Experimental Study on the Evolutionary Law of Transient Saturation Zones in a Red Mud Dam under Rainfall Conditions

Abstract

Utilizing a laboratory model test, this study seeks to evaluate the distribution patterns of volumetric moisture content, soil pressure, and pore water pressure within the body of a red mud dam, given varying initial conditions of slope types and ratios, during continuous heavy rainfall. The objective is to investigate the failure mechanisms of a red mud dam under distinct operational conditions during rainfall, thereby offering insights for landslide prevention and ensuring dam construction quality. The results suggest that a stepped red mud dam acts as a buffer platform, altering the seepage direction within the dam and minimizing the water seepage path. When the slope ratio is 1:1, the transient saturated zone is located on the slope face of the dam’s body, near the top of the slope, with the saturation time at the first monitoring point occurring 300 s earlier than in a dam with a slope ratio of 1:2. Rainfall affects the distribution of internal forces in the red mud dam body. After rainfall, in the transient saturated zone of the stepped dam body, vertical soil pressure decreases 25% and horizontal soil pressure decreases 6.5%; in the transient saturated zone of the dam with a slope ratio of 1:1, vertical soil pressure decreases 14.8% and horizontal earth pressure decreases 29%; in the transient saturated zone of a dam with a slope ratio of 1:3, the change in soil pressure is small.

1. Introduction

Red mud is a solid waste generated in alumina production, and approximately 0.8–1.5 tons of red mud is discharged for every ton of alumina produced [1]. Red mud is strongly alkaline and easily affects the surrounding environment. With attempts at continuous improvement and attention to environmental requirements [2], the comprehensive utilization of red mud has become a hotspot discussed in many research efforts [3,4,5]. However, most technologies related to the comprehensive utilization of red mud are still at the stage of indoor experimentation and cannot be realized within large-scale applications. The lack of effective methods for the large-scale integrated utilization of red mud has resulted in large quantities of red mud being landfilled, mainly in natural gullies [6,7], resulting in the formation of red mud tailing ponds. With the continuous production of alumina, the number of red mud tailings ponds continues to increase [8]. Red mud tailing ponds are potentially man-made mudslide hazards, with high potential energy and risk of dam failure, which can easily cause irreversible environmental damage in cases of failure [9,10]. Therefore, safety evaluation and prevention efforts involving red mud tailing ponds have become the key to the sustainable development of the environments surrounding red mud tailing ponds.

Rainfall is one of the main factors causing landslides [11], and rainfall-induced shallow landslides usually show clustering characteristics [12,13]. When the slope is steep or certain geologic conditions are present, the clustered landslides may evolve into debris flows [14]. Rainfall-induced slope instability is usually related to slope type, gradient, soil properties, and rainfall intensity [15,16]. Controlling the gradients and types of slopes is one of the effective measures used to reduce rainfall damage [17,18], and the flatter the slope is, the higher the safety factor is during rainfall events [16]. For man-made dammed stockpiles, the flatter slope means less stockpile quantity within a certain range, and a reduction in slope gradient can harm the economic and social value of man-made stockpiles. Soil properties are very important for slope stability [16], and soils with different properties have different slope failure forms during rainfall events [19,20,21]. Rainfall-induced slope failure is mainly caused by the transient saturated zone of the slope, where the increase in pore water pressure, the decrease in shear strength, the seepage force, and the decrease in the suction force of the substrate are the main causes of landslides on slopes [22,23,24,25,26]. In a heavy rainfall event, the intensity and duration of rainfall are critical factors for slope damage [27,28,29,30]. Red mud belongs to the group of fine-grained tailings, which are characterized by a large specific surface area, large pore ratio, poor permeability, and rapid reduction in strength upon exposure to water [31,32,33]. There are numerous research papers examining rainfall-induced landslides at present; the damage mechanisms of red mud dams in heavy rainfall events are still not clear. Therefore, it is of great significance to investigate the evolutionary law of transient saturated zones and the damage mechanisms of red mud dams under rainfall conditions.

