Effects of the meteorological data resolution and aggregation on the optimal design of photovoltaic power plants

The accuracy of the simulation and optimization of ground-mounted photovoltaic plants depends on the reliability of the meteorological datasets, which is affected by their source, length, and resolution. Quantifying the effect of the irradiance data temporal resolution on the optimal design parameters and the expected profitability is important to improve the credibility of photovoltaic design simulations. The optimal values of nine important design parameters, the annual energy yield, and the levelized cost of electricity are calculated for 50 Baseline Surface Radiation Network stations by an automatic photovoltaic optimization method based on datasets with six different resolutions created by two aggregation methods. The aggregation by averaging suppresses the high irradiance values resulting from the transient cloud enhancement effect, while the aggregation by sampling a single value from the middle of each aggregation interval can retain the original distribution of minute-resolution data. The optimization based on averaged hourly data underestimates the levelized cost of electricity by up to 3%, overestimates the inverter sizing ratio by 0.05 on average, and the tilt angle by up to 5 degrees compared to the high-resolution datasets. The datasets aggregated by sampling provide more reliable results even at lower resolutions; therefore, it can be an effective technique for accurate photovoltaic optimization without the long calculation time of the minute-resolution simulation.

Automated summary

The accuracy of ground-mounted photovoltaic plant simulation and optimization is dependent on the reliability of meteorological datasets, which are influenced by their source, length, and resolution. Quantifying the impact of irradiance data temporal resolution on design parameters and profitability is crucial for enhancing the credibility of photovoltaic design simulations. Aggregating datasets by sampling can provide more reliable results even at lower resolutions, making it an effective technique for accurate photovoltaic optimization without the long calculation time of minute-resolution simulations.