Towards a semantic digital twin for construction site PM emission management: threshold modelling and validation for alerting workflows

Particulate matter (PM) emissions from construction activities affect urban air quality, worker exposure, and surrounding communities. Improving their management is therefore a relevant contribution to sustainability and reflects UN Sustainable Development Goals (SDGs) related to health and well-being (SDG 3), sustainable infrastructure (SDG 9), and urban liveability (SDG 11). Digital Twins (DTs) can support PM emission management by connecting monitoring data, site context, threshold information, and decision-support services. However, existing approaches still lack structured semantic workflows able to represent threshold entities, evaluate PM observations, and retrieve alert-relevant context. Building on a previous high-level DT framework for PM emission management on urban construction sites and on the Construction Site Emission Management Ontology (XEMO), this paper proposes and evaluates a semantic alerting workflow based on threshold modelling and SHACL validation. Thresholds derived from current and forthcoming European environmental air quality reference values are represented as semantic entities linked to pollutants, averaging periods, units, values, and regulatory sources. A demonstrator is then developed in which synthetic PM10 and PM2.5 observations are generated, stored as time-series data, transformed into semantic data according to the Resource Description Framework (RDF), integrated into a knowledge graph, and validated against threshold-based warning conditions through SHACL shapes. The validation reports are queried to retrieve the related alert context, including pollutant, measured concentration, threshold condition, sensor, construction area, and emission source. The results show the prototype-level feasibility of using semantic modelling and SHACL validation to support traceable PM alerting, providing a reusable basis for future decision-support services within semantic DTs and real-site implementation.