Contents of the thesis

Motivation

The residential sector plays an increasingly important role in the energy transition. Electricity demand in households is rising due to the growing adoption of heat pumps, electric vehicles (EVs) with home charging, and other electric applications. This also increases the potential for load shifting, enabling better integration of renewable energy and more efficient use of time-variable electricity tariffs.

A major challenge is the limited availability of detailed consumption data at the individual household level, due to the slow rollout of smart meters in many countries. A promising alternative to direct measurement is the use of time-use surveys: These datasets record household activities in time resolution and allow the derivation of device-level electricity demand profiles.

Since time-use data is subject to privacy constraints and often cannot be used directly, synthetic modeling approaches are commonly applied in research. These approaches generate realistic profiles based on the characteristics of the original survey data without identifying individual households.

Several such models exist in the literature, but a comparative analysis – particularly evaluating how closely the synthetic demand resembles demand based on direct time-use survey application – is still largely missing.

Objective

This thesis aims to compare multiple existing models that synthetically generate household electricity demand profiles based on activity data from time-use surveys. The models (partially based on existing code) shall be implemented and applied using a freely available time-use dataset.

The core of the thesis is a comparative evaluation of the generated results, focusing on how the modeled demand profiles affect the assessment of time-variable electricity tariffs. The study will also examine how synthetic modeling results differ from those based on direct use of time-use data. Various metrics shall be applied for this comparison.

Possible Research Questions

• Which activity-based models are particularly suitable for generating synthetic household load profiles? (Existing work available)
• What are the differences between the models in terms of the resulting metrics?
• How do the generated profiles influence the assessment of different electricity tariff structures?
• What differences arise when comparing synthetic profiles with those based on the direct use of time-use data?

Methodology

• Review of and familiarization with existing models for household electricity demand modeling based on time-use data
• Selection and implementation of several modeling approaches, potentially extending existing open-source tools
• Application to a freely available time-use dataset
• Comparison of the resulting load profiles using structured metrics
• Simulation of different electricity tariff scenarios (flat, TOU, dynamic) to assess flexibility and cost implications
• Analysis of the differences between synthetic modeling and direct use of time-use survey data

Requirements

• Strong Python programming skills
• Solid mathematical background
• Willingness to work with and extend existing models and codebases
• Interest in household-level energy systems and electricity demand modeling
• Responsible and self-motivated working attitude
• Good German or English language skills (written and spoken)

Formal Aspects

• Language: German or English
• Application (in German or English):
  • Transcript of records
  • CV
  • Brief description of prior programming experience (esp. Python)