Evolution Law of Acoustic–Thermal Effect of Freeze–Thaw Sandstone Failure Based on Coupling of Multivariate Monitoring Information

Abstract

The instability and failure of rock that has been frozen and thawed cause serious rock engineering accidents in cold regions. Exploring the precursor information of freeze–thaw rock failure is of great theoretical value and engineering significance. Real-time uniaxial compression acoustic thermal monitoring experiments were conducted on freeze–thaw sandstone, and non-contact rock fracture precursor warning indicators were proposed. According to the coupled analysis of acoustic–thermal monitoring information, a precursor information chain for freeze–thaw rocks was established in time and space, and the spatiotemporal evolution of damage and acoustic thermal effects of freeze–thaw sandstone under compressive load was studied. The freeze–thaw cycle enhances the sensitivity of acoustic–thermal precursor information. Significant synchronous changes in ring count often occur during the rapid expansion period of damage, which can provide an essential reference for the occurrence and intensification of damage. The sequence of precursor warning information during the process of freeze–thaw sandstone compression failure is in the order of thermal infrared temperature → acoustic emission ringing count → acoustic emission energy → infrared thermal image. Thermal infrared temperature and acoustic emission precursor information can help in prioritizing early warning of rock damage in terms of time. At the same time, thermal image anomalies can predict potential fracture areas of rocks.

1. Introduction

The western region of China is rich in mineral resources, and the construction of infrastructure in the region is of great significance for the development and utilization of resources. However, there are many cold areas in the western region. Under the influence of a freeze–thaw environment, the mechanical properties of rock deteriorate, resulting in decreased strength. The failure process of rock seriously affects the stability of rock engineering in cold regions. Safe and stable infrastructure is the standard basis for ensuring production, operation, work, and living conditions for residents. It guarantees the regular operation of main urban facilities. It is an essential condition not only for material production but also for labor reproduction. Having a safe and stable infrastructure can promote the rapid economic development of Western China, narrow the gap with the Eastern region, and achieve sustainable development. Therefore, it is of great theoretical value and engineering significance to explore the precursor information of freeze–thaw rock failure, to provide warnings of rock failure, and to ensure the stability of rock engineering (tunnels, rock slopes, etc.) in cold regions.

At present, many experts and scholars at home and abroad have conducted in-depth research on rock instability and failure. Various experimental monitoring instruments, such as acoustic emission, high-speed cameras, and infrared thermal imagers, have been used to study the rock failure process, failure characteristics, and precursor information [1,2,3,4,5,6,7,8,9,10]. Gao [11] studied the mechanism of trans-scale fracture behavior of sandstone, granite, and gabbro in the microwave field on the macro to fine-micro scale. Zhang et al. [12] made rock-like specimens with 20 random joints by using 3D printing technology with a fracture network model. The digital image correlation method (DIC) was used to study the evolution of the strain field in the process of specimen failure, and the expansion process of cracks was analyzed. The deformation and failure law and crack propagation characteristics of complex fractured rock were quantitatively studied. Li et al. [13] obtained the spatial three-dimensional CT characteristics of the internal crack morphology of rock under different working conditions by using self-developed true triaxial loading test equipment supporting a CT scanning system and determined the internal crack propagation and evolution law in the whole process of rock failure in real-time. A study by Tronin [14] found that the thermal infrared radiation signal mutation during rock failure was closely related to the mechanical properties of rock, fracture forms, and loading methods. Archana et al. [15] used thermal infrared spectroscopy to quantitatively analyze the thermal infrared temperature variation during the rock compression fracture process. Wu et al. [16] proposed a quantitative analysis of the rock fracture process using thermal infrared parameters such as average infrared radiation temperature (AIRT) and infrared radiation temperature variance.

Fu [17] studied the acoustic emission characteristics of marble under cyclic loading and unloading and pointed out that the b value of acoustic emission could improve the accuracy of rock failure prediction. Wang et al. [18] analyzed the fracture and acoustic emission characteristics of a specimen using a high-speed camera and an acoustic emission monitoring device, and they discussed the fracture mechanism of the specimen and the influencing factors of fracture strength. The acoustic emission ring count was closely related to the failure mode of the rock sample. Chen [19] carried out uniaxial compression tests on granite, basalt, red sandstone, limestone, and marble samples; monitored the free surface temperature and internal acoustic signals during the sample loading process; and proposed a joint analysis index of thermal–acoustic sensitivity. Ren [20] analyzed the failure mechanism of fractured coal rock by using an acoustic emission monitoring system combined with a failure diagram and the stress–strain relationship of fractured coal rock. Li et al. [21] monitored the uniaxial compression failure process of gabbro, granite, and sandstone by using a far-infrared instrument and a strain gauge, and they found a positive correlation between rock temperature and volumetric strain. Liu et al. [22] studied the tensile acoustic emission characteristics of sandstone after freeze–thaw cycles and noted that the damage degree of rock samples could be judged by the peak value of the acoustic emission ring count.

