Non-normal distribution of residual flexural strengths of steel fiber reinforced concretes and its impacts on design and conformity assessment

• Main Authors: He, Shan, Yang, En-Hua
• Other Authors: School of Civil and Environmental Engineering
• Format: Article
• Language: English
• Published: 2022
• Subjects: Engineering::Civil engineering; Flexural Strength; Variability
• Online Access: https://hdl.handle.net/10356/160667
• Institution: Nanyang Technological University

Variability of concrete properties has strong influences on design and conformity assessment. However, variability of residual flexural strengths of steel fiber-reinforced concrete (SFRC) has yet been statistically investigated. In this study, a large number of SFRC specimens was sampled from a central mix concrete plant manufacturing precast SFRC elements for a real civil infrastructure project. Variability of residual flexural strengths of SFRC are reported. Results show that the residual flexural strengths of SFRC follow the log-normal distribution at large crack month opening displacement (CMOD). The origin that leads to the log-normal distribution of the residual flexural strengths of SFRC is discussed. The use of an assumed normal distribution of the residual flexural strengths of SFRC in the current design approach and conformity assessment results in more than 30% overdesign and more stringent acceptance criteria. This leads to higher materials cost and higher producer's risk, which discourage the wider adoption and application of SFRC. The logarithmic transformation provides a practical and convenient approach to design and to check the conformity of log-normally distributed SFRC properties adopting existing criteria. Compared to the typical lab-based study with limited number of specimens and test results from a well-controlled environment for material processing and specimen preparation, the field study reflects more closely the true material variability in real construction. Thus, the statistical analysis obtained and conclusions drawn from the current work are more representative.