IAGO CUPEIRO FIGUEROA WILL BE PRESENTING HIS PHD DEFENCE 17 MARCH 2021 15.00 CET

Our consortium partner Iago Cupeiro Figueroa will be presenting his PhD defence,  "Short- and Long-Term Optimal Control of Hybrid GEOTABS Buildings" on Weds 17 March at 15.00 CET. Register here: bit.ly/3ewte7p

This thesis investigates the optimal control of hybridGEOTABS buildings in both the short and long term and with the focus on the geothermal drilling field aspect. The energy intensity of buildings has decreased since the 1990s, but is not yet sufficient to compensate for the sharp increase we are seeing in the floor space of buildings. As a result of this we see a strong increase in the global energy use in buildings as well as in the related CO₂ emissions. More efforts towards energy-efficient buildings are therefore necessary. GEOTABS is a very efficient building concept that consists of a geothermal heat pump (GEO) that is connected to a thermally activated building structure (TABS) and is expanded with a fast-reacting supplementary production and / or delivery system. However, the anticipated savings of this concept are offset by the operational complexity, making optimal controllers such as Model Predictive Control (MPC) highly recommended. However, the time constant of the dynamic processes in the ground of the geothermal drilling field is much larger than the typical MPC prediction horizon. Therefore it is uncertain whether MPC (i) provides the optimal solution and (ii) will possibly deplete the soil in the long term. The first part of the thesis discusses in detail the problem mentioned above and the motivation that led us to this research. Furthermore, it introduces the reader to the basic concepts and tools we have used during the research.

The second part of the thesis focuses on methodological developments. In a first step, the MPC formulation is extended with the short-term drilling field dynamics by integrating a variable COP formulation and a dynamic drilling field model. The drill field ground model was adapted from an event-based load aggregation scheme to a resistance-capacitance network that represents the thermal diffusion in the ground. We were able to establish that the use of a variable COP leads to better control and smarter use of the heat pump and that peak loads are avoided. A drilling field model is necessary, especially when there is a risk of soil exhaustion. To investigate the long-term dynamics of the geothermal drilling field, a so-called “shadow cost” has been added to the objective function of the MPC. The drill field ground model has been adapted from an event-based load aggregation scheme to a continuous scheme. Using predefined heat and cooling needs, the optimisation has been expanded to include energy balance equations for each specific case, in order to calculate the optimal load split between the different systems. The drilling field fluid temperature is affected by the actions of the foregoing predictions and the short-term optimisation. The methodology has been validated and demonstrated, showing that there is potential in a step beyond the standard short-term MPC formulation. Since some states are hidden or unknown within the developed drilling field models, the accuracy of the models in an actual application remains to be seen. Therefore, condition estimators were tested and evaluated in both drill field controller models. A simple 1-step Kalman filter provides accurate results for the fast processes in the heat transfer fluid and well fill, while a more complex multi-step algorithm, such as “Moving horizon estimation”, is more suitable for the slower processes in the soil around the wellbore. 

With the aim of testing the practical applicability of the methods developed in the second part of this thesis, the third part presents the application of these methods in an emulator of a real building currently working with the short-term MPC. The current MPC implementation of the building assumes a constant temperature for the drilling field, leading to a thermal imbalance. The addition of a drilling field model in the controller and the shadow cost in the MPC formulation further energy savings and reduce the thermal imbalance. The system was able to operate with a drill field reduced by 72.3% compared to its original size. The new methodologies worked well even under the limits of computational power constraints and model mismatch, demonstrating the flexibility and robustness of the developed methodologies. 

The fourth and final part summarizes the main findings of the study and makes new research proposals for the future that, from the author's perspective, should follow this study.