Empirical Quantification and Prediction of Building Component Lifespans with Graph Neural Networks
(2025) Sustainable Built Environment Conference, SBE 2025 Zurich 1554.- Abstract
Extending the service life of existing buildings and their components is essential for slowing resource loops in circular construction. While overall building lifespans are relatively well understood, the actual service lives of individual components remain underexplored due to limited empirical data. This study addresses this gap by analyzing and modeling building component lifespans using real-world damage investigation records and condition assessments from Swedish buildings. Descriptive statistics reveal a wide range of lifespans across six major component categories, highlighting materials linked to damage from poorly designed construction details as a critical factor. Additionally, most residential components have markedly shorter... (More)
Extending the service life of existing buildings and their components is essential for slowing resource loops in circular construction. While overall building lifespans are relatively well understood, the actual service lives of individual components remain underexplored due to limited empirical data. This study addresses this gap by analyzing and modeling building component lifespans using real-world damage investigation records and condition assessments from Swedish buildings. Descriptive statistics reveal a wide range of lifespans across six major component categories, highlighting materials linked to damage from poorly designed construction details as a critical factor. Additionally, most residential components have markedly shorter lifespans-about 25-30 years-compared to non-residential ones. To predict component service lifespans, we configure building envelope deterioration factors as a generic acyclic graph and employ Graph Neural Networks for node-level regression. Among the tested models, Graph Convolutional Network achieved the highest performance (R2 = 0.83), outperforming GraphSAGE and Graph Attention Network due to its effective uniform aggregation across neighboring nodes. Our findings provide a high-granular, data-driven approach to estimating component lifespans, which can enhance predictive maintenance strategies and optimize the reuse potential of reclaimed materials in their next lifecycles.
(Less)
- author
- Wu, P. Y. LU and Mundt-Petersen, S. O. LU
- organization
- publishing date
- 2025
- type
- Chapter in Book/Report/Conference proceeding
- publication status
- published
- subject
- host publication
- IOP Conference Series: Earth and Environmental Science
- volume
- 1554
- article number
- 012036
- publisher
- IOP Publishing
- conference name
- Sustainable Built Environment Conference, SBE 2025 Zurich
- conference location
- Zurich, Switzerland
- conference dates
- 2025-06-24 - 2025-06-27
- external identifiers
-
- scopus:105027113120
- DOI
- 10.1088/1755-1315/1554/1/012036
- language
- English
- LU publication?
- yes
- id
- e75bbd14-3e51-4943-bf15-a5ef0e7bf721
- date added to LUP
- 2026-02-13 15:48:39
- date last changed
- 2026-02-13 15:49:42
@inproceedings{e75bbd14-3e51-4943-bf15-a5ef0e7bf721,
abstract = {{<p>Extending the service life of existing buildings and their components is essential for slowing resource loops in circular construction. While overall building lifespans are relatively well understood, the actual service lives of individual components remain underexplored due to limited empirical data. This study addresses this gap by analyzing and modeling building component lifespans using real-world damage investigation records and condition assessments from Swedish buildings. Descriptive statistics reveal a wide range of lifespans across six major component categories, highlighting materials linked to damage from poorly designed construction details as a critical factor. Additionally, most residential components have markedly shorter lifespans-about 25-30 years-compared to non-residential ones. To predict component service lifespans, we configure building envelope deterioration factors as a generic acyclic graph and employ Graph Neural Networks for node-level regression. Among the tested models, Graph Convolutional Network achieved the highest performance (R<sup>2</sup> = 0.83), outperforming GraphSAGE and Graph Attention Network due to its effective uniform aggregation across neighboring nodes. Our findings provide a high-granular, data-driven approach to estimating component lifespans, which can enhance predictive maintenance strategies and optimize the reuse potential of reclaimed materials in their next lifecycles.</p>}},
author = {{Wu, P. Y. and Mundt-Petersen, S. O.}},
booktitle = {{IOP Conference Series: Earth and Environmental Science}},
language = {{eng}},
publisher = {{IOP Publishing}},
title = {{Empirical Quantification and Prediction of Building Component Lifespans with Graph Neural Networks}},
url = {{http://dx.doi.org/10.1088/1755-1315/1554/1/012036}},
doi = {{10.1088/1755-1315/1554/1/012036}},
volume = {{1554}},
year = {{2025}},
}