The above studies show that various physical effects [23,24,25,26,27,28] occur in the process of rock disaster, such as stress [29], acoustic emission [30,31], infrared radiation [32,33,34], and others. Abnormal changes in these physical effects are related and sequential [3]. The occurrence of acoustic emission is mainly associated with the fracturing of the internal structure of the rock, and the infrared radiation signal reflects the distribution characteristics of the surface temperature of the rock. Monitoring this information can increase the reliability of identifying the precursors of rock fracture in engineering construction in cold regions.

Exploring methods that can comprehensively utilize precursor information of rock failure and accurately reflect rock failure at different stress stages is a challenge in current research on rock instability and failure. A non-contact multivariate information monitoring method is proposed that combines acoustic testing technology with infrared thermal imaging technology. Based on the joint acoustic–thermal index, the damage evolution and precursor information of the process of freeze–thaw sandstone compression failure can be studied. This non-destructive testing method provides a new approach for studying the precursor information of freeze–thaw sandstone failure and the macroscopic and microscopic damage evolution process.

2. Real-Time Acoustic–Thermal Load Monitoring Test of Freeze–Thaw Sandstone

2.1. Rock Sample Preparation and Test Conditions

The rock samples used in the test were taken from Xianyang, Shaanxi Province (Figure 1). The quartz sandstone of the Cretaceous Luohe Formation is brownish-red. The rock samples were complete and uniform, with a particle size between 0.05 and 0.25 mm, composed of gravel argillaceous cementation. The main components are quartz, feldspar, and mica. After drilling, cutting, and grinding, a standard cylindrical sample of Φ50 mm × 100 mm was made (Figure 2).

The wave velocity of the rock sample was measured by a wave velocity meter, and rock samples with similar wave velocity were divided into groups to ensure that the physical and mechanical properties of each group of rock samples were the same. The grouped rock samples were dried in an oven, and then their size and quality in the dry state were measured. The size and quality of rock samples under vacuum saturation were measured by a vacuum saturation instrument for 24 h. The rock samples were divided into 5 groups, and experiments with 0, 5, 10, 20, and 30 freeze–thaw cycles were carried out.

According to the temperature distribution of cold regions in China and the standard for methods of testing engineering rock mass [35], the freeze–thaw cycle temperature range was −20 to +20 °C in this test. The rock samples were placed in a programmed TMS9018-R20 low-temperature incubator to freeze at −20 °C for 4 h and melt at +20 °C for 4 h. At the end of each freeze–thaw cycle, the sample was saturated for 12 h to replenish the water. The sandstone samples were subjected to 0, 5, 10, 20, and 30 complete freeze–thaw cycles.

2.2. Multivariate Acoustic–Thermal Real-Time Monitoring Test of Freeze–Thaw Sandstone under Uniaxial Compression

Uniaxial compression tests of rock samples under different freeze–thaw cycles were carried out, and the failure process of freeze–thaw sandstone was monitored in real-time by an acoustic emission instrument and infrared thermal imager. The time adjustment of the mechanical testing machine, infrared thermal imager, and acoustic emission monitoring system was consistent, and the three instruments recorded the data synchronously during the test. The uniaxial compression test was carried out by a microcomputer-controlled TAW-1000 high-temperature rock triaxial creep test machine. The loading process was controlled by using the displacement control method at a rate of 0.06 mm/min, and the pace was constant during the loading process until failure. The infrared thermal imaging test used a FOTRIC 340 infrared thermal imager. The resolution of the imager is 384 × 288, the thermal pixel resolution (SR) is 765 × 576, and the frame rate is 16 Hz. The acoustic emission test system was a fully digital AE-8 acoustic emission monitoring system. To effectively filter the noise signal and reduce its influence on the test results, the threshold value of the acoustic emission system was set at 40 dB, and the sampling frequency was 1 MHz. Before the test, the signal acquisition of the acoustic emission channel was tested as normal, and the rock sample was gently tapped. The test was started when the signal could be monitored in the test system. Due to the sensitivity of acoustic and thermal signals in the failure process of rock samples, to ensure the accuracy of the test results, it was necessary to pay attention to the following points during the test: (1) Twenty-four hours before the start of the formal test, the rock sample was put in the laboratory to ensure that its temperature was similar to the ambient temperature and the instrument temperature and to reduce the interference of external factors during the test. (2) The emissivity of the thermal imager was set to 0.95, relative humidity was 50%, accuracy was 0.02 °C, temperature range was set to 1 °C, external optical transmittance was 1.0, and the infrared thermal imager was placed about 0.5 m away from the rock sample. The lens was aimed at the rock sample, and the focal length was adjusted to ensure that the image from the thermal imager was visible. (3) Before the formal test, the infrared thermal imager was preheated for 5–10 min until the temperature change of the rock surface stabilized. Once the test began, people were prohibited from walking back and forth, in order to ensure the stability of the test environment and the accuracy of the test results. The test equipment and process are shown in Figure 2.

