The understanding of tunneling rock failure characteristics under unloading conditions in the cold region is critical for the proper design of rock tunneling support and mining safety operation. Given this understanding is currently limited, this study aimed to investigate the characteristics of fractured sandstone samples in the microscale after cyclic freezing-thawing and triaxial unloading tests. The samples were first subjected to different cycles of freezing and thawing, followed by the triaxial unloading test and scanning electron microscopy imaging. The peak strength and damage dilatancy stress were measured from the stress-strain curves. The microcrack characteristics (number, length, and width) were obtained through the image analysis. The results show that the decrease in peak strength and damage dilatancy stress was more significant by the first 20 freezing-thawing cycles when the pore pressure gradient is maximum compared to the later freezing-thawing cycles. The mechanical properties also significantly deteriorated when the severe freezing-thawing treatments were performed. The fracture section mainly had the morphology of honeycomb-like microstructure, stripped microstructure, and flocculent microstructure. The cracking extent was mainly influenced by the freezing-thawing rather than the triaxial unloading test, but the azimuthal angle of microcracks was significantly altered by the triaxial unloading. To properly design rock tunneling support and safe operation of mining in the cold region, both impact of cyclic freezing-thawing and the excavation operation direction should be considered.

1. Introduction

During the excavation of deep tunnels under high-stress conditions, the stress concentration near the excavation surface reduces the radial stress component and increases the tangential stress component, leading to nonviolent or violent surface failure, i.e., spalling and strain burst, respectively [14]. These two significant geohazards could affect the safety of personnel and infrastructure in excavations and deep underground mines. For example, severe strain burst occurred along a considerable length of tunnel section during tunnel excavation of the China Jinping II hydropower station project [5].

With the implementation of national strategies such as Western Development and One Belt One Road in China, a variety of infrastructure engineering projects (e.g., tunneling) are carried out in the freezing-thawing zones. The rocks in these zones are subjected to frost pressure and frost-heaving force induced by the water freezing and thawing during freezing-thawing cycles [6]. This force complicates rock mechanics under unloading conditions. Note that the unloading-induced failures of rocks (i.e., spalling and strain burst) are different in characteristics from those caused by the loading on rocks. Recently, there has been a growing interest in the understanding of the freezing-thawing rock failure behaviour under unloading conditions as it is critical for the proper design of structures such as tunneling and underground mining in freezing-thawing zones. But this understanding is still limited as currently most unloading rock failure studies focus on the rock failure behaviour without freezing-thawing damage [79].

The static mechanical properties and durability of rocks degraded after freezing-thawing cycles [6, 1014]. The cyclic volumetric expansion and contraction accompanied by pore structure damage until the failure of limestone specimen were observed throughout freezing-thawing process [10]. The damage could decrease the rock dynamical peak stress and elastic moduli, increase rock porosity, and strengthen plasticity as reported by Zhang et al. [15], Li et al. [16], Ke et al. [12], Walbert et al. [13], and Khanlari et al. [14]. For example, reductions of 20.2% and 21.5% in static and dynamic compressive strengths of sandstones after freeze-thaw treatments were found in the study [12]; the static and dynamic elastic moduli were found to be impacted the most by the frost weathering and can be used as indexes of first microcrack initiation, whereas the uniaxial compressive strength and the porosity were found to be affected the least [13]. Also, both micropore and macropore sizes of rocks increase after freeze-thaw cycles [16]. This is because, under the freeze-thaw damage, micropores can become interconnected with each other and thus become mesopores and macropores. Several techniques have been applied to investigate the rock pore structure after freeze-thaw cycling. They mainly include computed tomography (CT), X-ray CT, nuclear magnetic resonance (NMR), scanning electron microscopy (SEM), and ultrasonic technology [1720]. The complex freeze-thaw damage to the rock also attracted research interest on the development of the damage constitutive model under freeze-thaw and loading [6].

