Research on interactions among parameters affecting dynamic mechanical properties of sandstone after freeze-thaw cycles

Abstract

In this study, the interactions among parameters affecting the dynamic strength of the sandstone after freeze-thaw (F-T) cycles are studied from the perspective of system. Based on the impact test results of F-T sandstone from different regions, a dataset is built to exclude the peculiar characteristics of certain kind of sandstone so that the interactions obtained are more universal. The improved Mahalanobis self-organizing map (SOM) algorithm is used to cluster the dataset, and global interaction matrix (GIM) in the rock engineering system (RES) is coded by the clustering result and Pearson correlation coefficient. The interactions among lithology, quality change (QC), porosity change (PC), strain rate (SR), number of F-T cycles (NFT), and F-T temperature difference (FTD) and their effects on the dynamic mechanical properties of the sandstone after F-T cycles are examined. The results show that the improved Mahalanobis SOM can eliminate the correlation interference between dimension and factors. The coding method based on Pearson correlation coefficient solves the problem of high subjectivity and low linearization quantization of GIM coded by subjective method. The system theory suggests that PC, lithology, NFT, QC, FTD, and SR have strong interaction intensity. NFT, PC, QC, and FTD are more dominant. The interaction path is divided into sensitive effect, adverse effect, and circulation effect. By analyzing these three effects, the deterioration mechanism of sandstone after F-T cycles is revealed from perspective of engineering system.