3. Analysis of Acoustic Emission Response Characteristics of Freeze–Thaw Sandstone under Uniaxial Compression

3.1. Variation Characteristics of Ring Count of Freeze–Thaw Sandstone during Compression Failure

The acoustic emission ring count is the acoustic effect of meso-structure changes in freeze–thaw sandstone compression failure. It reflects the strength of acoustic emission in the process of rock failure. It can characterize the real-time state of rock damage and failure and reveal the damage law of the internal meso-structure in the process of freeze–thaw sandstone failure.

According to the variation of stress and ring count of sandstone over time during different freeze–thaw cycles (Figure 3), it can be seen that acoustic emission ringing of rock samples after 0 and 5 freeze–thaw cycles did not appear before 900 s. This is because, at the initial loading stage, the internal structure of rock samples is relatively stable and has a strong bearing capacity. At this time, the inner cementation strength of sandstone is high, the cementation structure is stable, there is almost no acoustic emission signal generated by the change of internal microstructure, and the ring count is low. The sandstone ring count under different freeze–thaw cycles indicates the sandstone compression failure process under the synergistic effect of freezing–thawing–loading in real-time.

Under compressive load, the crystals inside the rock sample become misaligned, micro-cracks begin to form, initial damage gradually develops, and the ring count increases. With the continuous action of the load, micro-cracks and pores further develop and expand, the damage to the meso-structure inside the rock sample continues to increase, and a large number of acoustic emission signals are generated. When the peak stress is reached, the microcracks gather and penetrate to form a fracture surface, and the ring count and cumulative ring count show a sharp increase trend and noticeable Kaiser effect. After the peak, the stress decreases rapidly, and brittle failure of the rock occurs, but the internal structural damage of the rock sample continues to expand and there is still a ring count. It can be seen that the failure of freeze–thaw sandstone under compressive load is the result of the cumulative effect of continuous dislocation and damage expansion of the internal microstructure.

With increased freeze–thaw cycles, the peak value of the ring count rate gradually decreases. The reason is that the structure of rock samples without freeze–thaw cycles is relatively uniform and stable, and the degree of intergranular cementation is high. After repeated freeze–thaw cycles, the internal structure of red sandstone gradually deteriorates, the degree of cementation between particles is reduced, and there are weak parts. Therefore, the fewer the freeze–thaw cycles during rock failure, the more concentrated the acoustic emission signals of rock samples will be. It can be seen that the sensitive intensity of acoustic emission signal in the uniaxial compression failure process of freeze–thaw sandstone is in the order of 0 times > 5 times > 10 times > 20 times > 30 times.

In addition, from Figure 3c–e, it can be seen that after the 10, 20, and 30 freeze–thaw cycles, the sandstone has obvious acoustic emission signals before 900 s. With increased freeze–thaw cycles, the ring count also increases, which indicates that the freeze–thaw action aggravates the initial damage of sandstone at the mesoscale. At the same time, it is found that when the peak stress is reached, the internal cracks of the rock are penetrated, and a large number of acoustic signals are generated while the rock sample is destroyed, and the ring count rises sharply. The period of a sharp increase in ring count is consistent with the peak stage of rock failure. Therefore, the sharp increase in the ring count can be regarded as a sign of freeze–thaw sandstone failure.

