Dynamic Payback Period Estimation for Solar Power Plants Through ML Based Solar Power Generation Forecasts

Substantial capital is required to invest in solar power plants, which puts estimation of the payback period accurately at primary concern for stakeholders. In this paper, we proposed a novel method to estimate the payback period of a 10MW solar power plant through forecasting solar power generation by an optimized XGBoost Machine learning model. We began by forecasting the Global Horizontal Irradiance (GHI) and temperature with a temporal resolution of 1 hour for the upcoming year. Using these forecasted values, the expected power output of a single solar panel with given properties is calculated, which was further scaled for the entire plant to estimate the total power generated for entire year with a resolution of 1 hour. Monthly income from the plant is estimated by considering current electricity prices and the predicted power of each month. Plant investments accounted for various costs, including panels, land acquisition, maintenance, wiring, and equipment, while also considering compound interest on the initial investment (compounded monthly) with monthly loan repayments. By considering different scenarios, we determined the payback period by varying factors such as interest rates, land costs, panel efficiency, and maintenance expenses. Our findings offer offer valuable insights into the financial feasibility and potentiality of solar power plants and demonstrate how these factors influence the payback period of such plants.