The Role of Spatial Scale in Electricity System Optimisation Models

To investigate possible pathways to reduce greenhouse gas emissions in the electricity sector, researchers build optimisation models that typically minimise the total system costs such that all technical and physical constraints are met. For systems based on renewable energy, whose greatest expansion potentials are found for wind and solar generation, the chief challenge is dealing with their variability. To tackle this challenge, the optimisation models typically include large transmission networks to smooth renewable feed-in in space or storage technologies to smooth the variability in time. However, all aspects of the energy system at all levels of detail cannot currently be contained in a single model because of computational constraints. Instead, one must make simplifications and compromises that affect the optimality of the result from the point of view of the complete system. While reductions on the temporal scale and linearisation approaches of the model formulation have been previously analysed, in this thesis we focus on the quantification of the impact of the spatial scale. This is important because it is scientific practice to simplify models spatially while only little is known on the error made by the aggregation.
... mehr

The contents of this dissertations spatial scale analysis are three-fold and build upon one another: (i) A novel clustering methodology enables us to disentangle and quantify the error that is made by spatially aggregating generation sites where renewable electricity can be sourced versus the error made by aggregating transmission lines and, thus, electricity interactions between spatially distributed substations. By clustering the network on both features in tandem, we can verify the results and learn which of these two effects dominates the optimisation. (ii) Insights from (i) are used to improve existing spatial aggregation methods and to develop novel similarity measures to be applied for clustering electricity system models such that the spatially simplified model can better approximate the original, highly-resolved model with respect to renewable generation sites and the transmission grid. (iii) The prevailing best clustering method is applied on optimisation models with high shares of renewable generation to investigate if the spatially clustered low-resolved solutions are feasible with regard to the full, spatially highly-resolved model. To this end we propose novel inverse methods to spatially disaggregate the coarse optimisation solution in terms of the resulting, aggregated variables across the highly dimensioned model.