Do-a-thon: Modeling and optimization of flexibilities in power distribution systems



In this do-a-thon I would like to discuss challenges one faces when analyzing different flexibilities in power distribution systems, such as battery storages, flexible loads, or reactive power provision, and their impact on grid planning. These challenges are quite diverse and include amongst others

  • Open distribution grid data: Open distribution grid data or approaches to synthesize grid data, especially of urban grid structures where data is particularly rare
  • Assessment of flexibility: Open data and approaches to estimate available flexibility of e.g. battery electric vehicles
  • Optimization approaches: Finding the optimal combination of flexibilities through e.g. non-linear optimal power flow or heuristics
  • Complexity reduction: Choosing valid methods to reduce complexity on a temporal scale (e.g. through deriving relevant design cases for grid planning or probabilistic calculations), spatial scale (e.g. clustering of nodes) and/or with respect to the number of scenarios while still obtaining robust results

The topic is purposely kept quite broad as I would like to set the focus of the do-a-thon depending on the participants’ interests and backgrounds. Maybe we could even split up in focus groups or hold two consecutive do-a-thon sessions.
The format could be one where several people each have a chance to present their related work and discuss open questions. The desired outcome of this session is to have a good overview of relevant open data sources, helpful open source tools, and approaches to analyze flexibilities in distribution grids.
I could present two tools we developed in a former project that synthesize medium and low voltage grids [1] and implement heuristics on storage integration and grid supportive curtailment [2], as well as ongoing work on assessing flexibility of battery electric vehicles and non-linear optimal power flow using powermodels.jl.

[1] ding0
[2] eDisGo
[3] PowerModels

Electric distribution modeling
Do-a-thon pandapower / PowerModels

Quite a lot has been published on energy sector flexibility concepts. This position paper from Germany provides a good overview and might be a reasonable place to start?


acatech, Leopoldina, Akademienunion (February 2016). Flexibility concepts for the German power supply in 2050: ensuring stability in the age of renewable energies — Position paper. Germany: acatech, Leopoldina, Akademienunion. ISBN 978-3-8047-3549-1.


I just proposed a do-a-thon about pandapower/PowerModels that might fit well within this topic. Considering the broad nature of the topic it might still be worth it to split it up into several sessions. For example the topic about openly available networks (real/generic) might be specific enough for its own session?


At TH Köln we are currently working on a little storage model (python-based with the aim to integrate/couple it to pypsa) to integrate and evaluate also different operation strategies, which are not optimized. We would be interested in a focus group (as proposed here by Birgit) regarding that topic. I think it would be nice to exchange in advance, to know where exactly to set the focus. Maybe open_bea is heading in a similar direction? Any other opinions/hints on this topic?


I’m sorry for replying so late, I hadn’t had time to prepare much for the session yet… I really like your proposal @cbrosig to focus on integrating battery storages with different operation strategies into grid analyses, as that is something we will have to tackle in open_BEA as well. My project partner Daniel is doing a do-a-thon on battery modeling where he will present the battery model SimSES developed by the Technical University of Munich. In open_BEA we want to couple their model to our tool eDisGo.
For the power flow analysis we are currently using pypsa, but we are also working on implementing an OPF for which we will either directly integrate PowerModels or use pandapower. The optimization will be something we can talk about in Leon’s session.
However, the OPF won’t suffice as for example our battery models will most likely be too complex and the time frames too large to do an optimization for. This is where this session could come into play. We could each shortly present our existing approaches as well as some approaches from literature on how storages were considered in distribution grid analyses and then present our ideas for our current projects and discuss them. Questions I would like to discuss are what options are there to couple battery models and grid analyses tools and what are their advantages and disadvantages, what time frames do we need to analyse, how to integrate a Q(U) control. @cbrosig what do you think?
@leon.thurner it sounds like openly available networks will be discussed in this session. Is the integration of battery models into grid analyses something you are working on and/or would be interested in as well?