Experimental study and machine learning algorithms for evaluating the performance of U-shaped ultra-high performance reinforced fiber concrete under static and impact loads. (1st July 2023)

• Record Type: Journal Article
• Authors: Al-shawafi, Ali; Zhu, Han; Haruna, S.I.; Bo, Zhao; Laqsum, Saleh Ahmed; borito, Said Mirgan
• Is Part Of: Journal of building engineering. Volume 70(2023)
• Publication Date: 2023-07-01
• Subjects: Drop-weight impact test -- Ultra-high performance concrete -- U-shaped specimen -- Artificial intelligence; Building -- Periodicals; 690.05
• DOI: 10.1016/j.jobe.2023.106389
• Languages: English
• ISSNs: 2352-7102

Abstract

Abstract: The inherent brittle nature of ultra-high performance concrete (UHPC) increased with strength and is considered its shortcoming in structural applications. This study investigated the impact strength of U-shaped UHPFRC mixtures involving different fiber volume fractions (Vf ) under drop-weight impact loads. A U-shaped configuration was introduced in this research to modify the testing procedure of the ACI 544-2R committee for the drop-weight impact test aiming to minimize the scatter result of this testing method. The micro steel fiber (MSF) was incorporated into the UHPFRC mixes at 0–3% Vf (interval of 1%). Moreover, artificial intelligence (AI)-based techniques, including artificial neural network (ANN) and multilinear regression (MLR) model, were employed to train and test the experimental dataset for the prediction of energy absorption capacity (EI2 ) at the failure crack stage. The results indicated that micro steel has remarkably improved the absorbed energy absorption of the U-shaped UHPFRC mixtures. The U-shaped configuration controls the crack formation. The impact strength at the two cracking stages increased with micro steel fiber content. At the first crack stage, the impact strength of UHPFRC-2 and UHPFRC-3 is 100% and 408% higher than that of UHPFRC-1. The average ductility index value first increased and then decreased with high fiber content. The AI-based technique accurately predicts the energy absorption capacity of a U-shaped UHPFRC mixture at failure crack strength. The ANN and MLR model had an overall R 2 value of 0.8231 and 0.9502, respectively. Highlights: U-shaped UHPFRC specimens were developed. AI-based techniques were developed to evaluate the energy absorption capacity. The impact resistance of U-shaped UHPFRC was significantly improved due to the addition of MSF. The MSF improves the mechanical properties of UHPFRC mixtures. ANN and MLR models predict the energy absorption capacity with high accuracy.