Search

• CatBoost regression model for prediction of surface chloride level. • The regression model has MAE and RMSE scores of 0.11 and 0.16 respectively. • Increase in chloride ingress with the increase in temperature and humidity. • Exposure zone has a significant effect on chloride ingress. • Temperature and humidity of RCP 2.6 and RCP 8.5 considered for analysis. This paper presents the effect of climate change on the long-term reliability of concrete structures due to chloride ingress. The impact has been demonstrated for three regions across the globe for the best and worst-case climate change scenarios. The methodology adopted here can be divided into three main steps. The first one is the development of a machine learning regression model using CatBoost for estimation of surface chloride levels, which is a function of temperature, concrete material properties, exposure conditions, and chloride levels in the seawater. The CatBoost model fitted over the experimental data has a mean absolute error (MAE) and root mean square error (RMSE) value of 0.11 and 0.16, respectively, and a high R2 value of 0.82. The second step is solving the coupled non-linear partial differential equation for temperature, humidity, and chloride ingress in concrete with time. A parametric study considering the effect of external parameters (temperature, humidity, exposure condition, and chloride levels in water) reveals the complexity and the dependencies of the chloride ingress process on external parameters. The final step involves the reliability analysis to estimate the time for chlorides to reach threshold levels at the rebar level using the numerical model. The input for the numerical model is temperature and humidity condition for different representative concentration pathways (RCP) scenarios, the material properties of concrete, and the exposure conditions. The reliability analysis shows that, for the concrete structure located in Dubai, UAE, and Cascais, Portugal, the mean time to corrosion initiation decreases due to climate change and it increases for the structure located in Tasmania, Australia. Further, the structures that are built in the future may have lower service life compared to the structures that were built in the past. Based on the study it concluded that it is necessary to consider the impact of climate change on service life to design resilient structures. [ABSTRACT FROM AUTHOR]