Experimental and numerical studies on the deterioration and failure process of Tasmanian sandstone subject to freeze-thaw cycles

The physical-mechanical properties of rocks, especially those in regions that are frequently impacted by extreme climatic variations, are subject to degradation. Thus, it is crucial to quantify and predict the deterioration rate and failure process of rocks, particularly if they are subjected to various numbers of freeze-thaw cycles. Although various laboratory and field tests have been conducted in recent decades to study the fracture mechanism of rocks that undergo various freeze-thaw cycles, the underlying mechanisms of crack initiation, propagation and the resultant failure process are still not well understood, and therefore, deserve further study. Moreover, various numerical methods, including grain-based (GB) numerical techniques, have been implemented to study the failure of rocks under mechanical loads, although these numerical methods have seldom been applied to model the failure of rocks subjected to freeze-thaw cycles. Furthermore, due to the computationally expensive nature of these GB numerical methods, none of them have been applied to model the deterioration and failure behaviour of fine-grain rocks such as Tasmanian sandstone. Therefore, further developments are necessary to improve the computational efficiency of the numerical methods, especially the GB techniques. This study was aimed at investigating the deterioration and failure of Tasmanian sandstone subjected to various numbers of freeze-thaw cycles using both experimental and numerical methods. A series of laboratory tests, including P-wave velocity tests, freeze-thaw tests, uniaxial compression strength (UCS) tests, and Brazilian tensile strength (BTS) tests, were conducted to investigate the physical-mechanical properties and failure behaviour of Tasmanian sandstone subjected to various numbers of freeze-thaw cycles. The results of the experimental study demonstrated that the P-wave velocity, BTS and UCS of the sandstone decreased as the number of freeze-thaw cycles increased. The rate of decrease from 0 to 20 freeze-thaw cycles was more pronounced than from 20 to 40 and 40 to 60 freeze-thaw cycles. Moreover, the main failure mode of the sandstone changed from axial splitting to shearing along a single plane in the UCS tests, and from smooth central fractures to a central zigzag fracture in the BTS tests, as the number of freeze-thaw cycles increased. A theoretical damage model was then derived based on continuum damage mechanics to quantify the degradation of the physical-mechanical properties of the Tasmanian sandstone subjected to various numbers of freeze-thaw cycles. After that, a three-dimensional (3D) numerical modelling was conducted using a hybrid finite-discrete element method (HFDEM) parallelised on the basis of general-purpose graphic processing units (GPGPU) to further investigate the failure mechanism of Tasmanian 8 sandstone subjected to various freeze-thaw cycles in the UCS and BTS tests. The 3D numerical modellings of sandstone subject to various freeze-thaw cycles revealed the deterioration and failure mechanisms. For the sandstone that was subjected to various freeze-thaw cycles, the increasing number of freeze-thaw cycles caused the macroscopic cracks to propagate, interact and coalesce in the shear behaviour, resulting in the change of failure pattern from axial splitting to shear sliding and shear crushing as the number of freeze-thaw cycles increased in the UCS test. At the same time, tensile and mixed-mode damage were the dominant failure mechanisms in the UCS tests. The 3D numerical modelling of the BTS test showed that although a central fracture was the eventual failure pattern for both the models that were and were not subjected to freezing and thawing cycles, the failure surface became more zigzag as the number of freeze-thaw cycles increased. A comparison of the results of the laboratory experiments, theoretical damage models, and HFDEM numerical simulations validated the proposed theoretical and numerical models. The theoretical and numerical models were able to quantify the deterioration rate and reproduce the failure of the Tasmanian sandstone that was subjected to various numbers of freeze-thaw cycles and tested in the laboratory experiments, and to predict the behaviour beyond those in the experimental tests. To further investigate the deterioration and failure behaviour of sandstone subjected to various freeze-thaw cycles, a novel grain-based hybrid finite-discrete element method (HFDEM-GB) was developed to incorporate the granular-scale heterogeneities and to study their effects on the initiation and propagation of intragranular, intergranular and transgranular cracks during the failure of sandstone subjected to mechanical loads and freeze-thaw cycles. In the HFDEM-GB, the polycrystal grains, categorised into three types of mineral conglomerates based on the microstructure of the Tasmanian sandstone, namely, low-strength grain (LSG), medium-strength grain (MSG), and high-strength grain (HSG). To solve the expensive computational issues of the HFDEM-GB, a semi-adaptive contact activation approach (semi-ACAA) and a tetrahedron to point (TtoP) contact interaction algorithm were implemented in the HFDEM-GB to further speed up its computation. The detailed performance analyses using the newly developed HFDEM-GB showed that the semi-ACAA was about 20 to 2.21 times faster than the full contact activation approach (FCAA), depending on the stage of the simulation, while the TtoP was at least 1.4 times faster than the tetrahedron to point (TtoP) contact interaction algorithm. Finally, the HFDEM-GB, equipped with the semi-ACAA and TtoP, was employed to investigate the deterioration and failure of Tasmanian sandstone subjected to various numbers 9 of freeze-thaw cycles, with a special focus on the effect of the granular-scale heterogeneities on the initiation and propagation of intragranular, intergranular and transgranular cracks. Similar to the HFDEM simulations, the HFDEM-GB simulations confirmed that the dominant fracture pattern of the sandstone subjected to various numbers of freeze-thaw cycles in the UCS tests changed from axial splitting to shear sliding and shear crushing with increasing numbers of freeze-thaw cycles. Moreover, the HFDEM-GB simulations clarified that the HSG had a higher stress-bearing capacity than the MSG and LSG, and most of the grain boundaries and intragranular cracks occurred in the HSG and MSG, while the LSG failed due to intragranular and transgranular cracks. Furthermore, the HFDEM-GB modelling revealed that although the simulations for the sandstone subjected to 60 freeze-thaw cycles presented similar types of cracks compared to those of the control sample, the number of intragranular cracks was higher than in the control sample. The modelling of the samples subjected to 140 freeze-thaw cycles presented extensive transgranular cracks leading to crushing failure patterns