KIVi - Künstliche Intelligenz zur Untersuchung der Versorgungssicherheit mit Elektrizität

  Logo KIvi Copyright: © FCN-ESE

Project duration: 06/2020 – 05/2023

Funded by: Bundesministerium für Wirtschaft und Energie (BMWi)

Funding code: 03EI1022A

Consortium partners:

  • Centre of Innovative Energy Systems at Düsseldorf University of Applied Sciences (ZIES)

Assoziierte Partner:

Possible losses in the reliability of the electricity supply are frequently listed as important concerns against the phase-out of nuclear energy use and coal-fired power generation. However, current methods for investigating the impact on supply security due to changes in the German power plant portfolio in a consistent manner are extremely time-consuming and have therefore severely limited the number of scenarios that can be investigated.

An interdisciplinary team of scientists from RWTH Aachen University and Düsseldorf University of Applied Sciences is, therefore, developing new approaches based on methods from the field of artificial intelligence to accelerate such analyses of the security of electricity supply. In particular, methods for predicting model input data, data consolidation, and metamodeling are being investigated (see Figure 1). The main research questions in the KIVi project are:

  1. What methods from the field of data science are particularly suited in the context of supply security assessment?
  2. How can these methods best be transferred to:
    • Process data and estimate uncertainties?
    • Avoid the actual modeling process and thus reduce run times through metamodeling?
    • Enable more comprehensive security of supply analyses by avoiding the modeling effort?
  3. To what other uses in energy research can these methods be transferred?
  Meta modeling Copyright: © FCN-ESE Figure 1: Research framework in the KIVi project

The relevance of this topic is reflected by the establishment of a practical advisory board with the participation of representatives of all four German transmission system operators (TSOs) as well as the Federal Network Agency (BNetzA) for the regular transfer of the results into practical application.

First results of the metamodeling and artificial intelligence methods jointly developed by scientists from the Chair of Energy System Economics (FCN-ESE) at RWTH Aachen University and the Center for Innovative Energy Systems (ZIES) at Düsseldorf University of Applied Sciences are promising. In test runs, the computing cluster-based simulations, which originally took around ten hours, could be reduced to computing times of just under two minutes. The aim of the research project now underway is to transfer these initial approaches into robust analysis methods to be able to investigate multiple possible different energy scenarios and their implications for supply security in a fraction of the time originally required.


Selected publications of project results

Journal Articles (full scholarly peer-review)

Köhnen, C. S., Priesmann, J., Nolting, L., Kotzur, L., Robinius, M., Praktiknjo, A. (2021). The Potential of Deep Learning to Reduce Complexity in Energy System Modeling, International Journal of Energy Research. [Wiley]

Nolting, L., Praktiknjo, A. (2021). The complexity dilemma – Insights from security of electricity supply assessments. Energy, 122522. [ScienceDirect]

Behm, C., Nolting, L., Praktiknjo, A. (2020). How to Model European Electricity Load Profiles using Artificial Neural Networks, Applied Energy, 277, 115564. [ScienceDirect]

Nolting, L., Spiegel, T., Reich, M., Adam, M., Praktiknjo, A. (2020). Can Energy System Modeling Benefit from Artificial Neural Networks? Application of Two-stage Metamodels to Reduce Computation of Security of Supply Assessments, Computers & Industrial Engineering, 142. [ScienceDirect]



Aaron Praktiknjo

Head of Chair Energy System Economics (FCN-ESE)


+49 241 80 49870