Bayesian Linear Regression analysis of the effect of silica fume on concrete creep predictions and structures’ safety

Abstract

An important evolution in the construction field is noticed nowadays. New high-rise buildings, bridges and dams are taking place all over the world. This evolution entails the incorporation of admixtures and additives to the mix composition which affect concrete behavior. Therefore, an accurate prediction of the behavior of these constructions is a must. Since creep deformations have an important impact on long-term integrity and stability of structures, an accurate prediction of these deformations is needed. To study the impact of silica fume on creep predictions, a comparison between the creep compliance predicted by Eurocode 2 and experimental results is performed using CEB statistical methods. The large database used in this study is established from international laboratories and research centers. An inaccurate estimation of Eurocode 2 creep compliance for concrete with silica fume is noted. To overcome this difference, we have calibrated Eurocode 2 creep equations by adding correction coefficients that take the effect and percentage of silica fume into consideration. A linearization of the creep model is performed, then a Bayesian analysis is applied. Using these corrective coefficients help to predict accurately creep deformations in the design stage and hence, the long-term deflection. Therefore, the required measurements and precautions can be applied to avoid excessive deflections after construction. After implementing the correction coefficients to Eurocode 2 compliance formula, and to evaluate our updated Eurocode 2 creep model, the CEB statistical methods have been used. An improvement in the results is clearly shown. Using the large experimental database, this study demonstrates that Bayesian model assessment is an important procedure applied to update Eurocode 2 creep model. The long-term serviceability of structures subject to creep is well improved by adopting such a design approach.