Development of machine learning-embedded applications to support hazardous material assessment in renovation and demolition

The unexpected discovery of hazardous materials during renovation and deconstruction increases uncertainties concerning working safety and project economy.

Despite mandatory pre-demolition audits in Sweden, hazardous waste inventory is hard to do thoroughly due to inaccessibility and difficulty in distinguishing hidden contaminants.

To address the need for a circular built environment, developing a new approach to facilitate risk screening for hazardous materials is needed. Predictions can be achieved by systemizing past environmental records.

Machine learning modeling was found to be promising for identifying patterns of asbestos and PCB materials based on building registers and inventories. Its potential in characterizing buildings with high contamination risk at the component levels has been demonstrated.

Building on these results, the proposed study intends to investigate the feasibility of implementations and development of decision support tools for relevant actors.