The true triaxial unloading test where unloading of only one loading platen is applied has been carried out to study the dynamic rock failure behaviours under room temperature. Different testing parameters including sample material, height-to-width ratio, and the unloading rate have been investigated to understand their influence on failure behaviours [2123]. For example, Zhao and Cai [22] presented the dynamic failure process of the unloading surface transformed from local rock ejection to full-face burst when the height-to-width ratio decreased from 2.5 to 1.0, indicating the significant size effect on the dynamic failure behaviour. However, few systematic studies focus on the true triaxial unloading test on rocks after freezing-thawing cycles. Although Yu and Xu [24] showed the dynamic mechanical properties of sandstone after freezing-thawing cycles under triaxial confining pressure unloading, the mechanism was not thoroughly explored by analyzing the microscopic structure of sandstone samples. As the unloading-induced stress complicates the pore size and its distribution as well as the microstructure of rocks under freeze-thaw cycles, the rock mechanics and failure characteristics could be significantly different under unloading conditions and thus are worth for further investigation.

In this study, an experiment of axial compression and unloading confining pressure was performed to understand the microscopic characteristics of the fractured surface of sandstone after freezing-thawing cycles. The sandstone samples were first subjected to a series of freezing-thawing cycles, and the damaged samples were tested under the triaxial unloading condition. The microscopic characteristics of the fractured surface of sandstone cross section were then obtained by analyzing its SEM images. As one of the first attempts, our study helps to understand the dynamic failure characteristics of sandstones at the microscopic scale in the cold region; thus, the excavation engineering of deep tunnels experiencing freezing-thawing cycles could be benefited.

2. Materials and Methods

2.1. Sample Preparation

The rock samples used in this experiment were collected from the excavation face of a tunnel in the Altai region, Xinjiang Province, China. They were from one visually homogeneous sandstone zone to minimize the difference caused by the sampling process. According to the phase analysis, their compositions are mainly quartz, feldspar, and mica. As shown in Figure 1, the sandstone sample was off-white with a little argillaceous inclusion. The grain size of the sample was in a medium range.

The sandstone sample was first processed to produce the standard cylindrical samples with a diameter of 50 mm and a height of 100 mm [24]. The parallelism and the flatness of the end faces were within ±0.05 mm and ±0.02 mm, respectively. Note that the samples were manually grinded to the standard size to minimize the damage caused by the hydraulic cutting. After removing the samples with cracks and joints, the remaining ones with consistent mechanical properties were chosen by analyzing the longitudinal wave velocities through the samples. Note that the average value of longitudinal wave velocity is 3321 m/s.

2.2. Experimental Devices

Three consecutive tests were conducted on different groups of samples in the order of the freezing-thawing test, the triaxial unloading test, and the SEM imaging. The freezing-thawing test was performed through a thermostatic drying oven (101A-1) and an ultralow temperature freezer (DW-60W60) with a minimum temperature of −65°C and accuracy of 0.1°C.

The triaxial unloading test with constant axial compression load and decreasing confining pressure was conducted at room temperature through the MTS815 rock mechanics test system with a maximum axial load of 4600 kN and a maximum confining pressure of 140 MPa in the State Key Laboratory for Geomechanics and Deep Underground Engineering, China University of Mining and Technology (Figure 2). The test system is mainly composed of the loading system, the servo control system, and the confining pressure system. The control method of loading includes load control, axial displacement control, and circumferential displacement control.

The representative fracture sections from the samples were imaged through the Quante-250 scanning electron microscope (SEM) in the Advanced Analysis and Computation Center, China University of Mining and Technology (Figure 3). The resolution was 1.4 nm, and magnification could reach to ×1,000,000.

2.3. Freezing-Thawing Test

Temperature is an important parameter that should be controlled in the cyclic freezing-thawing test. Given the 3-year temperature records (see Figure 4) in Altai where the sandstone samples were collected, the minimum temperature and the maximum temperature of the freezing-thawing test were set to −30°C and 20°C, respectively.

The sample was first placed in a water tank with a constant temperature of 20°C. The air in the water tank was pumped for 2 h before injecting distilled water [25]. The evacuation continued until no bubbles were observed; then, the sample was immersed for 48 h in the water tank. Afterwards, the sample was placed into the ultralow temperature freezer at −30°C for 9 h and then in the water with a temperature of 20 ± 2°C for 5 h. A freezing-thawing cycle with a total period of 14 h was completed. The freezing-thawing cycles were repeated for 0, 20, 40, 60, and 80 times to produce five groups of samples.

