Time Series analysis for prediction of ASR-induced expansions

Abstract

The aim of this study was to predict and classify ASR reactivity of mortar bars submerged in 1,0.5 and 0.25N NaOH solutions. The experimental expansion data were recorded from the mortar specimens prepared with twelve different aggregate groups for the three alkali solutions. The dataset were then fitted in a Time Series (TS) regression model to predict the ASR-induced expansion of each aggregate group for up to 98days. The results showed that, regardless of aggregate source, the TS model was very suitable in prediction of mortar expansions when compared to those obtained experimentally. The ASR classification of the investigated aggregates obtained from the predicted expansions of the TS regression model was nearly identical to the classification determined by using experimental mortar expansions.