This repository contains all the code and documentation from my PhD research on data-driven methodologies for building efficiency in the district heating sector. It includes R scripts for time series analysis, machine learning, and energy optimization using smart heat meter data. Below you can read the abstract of my thesis.
The urgency to address environmental sustainability has intensified the focus on decarbonizing energy sources across various sectors, including district heating (DH) systems. As these systems are integral to urban infrastructure, providing heating solutions that are efficient and potentially environmentally friendly when configured and optimized correctly. However, although under transformation many existing DH networks still rely on carbon-intensive energy sources, which contributes significantly to large carbon emissions. Therefore, there is a pressing need to enhance the efficiency and operational effectiveness of these systems to replace their energy sources with renewable solutions. Innovations such as smart heat meters (SHM) offer potential breakthroughs in managing and optimizing DH systems, aiming to reduce their carbon footprint while improving overall grid performance. This dissertation investigates the innovative use of SHM data in buildings connected to the DH grid while emphasizing advancements in energy efficiency and fault detection and diagnosis. It is structured across multiple key chapters, each responding to specific research questions outlined in Chapter 1.
The main contributions of this Ph.D. work are:
- Validated the effectiveness of SHM in assessing buildings within the DH grid, demonstrating that incorporating weather data can help operators better predict and manage their energy usage. Additionally, using information from EPC helps identify buildings needing efficiency improvements, while visualization tools assist in interpreting data to make informed decisions.
- Developed a methodology using machine learning models based on SHM data to accurately disaggregate and estimate energy demands for space heating and domestic hot water. This work also addressed the impacts of data rounding in utility data collection, proposing methods to mitigate these issues.
- Examined the influence of occupant behavior on energy usage, noting that SHM data alone are insufficient for capturing all behavioral nuances. Incorporating detailed indoor condition data and occupant interviews provides a more comprehensive understanding, leading to improved energy management strategies.
- Investigated the efficacy of SHM in fault detection and diagnosis within DH systems. Demonstrated that continuous monitoring allows for the application of advanced analytical methods for automated fault diagnosis, hence improving operational efficiency and intervention costs.
Overall, this dissertation provides a foundational framework for utilizing SHM data in buildings with other sources of data to enhance the efficiency and sustainability of DH systems, highlighting the significant potential for future advancements through continued research and technology integration.
District heating networks; Smart heat meters; Building energy efficiency; Fault detection and diagnosis; Machine learning; Heating systems performance.
