学术文献库

Energy system models are increasingly being used to explore scenarios with large shares of variable renewables. This requires input data of high spatial and temporal resolution and places a considerable preprocessing burden on the modeling team. Here we present a new code set with an open source license for automatic generation of input data for large-scale energy system models for arbitrary regions of the world, including sub-national regions, along with an associated generic capacity expansion model of the electricity system. We use ECMWF ERA5 global reanalysis data along with other public geospatial datasets to generate detailed supply curves and hourly capacity factors for solar photovoltaic power, concentrated solar power, onshore and offshore wind power, and existing and future hydropower. Further, we use a machine learning approach to generate synthetic hourly electricity demand series that describe current demand, which we extend to future years using regional SSP scenarios. Finally, our code set automatically generates costs and losses for HVDC interconnections between neighboring regions. The usefulness of our approach is demonstrated by several different case studies based on input data generated by our code. We show that our model runs of a future European electricity system with high share of renewables are in line with results from more detailed models, despite our use of global datasets and synthetic demand.