used for decision-making and planning purposes towards the achievement of universal access to electricity. This dissertation aims to advance the state of the art in geospatial electrification modelling to support electrification efforts. In particular, the thesis examines the dynamics between the three types of electricity supply technologies (grid-extension, mini-grids and stand-alone technologies) under different modelling approaches, timelines and scenarios. Three research questions based on gaps in existing literature and applications are studied and explored through four publications. Furthermore, each publication provides a case study on one of the countries with the largest electricity access gap globally, namely Burkina Faso, Ethiopia, Somalia and the Democratic Republic of the Congo (DRC).The first research question explores how the use of scenarios and simulations in geospatial electrification modelling can be improved to better inform policy- and decision making in the field of electricity access. Lack of data is widely recognized as a key challenge in the field, as important datasets are missing, incomplete or of poor quality in many geographies. Combined with the difficulty of predicting latent electricity demand in currently underserved areas, and the numerous stakeholders in the field of electricity access, designing useful and informative scenarios can be challenging. In response to this, the first paper presents the first scenario discovery analysis in geospatial electrification modelling. In the scenario discovery approach, a large set of simulations based on variations of model parameters are computed. Next, statistical data-mining algorithms are applied to identify candidate scenarios of interest among these simulations. Using this approach, key scenarios that have the highest risk of leading to high electrification costs and scenarios that have the highest chance of low costs in Burkina Faso are identified.The second research question focuses on the time-aspect of geospatial electrification modelling, seeking to understand how the time-line selected changes the dynamics between the least-cost electrification technologies. With few exceptions, geospatial electrification models have focused on identifying the least-cost technologies by a single year, either 2030 or earlier. However, this provides limited insight on how the system may evolve over time. In paper II, least-cost electrification options in Ethiopia are modelled in 10-year intervals until 2070. The transition between technologies over this longer time-frame are studied under different constraints and demand levels. Furthermore, paper III focuses on how time is incorporated in the model. Through a case study of Somalia, least-cost technology options are explored both until 2030 and 2040. First, the model is run similar to a perfect foresight model, identifying the least-cost solutions directly for the population and demand by 2030 and 2040 respectively. Next, the model is run myopically, first in five-year time-steps and then in one-year time-steps, to explore how the least-cost solutions by the end year of the analysis change, and the implications this has for electricity access planning. The results of both case studies highlight that shorter term planning led to relatively higher levels of stand-alone technologies, whereas longer-term planning favors mini-grids and the grid to a larger extent.The third research question aims to shed light on the effects of different modelling approaches and model complexity in geospatial electrification modelling. Several geospatial electrification tools and frameworks have been developed and applied to inform decisions and planning towards increased electricity access. Naturally, these tools and frameworks differ in terms of modelling complexity. A comparison of published results from geospatial electrification models reveals that even in cases where these are studying the same region and similar demand levels, they identify different mixes of least-cost technology options. The fourth paper presents the first flexible geospatial electrification tool, which can provide both rapid first-pass assessments as well as more detailed analysis. Through a case study of the DRC, the effects on geospatial electrification modelling from the first-pass assessment and more detailed versions of the tool are explored. Differences in the least-cost technology mix using different algorithms in the OnSSET tool are explored, as well as the difference in data and computational requirements.