3.2. Influence Mechanism of Freeze–Thaw Cycles on Mesoscopic Damage of Rock and Acoustic Emission Parameters

Under long-term geological processes, different types of initial damage occurs within natural rocks, such as pores and microcracks. Because of the influence of external temperature and load, the initial damage continues to expand, resulting in new microcracks. Through continuous evolution, the initial damage ultimately develops into macroscopic cracks. The expansion of initial damage causes deterioration of the microstructure, which alters the internal structure of the rock sample. Therefore, under compressive load, the peak acoustic emission ring count of rock samples under different freeze–thaw cycles decreases, and the sensitivity is closely related to the microstructure of the rock.

The mechanism of the influence of freeze–thaw cycles on the variation of acoustic emission characteristic parameters is shown in Figure 4. Pores and cracks naturally occur in sandstone (Figure 4a,b). When sandstone is in a water-rich environment, the water enters the pore network until it is fully saturated. In the freezing process (Figure 4b), the water in the pores freezes, which results in volumetric expansion and frost heaving. The pore ice formed from the pore water expands by approximately 9% and can generate significant frost heave stress. If more than 90% of the pore volume is filled with water, expansion of the water during freezing will cause pressure. This stress acts on the pore walls, and when the pressure exceeds the tensile strength of the rock, cracking occurs and the pore walls are destroyed, potentially propagating the existing crack tips. This is illustrated in Figure 4c, which shows the expansion and elongation of existing cracks. In addition, the effects of low temperatures can cause thermal stress, which can initiate new cracks oriented perpendicular to the existing cracks (Figure 4a). Therefore, under low-temperature conditions, frost heave and thermal stress are the leading causes of damage to sandstone structures. Under repeated frost heave, deformation occurs, and such deformation cannot be restored during the melting stage. Therefore, the damaged structure within the rock sample continues to evolve and expand. Under compressive loads, changes in the microstructure inevitably result in changes in the acoustic emission parameters.

4. Analysis of Thermal Infrared Radiation Response Characteristics of Freeze–Thaw Sandstone Failure Process

4.1. Characteristics of the Relationship between Thermal Image Anomalies and Mesostructure Damage Propagation

According to Thomson’s research [36], the temperature of the rock sample is proportional to the principal stress, and the stress concentration in the area near the fracture causes the temperature to rise. Assuming that the crack in freeze–thaw sandstone under compressive load is a I-II composite crack, the fracture strength factor $K_j$ [37] can be expressed as

$$\left.\begin{array}{l}{K}_2=\tau \sqrt{\pi a}\\ K_1=K_{1N}+K_{1T}\\ K_{1T}=0 ({\sigma }_T>0)\\ K_{1T}=\frac{1}{2}{\sigma }_T\sqrt{\rho /a}\sqrt{\pi a} (\rho /a\to 0,{\sigma }_T<0)\\ K_{1N}={\sigma }_N\sqrt{\pi a}\end{array}\right\}$$

(1)

$$\left.\begin{array}{l}{\sigma }_T={\sigma }_y{\cos }^2\beta +{\sigma }_x{\sin }^2\beta \\ {\sigma }_N={\sigma }_y{\sin }^2\beta +{\sigma }_x{\cos }^2\beta \\ \tau =({\sigma }_y-{\sigma }_x)\sin \beta \cos \beta \end{array}\right\}$$

(2)

In the formula, σ_T, σ_N, and τ are the transverse compressive stress, normal stress, and shear stress of the fracture, respectively; β is the angle between the crack tip and the vertical pressure direction; ρ is the radius of curvature of the crack tip; and a is the fracture length.

Based on the relationship between the thermal infrared image signal and tip stress, we have the following [38]:

$$\Delta T=-\frac{\alpha }{\rho C_p}T\Delta ({\sigma }_N+{\sigma }_T)$$

(3)

In the formula, ΔT is the temperature variation, T is the absolute temperature value of the rock surface, α is the thermal expansion coefficient, and $C_p$ is the heat capacity of the rock under normal pressure.

From the above formula, we obtain

$$\Delta T=-\frac{\alpha }{\rho C_p}T\Delta (\frac{2K_{1T}+K_{1N}}{\sqrt{\rho /a}\sqrt{{\displaystyle \prod a}}})$$

(4)

It can be seen from Equation (4) that the stress intensity factor of the damaged area is proportional to the temperature, and the temperature at the stress concentration is relatively high. In a thermal infrared image, the stress concentration area appears as an anomaly. The original cracks continue to expand under freeze–thaw and load, resulting in new cracks and other damaged areas, and the temperature also increases. The temperature distribution law is consistent with the change law of the mesoscopic damage inside the rock sample, and the abnormal change position of the thermal infrared image is closely related to the propagation direction of damage such as cracks.

