Utilizing Data-Driven Methods to Predict the Fatigue Life of Cement Concrete Considering Corrosive Environmental Factors

Abstract

The primary objective of this study is to assess the fatigue resistance of cement concrete when exposed to corrosive environments. To achieve this, experimental results from high-cycle fatigue (HCF) and low-cycle fatigue (LCF) tests conducted on cement concrete samples subjected to various corrosive conditions were used. Various data-driven techniques, including multiple linear regression (MLR), Taguchi sensitivity analysis (TSA), and response surface method (RSM) were utilized. The aim was not only to identify the most influential parameter affecting fatigue life but also to offer a simpler and cost-effective alternative to experimental approaches. Consequently, two key parameters related to the corrosive environment: pH value and immersion time, along with the cyclic force applied to the concrete samples as input variables across different approaches were considered. The number of cycles until sample failure regarded as the output variable in all analyses. Furthermore, the analyses were conducted with the assumption that longer fatigue life is preferable. The findings revealed that the fatigue life of Portland cement concrete consistently decreased with increasing immersion time. Notably, the pH value emerged as the most significant parameter, while the other two factors exhibited equivalent impacts.