This paper focus on analysis of the red mud dam’s characteristics (the distribution patterns of volumetric moisture content, soil pressure, and pore water pressure), given varied initial slope types and ratios, in heavy rainfall. The objective of this analysis is to elucidate the failure mechanisms of red mud dams under distinct operating conditions amidst persistent heavy rainfall. This exploration could serve as a valuable reference in preventing landslides and ensuring dam quality during the construction of red mud dams.

2. Experimental Model Design

2.1. Introduction of the Test System

The laboratory test tank, welded from a steel frame, has the dimensions of 3.5 m length, 1.2 m width, and 1.2 m height. To improve visibility, the tank is encased by five tempered glass panels covering the four sides and the bottom of the model groove, as depicted in Figure 1.

To mitigate challenges such as intense concentrated rainfall and excessive raindrop diameters exacerbating surface erosion in rainfall simulations, a micro-sprinkler commonly used in agricultural irrigation is incorporated to simulate natural rainfall. An interactive array of multiple sprinklers is employed in this experiment to address the uneven dispersion typical of a single sprinkler. The monitoring components include data acquisition systems, pore water pressure sensors, soil pressure sensors, and soil moisture sensors. The water supply equipment in this study comprises a water storage tank and a self-priming pump, with water pressure manually regulated using a valve.

2.2. Rainfall Uniformity Tests

In consideration of the local climate conditions and rainfall patterns during the wet season, the intensity of the simulated rainfall, at 85 mm/h, is determined by adjusting rainfall bracket height, flow rate, and the number of nozzles, as well as their arrangement.

Based on the predetermined rainfall intensity and the results of multiple tests, the designated configuration includes a total of 14 rainfall nozzles. To ensure uniform rainfall distribution, 7 nozzles are placed on each side of the median line in an integrated arrangement. (Figure 2)

Rainfall uniformity is an important indicator for assessing the accuracy of rainfall tests, and it is generally recognized that rainfall uniformity greater than 80% meets the test requirements. In this test, 18 measurement points are arranged in the model tank to determine the test’s uniformity. The measurement point arrangement is shown in Figure 3. After the rainfall uniformity test, the measured rainfall uniformity was 82%, which meets the test requirements. The rainfall in this experiment is uniform rainfall.

The rainfall uniformity U can be expressed as follows:

$$U=1-\frac{{\displaystyle \sum \left|R_i-\overline{R}\right|}}{n\overline{R}}$$

R_i—the amount of rainfall measured at point i in the rainfall range during the same time period; n—number of measurement points; $\overline{R}$—average rainfall for the same time period within the rainfall range.

2.3. Experimental Similarity Ratio Design

The red mud used in the test is taken from a red mud dump; the discharge of red mud to the dump is carried out by automobile transportation, belt transportation, and dry stockpiling. According to the laboratory test, the dry density of a red mud dam body is 1.28 g/cm³, and the water content is about 25% when laying red mud.

Upon analyzing the model groove size relative to the prototype’s dimensions, it was determined that the similarity ratio of the model to the prototype stands at 1:200. Using the principles of elastomechanics and corresponding boundary conditions, a similarity relationship for the slope model is derived. The similarity relationships across various physical quantities are then established in line with the second theorem of similarity, employing a dimensional matrix analysis method, as presented in

Table 1. Model similarity ratios.

Physical Quantity	Similarity Ratio	Physical Quantity	Similarity Ratio	Physical Quantity	Similarity Ratio
Length	1:200	Rate Of Flow	1:2005/2	Volume	1:2003
Cohesiveness	1:1	Density	1:1	Internal Friction Angle	1:1
Acceleration	200:1	Force	1:2002	Stress	1:1
Strain	1:1	Slope	1:1	Displacement	1:200
Permeability Coefficient	1:1	Moisture Content	1:1	Rainfall Intensity	1:1

.

