Dr. S.H. (Simon) Tindemans
Simon Tindemans is an Assistant Professor in the Intelligent Electrical Power Grids group at TU Delft. Previously, he was with the Control and Power group at Imperial College London as a Research Fellow, Marie-Curie Intra-European Fellow and Research Associate. Before switching his research focus to energy systems, he performed his PhD research at AMOLF (Amsterdam, NL), furthering understanding pattern formation in biomolecular systems. He obtained an MSc in theoretical physics from the University of Amsterdam.
The ambition to drastically reduce greenhouse gas emissions is driving unprecedented changes to the way we design and operate the electricity grid. My research targets three challenging areas of interest for current and future electricity grids: data analytics, efficient simulation methods and control of decentralised loads - with a particular interest in the overlaps and interactions between these areas.
In June 2017, I gave a talk at Tech Foresight 2037 on the future of decentralised electricity grids. You can watch it here. More technical talks about smart fridges and optimal use of heterogeneous battery sets are also available online.
In modern electrical power systems, sensors and other sources generate a growing stream of data. Efficient operation of the system requires extracting useful information and actionable insights from this data. My work in this area focuses on statistical learning of predictive models. Examples include the learning of data-driven surrogate models (proxies) that approximate complex elaborate models, and the quantification of uncertainty in predictive models.
Efficient computational methods
Accurate models of large, intelligent grids are rapidly increasing in complexity. Even if they are accurately validated, exploring the vast space of future events with such models is often unacceptably slow. Moreover, we are often interested in rare high-impact low-probability (HILP) events, where computational bottlenecks are especially prominent. I am working on various Monte Carlo methods that improve sampling efficiency, including importance sampling, active learning and multi-level Monte Carlo.
Control of decentralised flexible loads
Smart appliances and responsive end users provide a significant potential for demand response, but it is not clear what the best approach is for unlocking this potential. Decentralised control with minimal communication requirements is an attractive proposition from the perspective of practical implementation (communication requirements) and privacy (amount of information exchanged; local control decisions). I investigate decentralised aggregate control strategies for smart thermal loads, e.g. refrigerators, air conditioners.
- ancillary activities
- No secondary work - Alan Turing Institute
2018-06-01 - 2021-06-01
Consultancy / research and other