2.4. Triaxial Unloading Test

After freezing-thawing cycles, the triaxial unloading test was conducted on the samples. Based on the hydrostatic pressure condition (the axial pressure equals to the confining pressure), 14 MPa of the hydrostatic pressure was applied to the sample by controlling the stress rate. The axial pressure was gradually increased to about 75 MPa at the displacement rate of 0.004 mm/s, while the confining pressure (14 MPa) remained constant. The confining pressure was then decreased at the rate of 0.05 MPa/s in the unloading process, causing damages in the sample. Note that the proper value of maximum axial pressure was controlled at 75 MPa which is slightly higher than the uniaxial compression strength of samples after 80 freezing-thawing cycles and less than the triaxial compression strength of samples after 80 freezing-thawing cycles. This critical value should be high enough to ensure the sample would be damaged during the unloading process and should not be too high to cause the compression failure.

Figure 5 shows the typical damage patterns after the unloading tests for sandstone samples under different freezing-thawing cycles. The results show that the fracture appeared as shear failure in all samples. A main fracture would cut through both end faces of the sample; thus, a section that intersected with the central axial was formed. More severe main failure was produced with the increasing number of freezing-thawing cycles. Subsequently, spalling occurred at the end faces of the sample. Besides, macrocracks were found to propagate from the main fracture surface.

2.5. Scanning Electron Microscopy Test

The microstructure of rock has a strong influence on its deformation and mechanical characteristics. Especially, the material properties of mineral crystals, the spatial arrangement, and the agglomeration of mineral crystals strongly affect the stress-strain curves. Therefore, studying microstructure characteristics of fractures can help to understand the mechanical characteristics of rocks.

At least six representative sections (10 mm × 10 mm × 5 mm) were obtained from different locations in the fracture surface of each sample after the triaxial unloading test. The sections were first cleaned, dried, and gilded and then placed on the SEM chamber for taking images. Each section was scanned row by row to avoid the repetition or the omission of key sample position. Most SEM images were taken by using ×1000 magnification, and some of them were taken by using ×500 and ×2000 magnification. The sample damage characteristics at the microscopic scale were then analyzed through these SEM images. The number, azimuthal angle, length, and width of the microcracks were measured from the SEM images by using ImageJ.

3. Experimental Results and Discussion

3.1. Unloading Stress-Strain Curve Results

Figure 6 shows the stress-strain curves of sandstone samples after 0, 20, 40, 60, and 80 freezing-thawing cycles. The key mechanical parameters (see Table 1) including peak strength and damage dilatancy stress were then obtained from the above stress-strain curves. The decreasing rate of peak strength and damage dilatancy stress with increasing cycles of freezing-thawing is also presented in Table 1.

The results show that the peak strength and the damage dilatancy stress of the samples both significantly decreased with increasing freezing-thawing cycles. After the first 20 freezing-thawing cycles, the peak strength decreased by 10.7% and the damage dilatancy decreased by 5.9%, compared to that of samples without experiencing freezing-thawing treatment. While the decreasing rate of the peak strength was only 2.3% after 40 freezing-thawing cycles and 5.5% after 60 cycles, the decreasing rate of the damage dilatancy stress was only 4.7% after 40 freezing-thawing cycles and 6.2% after 60 cycles. It is interesting to find that the decreasing rate of the peak strength and the damage dilatancy stress both again increased greatly after 60–80 freezing-thawing cycles. Thus, the effect of freezing-thawing cycles on mechanical characteristics of the sandstone samples was assumed to be more significant for the first several freezing-thawing cycles or the large number of freezing-thawing cycles, i.e., 80 cycles of freezing-thawing in this study.

The above results are in agreement with other studies. For example, Eslami et al. [10] showed a volumetric expansion during the freezing phase accompanied by very important damage from the first cycle until the failure of specimen for Migne and Saint-Maximin limestones in French. As the pore structure is subjected to the maximum pressure gradient due to the ice expansion in the first cycle when the pores are still relatively small, the pore structure in the rock could be deteriorated at a large extent. With the increasing size of pores, the pressure gradient due to freezing-thawing decreases. However, the pore could be collapsed when more and more pores are interconnected. This is probably why the peak strength value started to significantly drop after 60 cycles of freezing-thawing in our study.

3.2. Micromorphology Characteristics

After freezing-thawing cycles, microcracks were observed in the sandstone samples under unloading conditions. The mineral crystal structure was broken, and the cementing material was cracked. Some of the microcracks eventually combined to macrocracks until the sample was failed. The morphologies of the fracture section are displayed in Figure 7.

