A data-driven approach to optimize urban scale energy retrofit decisions for residential buildings. (1st June 2020)

• Record Type: Journal Article
• Authors: Ali, Usman; Shamsi, Mohammad Haris; Bohacek, Mark; Hoare, Cathal; Purcell, Karl; Mangina, Eleni; O'Donnell, James
• Is Part Of: Applied energy. Volume 267(2020)
• Publication Date: 2020-06-01
• Subjects: Building energy retrofit -- Building energy performance -- Urban building energy analysis -- Building energy stock -- Retrofit modeling -- Data-driven approach; Power (Mechanics) -- Periodicals; Energy conservation -- Periodicals; Energy conversion -- Periodicals; 621.042
• DOI: 10.1016/j.apenergy.2020.114861
• Languages: English
• ISSNs: 0306-2619

Abstract

Highlights: Evaluation of existing approaches of buildings energy retrofit decisions. Generalized methodology for buildings' energy performance knowledge base. Key retrofit recommendations identified for urban scale residential buildings. Study of reduction in energy demand and emissions at a large scale. Abstract: Urban planners face significant challenges when identifying building energy efficiency opportunities and developing strategies to achieve efficient and sustainable urban environments. A possible scalable solution to tackle this problem is through the analysis of building stock databases. Such databases can support and assist with building energy benchmarking and potential retrofit performance analysis. However, developing a building stock database is a time-intensive modeling procedure that requires extensive data (both geometric and non-geometric). Furthermore, the available data for developing a building database is sparse, inconsistent, diverse and heterogeneous in nature. The main aim of this study is to develop a generic methodology to optimize urban scale energy retrofit decisions for residential buildings using data-driven approaches. Furthermore, data-driven approaches identify the key features influencing building energy performance. The proposed methodology formulates retrofit solutions and identifies optimal features for the residential building stock of Dublin. Results signify the importance of data-driven retrofit modeling as the feature selection process reduces the number of features in Dublin's building stock database from 203 to 56 with a building rating prediction accuracy of 86%. Amongst the 56 features, 16 are identified to be recommended as retrofit measures (such as fabric renovation values and heating system upgrade features) associated with each energy-efficiency rating. Urban planners and energy policymakers could use this methodology to optimize large-scale retrofit implementation, particularly at an urban scale with limited resources. Furthermore, stakeholders at the local authority level can estimate the required retrofit investment costs, emission reductions and energy savings using the target retrofit features of energy-efficiency ratings.