2.4. Red Mud Filling Process

In conjunction with the on-site landfill construction process, the red mud filling program is specified as follows:

Initially, a layer of petroleum jelly should be applied to the interior wall of the model. The red mud is then prepared at the desired moisture content and layered, with each layer at first measuring 15–20 cm in height, and uniformly compacted to 12.17 cm using a compactor, achieving a compaction density exceeding 0.95. The sensors are subsequently embedded upon compaction completion. To ensure optimal control over compaction density and moisture content within the red mud, density and moisture tests are conducted at five random locations following the completion of each layer. This layering and testing process continues in sequence until the structure reaches a height of 73 cm.

A model test was conducted based on the #2 red mud dam, the most dangerous in the stackyard, which has a maximum stacking height of 146 m and an outer slope stacking ratio of 1:3.0. After a downsizing scale of 1:200, the maximum stacking height in the experiment was 0.73 m, the width of the bottom of the downstream dam ditch was approximately 1.2 m, and the slope ratio was set to 1:1 or 1:3, as per the experiment’s requirements. The length of the dry beach measured 0.5 m, and the bottom length of the dam body was calculated based on the slope ratio. Where stairs were present, their length was set at 0.1 m.

The experimental compactor is a small electric tamper with a rated power of 990 W (model number: 0875), weighing roughly 6 kg, and equipped with a tamper plate with a size of 150 × 150 mm. The process of red mud compaction and sensor embedding is illustrated in Figure 4.

2.5. Experimental Scheme

This study aims to examine the effects of transient saturation zone evolution patterns on the stability of red mud dams. The experiment is conducted under a rainfall intensity set at 85 mm/h to investigate the variation rules of parameters such as moisture content, soil pressure, and pore water pressure in red mud dams. Different starting conditions of sloping types and slope ratios were considered while simulating the effects of heavy rainfall.

In the experiment, 2 groups were designed. The influence of slope type on the stability of the dam under rainfall conditions is evaluated in cases L1 and L2, while the impacts of varying slope ratios on the stability of a red mud dam under similar conditions are assessed under cases L2 and L3. Details of the experimental plan are presented in

Table 2. Experimental scheme.

Sequence Number	Initial Dry Density	Initial Water Content	Rainfall Time	Rainfall Intensity	Slope Type
L1	1.281 g/cm3	25%	100 min	Heavy rain	Step-structure
L2	1.277 g/cm3	25%	116 min and 13 s	Heavy rain	1:3
L3	1.275 g/cm3	25%	74 min and 52 s	Heavy rain	1:1

.

3. Results and Discussion

3.1. Effect of Slope Type on Dam Stability under Rainfall Conditions

3.1.1. Distribution of Moisture Content in Red Mud Dams under Rainfall Conditions

The soil moisture sensor burial location for case L1 is shown in Figure 5. The rainfall duration of case L1 is 100 min, and the distribution of volumetric moisture content at the monitoring points of the red mud dam body for case L1 is shown in Figure 6.

As depicted in Figure 6a, the volumetric moisture content at monitoring point 3 on the secondary bench platform reaches 100% first. This is because monitoring point 3 is impacted not only by the change in the volumetric moisture content of red mud induced by rainfall infiltration, but also by the drainage influence of the runoff from the primary slope. Since the secondary bench platform’s slope is flat, it can store runoff water, causing the volumetric moisture content at monitoring point 3 to change at the fastest rate. The secondary slope platform corresponds to the toe of the primary slope, an area which was shown to be susceptible to water ponding in the Orense et al. (2004) study. After a period of rainfall, the volumetric moisture content of monitoring point 5, located at the base of the primary slope near the secondary bench platform, quickly reaches 100%. Additionally, the volumetric moisture content at monitoring point 1 on the secondary slope reaches 100% faster than that at monitoring point 7 on the primary slope; this is mainly due to the effect of the accumulated water on the secondary bench platform, which accelerates the infiltration of rainwater. Concerning the primary slope, volumetric moisture content at monitoring point 9, located at the hilltop, reaches 100% last. Portions of the post-rainfall water infiltrate into the red mud on the hilltop, while the remainder flows as runoff. This runoff accelerates the surface infiltration of rainwater on the slope, leading to earlier moisture changes at monitoring points 5 and 7 in comparison to point 9.

