Graph framework for automated urban energy system modeling

  • Graphen-basiertes Framework zur automatisierten Modellerstellung für Urbane Energiesysteme

Fuchs, Marcus; Müller, Dirk (Thesis advisor); Saelens, Dirk (Thesis advisor)

1. Auflage. - Aachen : E.ON Energy Research Center, RWTH Aachen University (2017, 2018)
Book, Dissertation / PhD Thesis

In: E.ON Energy Research Center ; 57. Ausgabe : EBC, Energy efficient buildings and indoor climate 57. Ausgabe : EBC, Energy efficient buildings and indoor climate
Page(s)/Article-Nr.: 1 Online-Ressource (xii, 120 Seiten) : Illustrationen, Diagrammen

Dissertation, RWTH Aachen University, 2017


Current efforts to decarbonize energy supply chains lead to new challenges for urban energy system modeling. Among the key challenges are increasing model integration to leverage synergies between different subsystems, better prediction of the dynamic system behavior in time-varying operation conditions and workflow automation to handle the increasing system complexity. To help address these challenges, this thesis presents the graph framework uesgraphs for model representations of urban energy systems. For the application of district heating systems, the thesis demonstrates how this framework facilitates automating modeling-related workflows like the creation of scalable use cases and the generation of system models with different modeling approaches, which enables model comparisons and thus supports the development of new dynamic modelling approaches to better address the needs of future urban energy systems. The presented graph framework uesgraphs separates the system description into a model-neutralsystem description layer and leaves the energy modeling itself to an additional model layer to be defined in separate applications. To this end, uesgraphs defines a Python package that extends existing general graph methods to represent different energy networks, buildings and the street network inform of a geo-referenced system graph. An automated management of nodelists allows users to flexibly extract subgraphs of certain energy networks and recombine them into an urban energy systemgraph. Based on this system description in the graph layer, two additional Python packages introduce methods for automated model generation for district heating networks. The package dhcstatic defines methods for quasi-static district heating modeling and simulation in Python while the package uesmodels enables automated generation of dynamic district heating system models in Modelicacode. These applications build a toolkit for in-depth analyses of different approaches to districtheating modeling. A comparison between both modeling approaches shows that a dynamic pipe model accounting for temperature wave propagations through the network can add accuracy regarding the short-term prediction of building substation supply temperatures as well as of the overall network efficiency. In addition, a dynamic substation model not only improves short-term simulation results but also has significant effects on the predicted overall network performance over longer simulation periods. Together, the developed methods and models demonstrate a graph framework for automated urban energy system model generation that facilitates analyses, modelling and simulation and significantly reduces manual effort in the process.


  • Chair of Energy Efficient Buildings and Indoor Climate [419510]
  • E.ON Energy Research Center [080052]