4.2. Temperature–Stress–Time Variation during Compression Failure Process

According to the results of acoustic–thermal real-time monitoring of freeze–thaw sandstone under uniaxial compression (Figure 5), the temperature will fluctuate before the rock samples fail under different freeze–thaw cycles, and the time when the temperature reaches the maximum value is consistent with the time when the yield strength of rock samples occurs. However, with increased freeze–thaw cycles, the frequency of temperature fluctuation with time increases during the compression process, and the fluctuating amplitude first increases and then decreases when the rock sample is destroyed. The temperature fluctuates by 0.3 °C when the rock sample is destroyed by 0 freeze–thaw cycles, 0.9 °C when the rock sample is destroyed by 10 freeze–thaw cycles, and 0.6 °C when the rock sample is destroyed by 30 freeze–thaw cycles. The average temperature–time and stress–time curves correspond well with each other, especially when the load reaches peak strength. The evolution of the mesoscopic damage caused by the freeze–thaw cycles results in reduced strength of the rock sample, which leads to differences in temperature change amplitude and time sequence in the process of freeze–thaw rock compression. Therefore, measuring the average temperature is a non-contact way to reflect the damaging effect of the rock fracture process and freeze–thaw cycles on rock samples.

4.3. Three-Dimensional Thermal Image Change Law of Freeze–Thaw Sandstone Failure Process

During the uniaxial compression acoustic–thermal multivariate monitoring test, thermal images of freeze–thaw sandstone at five time points were recorded: the beginning of loading (t = 0 s), 0.5 σmax, the beginning of the abnormal thermal image, the peak stress, and the end of loading. The statistical analysis (Figure 6) intuitively reflects the spatial distribution of rock temperature changes in different loading stages. In freeze–thaw sandstone during loading and damage, a temperature anomaly area gradually develops from the edge to the middle of rock samples. The high-temperature area is located in the center of the rock sample when the sample is destroyed. The temperature difference on the surface of the rock sample is significant. With increased freeze–thaw cycles, the damage to the rock sample is aggravated. During the compression process, the high-temperature area of the rock sample gradually increases, the thermal image evolves emphatically, and the difference in the three-dimensional thermal image is greater. The abnormal change position of the thermal infrared image is closely related to the damaged area and expansion direction. According to the anomalies in three-dimensional thermal images of freeze–thaw sandstone, areas of increased temperature can be judged, which reflects the precursor information of damage from the three-dimensional perspective and can indicate the damaged areas of rock samples in space.

The first appearance of the precursor of the temperature anomaly in rock samples under 30 freeze–thaw cycles was the earliest. Rock samples that undergo many freeze–thaw cycles are more sensitive to temperature. The first appearance of the temperature anomaly precursor in an unfrozen rock sample was the latest; the freeze–thaw cycle affects the sensitivity of rock to temperature. The temperature variation range of rock samples under freeze–thaw cycles is 2–3 times that of rock samples that do not undergo freeze–thaw cycles, indicating that the energy released by frozen and thawed rock samples during fracture is greater than that of rock samples that have not been frozen and thawed. Therefore, the temperature sensitivity of different freeze–thaw rock samples is in the order of 30 times > 20 times > 10 times > 5 times > 0 times.

4.4. Analysis of Thermal Infrared Early Warning Index in the Process of Freeze–Thaw Sandstone Failure

To examine the abnormal temperature change in the process of freeze–thaw rock failure, a precursor warning index of rock instability and failure is proposed [39]. Based on the theory of statistical damage mechanics, the early warning index K is defined to characterize the prediction of thermal infrared temperature during compression failure of freeze–thaw sandstone, which provides precursor information for such failure. The initial damage distribution of freeze–thaw rock is random. Assuming that the temperature change and damage variable of each micro-element conform to Weibull distribution, and the deformation of each micro-element conforms to Hooke’s law before it ruptures, the probability of damage of freeze–thaw rock can be expressed as follows [39]:

$$f(x)=\frac{k}{\lambda }{\left(\frac{x}{\lambda }\right)}^{k-1}\exp \left[-{\left(\frac{x}{\lambda }\right)}^k\right]$$

(5)

where x is the standard deviation of rock surface temperature at a particular moment, λ is the average temperature of the rock surface, K is the inhomogeneous index of rock, and x and λ can be obtained by Equations (7) and (8):

$$x=\sqrt{\frac{1}{n-1}{\displaystyle {\sum }_{i=2}^n{(T_i-\overline{T})}^2}}$$

(6)

$$\lambda ={\displaystyle {\sum }_{i=1}^n(T_i/n)}$$

(7)

In the formula, $T_i$ is the temperature at any point on the surface of the rock sample at time t, and $\overline{T}$ is the average temperature of the surface of the rock sample in the period 0 to t.

