Scholarlink

<p>Material stocks (MS) represent the amount and composition of construction resources embodied in the built environment, and are critical to understanding long-term resource use, waste generation, and circular economy strategies. The reliability of building MS modelling depends largely on the appropriateness of archetype classification, with building structure being a key descriptor as it reflects the core materials used. Yet, structural information is often missing from building inventories, particularly in Europe. In contrast, building features such as shape (e.g., GIS polygons), age, and height can be derived from increasingly widely available data. This study develops and tests a model that predicts building structures based on building features—particularly morphological features, using Gothenburg as a case study. A dataset of 331 multi-family residential buildings was compiled, combining building features with manually identified structural types. Multinomial logistic regression was applied to examine the relationship between building features and building structures. The results show that morphological indicators and building age significantly influence structural prediction, with building height and age exerting the greatest impact. The model demonstrated strong predictive performance, with an accuracy of 83%. Based on the predicted building structures, we estimated the MS of multi-family buildings in Gothenburg, amounting to a total of 17.2 megatons, with an uncertainty from structure classification amounting to 5.8 megatons. This study supports the enrichment of building inventory with structural information, thus supporting a more robust accounting of MS.</p>

错误：搜索内容不能为空，请输入英文关键词
错误：关键词超出字数限制，请精简
高级检索

Enhanced material stock accounting by predicting building structures with morphological indicators

  • 期刊论文

  • 出版模式：

    开放获取

  • 发表日期：

    2026年5月10日

查看全文
  • Xin Bian,
  • Meta Berghauser Pont,
  • Jonathan Cohen,
  • Maud Lanau

摘要

Material stocks (MS) represent the amount and composition of construction resources embodied in the built environment, and are critical to understanding long-term resource use, waste generation, and circular economy strategies. The reliability of building MS modelling depends largely on the appropriateness of archetype classification, with building structure being a key descriptor as it reflects the core materials used. Yet, structural information is often missing from building inventories, particularly in Europe. In contrast, building features such as shape (e.g., GIS polygons), age, and height can be derived from increasingly widely available data. This study develops and tests a model that predicts building structures based on building features—particularly morphological features, using Gothenburg as a case study. A dataset of 331 multi-family residential buildings was compiled, combining building features with manually identified structural types. Multinomial logistic regression was applied to examine the relationship between building features and building structures. The results show that morphological indicators and building age significantly influence structural prediction, with building height and age exerting the greatest impact. The model demonstrated strong predictive performance, with an accuracy of 83%. Based on the predicted building structures, we estimated the MS of multi-family buildings in Gothenburg, amounting to a total of 17.2 megatons, with an uncertainty from structure classification amounting to 5.8 megatons. This study supports the enrichment of building inventory with structural information, thus supporting a more robust accounting of MS.