As depicted in Figure 6b, the volumetric moisture content at deeper monitoring points displays a less significant variation, compared to that found at shallower points. The most noticeable fluctuation of volumetric moisture content can be seen at monitoring point 4, partly due to the volumetric moisture content at monitoring point 5 above it having already achieved 100%. The variability at monitoring point 4 is also influenced by the downward seepage of rainwater on the primary slope’s surface. Conversely, the changes in volumetric moisture content at monitoring points 6 and 8 are minimal, which can be attributed to the fact that slope infiltration of rainwater does not follow a direct vertical path, but rather exhibits a J-shaped pattern. He noted a similar pattern during heavy rainfall experiments [27]. This J-shaped infiltration line explains the significant moisture content fluctuation at monitoring point 4.

The soil moisture sensor burial locations for case L2 are shown in Figure 7. A rainfall duration of 116 min and 13 s for case L2, and the distribution of volumetric moisture content at the monitoring points of the red mud dam body, are shown in Figure 8.

Figure 8a indicates that monitoring points 4 and 7 exhibit the initial inflection points. Monitoring point 4 is located towards the middle and lower parts of the red mud slope. Surface runoff on the slope gradually increases with ongoing rainfall. Simultaneous with the runoff at monitoring point 4, some surface rainwater infiltrates into the slope’s interior. Due to the combined effects of surface runoff and internal seepage in the red mud slope, the moisture content at monitoring point 4 reaches the inflection point first. Monitoring point 7 is located at the top of the slope. During the early stages of rainfall, some of the rain falls onto the top platform, causing the downward infiltration of this rainwater portion. Therefore, monitoring point 7 reaches the inflection point early in the initial stage. Subsequently, monitoring point 5 reaches the inflection point. This is attributed to the increasing duration of rainfall, and the resulting rainwater infiltration towards the slope’s depth near the top. The elevated moisture content at monitoring point 7 contributes to rainwater infiltration in the direction of the slope, leading monitoring point 5 to reach the inflection point after monitoring points 4 and 7. Similarly, monitoring point 3 exhibits an inflection point shortly after monitoring point 5, rapidly reaching a moisture content of 100%. Monitoring point 10 exhibits the inflection point last. Zhao concluded in a study that the effective infiltration of the slope depends on factors such as rainfall intensity, time, slope characteristics, and surface runoff. Since red mud has poor permeability, a significant portion of rainwater flows away through runoff, affecting the surface infiltration on slopes [34].

The moisture content variations at monitoring points with greater depths are inconsistent with those at monitoring points with shallower depths, as illustrated in Figure 8b. The volumetric moisture content at monitoring point 6 reaches 100% first, mainly because monitoring point 7 attained a high moisture content beforehand. During continuous rainfall, the increase in moisture content at monitoring point 6 is mainly attributed to the downward infiltration of rainwater from the upper part. Subsequently, after 1145 s, monitoring points 8 and 1 also reach turning points, which is due to the continuous rainfall, which leads to the inward infiltration of rainwater in the direction of the slope within the body of the red mud dam, gradually saturating monitoring points 8 and 1. As the volumetric moisture content at monitoring points 6, 8, and 1 gradually reached 100%, the volumetric moisture content at monitoring point 2 continued to increase steadily after the turning point. In contrast, the volumetric moisture content at monitoring point 9 exhibits minimal changes, indicating a lack of rainwater infiltration at that depth. The infiltration at the upper layer monitoring point will impact the water content at the lower-level monitoring point associated with the same location.

