Data-driven classification of Urban Energy Units for district-level heating and electricity demand analysis

Abstract

The building sector is a significant contributor to global energy consumption and accounts for approximately one-third of total greenhouse gas emissions. While building energy analysis has traditionally focused on individual buildings, analyzing larger settlements, such as districts or neighbors, offers additional opportunities. The objective of this study is to define and classify typical urban areas for energy analysis, referred to in this paper as Urban Energy Units (UEUs), which represent geographical regions within a city with specific building characteristics, settlement patterns and energy demand. Sixteen different UEUs were classified using literature and open data. The proposed methodology leverages open-source data and uses a random forest model to enhance missing building properties of the building stock such as building age and construction type. It further subdivides the study area into geographically defined sections, and deploys a decision tree model to classify these sections into the sixteen different UEUs. These UEUs enable the creation of energy districts in a modular manner and flexible for its use in any given area. This study demonstrates the practical implications related to the 2023 german municipality heating plan. The methodology was applied in Oldenburg, a mid-sized German city. The city was subdivided into a total of 8249 UEUs, with the detailed results for energy demand presented in this report.