Cluster monitoring of Macedonian concrete bridges – Results, trends and lessons learned

Abstract Effective bridge management depends on transforming heterogeneous infrastructure and monitoring data into actionable intelligence. A data‐driven workflow is presented for the prestressed concrete bridge stock in North Macedonia. Three datasets are utilised: (i) a national inventory of 1,310 structures, from which 122 precast prestressed concrete T‐girder bridges are classified into four cross‐sectional families using agglomerative clustering; (ii) dynamic load testing of 27 bridges with 146 controlled passages at 10–80 km/h; and (iii) a 30‐day Bridge Weigh‐in‐Motion (B‐WIM) campaign recording 22,006 heavy‐vehicle events, segmented into eight vehicle typologies. B‐WIM analysis identifies an inverse relationship between vehicle weight and Dynamic Amplification Factor (DAF), with 4‐axle trucks showing the strongest association with upper‐tail dynamic amplification in the monitored dataset. The bridge inventory analysis indicates that four clusters consolidate into two broader groups. DAF is influenced by multiple interacting parameters, and sensitivity analysis shows that the choice of signal‐processing cut‐off alone accounts for variations of up to 18.5 % at highway speeds. The envelope‐based lower cut‐off exhibits the strongest agreement with reference measurements, establishing a reliable extraction method for DAF. This workflow offers an empirical basis for defining bridge and vehicle typologies applicable to similar inventories, subject to local validation.