Predicting the technical reusability of load-bearing building components: A probabilistic approach towards developing a Circular Economy framework

• Published in: Journal of Building Engineering 2021-10, Vol.42, p.102791, Article 102791
• Main Authors: Rakhshan, Kambiz, Morel, Jean-Claude, Daneshkhah, Alireza
• Format: Article
• Language: English
• Subjects: Building structure; Gaussian process; K-nearest neighbours; Random forest; Reuse; Supervised machine learning
• ISSN: 2352-7102; 2352-7102
• DOI: 10.1016/j.jobe.2021.102791

The construction sector is the largest consumer of raw materials and accounts for 25%–40% of the total CO2 emissions globally. Besides, construction activities produce the highest amount of waste among all other sectors. According to the waste hierarchies, reuse is preferred to recycling; however, most of the recovery of construction and demolition wastes happens in the form of recycling and not reuse. Part of the recent efforts to promote the reuse rates includes estimating the reusability of the load-bearing building components to assist the stakeholders in making sound judgements of the reuse potentials at the end-of-life of a building and alleviate the uncertainties and perceived risks. This study aims to develop a probabilistic model using advanced supervised machine learning techniques (including random forest, K-Nearest Neighbours algorithm, Gaussian process, and support vector machine) to predict the reuse potential of structural elements at the end-of-life of a building. For this purpose, using an online questionnaire, this paper seeks the experts’ opinions with actual reuse experience in the building sector to assess the identified barriers by the authors in an earlier study. Furthermore, the results of the survey are used to develop an easy-to-understand learner for assessing the technical reusability of the structural elements at the end-of-life of a building. The results indicate that the most significant factors affecting the reuse of building structural components are design-related including, matching the design of the new building with the strength of the recovered element. •Supervised machine learning to predict the reusability of the building structure.•A random forest model produces the most reliable predictions.•Feature selection/sensitivity analysis methods used to open the random forest model.•Results show the most significant factors affecting reuse are design-related.•An easy-to-understand model is developed that determines the technical reusability.