Samples without being experienced freezing-thawing treatment have typical striped, step-like, flocculent, and smooth-joint microstructures under the unloading condition. With the increase of freezing-thawing cycle, the honeycomb-like microstructure, stripped microstructure, and flocculent microstructure dominated on the sectional surface of samples. Only a few step-like microstructure and smooth-joint microstructure were observed. This phenomenon can be explained as follows. The microstructure of the sandstone samples consisted of different mineral crystals, crystal interfaces, impurities, water, pores, and microcracks. During the freezing-thawing process, water repeatedly transformed between the liquid and solid state, resulting in the fatigue damage under repeat frost heave forces. Thus, the damaged microstructures such as honeycomb-like structures increased with the number of freezing-thawing cycles.

3.3. Microcracks

The typical microcracks inside and between the mineral crystals in the sandstone samples were composed of intercrystalline crack that occurred along the boundary of mineral crystals, transcrystalline crack that propagated through the mineral crystals, and intracrystalline crack that occurred inside the mineral crystals, as shown in Figure 8. These characteristics of microcracks are in agreement with the results of other researchers, such as Ji et al. [26], when the sandstone samples are not subjected to freezing-thawing treatment. The observed microcracks initiated when the sandstone samples were subjected to the first several cycles of freezing-thawing treatments and developed further with more cycles of freezing-thawing under the triaxial unloading tests as shown in Figures 9 and 10. Note that the total number, median length, and median width of the microcracks in sandstone samples with different freezing-thawing cycles are shown in Figure 9.

When the sandstone samples were not under freezing-thawing condition, they still presented few microcracks but which are far less than those under severe freezing-thawing treatments, particularly for intercrystalline cracks and transcrystalline cracks. These microcracks mainly resulted from the triaxial unloading test, indicating that the triaxial unloading condition could change the stress conditions of sandstone samples and cause damage if the stress reached a critical strength limit. This is consistent with other studies such as Dyskin and Germanovich [3] and Wang et al. [4]. It should be noted that these microcracks were caused by triaxial unloading condition when the sandstone samples were less than 75 MPa in this study. For the practical excavation of deep tunnels under much higher stress conditions, the unloading condition could thus cause more severe damage. Future studies are underplanned to explore the influence of axial compression on the dynamic failures of rocks under the triaxial unloading test.

Due to the freezing-thawing effect, the frost heave force caused by the repeated transformation of liquid and solid state of water in pores would greatly damage the microstructure of the sandstone sample. After 20 freezing-thawing cycles, the number of intercrystalline cracks, transcrystalline cracks, and intracrystalline cracks significantly increased by 113%, 250%, and 50%, respectively. The transcrystalline crack was the most and the intracrystalline crack was the least. Thus, the majority of the microcracks occur either at the interface between the cementing material and the mineral crystal or through the cementing material, propagating from the surface of one crystal to the surface of a neighboring crystal. The microcracks are either arrested at the crystal or travel around the crystal. It may also propagate through the crystal, although this was very rare. These results are expected as the freezing-thawing process mainly occurred in the porous cementing material rather than the dense mineral crystals, which have a higher resistance of cracking than the cementing material. Our results are consistent with those in quasi-brittle materials such as mortar and concrete [12, 27, 28], which showed that the microcracks propagated in matrix, or were arrested at the aggregate, or travelled around it; few microcracks penetrated through the aggregate.

With more freezing-thawing cycles from 20 to 60 cycles, the microcracks continued to propagate. However, the increasing rates of intercrystalline crack, transcrystalline crack, and the intracrystalline crack are much lower than that when freezing-thawing cycles increased from zero to 20. For example, they are, respectively, 29%, 43%, and 33% after 40 freezing-thawing cycles compared to 20 freezing-thawing cycles and are, respectively, 14%, 10%, and 50% after 60 freezing-thawing cycles compared to 40 freezing-thawing cycles. This observation matches well with that found by Eslami et al. [10] who showed the majority of damage occurred in the first freezing-thawing cycle. Further increase of freezing-thawing cycle number from 60 to 80 causes a significant increase in all three types of microcracks. This is probably because the pores became larger and interconnected, resulting in the collapse of pores and microstructures. More cracks could propagate through the crystal, causing more transcrystalline cracks and intracrystalline cracks. Also, the above microcracking variations in terms of number, length, and width of microcracks (solid lines) are generally proportional to the decreasing rate of peak strength (dash line) in this study, as shown in Figure 9. This indicates that the cracking caused the deterioration of sample strength as thus the decrease of peak strength.