3.1.2. Distribution of Infiltration Peaks in the Body of the Dam

The volumetric moisture content trend analysis for various monitoring points in Figure 6 and Figure 8 reveals the infiltration line trends for the operating conditions of cases L1 and L2, as shown in Figure 9a and Figure 9b, respectively. From the infiltration line trend for the operating conditions of case L1 in Figure 9a, it is evident that volumetric moisture content at monitoring point 3 in the stepped red mud dam reaches the inflection point first, followed sequentially by monitoring points 5, 1, 7, 4, and 9. The volumetric moisture content at monitoring points 6, 8, and 2 experiences minimal changes before the end of the rainfall. Analysis of the infiltration line trend for the operating conditions of case L1 indicates that at the stepped platform, the infiltration line gradually flattens and tends towards a horizontal orientation. This phenomenon is attributed to the presence of the stepped platforms in the red mud dam, which act as buffer platforms, altering the water seepage direction within the dam.

The change trend of the infiltration line under the conditions of case L2 in Figure 9b indicates that in the red mud dam body without terraces, the volumetric moisture content at monitoring points 4 and 7 reaches the inflection point first, followed by monitoring points 5, 3, 10, 6, 8, 1, and 2 successively reaching their inflection points. It can be observed from the change trend of the infiltration line under the conditions of case L2 that in the red mud dam body without terraced platforms, the infiltration line direction consistently slopes downward in the direction of the slope, unlike in terraced red mud dam bodies, where the infiltration line gradually levels off towards a horizontal plane.

By comparing the monitoring points of cases L1 and L2 under the same rainfall conditions, it was determined that the location of the earliest appearance of the moisture content inflection point differed between the stepped red mud dam and the non-stepped red mud dam. The stepped red mud dam shows the earliest inflection point of volumetric moisture content at the stepped platform. Contrastingly, the non-stepped red mud dam exhibits a first inflection point of volumetric moisture content slightly lower and in the middle of the slope’s surface. Analysis of the volumetric moisture content variations at the monitoring points of cases L1 and L2 indicates that a stepped red mud dam serves as a buffer platform when it comes to water seepage. This is because the presence of the platform changes the direction of water seepage in a red mud dam. Consequently, the potential energy of the water in a red mud dam decreases as the height of seepage diminishes. In contrast, red mud dams without stepped platforms always experience internal seepage in the direction of the slope, leading to heightened risks of landslides, especially considering the fact that red mud dam heights commonly exceed 100 m. Therefore, it is crucial to incorporate stepped designs in red mud dam structures. Furthermore, observations from the monitoring points under case L2 conditions show that as rainfall persists, the rate of moisture increase near the top of the slope surpasses the rate in the middle-to-lower section, creating conditions unfavorable for the safe operation of the red mud dam.

3.1.3. Distribution Pattern of Pore Water Pressure

The locations of the pore water pressure sensors for case L1 are shown in Figure 10a. From the distribution of volumetric moisture content under the working conditions of case L1, as shown in Figure 6, it can be inferred that the volumetric moisture content at monitoring point 3 initially reaches the inflection point at 720 s. The trends of pore water pressure variation under the working conditions of L1, as illustrated in Figure 11a, indicate that the pore water pressure at monitoring point A6123 remains relatively stable between the points of 0 and 2509 s, showing a growth trend only between 2509 s and 3285 s, during which the pore water pressure at 3285 s is 0.035 kPa. Subsequently, it remains constant until 6706 s, when it rapidly increases, reaching 0.23 kPa at the end of the rainfall. Similarly, monitoring point A6126 also reaches the inflection point relatively early, with a pore water pressure trend resembling that of monitoring point A6123. Considering the locations and moisture content distributions of monitoring points A6124, A6125, and A6127, it can be concluded that the red mud structure at these points has not been damaged yet. Therefore, the pore water pressures at these monitoring points are all slowly increasing due to rainwater infiltration.

