Computationally scalable geospatial network and routing analysis through multi-level spatial clustering.

Network analysis finds natural applications in geospatial information systems for a range of applications, notably for thermal grids, which are important for decarbonising thermal energy supply. These analyses are required to operate over a large range of geographic scales. This is a challenge for existing approaches, which face computational scaling challenges with the large datasets now available, such as building and road network data for an entire country. This work presents a system for geospatial modelling of thermal networks including their routing through the existing road network and calculation of flows through the network. This is in contrast to previous thermal network analysis work which could only work with simplified aggregated data. • We apply multi-level spatial clustering which enables parallelisation of work sets. • We develop algorithms and data processing pipelines for calculating network routing. • We use cluster-level caching to enable rapid evaluation of model variants.