The probability density function of the micro-element damage variable and temperature variable based on the thermal infrared temperature can be expressed as

$$\frac{dD}{dx}=f(x)$$

(8)

The micro-element damage variable D is calculated as

$$D=\frac{\left|A_T-A_{Ti}\right|}{A_T}$$

(9)

In the formula, $A_{Ti}$ is the temperature at any point on the rock surface, and $A_T$ is the overall average temperature of the rock surface. The early warning index K is

$$K=\frac{\ln \left(\frac{x}{\lambda }\right)}{\ln \left|\ln (1-D)\right|}$$

(10)

According to Equation (10), the K–temperature–time curve during the failure process of freeze–thaw sandstone can be obtained (Figure 7). After performing statistical analysis,

Table 1. Schedule of steep increases of K value and temperature in the process of freeze–thaw sandstone failure.

Freeze–Thaw Times	Time When K Value Steeply Increases	Time When Temperature Surges	Lead Time
0	903 s	916 s	13 s
5	949 s	966 s	17 s
10	722 s	748 s	26 s
20	719 s	746 s	27 s
30	729 s	764 s	35 s

was obtained. It can be seen that in the whole process of loading until failure, the value of early warning index K is consistent with the changing temperature trend. When the overall temperature of the rock changes little, K tends to be stable, and the value will increase significantly in the first few seconds of the sudden increase in temperature. It can be seen that the early warning index has a consistent change trend with the temperature change but has a certain advanced appearance, which can better realize the extraction of information about thermal infrared damage to freeze–thaw rock.

5. Analysis of Acoustic–Thermal Information Chain for Freeze–Thaw Sandstone Damage

5.1. Coupled Acoustic–Thermal Information Characteristics of Freeze–Thaw Sandstone Failure Process

The acoustic emission and infrared radiation information have different degrees of response during the failure process of freeze–thaw sandstone. The characteristic parameters of real-time acoustic–thermal monitoring and the response to stress change were summarized and analyzed to obtain Figure 8. Comparing the characteristics of acoustic emission and infrared radiation parameters in the process of compression failure, the evolution of freeze–thaw sandstone under compression load shows good coupling characteristics, which can be divided into three stages:

(1) Initial damage closure stage: The initial damage (micropores and cracks) in the rock sample close under uniaxial compression load. As the load increases, the initial pores and cracks begin to expand under continuous compression.

(2) Micro-damage stage: The cracks and pores inside the rock sample are closed under the action of external forces, and the stress increases approximately linearly. After 0 and 5 freeze–thaw cycles, the ring count of the rock sample has no signal, the acoustic emission energy is in the accumulation stage, and the thermal infrared temperature fluctuates. The rock samples after the 10, 20, and 30 freeze–thaw cycles have more active acoustic emission activity, the ring count shows a slow upward trend, less acoustic emission energy is released, and the process is in the energy storage stage. The thermal infrared temperature fluctuates more compared to rock samples with fewer freeze–thaw cycles, without abnormal phenomena.

(3) Damage propagation stage: The slope of the stress curve becomes larger and the growth rate accelerates. After 0 and 5 freeze–thaw cycles, the rock sample produces a small ring count, less acoustic emission energy is released, and the thermal infrared temperature fluctuates. After 10, 20, and 30 freeze–thaw cycles, a large number of acoustic emission signals are generated, the growth rate of ring count and energy is accelerated, the thermal infrared temperature fluctuates frequently, and there are no abnormal phenomena on the whole.

(4) Failure stage: Under the continuous action of load, many cracks are generated inside the rock sample, and the original cracks expand, cluster, and penetrate to form prominent macrocracks. This stage is the critical period for capturing the precursor information of rock sample fractures. The thermal infrared temperature first showed a rapid growth trend, and then the stress peaked. During this period, the microfracture activity inside the rock sample was intense, releasing a large amount of acoustic emission information. The ring count surged, energy was released, and the change of acoustic–thermal information and the shift in stress responded synchronously in time.

5.2. Time Series Characteristics of Acoustic–Thermal Precursor Information of Freeze–Thaw Sandstone Failure Process

Before compression fracture of freeze–thaw sandstone occurs, the acoustic emission signal increases abnormally, an anomaly appears on the thermal infrared image, and the temperature changes greatly. These phenomena can be regarded as the precursor information of damage.

