`Solving' the energy transition - Research on the interface between optimization and application

28^th January 2025, 13:00
Leoni Winschermann
University of Twente

Abstract

The on-going energy transition gives rise to a number of fascinating optimization problems. The content of this talk is motivated by a real-life use case, from which we dive deeper into a concrete algorithm using methods from discrete optimization, flow networks and scheduling in general.

In particular, we consider an office parking lot where employees charge their electric vehicles. In order to prevent power outages and damage to the grid infrastructure, we have to coordinate (or schedule) charging in such a way that existing grid capacity limits are respected. Furthermore, we consider two complicating factors:
(i) From the user-perspective, charging cannot be infinitely delayed. No-one wants to wait around after work for their car to finish charging. Furthermore, users tend to mistrust charging systems that act as a blackbox, i.e., where they don't know what to expect.
(ii) We lack the input data needed for our optimization algorithms. Client visits, part time work, emergencies with the kids are just some of the reasons one may want to leave before the end of the working day. This makes it hard to predict the exact moment in time that a vehicle wants to leave.

This talk takes a real-life office parking lot in the Netherlands as an example, where both user needs and information gaps are currently addressed using (among others) so-called charging guarantees. The employer clearly communicates to the employees that they will have charged enough for their work-home commute by 4pm. Starting from a description of the use case, we dive deeper into the mentioned problems, and (on a high level) discuss optimization algorithms we developed to support this and similar use cases.

Biography

Leoni Winschermann is researcher at University of Twente in the Netherlands that applies her mathematical background to problems related to the energy transition. Her main focus is on optimization algorithms for scheduling large groups of electric vehicles. The goal is to on the one hand develop provably optimal or (approximately) good algorithms, and on the other hand to validate their performance with real-world data and to take system users (and their concerns/interests) into account. She conducts her work with support of the SmoothEMS met GridShield project, and is in the final stages of her doctoral degree at the University of Twente.