Tobias Gebhard

M.Sc. Tobias Gebhard

Residential Power Demand Modeling

S3|10 310
Landgraf-Georg-Str. 4
64283 Darmstadt

Research Interest

  • Modeling of Residential Energy Demand
  • Modeling of Consumer Dependencies/Synchronization
  • Probabilistic Modeling
  • Data Analytics
  • Digital Twins

 

The modeling of energy demand is important for predictions and ensuring power grid stability. The residential sector exhibits much uncertainty created by human behavior and increasing electricity demand, e.g. due to more and more heat pumps (HPs) and electric vehicles (EVs). In demand modeling, consumers are often considered independent from each other because peaks at different times mostly balance out. However, the synchronized change of power demands can quickly pose a problem to power grids (also known as the "TV pick-up" effect). This could potentially lead to blackouts, as distribution grids are not designed for every household to receive maximal power simultaneously. Therefore, a better understanding of consumer dependencies and the early detection of such anomalies is necessary.

These ideas are investigated in an interdisciplinary context together with the research center emergenCITY, funded by the LOEWE initiative (Hesse, Germany) and the DLR, Institute for the Protection of Terrestrial Infrastructures. Based on the digital city of the future, we tackle the questions on how to maintain its functionalities even in extreme situations, crises and disasters and to increase the reliability and resilience of critical infrastructures. We build demonstrators that simulate a city (for example Darmstadt) as a socio-technical system, considering the interdependencies of energy, communication and water infrastructure.

 

 

Open theses

Supervisor: Tobias Gebhard
Earliest start: Immediately
Type: Master Thesis


The efficient design of low-voltage networks is a difficult task. Over-dimensioning of electrical equipment is economically costly, especially in times of high transformer prices. However, under-dimensioning can lead to overloads, equipment damage, and local blackouts. Current practices for grid planning and capacity design have two drawbacks. First, they usually consider a fixed value as limit for the aggregated load, whereas electric loads need to be seen as probabilistic in nature. Second, the type of connected consumers, their individual consumption patterns (e.g. daily/seasonal), and load correlation with other consumers are often neglected, but have a significant impact on the maximum load.
In this thesis, a new methodology for optimal grid planning and configuration is developed. This concerns for example the capacity design, the geographical transformer placement, and switch placement/configuration. The approach is based on analytical, multivariate statistical modeling. To test and evaluate the method, a data analysis of electricity usage patterns of diverse consumers (e.g. residential, commercial, retail, etc) is carried out. Different datasets can be used for timeseries and correlation analysis, e.g. [1]. The developed algorithm shall yield an optimal solution based on a suitable chosen metric.
Requirements:
  • Basic Knowledge in (multivariate) statistics
  • Interest in statistical modeling and data analysis
  • Attendance in lecture “Data-driven Modeling / Machine Learning & Energy” helpful
  • Attendance in lecture “Energy management & Optimization” helpful
[1] Schlemminger et. al.: Dataset on electrical single-family house and heat pump load profiles in Germany Link

Short Bio

  • Since 2021: PhD Student at EINS and researcher at DLR
  • 2018 - 2021: M.Sc. Mechatronik at TU Darmstadt
  • 2014 - 2018: B.Sc. Mechatronik at TU Darmstadt
     

Publications

Explainable Anomaly Detection for Grid Monitoring using Probabilistic Load Forecasting

[Conference]
Tobias Gebhard, Andrea Tundis, Florian Steinke:
*accepted* Explainable Anomaly Detection for Grid Monitoring using Probabilistic Load Forecasting.
To appear in: IEEE SmartGridComm 2025, Toronto, Canada, 2025

Modeling and monitoring social media dynamics to predict electricity demand peaks

[Journal]
Isabella Nunes Grieser, Tobias Gebhard, Andrea Tundis, Jens Kersten, Tobias Elßner, Florian Steinke:
Modeling and monitoring social media dynamics to predict electricity demand peaks.
In: Elsevier Energy Reports 13 , P. 1548-1557, 2025

Automated Generation of Urban Medium-voltage Grids using OpenStreetMap Data

[Conference]
Tobias Gebhard, Andrea Tundis, Florian Steinke:
Automated Generation of Urban Medium-voltage Grids using OpenStreetMap Data.
In: IEEE PES Innovative Smart Grid Technologies Europe (ISGT Europe), Dubrovnik, Croatia, 2024

Monitoring Electricity Demand Synchronization Using Copulas

[Conference]
Tobias Gebhard, Eva Brucherseifer, Florian Steinke:
Monitoring Electricity Demand Synchronization Using Copulas.
In: IEEE PES Innovative Smart Grid Technologies Europe (ISGT Europe), Novi Sad, Serbia, 2022