Author:

Abstract:

This dissertation investigates the optimal control of hybrid GEOTABS buildings both in the short- and the long-term, and with a focus on the geothermal borefield aspect. Building energy intensities have decreased since the 90s, but not enough to offset the strong growth in building floor area. As a result, global building energy use and related CO2 emissions continue to rise. More efforts towards energy efficient buildings are necessary. One highly-efficient building concept is the hybrid GEOTABS building, which is comprised of a geothermal heat pump (GEO) connected to a thermally activated building structure (TABS) and extended with a fast-reacting auxiliary production and/or emission system. The envisaged savings of this concept are jeopardized by its operation complexity, therefore optimal controllers such as Model Predictive Control (MPC) are desired. However, the time constant of ground dynamics of the geothermal borefield is too large to be captured by the typical prediction horizons of MPC. In that regard, it is uncertain whether MPC (i) is achieving an optimal solution and (ii) will deplete the ground-source in the long-term. The first part of the dissertation explains in detail the aforementioned problem and the motivations that lead us to this investigation, and introduces the basic concepts and tools used through the dissertation to the reader. The second part of this dissertation elaborates on the methodological developments. In a first step, the borefield short-term dynamics are included in the MPC formulation by incorporating a variable COP formulation and a borefield model. The borefield ground model is modified from an event-based load aggregation scheme to a resistance-capacitance network that represents the thermal diffusion in the ground. It is found that using a variable COP leads to a better control performance and a smarter use of the heat pump, avoiding ground peak loads. A borefield model is necessary, especially when there exists risk of depleting the ground source. To evaluate the long-term dynamics of the geothermal borefield, a so-called 'shadow-cost' is added to the MPC objective function. The borefield ground model is modified from an event-based load aggregation scheme to a continuous one. Using a set of prescribed heating and cooling needs predictions, the optimization is extended by including energy balance equations specific for each case to compute the optimal load split between the different systems. The borefield fluid temperature is affected by the actions of the previous predictions and the short-term optimization. The methodology is validated and demonstrated, showing that there is potential beyond a standard short-term MPC formulation. Since the states within the developed borefield models are hidden or unknown, the accuracy of the models is compromised in real applications. Therefore, state estimators are tested and evaluated in both borefield controller models. It is found that a simple 1-step ahead Kalman Filter algorithm provides accurate results for the fast fluid and grout dynamics, whereas a more complex multi-step algorithm such as Moving Horizon Estimation is more suited for the slower ground dynamics. With the aim of testing the practical application of the methods developed in the second part of the dissertation, the third part presents their application to the emulator of a real building currently operating with short-term MPC. The current MPC implementation of the building assumes a constant temperature extracted from the borefield, leading to thermal imbalance. By incorporating a borefield controller model and the shadow cost into the MPC formulation further energy savings are obtained and thermal imbalance is reduced. The system worked with a borefield downsized by 72.3 % compared to its original size. The novel methodologies worked properly even under limitations set by computational power constraints and model mismatch, demonstrating the flexibility and robustness of the developed methodologies. The fourth and final part concludes the dissertation summarizing the main findings of the investigation and proposing some future lines of research that should be followed from the author's point of view.