Comparative study of flax, glass, and carbon textile reinforced concrete: experimental and machine learning evaluation of tensile and flexural behaviour

Direct comparative studies on the mechanical performance of flax textile-reinforced concrete and conventional textile systems under consistent experimental conditions remain limited. This study investigates the tensile and flexural behaviour of concrete reinforced with flax, carbon, and alkali-resistant (AR) glass textiles. Experimental testing was combined with machine learning approaches, including artificial neural networks (ANN) and random forests (RF), to assess the relative importance of key input parameters. Uniaxial tensile tests showed that mesh size had no significant effect on tensile properties when reinforcement ratios were constant. Flax- and AR-glass reinforced concretes exhibited comparable tensile strengths. ANN analysis further indicated that reinforcement arrangement, particularly the number of layers and reinforcement ratio, had a greater influence on tensile strength than mesh size or the intrinsic tensile properties of the reinforcement. Four-point bending tests highlighted the dominant role of textile type in flexural performance, with flax textile-reinforced concrete exhibiting significantly lower flexural strength and toughness than carbon- and AR-glass systems. Consistent with experimental observations, RF modelling identified reinforcement tensile strength as the most influential parameter governing flexural strength. Overall, flax textile reinforced concrete shows promise for tensile applications but remains less suitable for flexural structural use within the scope of this study.