Transforming the built environment: Leveraging large-scale retrofit and building-level demand flexibility
Open access
Author
Date
2023Type
- Doctoral Thesis
ETH Bibliography
yes
Altmetrics
Abstract
Our world is undergoing a significant energy transition, shifting from fossil fuels to renewable energy sources. This transition is driven by factors such as population and economic growth, depletion of fossil fuel reserves, and higher living standards. Simultaneously, this transition is aimed towards a more sustainable built environment. Buildings, as an essential part of our society, and power grids are expected to transform and play a crucial role in facilitating this transition.
Power grids are evolving to interact with the demand-side for frequency regulation, mitigating the challenges associated with integrating volatile renewable energy sources. Meanwhile, buildings are transforming into "prosumers", regulating their power consumption to support grid operation
and planning while maintaining occupant thermal comfort. Buildings hold a considerable energy load that can be reduced to meet ambitious energy targets in Switzerland and Europe. On the other hand, this energy can serve as a flexible load, providing ancillary services through demand-side
management.
Building retrofit is recognized as a key solution for making the existing building sector more environmentally friendly. However, identifying optimal retrofit solutions involves many complex challenges due to typically conflicting objectives and the utilization of highly heterogeneous data. At the same time, adjusting the various flexible sources within a building through retrofitting can impact the potential for demand response and vice versa. Consequently, integrating building retrofit and demand flexibility to quantify this impact highlights a pressing research need.
This dissertation aims to support this energy transition by evaluating grid-responsive, environmentally friendly, and computationally efficient solutions for the building stock. The first part focuses on improving the identification of near-optimal retrofit solutions at both the building level and large scale. This involves training a building-level retrofit model using artificial neural networks with real building and retrofit data from a conventional method. The aim is to develop a retrofit model that offers ease of application, efficient data collection, a great balance between accuracy and computational cost, and scalability. The next step involves
enhancing the scalability of this model to create a bottom-up large-scale retrofit framework, incorporating building archetypes and climatic data for improved generalization. The building-level model and the large-scale framework are then applied and tested on the Swiss residential building stock, showcasing their performances through real case studies.
The second part of the thesis introduces a demand flexibility quantification methodology, leveraging a conventional retrofit model for generating optimal building retrofit solutions. The methodology employs co-simulation with a white-box building model and a predictive control algorithm to quantify demand flexibility while ensuring occupants’ thermal comfort. Through
this approach, our objectives are to assess the impact of building retrofit on demand flexibility, enhance the retrofit decision-making process, and provide necessary data for reducing grid reinforcement costs, overloading issues, and facilitating balancing services.
The main findings of this thesis highlight the advantages of utilizing artificial neural networks for building-level and large-scale retrofit analysis. Notably, near-optimal retrofit solutions are calculated swiftly, eliminating the need for parameter calibration and extensive input data collection. Additionally, results demonstrate the effectiveness of the proposed
machine-learning-based retrofit methods in efficiently identifying near-optimal retrofit solutions across large geographical areas and numerous buildings. Furthermore, the results emphasize the significant but building-specific influence of change on the building envelope and energy systems on the potential for providing flexible reserves. They also underscore their substantial impact on the retrofit decision-making process, especially when considering multiple decision criteria.
Overall, this thesis highlights the importance of investigating both building retrofit and demand-flexibility potential for a smoother energy transition. Precisely, it provides specific methodologies and tools to improve accessibility to the identification of retrofit solutions, with the aim of accelerating retrofit rates and supporting informed retrofit decision-making and effective grid planning. Show more
Permanent link
https://doi.org/10.3929/ethz-b-000647401Publication status
publishedExternal links
Search print copy at ETH Library
Contributors
Examiner: Hug, Gabriela
Examiner: Orehounig, Kristina
Examiner: Nagy, Zoltan
Examiner: Miller, Clayton
Publisher
ETH ZurichSubject
Energy efficiency; Energy system modeling; Urban energy systems; Building retrofit; Machine Learning; Surrogate model; Pareto optimal; Energy hub; Building simulation; Planning; retrofit decision-making; Multi-criteria decision making; Demand response; Demand-side management; Building envelope; Scalability; Data-driven models; Artificial neural networks; Co-simulation; Grid planning; Balancing services; Large-scale studyOrganisational unit
09481 - Hug, Gabriela / Hug, Gabriela
More
Show all metadata
ETH Bibliography
yes
Altmetrics