Solar Energy Assessment using Data-driven and Physical  Models: Application for Crystalline Silicon Photovoltaic  Systems in Ethiopia

Assaye, Gedifew Yismaw

dc.contributor.author	Assaye, Gedifew Yismaw
dc.date.accessioned	2026-08-20T06:55:58Z
dc.date.available	2026-08-20T06:55:58Z
dc.date.issued	2025-11
dc.identifier.uri	http://ir.bdu.edu.et/handle/123456789/17047
dc.description.abstract	Solar radiation, the electromagnetic energy emitted from the Sun, is a fundamental driver of Earth's weather and climate systems and represents a vast, clean energy source. Global solar radiation (GSR) is the total amount of solar radiation (both direct and diffuse) reaching a horizontal surface on Earth. It is a key measurement for evaluating the solar energy potential of a specific location, which in turn is essential for assessing the expected performance and efficiency of solar photovoltaic (PV) system. Thus, the accurate measurement and estimation of GSR is crucial for assessing and utilizing solar energy resources at both global and local scales. Hence, this study investigates estimations of GSR and assesses the performance of crystalline silicon (c Si) PV cells/modules across Ethiopia, by utilizing a comprehensive approach that integrates data driven and physical models. The study also implemented various optimization techniques such as determining the optimum tilt angle and tracking mechanisms. For this purpose, twelve machine learning (ML) and one stacked/ensembled model were trained and validated with hourly, daily and monthly ground-based global solar radiation data from 16 synoptic weather stations (2020-2022), supplemented by meteorological, aerosol, and sky condition data from MERRA-2 and NASA POWER archives. The three stations with distinct weather patterns were withheld from the model development process for model transferability/generality test. A stacked/ensemble model (i.e., constructed by stacking better performing separate models) showed exceptional predictive performance with error metric values ranging (R²: 0.956-0.963; RMSE: 9.938-11.784 W/m²) for all time scales. With this performance capability we generated a high-resolution (1° x 1°) global solar radiation data across Ethiopia for the year 2022, and the distribution showed a precise spatial and seasonal dependence with the highest in spring (i.e., 594 - 641 W/m2; eastern and northeastern) and lowest in summer (i.e., 359 – 405 W/m2; western and southern parts of the nation). Such analogs were also observed on the peak sun hours and plane-of-array (POA) irradiance distribution with their annual value ranging from 4.83 – 6.57 kWh/m2/day and 0.65 – 1.05 kW/m2, respectively, across the nation. Here it’s worth noting that to model POA irradiance, we implemented five decomposition and six transposition models (i.e., thirty different independent combinations). Furthermore, we incorporated POA irradiance into a single diode PV cell model to evaluate c-Si PV cell performances. Consequently, the annual PV cell temperature, ranging	en_US
dc.language.iso	en_US	en_US
dc.subject	Physics	en_US
dc.title	Solar Energy Assessment using Data-driven and Physical Models: Application for Crystalline Silicon Photovoltaic Systems in Ethiopia	en_US
dc.type	Dissartation	en_US