Developing a model for estimating global in-plane solar irradiance for Uganda.

Abstract

The project was focused on evaluation of the accuracy of combination of models that estimate plane-of-array (POA) irradiance from measured global horizontal irradiance (GHI), developing a decomposition model to estimate beam and diffuse horizontal irradiance and ascertaining optimum orientation of a solar collector for any given location in Uganda. The most widely used decomposition models were developed using empirical formulas and several parameters which were determined using datasets of a single location, these models do not necessary fit in every datasets of different locations due to discrepancies in environmental conditions. This estimation involves decomposition of global horizontal irradiance (GHI) into beam horizontal irradiance (BHI) and diffuse horizontal irradiance (DHI) components, then transposition of beam and diffuse global horizontal components into POA irradiance. Developing a model for estimating global in-plane irradiance involves two steps: (1) Developing decomposition model using diffuse fraction and clearness index with BHI, GHI and extraterrestrial irradiance as inputs. (2) Coupling transposition models with developed decomposition models to predict plane of array irradiance. Measured GHI and respective measured POA irradiance were used to evaluate combination of clear-sky, decomposition and transposition models to quantify relative mean bias deviation and root mean square deviation and the combination with the smallest relative mean square deviation was selected for further optimization analysis to obtain tilt angle and surface azimuth that yields maximum annual irradiation. Results suggests that for collector facing south and tilted should be left flat to yield maximum irradiation and the optimal tilt and surface azimuth angle was about 11 and 82 degrees respectively for regions near the equator. The best performing combination of models consisted clear-sky model of Duffie and Beckman, the decomposition model that was developed using local datasets and transposition model of Perez. This combination had 2.2% and 35% relative Mean bias deviation and relative root mean square bias deviation respectively. The optimal parameters that yield maximum deviation applies for north of equator regions since this project never had enough data for regions in south of equator to quantify optimal parameters that yield maximum annual irradiation. The developed model performed well compared to other models on basis of relative root mean bias deviation and relative root mean square deviation and it performs well annual data with R2 of 0.98.