The locations of the pore water pressure sensors in case L2 are shown in Figure 10b. Analysis of the pore water pressure variation trends for case L2, as shown in Figure 11b, reveals a rapid increase in pore water pressure at both monitoring point A6125 and monitoring point A6126. Monitoring point A6125, located near the slope’s base, easily accumulates rainwater, leading to a rapid rise in pore water pressure due to the continuous rainwater accumulation at this location. In contrast, monitoring point A6126, located lower on the slope, reaches the inflection point first, as indicated by the volumetric moisture content distribution under the circumstances of case L2, as shown in Figure 11. With prolonged rainfall, rainwater accumulates at monitoring point A6126, causing structural damage to the red mud dam and a rapid increase in pore water pressure. In contrast, the red mud dam’s structure at monitoring points A6123, A6124, and A6127 remains intact, resulting in a gradual increase in pore water pressure at these monitoring points due to rainwater infiltration. Pore water pressure determines the state of the slope’s surface after rainfall. The more the pore water pressure increases, the more the shear strength decreases; the seepage force increases while matrix suction decreases; and the area with the highest pore water pressure is more prone to surface landslides [22,23,24].

3.1.4. The Variation Patterns of Vertical and Horizontal Soil Pressure over Time

The Variation in Vertical Soil Pressure with Respect to Time in Case L1

The locations of the soil pressure boxes for case L1 are shown in Figure 12. The vertical soil pressure distribution of the first row for case L1 is illustrated in Figure 13a. It reveals that before the rainfall, the maximum vertical soil pressure is observed in the middle section of the secondary slope, followed by the vertical soil pressure at the bottom of the primary slope, with the minimum vertical soil pressure found at the top of the primary slope. Subsequent to rainfall, monitoring points A6118, A6119, and A6111 show initial increases in vertical soil pressure, followed by decreases. By examining the moisture content distribution for case L1 in Figure 6, it becomes apparent that this phenomenon can be attributed to the continuous rainwater infiltration in the overburdened soil proximate to the soil pressure box post-rainfall. The increasing water content of the red mud and the corresponding increase in the weight of the overburdened soil lead to an escalation in soil pressure at the monitoring points. However, after a certain period, a decrease in soil pressure is observed, one which can be attributed to the buoyancy effect of the pore water in the red mud particles acting on the soil pressure box.

The vertical soil pressure distribution of the first column for case L1 is shown in Figure 13b; it demonstrates a positive correlation with depth before rainfall, that is, vertical soil pressure increases as the depth increases. Similarly, Figure 13c illustrates that the instances of vertical soil pressure under condition L1 for the third row are sequentially represented by monitoring points A6112, A6113, and A6118, respectively.

After the rainfall, monitoring point A6111 exhibits an initial increase in vertical soil pressure, followed by a decrease. In contrast, the vertical soil pressures at monitoring points A6120, A6112, A6118, and A6113 demonstrate generally upward trends. Since the monitoring points A6120 and A6112 did not experience rainfall infiltration, they were not subjected to the buoyancy generated by the pore water.

The Variation in Horizontal Soil Pressure with Respect to Time in Case L1

Analysis of the horizontal soil pressure distribution of the first row for case L1, as described in Figure 14a, reveals that before the rainfall, the maximum horizontal soil pressure is observed in the middle part of the secondary slope, followed by values at the bottom of the primary slope, while the minimum horizontal soil pressure is found at the top of the primary slope. This is attributed to the combined effects of horizontal gravitational stress and the slope’s sliding force relative to monitoring points A6116 and A6117.

Analysis of the horizontal soil pressure distribution of the first column for case L1, as described in Figure 14b, reveals that before the rainfall, the magnitude of horizontal soil pressure in the first column is positively correlated with the depth of the monitoring points; that is, the deeper the depth is, the greater the horizontal soil pressure is. After the rainfall, there are no significant changes in the horizontal soil pressure at monitoring points A6114, A6121, and A6115.

