Online Learning for an Energy Storage Management Problem
Dr. Stephan Meisel
Associate Professor, IEBIS Department, University of Twente.
We consider a periodic energy storage management problem under unknown uncertainty. In this problem, sequential decisions about (dis-)charging the storage and about buying or selling energy at the spot market must be made during each of a number of subsequent periods. The market price of energy is the main source of uncertainty, and the goal is to maximize the expected profit of the storage management process over all periods. Such a problem occurs, e.g., with energy storage management service providers that do not have an accurate model of the spot market price processes. We formulate the problem as an online learning problem, and we distinguish between two types of learning approaches to solve the problem. We show that choosing one of the approaches over the other may, in practice, result in significant performance loss, and we propose a method for solving the learning problem with performance guarantees
Stephan Meisel is an Associate Professor of Stochastic Operations Research in the Industrial Engineering and Business Information Systems Department (IEBIS) at the University of Twente. Before joining the department, he was an Assistant Professor of Quantitative Methods for Logistics at the University of Münster, Germany. Prior to that, he was a post-doctoral research associate in the Department of Operations Research and Financial Engineering at Princeton University. At the University of Münster, he was leading the research group Quantitative Methods for Logistics. His main research focus is on sequential decision making under uncertainty with applications in logistics and energy systems.
Circular (De-)Construction Matchmaking
Yifei Yu
Ph.D. Candidate, IEBIS Department, University of Twente.
Aligned with the European Circular Economy (CE) action plan, the construction industry strives for a sustainable and circular future. Industrial Symbiosis is introduced to foster the emergence of circular construction ecosystems where heterogeneous projects are symbiotically connected via waste-to-resource matchmaking. However, it is challenging to ensure efficient matchmaking due to quantity, quality, and spatial-temporal uncertainties embedded in construction and demolition projects. The development of methods to identify, coordinate, and evaluate symbiotic matchmaking remains an open issue in the built environment. This research proposes an agent-based simulation model for (de-)construction matchmaking from a complex adaptive system perspective. A conceptual model framework is developed based on the theory of shearing layers. The framework is used to instantiate a simulation model demonstrating matchmaking dynamics over time and space. The results indicate that waste transfer hubs could provide extra storage time and allow larger transportation distances, thus, potentially contributing to successful matchmaking. In future research, the model’s applicability and validity will be illustrated through urban mining cases in the city of Enschede, the Netherlands.
Yifei Yu is a full-time PhD student at IEBIS. He obtained his Master's degree in Construction Management and Engineering specializing in digital technology design at the Faculty of Engineering Technology in 2020. His current PhD project is supported by BMS Signature PhD Grant aiming to foster inter-disciplinary research collaborations. He investigates the scientific nexus of Circular Economy (CE), Construction Supply Chain Management, Information Systems, and Public Policy. He aims to provide scientific theories for designing a Circularity Information Platform to support CE business- and policy-making in the built environment. Methodologically, he focuses on the integration of Agent-Based Modelling, Building Information Modelling, and Geographic Information Systems, blending with Enterprise Architecture Design. He aims to improve construction resource efficiency and ultimately gravitates a paradigm shift toward Smart Circular Construction Ecosystems.