Abstract
Greenhouse gas emission reduction is a world-wide challenge that encourages most research in the building sector. Many sources in Europe, assess the penetration of housing final energy needs (Heating Cooling and Domestic Hot Water – DHW) as up to 40% of the total energy end use. To reduce gas emissions effectively, it is necessary to improve the efficiency of the overall building system. This means that both passive solutions (envelope and windows) and active solutions (HVAC systems) have to be considered for improvements. Due to the nature of the European building stock and since the market of new buildings has faded away and reached saturation, deep refurbishment is promoted as one of the major solutions for addressing the building energy problem. For clarity, deep renovation attempts to improve a building as an overall unit, considering both the HVAC system and envelope solutions as fundamental actions to work with to obtain better results.
In this context, this research is focused on the improvement of HVAC systems for a deep renovation solution with a specific orientation on their control strategies. Recently, to increase the efficiency of these complex assembly of components new design solutions have been developed: the so-called “hybrid” solutions. These augment the capacity of an HVAC system to gather and transform energy from different energy sources (solar, air, ground). Unfortunately, in multi-family buildings the system centralization required to ease the multi-source management, causes inefficiencies related to energy distribution. These inefficiencies can be reduced by means of using advanced strategies for optimizing the use of such complex centralized systems.
This work is structured in three parts. After the initial contextualization and identification of the problem, the second part faces the issue about the HVAC system for residential multi-family homes (MFH) complexity. This part simulates, through the implementation of conventional control strategies, many cases under different HVAC design and boundary conditions (various climatic conditions and building typologies). Furthermore, among the studied cases, one more detailed model has been developed and validated according to an existing building. The outcomes of this first part identify and assess, in addition to the performance of the system, the main inefficiencies which lead to HVAC system high energy losses. For MFH, the use of DHW recirculation can lead to many energy losses which could be avoided with more demand-oriented strategies. The energy losses observed in numerical models have been supported by means of monitoring data of an existing building.
The third part of this work aims to find out solutions for the gaps discovered or revealed in the first part. This section explains the development of a novel algorithm derived by the combination of a non-homogeneous Markov chain model and with reinforcement learning automatic tuning. The algorithm performance, has been shown for the sake of completeness, an energetic analysis has been performed to quantify the energy savings that is possible to obtain through its use. Therefore, energy savings have to find a trade off with the discomfort hours.
Finally, considerations in terms of robustness and potential application in real case scenarios are discussed as a conclusion to this work.
The objective of this work is to assess and understand if the advanced strategy developed during the doctorate can be an effective solution to improve the efficiency of complex HVAC system.