Long-term energy system modelling: the impact of different time-series clustering algorithms

ABSTRACT Climate change poses severe challenges to human and environmental the environment around them. Therefore, it is necessary to act in the energy domain especially as a significant source for climate change is greenhouse gas emissions. Effective policy measures are imperative for addressing this critical issue. A key strategy is the transition from conventional fossil fuels to renewable energy sources. This strategy involves a fundamental shift from carbon-based electricity to a diversified energy portfolio predominantly constituted of renewables, complemented by a minor share of gas. The thesis at hand aims to conduct extensive energy system modelling for the island of Favignana over three different scenarios, identifying the most efficient clustering algorithm to reduce computational demands while maintaining precise outcomes. Utilizing an advanced iteration of the OpenSource energy Modelling System (OSeMOSYS), the energy systems are constructed through the clustering method applied to time-series data. This approach uses representative days (RDs) for various years, considering critical attributes on both demand and supply fronts. These RDs are to be clustered using diverse algorithms, with the objective of pinpointing the optimal one. The scenarios developed characterize distinct operational conditions of the island's energy systems, which are sole utilization of photovoltaic systems, sole dependence on wind energy, and a mixture of both, alongside the incorporation of various storage technologies such as lithium-ion batteries and hydrogen storage. The findings underscore that an aggressive decarbonization strategy is not only viable but also advantageous, and that different clustering algorithms exhibit varying degrees of suitability across different scenarios.