Postdoc position on control algorithms for energy flexibility in industry

This post-doc position is framed in and funded by the VLAIO-Moonshot-InduFlexControl II project, which is a collaboration between KU Leuven, VITO and UGent. The main research question is: How can we unlock and enable increasing amounts of flexibility in the energy-intensive industry by developing ground-breaking control techniques which will allow the incorporation of new sources of flexibility while remaining suitable for the overall eco-system?

In general terms, flexibility refers to the ability of a system to deviate from a given plan and respond to short term changes. However, in the energy-intensive industry, most of the processes are foreseen to run at almost full capacity to achieve the maximum profit. Thus, although flexibility is present in these systems its use in an energy context is not always obvious, as on most occasions that flexibility was conceived for a different purpose (e.g., reliability, processes interrelationship, etc.). 

To transform our energy system and industry into a sustainable, low carbon and climate-friendly eco-system we need to look into new alternatives. One enabler is the incorporation of increasing capacities of renewable energy sources (RES) into the grid or within industrial facilities. This comes at the cost of higher uncertainty in the energy supply due to the RES’ dependency on environmental factors. A solution to this is unlocking a higher level of flexibility at the consumer side, a flexibility that will be necessary to support the stable and secure operation, as well as guaranteeing system adequacy and resilience, while benefiting from low carbon emissions. 

With this fundamental idea, the preceding sprint-cSBO project investigated the exploitation of underlaying flexibility that could be found in the energy-intensive processes. The results have demonstrated the existence of large amounts of inherent flexibility in main processes (such as furnaces or electrolysers), as well as auxiliary system (e.g., cooling towers, compression systems, etc.). Nevertheless, when considering the expected profit, reliability of the system and production quality, further flexibility is needed to accommodate increasing levels of low emission energy sources. Within the range of available flexibility options, power-to-X solutions provide the highest flexibility potential, because of their large energy volumes (hundreds of MWh to tens of GWh), high power ratings (tens to hundreds of MW) and long-term storage potential (weeks, months, seasons). 

To unlock this flexibility potential, we aim to combine the unique potential of model-based and data-driven modelling and control approaches to provide a feasible solution. To guarantee the valorisation of the fundamental research question, these aspects are not considered in isolation, but strategic contextual factors are incorporated in the research plan: (i) the energy market design, and (ii) the power/energy network configuration, with their relevant constraints. 

The post-doctoral researcher in The SySi Team will focus on learning-based MPC to design a robust controller that handles uncertainties and disturbances. The main task is to design and develop robust, stable, lightweight, and self-tuning Deep Learning (DL)-based MPC which will handle constraints on input, output, and states, plant-model mismatch, disturbances on inputs and outputs, and guarantee closed-loop stability with less computational complexity. The potential of this novel approach needs to be (virtually) tested (incl. benchmarking with respect to conventional control strategies) and demonstrated for multi-carrier energy systems in industry and different types of storage, incl. power-to-X.