Development of a Modeling Framework to Predict Energy Demands in Developing Nations

Accurate and detailed energy demand estimates are crucial to achieving adequate energy infrastructure planning. These estimates are often non-existing or deficient in many developing countries, and consequently, electricity supply is unreliable and outages are unpredictable. To this end, we present a novel approach for modeling and prediction of electricity demand using LandScan Global as the foundational input data. LandScan is a global population database which provides population counts at approximately 1 km² spatial resolution; its use is based on the premise that electricity consumption is dependent on where people are located. These population counts can then be converted to electrical customers using a conversion factor developed from census and other survey count data for each spatial cell. Next, based on characteristic parameters such as building type, seasonality, standard of living, location (urban vs. rural), literacy levels, and justifiable scenarios for intensity of parameters, a daily electricity usage estimate is computed for each customer. This process is repeated in an iterative fashion over each customer in the cell to obtain a total value for the cell, and then over each cell in the study to obtain a total value for the region of interest. This spatial energy demand data can be used to develop electricity demand maps that could be valuable for energy infrastructure planning. In this study, the East-African nation of Uganda is used as a proof-of-concept, and the results are presented. Because LandScan Global is a global database, the methodology and framework developed in this study can be extended to other countries/regions of interest.