Reviewing approaches for thermal loads modelling

Following up the positive feedbacks from the post in the Modelling section (here), I propose a discussion to review the existing models/approaches available for the generation of thermal loads that have been reported there. In fact, thermal loads (including both residential and industrial users) are critically needed to fully exploit the potentiality of energy models that allow for the integration of heat and electricity but, unlike electric loads, they are scarcely available from system operators. Therefore, appropriate models are needed to generate them with good approximation.

The objective of the discussion would be to identify the pros and cons of the available approaches, eventually coming up (if needed) with a plan to continue the work after the meeting in the direction of updating/better documenting what is already there or creating something new.


Hello Francesco. Seeing thermal demand is closely related to weather, the following publication on open weather data (and analysis and prediction) might be of use. (Note that some closed energy system models use proprietary weather datasets and are unable to share this information even with their clients!). Robbie.

Dodds, Leigh, Alex Longden, Simon McLellan, and Amanda Smith (April 2017). The state of weather data infrastructure — White paper — ODI-WP-2017-003. London, United Kingdom: Open Data Institute (ODI). Scribd registration required.

Hi Robbie, thanks a lot for sharing this dataset! You’re right, it could be useful especially since most of current approaches are totally based on elaborations of outdoor temperature.

Hi Francesco, hi all,

I will prepare a short tutorial on how to use the demandlib (check out the [github repo]( and the documentation ).

I will show which data is needed and how to get hourly heat profiles with the BDEW approach.

Looking forward to the do-a-thon!


Hi Jann,

sounds great! Very nice idea, thank you!

See you soon,



How to deal with stochasticity when dealing with small systems?
-> Check out enlopy

Data sources:

Make Demandlib adaptable to other contexts:

  • substituting empirical hourly profiles with bottom-up generated ones
  • simplifying the sigmoid function without relying on empirical data, or checking its adaptability to other contexts against other data

People willing to work on creating alternative adaptable versions of current demandlib code:

  • Francesco Lombardi
  • Gabriele Cassetti
  • ?

Collecting open datasets against which models can be tested and validated:

  • Germany (Jann)
  • top-down approach based on Heating Degree Hours (Marta Victoria, Jonas Dahlbæk)

Collecting data across Europe for approaches validation, regarding:

  • district heating systems
  • ext temperatures
  • building clusters (age mainly)
1 Like
Text and images licensed under CC BY 4.0Data licensed under CC0 1.0Code licensed under MITSite terms of serviceOpenmod mailing list.