A predictive modeling approach for determining optimal PPF content in concrete under specific mixture parameters using ANN and RSM

The incorporatison of polypropylene fibers (PPF) enhances concrete’s flexural performance but often reduces workability, creating a critical trade-off in mix design. This study employs ANN and RSM to predict FSV and SV, comparatively evaluate the two modeling approaches, and determine the optimal PPF content by balancing flexural strength enhancement and workability retention. During cross-validation, the ANN models achieved R 2 of 0.887 for FSV and 0.842 for SV, compared to 0.720 and 0.773 for RSM, respectively; the final models fitted to the complete dataset achieved R 2 of 0.935 and 0.852 (ANN) and 0.885 and 0.896 (RSM). A multi-objective optimization approach identifies PPF contents that balance strength improvement with minimal workability loss, providing practical guidance for the design of fiber-reinforced concretes in construction applications.