Table 2. Comparative analysis of acoustic–thermal precursor information chain during freeze–thaw sandstone failure.

Freeze–Thaw Cycles	Acoustic Emission		Thermal Infrared		Comparative Analysis of Acoustic–Thermal Precursors
	Acoustic Emission Energy Increases Sharply	Acoustic Emission Ring Count Increases Sharply	Temperature Jumps	Thermal Image Mutates
0	998 s	995 s	916 s	1005 s	Temperature jump is 79 s earlier than ring count, 82 s earlier than energy, 89 s earlier than thermal image mutation, and 95 s earlier than peak stress.
5	1055 s	1051 s	966 s	1063 s	Temperature jump is 85 s earlier than ring count, 89 s earlier than energy, 97 s earlier than thermal image mutation, and 106 s earlier than peak stress.
10	841 s	834 s	748 s	852 s	Temperature jump is 86 s earlier than ring count, 93 s earlier than energy, 104 s earlier than thermal image mutation, and 115 s earlier than peak stress.
20	828 s	811 s	746 s	851 s	Temperature jump is 65 s earlier than ring count, 82 s earlier than energy, 105 s earlier than thermal image mutation, and 118 s earlier than peak stress.
30	840 s	802 s	764 s	872 s	Temperature jump is 38 s earlier than ring count, 76 s earlier than energy, 108 s earlier than thermal image mutation, and 125 s earlier than peak stress.

Note: The time in the table is the time for the loading process.

shows a time series comparison of acoustic–thermal precursor information in the process of freeze–thaw sandstone compression failure. It can be seen that in the process of stress failure, the precursor information of rock samples under different freeze–thaw cycles occurs in the same order: thermal infrared temperature → acoustic emission ring count → acoustic emission energy → thermal infrared image. The thermal image anomaly appears in the shortest time from rock failure, and its precursor warning is somewhat weak. On the other hand, the turning point of thermal infrared temperature anomaly occurs in the longest time from rock failure, and its early warning is relatively the best. The time when the acoustic emission ring count and high-frequency energy appear is between the time of rock failure and the transition of the thermal image anomaly and the appearance of the thermal infrared temperature anomaly. The early warning is better than that of the thermal infrared image anomaly, but worse than that of the thermal infrared temperature transition.

The main reasons for the difference in the time sequence of the precursor information of acoustic thermal early warning are as follows: (1) Because the induction of the thermal infrared temperature is sensitive and can detect the temperature transition before the freeze–thaw rock failure occurs, the thermal infrared temperature precursor information appears earliest. However, the temperature change of the rock surface will cause an anomaly in the infrared thermal image, which requires reaching a certain threshold, so the precursor information of the thermal image abnormality is relatively slow. (2) Crack development in the freeze–thaw rock sample will cause the ring count to fluctuate, and the microcracks inside the rock will greatly expand. Such penetration indicates that it is on the verge of failure. At this time, the acoustic emission precursor information appears in a relatively short time from rock failure, so the acoustic emission warning is relatively later than the thermal infrared temperature warning.

In summary, as the number of freeze–thaw cycles increases, the emission signal increases, and it shows an eccentric increase. At the same time, the phenomenon of abnormal temperature change on the surface of rock will occur. When the rock is on the verge of failure, anomalies appear on the thermal image at the surface microcracks. The precursor acoustic thermal warning information arises from the change of mechanical properties caused by the expansion of meso-damage in the structure when the freeze–thaw rock is loaded to a certain stage. Therefore, it is feasible to determine the evolution process of freeze–thaw damage through the abnormal changes of acoustic and thermal infrared signals to achieve the purpose of damage warning.

5.3. Precursor Information Chain of Rock Failure

A comparison and analysis of the occurrence time and characteristics of acoustic–thermal precursor information shows that the information chain of freeze–thaw sandstone compression failure (

Table 3. Precursory information chain during freeze–thaw sandstone failure.

