A two-step cascade modelling between EnergyScope Pathway-BO and PyPSA-BO for energy transition planning. Part B: Geo-spatial characterization and effects

As the transition of energy systems becomes more urgent worldwide, countries and communities are searching for pathways toward sustainable energy systems, which involves developing long-term energy plans and deciding on the key resources and technologies required to meet their future energy needs. In this context, bottom-up models are commonly used to analyze scenarios that assess the development of energy systems. Nevertheless, the technical, temporal, and spatial detail levels vary as systems and cases become more complex.
Various modelling tools are currently available, each tackling issues using distinct approaches or addressing specific aspects of the system’s behaviour. Thus, considering a common objective and structure, redundancies among models could be used to check consistency across models and provide an additional validation layer. Alternatively, models can be enriched by including inputs or supplementary information from other tools, considering the aspects each tool focuses on.
This study is one of a two-part study and addresses the coupling coupling, synergies and complementarities of two models to tackle the issue of planning the development of an energy system in the long term. The first part, focuses on optimizing the investment strategy path from 2021 to 2050 (EnergyScope Pathway), while the second, detailed in this article, concentrates on a detailed characterization of the electrical network and dispatch (PyPSA-Earth). This paper (Part B) analyses the electric system development of Bolivia taking into account the geo-spatial factor, which until this point has been simplified by disregarding the complexities of the transmission system. For the future scenario, the model makes use of the electrical demands expected in 2050 calculated by EnergyScope-Bolivia (Part A of this study).
In this sense, alternative clustering levels are explored between 2 and 62, using the k-means clustering method available in PyPSA-Earth to aggregate network components. An analysis is conducted across over 25 alternative node configurations where total installed capacities, generation, transmission capabilities, and computational runtime are considered. The analysis of optimization runtime and number of nodes considered confirms that a power-law relation exists between these variables, justifying the necessity
of defining an cost-optimal configuration of a modelled network. A Pareto analysis comparing the expansion for solar technologies and the number of nodes shows that a 22 node configuration, would suffice to accumulate over 80% of the deviations compared to the maximum resolution possible, with only a 3% divergence for this technology. Finally, general results from PyPSA-BO are consistent with those from EnergyScope-Bolivia, as both tools provide a titular role to solar PV in the transition of the system. However, discrepancies can be found in the total installed capacities and the inclusion of storage technologies required. These differences open the door for future work where a feedback loop is implemented in the coupling of both tools, providing inputs from PyPSA-BO to EnergyScope-Bolivia.