Architecture driven design of continuous steel wire reinforced FDM composites: Experimental investigation and mechanics-informed machine learning of structure–property relationships

Continuous fiber reinforced FDM composites treat reinforcement content as the main design variable, while through thickness placement remains unexplored. This study establishes reinforcement architecture as a design parameter governing tensile and flexural performance of 0.20 mm stainless steel wire reinforced ABS composites. An in situ FDM platform embedded wires while varying fiber volume fraction (V f ), through thickness position, stacking sequence, symmetry, raster orientation, and build orientation. At 8.73 vol% reinforcement, tensile strength increased from 26.6 to 99.3 MPa (3.73×), while flexural strength increased from 37.6 to 122.8 MPa (3.27×) relative to neat ABS. Increasing reinforced layers produced a near linear rise in tensile strength, with strain at maximum tensile strength increasing from 2.7% to 8.9%, indicating progressive strengthening and toughness rather than embrittlement. Controlled comparisons at matched V f isolated architecture effects, producing increases of up to 53.7% in tensile strength, 14.2% in tensile modulus, 31.8% in flexural strength, and 80.0% in flexural modulus, confirming that placement can rival content. Raster orientation and stacking sequence affected tensile and flexural behavior differently: alternating layer arrangements performed better under tensile loading, whereas clustered arrangements were more effective under flexural loading. Matrix cracking was followed by wire bridging, with intact wires sustaining load, suggesting potential for strain based structural health monitoring. A mechanics informed machine learning framework, benchmarked across thirteen algorithms using leave one out cross validation, predicted tensile modulus with R 2  = 0.959 and flexural strength with R 2  = 0.777. These findings provide a strategy for enhancing composite performance without increasing reinforcement content.