A Techno-Economic Framework for Divisional Renewable Energy Cost Analysis across Bangladesh’s Grid Zones Using Machine Learning

Bangladesh aims to source 20% of its installed generating capacity from renewable energy by 2030, yet the delivered cost of renewable electricity across its eight administrative divisions has never been established on a common basis. This study
develops a zone-wise levelized cost of energy (LCOE) framework that integrates division-specific distribution loss rates and the regulated retail tariff with solar irradiance and wind speed, and maps the resulting cost surface using regression-based sensitivity analysis. A discounted cash flow model with a delivered-energy correction is parameterised from verified sources and evaluated for utility-scale solar photovoltaics (PV) and onshore wind in each division. Three machine-learning regression models – linear, polynomial and random forest – are trained on Latin-hypercube samples and compared against one-at-a-time sensitivity analysis.
Solar PV LCOE is consistent across the eight divisions (0.092–0.099 USD/kWh) and reaches near-parity with the tariff in four of them, while onshore wind is competitive only in the three coastal divisions and is costlier than PV everywhere. Divisional irradiance varies by only 7.7%, whereas a 2% point reduction in the cost of capital lowers LCOE by 13.2%. The results show that financing conditions, not resource geography, constrain renewable cost competitiveness in Bangladesh.