The above observation is generally consistent with the variation trend of microcrack length and width. Within the first stage (before 20 cycles of freezing-thawing), the first 20 freezing-thawing cycles increased the lengths greatly, while the widths were moderately affected. For the second stage (20–60 cycles of freezing-thawing), both crack length and width increased gently. In the third stage (60–80 cycles of freezing-thawing), they increased significantly as they did in the first stage.

Figure 10 shows the azimuthal angles of three types of microcracks in sandstone samples after the triaxial unloading tests. The axial loading direction was ±90°. With the increasing freezing-thawing cycle, the azimuthal angle of the intercrystalline crack was more concentrated at ±70–80°, that of the transcrystalline crack was distributed between ±40 and 80°, and that of the intracrystalline crack was irregularly distributed between ±0 and 90°. As the exterior surface of sandstone samples was under uniform freezing-thawing treatment, the initial cracks produced by the freezing-thawing process would distribute homogeneously. The heterogeneous distribution of cracks was thus caused by the triaxial unloading test. Among the three types of crack, the intercrystalline crack was mostly impacted, which is consistent with the observation in Figure 9.

When the samples were not under freezing-thawing treatment, i.e., the number of freezing-thawing cycles is zero as shown in Figure 9, the total number of intracrystalline cracks, transcrystalline cracks, and intercrystalline cracks is 10, 18, and 37, respectively. They are much less than that when the samples were subjected to severe freezing-thawing treatments. Thus, our results generally show that the triaxial unloading test has an insignificant impact on the cracking extent such as crack number, length, and width but has a significant impact on the azimuthal angles of three types of microcracks (intercrystalline crack, transcrystalline crack, and intracrystalline crack). The main reason caused the significant cracking (as shown in Figure 9) is due to the freezing-thawing treatment, indicating that the dynamic failure characteristics of sandstones in a cold region where cyclic freezing-thawing occurs are rather different than rocks in a warm region. Also, since the cracking direction could be altered by the influence of unloading, the design of rock tunneling support and safe operation of mining in cold region should both consider the impact of cyclic freezing-thawing and the excavation operation direction. Further SEM imaging analysis can be carried out to isolate the effect of freezing-thawing treatment and unloading test on the microcracking characteristics of rocks. Work is currently being carried out to perform three times of SEM imaging on samples: the first one is performed before freezing-thawing treatment to measure the undamaged samples without freezing-thawing treatment; the second one is carried out after freezing-thawing treatment to understand the true impact of freezing-thawing treatment on cracking; the third one is conducted after the unloading test to understand the true impact of the unloading test after excluding the influence of freezing-thawing treatment.

4. Conclusions

To understand the dynamic failure characteristics of tunneling rocks in cold regions, sandstone samples were subjected to different cycles of freezing-thawing (0, 20, 40, 60, and 80 cycles) and their stress-strain curves with peak strength and damage dilatancy stress values were measured under triaxial unloading tests. The following conclusions can be made:(1)The first several freezing-thawing cycles had a more significant influence on the mechanical characteristics of the sandstone samples (the peak strength and damage dilatancy stress) than the later freezing-thawing cycles unless the freezing-thawing cycle became sufficiently high (i.e., 80 cycles in this study).(2)With the increasing number of freezing-thawing cycle, the morphology of the fracture section mainly appeared as honeycomb-like microstructure, stripped microstructure, and flocculent microstructure. Few step-like microstructure or smooth-joint microstructure pattern was observed.(3)The triaxial unloading test insignificantly impacted the cracking extent such as crack number, length, and width but significantly impacted the azimuthal angles of three types of microcracks, i.e., intercrystalline cracks, transcrystalline cracks, and intracrystalline cracks.(4)Future study can be carried out to understand the influence of axial pressure on the dynamic failure characteristics of tunneling rocks under the triaxial unloading test, particularly when the axial pressure is rather high.

Data Availability

The data used to support the findings of this study are available from the corresponding author upon request.

Conflicts of Interest

The authors declare that there are no conflicts of interest regarding the publication of this paper.


This work was supported by the National Natural Science Foundation of China (grant nos. 51409122 and 51878249) and the Natural Science Foundation of Zhejiang Province (grant no. LQ12E08004).