Comparison of PV potential models for africa and their potential cost implications.
<p>We currently have more than 7500 planned mini grids, most of them in Africa. These will soon connect more than 27 million people and cost about 12 billion dollars <sup>[1]</sup>. Africa is in a good position for Photo voltaic (PV) mini grid optimization, receiving more than 1800 KWh/m<sup>2</sup> Global Horizontal Irradiation (GHI) every year <sup>[2]</sup>, for most parts of the continent. However, the lack of a coordinated renewable energy monitoring and distribution network works against optimization of PV potential models <sup>[3]</sup>. This study shows the accuracy of existing photo voltaic potential estimators like renewables ninja <sup>[3]</sup>, the National Renewable Energy Laboratory (NREL), International Renewable Energy Agency (IRENA), and the global solar atlas <sup>[2]</sup>, by comparing the modeled values with long term measurements from ground solar stations. This is done for more than 20 stations distributed over Africa. Our results show best correlations <sup>[4]</sup> of up to 65.3% from version 2 of the Surface Radiation Data Set from Heliosat (SARAH) derived from the Photovoltaic Geographical Information System (PVGIS). However, we also have correlations as low as 16.2% for models commonly used in off grid simulations. The sensitivities of the modeled cost of a mini grid to the variation in PV potential were tested <sup>[5][6]</sup> using the statistical range in sourced PV potential from the different estimators, giving us cost variation of more than 2.8% that may arise from the different sources.</p><p><strong>References</strong></p><p>1. World Bank, ESMAP - Mini grids for half a billion people</p><p>2. https://globalsolaratlas.info/map</p><p>3. doi: 10.1016/j.energy.2016.08.060</p><p>4. Wikipedia contributors. (2021, January 7). Pearson correlation coefficient. In Wikipedia, The Free Encyclopedia. Retrieved 09:00, January 20, 2021, from https://en.wikipedia.org/w/index.php?title=Pearson_correlation_coefficient&oldid=998963119</p><p>5. Cader. 2018</p><p>5. Hoffmann. 2019</p><p>7. https://doi.org/10.2136/vzj2018.03.0062</p>