Freeze–Thaw Cycles		Temperature Jumps	Acoustic Emission Ring Count Increases Sharply	Acoustic Emission Energy Increases Sharply	Thermal Image Mutates	Peak Stress
0	Stress	16.28 MPa	18.94 MPa	18.96 MPa	17.95 MPa	19.12 MPa
	Time	95 s	16 s	13 s	6 s	0 s
	Performance form
5	Stress	15.73 MPa	17.03 MPa	17.24 MPa	17.30 MPa	17.37 MPa
	Time	106 s	21 s	17 s	9 s	0 s
	Performance form
10	Stress	12.76 MPa	14.46 MPa	14.67 MPa	14.89 MPa	14.96 MPa
	Time	115 s	29 s	22 s	11 s	0 s
	Performance form
20	Stress	12.51 MPa	13.62 MPa	13.77 MPa	13.83 MPa	14 MPa
	Time	118 s	53 s	36 s	13 s	0 s
	Performance form
30	Stress	9.17 MPa	9.65 MPa	10.33 MPa	10.75 MPa	10.88 MPa
	Time	125 s	87 s	49 s	17 s	0 s
	Performance form

Note: The time in the table is the time from peak stress during the loading process.

and Figure 9) includes the following: monitoring of the tensile or shear failure inside the rock sample indicates the abnormal thermal infrared temperature and the distance from the peak stress are about 95–125 s; subsequently, the fracture caused by crack propagation in the rock sample causes the acoustic emission ring count to increase sharply, and the time from the peak stress is about 16–87 s. During the loading process, the energy of the freeze–thaw rock sample accumulates. When it is close to the peak stress, the energy is released, causing a sharp increase in acoustic emission energy. The time from the peak stress is about 13–49 s. Second, with the continuous action of load, the crack further expands and the rock breaks and the infrared thermal image is abnormal. At this time, it is close to the peak stress time, about 6–17 s.

In the process of freeze–thaw sandstone compression failure, with increasing freeze–thaw cycles, the precursor characteristics become increasingly obvious. The precursor information chain links the acoustic–thermal precursor information in the process of freeze–thaw sandstone damage under load and constructs a relatively complete precursor information warning chain in time and space. This information chain can provide warnings about damage to the rock in time and space. The thermal infrared temperature and acoustic emission precursor information can be used as the early warning in the information chain, in order to prioritize warning of damage to freeze–thaw rock samples in time. The precursor thermal image anomaly and macroscopic crack propagation information can be used as a later warning for the potential fracture position of rock in space.

6. Discussion

There are thermal infrared and acoustic emission signal changes during rock fracture. Therefore, it is of great significance to study the thermal–acoustic signals in the process of the rock fracture process. Red sandstone is a widely distributed rock used in cold engineering. The acoustic thermal precursor signals of freeze–thaw red sandstone have significant reference significance for evaluating rock stability in cold areas. However, due to the complex distribution of geological conditions of rock engineering in cold regions, failure modes, and processes vary greatly across lithologies, which significantly affects the evaluation of rock engineering stability in cold areas. Therefore, the thermal–acoustic sensitivity of the failure process of different lithological rocks, such as granite, basalt, limestone, marble, etc., should also be investigated.

In this study, uniaxial compression experiments on rock samples that had undergone 30 freeze–thaw cycles were carried out. More freeze–thaw cycle experiments need to be performed for rock engineering in permafrost. This is also the work we intend to carry out in future research.

7. Conclusions

A non-contact multivariate information monitoring method is proposed that combines acoustic testing technology with infrared thermal imaging technology. The influence of freeze–thaw cycles on sandstone damage and failure was explored with this new approach. The joint acoustic–thermal index proposed in this paper has practical engineering value for evaluating the stability of rock engineering in cold regions. The conclusions are as follows:

(1) Acoustic emission characteristic parameters and thermal image information link precursor information during the process of freeze–thaw sandstone failure. A relatively complete warning chain is constructed in time and space.

(2) The period of sharply increased ring count is consistent with the peak stage of rock failure, and the peak value of acoustic emission r increases with increasing freeze–thaw cycles. It was found that 10 freeze–thaw cycles were needed to change the internal microstructure of rock samples. The freeze–thaw cycles promote the advancement of the occurrence of parameters such as acoustic emission ring count.

(3) The thermal infrared temperature steadily increases before instability, and the thermal image shows apparent abnormal phenomena before the damage. The abnormalities in a three-dimensional thermal image of freeze–thaw sandstone can determine the location of higher temperatures on the rock sample and indicate the damaged area of the rock sample in space.

(4) The thermal–acoustic precursor information chain can provide spatiotemporal warning of rock damage, with thermal infrared temperature and acoustic emission parameters serving as early warnings to prioritize the damage of freeze–thaw rock samples in terms of time, and thermal image anomalies and macroscopic crack propagation serving as a late warning of potential rock fracture locations.