GeoPython 2024: Using geospatial data to solve Germany's heat transition problem

Mirjam Kirchner, Lin Ao
Using geospatial data to solve Germany's heat transition problem

To enable the German heat transition (Wärmemwende), we identify heat-network potential areas based on geospatial information about the local heating demands. To answer the question whether it is profitable to connect these potential areas to a heat network, we integrate the locations of heat sources and maximize the capitalized value of the related heat network investment using a linear mixed integer programme.

To enable a smooth heat transition (Wärmewende) for Germany, we help energy providers to identify potential district heating networks based on geospatial information about the local heating demands.
For our input data, we use the results from our heat concept application, which generates simulated results for heating demand as well as heating technologies and renovations present in all 20 million buildings in Germany. From an economics perspective, district heating benefits greatly from economies of scale, i.e., supply as many households as possible at once with the caveat that the underlying infrastructure require a continuous connection to the local source(s) of heat generation.
Our approach first groups existing buildings into areas in the shapes of hexagons based on the H3 approach developed by Uber, hereby creating uniformly shaped smallest units of area for us to work with. Together with our heat experts, we defined a set of properties that identify hexagons of high heat-network potential that we merge into cohesive areas if they are direct neighbors to each other. To answer the question whether it is profitable to connect these potential areas to a heat network, we integrate the locations of heat sources and maximize the capitalized value of the related heat network investment using a linear mixed integer program.
Our results are detailed geospatial datasets consisting of large networks for potential district heating. Additionally, we calculate hundreds of economic, technological and social economical key performance indicators for each of these areas, ensuring that the end users have all necessary information available for further actions.
All of this is made possible by the amazing Python packages for geospatial data such as GeoPandas, shapely, and h3-py. Further, we used Pyomo as a Python-based optimization modeling language.