Compilation and Analysis of Thaw Settlement Test Results: Implications for Prediction Tools and Stress-Strain Characterization in Permafrost

Abstract. Climate change has already significantly impacted infrastructures built on permafrost, with thaw settlement as the most frequently reported issue. By characterizing thaw settlement tests, an improved understanding of the thaw settlement properties of permafrost sediment can be obtained. Thaw settlement tests involve thawing permafrost samples under an initial load, followed by applying additional load to the thawed sample to characterize its volume change behaviour upon thawing. In the absence of a standardized procedure for conducting thaw settlement tests, characterizing thaw settlement properties has been done using various methods in the existing literature; however, to date, these data have not been broadly compared. This is in part because they had not previously been compiled in a single dataset. This study presents a comprehensive dataset of thaw settlement test results, digitized from published papers and reports. The data are standardized and stored in an open-source repository. Aggregating the data enabled a cross-comparison of thaw settlement properties for different soil types. This was achieved by constructing an idealized stress-strain curve for each test and deriving thaw settlement parameters from the curves. These parameters were then used to compare the thaw settlement behaviour of fine-grained, coarse-grained, and highly organic permafrost samples. Additionally, the compiled data was used to evaluate the effectiveness of various empirical tools developed to predict thaw strain from index properties. The predicted thaw strain values were compared with the measured thaw strains to determine which tool provided the most accurate and reliable predictions for each soil type. The results suggest that a correlation developed by Nixon and Ladanyi (1987) for estimating thaw strain based on frozen bulk density shows the smallest deviation from actual values and exhibits the least bias in its predictions. This dataset is expected to enhance the understanding of thaw settlement and improve its estimation. Used alone or in conjunction with localized data, it can contribute to developing new empirical tools for predicting thaw strain from index properties. Additionally, this dataset aids enhanced characterization of stress-strain behaviour upon thawing, facilitating future efforts in the numerical modelling of the thaw settlement process. The dataset is accessible at https://doi.org/10.5281/zenodo.14538524, and the results presented in this paper are based on version 1.2.0 of the dataset (Mohammadi and Hayley, 2024c).