Probabilistic Estimation of Structural Elements in Timber Buildings for Circular Construction Planning
Circular economy strategies in the construction sector increasingly emphasize the reuse of structural elements from obsolete buildings to reduce environmental impact and resource consumption. However, implementing these strategies is often constrained by limited information on building composition, particularly at the structural element level. Conventional estimation approaches are typically deterministic, subjective, or lack sufficient detail to support accurate material recovery planning. This study introduces a probabilistic modelling framework to improve the estimation of structural element dimensions and material quantities in residential timber buildings.
Bayesian Network Framework for Structural Estimation
The research proposes a probabilistic modelling approach based on Bayesian Networks (BNs) to estimate structural characteristics under uncertainty. Bayesian Networks provide a structured framework that represents relationships among variables through probabilistic dependencies, enabling inference even when data are incomplete. By modelling structural relationships probabilistically, the framework generates uncertainty-aware predictions of beam dimensions and material quantities, which are essential for planning the reuse of load-bearing components in circular construction systems.
Integration of Building Data and Structural Design Rules
The BN model integrates building-related variables derived from a dataset of approximately twenty thousand residential buildings with structural design equations derived from historical design practices and modern structural codes. These structural equations guide the estimation of load-bearing element dimensions based on typical design requirements. The combined dataset and design knowledge establish prior modelling assumptions that enable robust probabilistic inference for building composition analysis.
Model Application and Validation on Real Buildings
The developed model was applied to four real residential timber buildings to test its predictive capability. Using partial input data, the BN generated probabilistic estimates of structural beam dimensions and associated material quantities. The results demonstrated that the model can produce reliable predictions even when key building information is missing, making it suitable for early-stage assessments where detailed structural documentation is unavailable.
Advantages Over Deterministic Estimation Methods
Compared with traditional deterministic approaches, the Bayesian framework significantly expands the analytical scope by incorporating a larger set of variables and a substantially larger dataset. This enables more accurate modelling of uncertainty and structural variability at the level of individual building components. The probabilistic representation also allows decision-makers to evaluate confidence ranges rather than relying on single-point estimates, improving the reliability of planning outcomes.
Implications for Circular Construction and Urban Sustainability
The proposed methodology supports urban planning and circular construction strategies by enabling early estimation of reusable structural materials within building stocks. By quantifying variability in structural properties before conducting detailed inspections, planners can better assess future material availability and potential greenhouse gas emission reductions associated with reuse strategies. The element-level modelling capability strengthens sustainability planning and provides a valuable decision-support tool for managing existing building resources.
Comments
Post a Comment