The horizontal soil pressure distribution of the third row for case L1 is shown in Figure 14c. It can be seen that, before the rainfall, monitoring point A6122 exhibited the maximum horizontal soil pressure, followed by A6117, while A6114 recorded the minimum value. These patterns are mainly influenced by the sliding forces exerted upon A6117 and A6122. After the rainfall, the horizontal soil pressure readings from A6114 and A6122 remained relatively stable overall. Conversely, a pattern of initial increase followed by subsequent decrease was observed at A6117. By correlating with the volumetric moisture content distribution under the working conditions of case L1 (Figure 6), it becomes evident that, at A6117, the pore water in the red mud exerts a buoyant force on the soil pressure box following a period of persistent rainfall.

The Variation in Vertical Soil Pressure with Respect to Time in Case L2

The locations of the soil pressure boxes for case L2 are illustrated in Figure 15. The distribution of vertical soil pressure under the operating conditions for case L2, as illustrated in Figure 16, indicates that before the rainfall, the highest vertical soil pressure is recorded at monitoring point A6114, which is located in the middle of the slope. Subsequently, the vertical soil pressure at monitoring point A6120, located at the base of the slope, is lower, while the minimum value for vertical soil pressure is detected at monitoring point A6118, which is located at the crest of the slope.

After the rainfall, there are minimal changes observed overall at monitoring points A6114 and A6120. Analyzing the volumetric moisture content distribution in Figure 8 for case L2, it can be noted that monitoring point A6114 corresponds to monitoring point 3, while monitoring point A6120 corresponds to monitoring point 10. Monitoring points 3 and 10 reach their turning points at 3478 s and 4125 s, respectively, indicating delayed turning points compared to other monitoring points. Due to the lower infiltration of rainfall and the majority of the rainwater being runoff, monitoring points 3 and 10 exhibit lower initial infiltration volumes, resulting in minimal changes in vertical soil pressure at A6114 and A6120. The vertical soil pressure at monitoring point A6121 shows an increasing trend before the end of the rainfall, with an initial vertical soil pressure of 7.1974 kPa and a final vertical soil pressure of 8.51316 kPa. Corresponding to monitoring point 4, the volumetric moisture content at monitoring point A6121 reaches 100% first, representing a higher level of rainfall infiltration, leading to increases in soil weight and vertical soil pressure at A6121. The vertical soil pressure at monitoring point A6112 remains relatively constant, between 6.6298 kPa and 6.9981 kPa, from 0 to 3341 s, and starts to decrease after 3341 s, reaching 5.5248 kPa at the end of the rainfall. Corresponding to monitoring point 5, which reaches the turning point at 3240 s, the vertical soil pressure at A6112 gradually decreases at 3341 s, due to the buoyancy effect of the pore water. The vertical soil pressure at monitoring point A6119 gradually increases from the initial soil pressure of 5.0849 kPa to 5.4466 kPa between 0 and 4593 s, then starts to decrease, reaching 4.7233 kPa at the end of the rainfall. Monitoring point A6119 corresponds to monitoring point 6, which approaches a turning point at 4593 s, leading to a gradual decrease in vertical soil pressure due to pore water buoyancy. Vertical soil pressure at monitoring point A6118 experiences a gradual increase, from 3.7513 kPa to 5.5026 kPa, between 0 and 2241 s, followed by a gradual decrease to a minimum of 4.4518 kPa after 2241 s. The vertical soil pressure at monitoring point A6118 exhibits a trend of initially increasing and then decreasing. Monitoring point A6118 corresponds to monitoring point 7, for which a turning point in volumetric moisture content appears earliest at 1416 s, indicating significant rainfall infiltration. Subsequently, the vertical soil pressure at monitoring point 7 notably increases due to the high rainfall infiltration and the increased weight of the red mud, resulting in a significant rise in vertical soil pressure. However, after 2241 s, the vertical soil pressure begins to decrease gradually due to the increased presence of pore water in the red mud, which exerts a buoyant force on the soil pressure box, leading to a reduction in vertical soil pressure.

The Variation in Horizontal Soil Pressure with Respect to Time in Case L2

The distribution of horizontal soil pressure in case L2 can be seen in Figure 17. Before the rainfall, the horizontal soil pressure is highest at monitoring point A6115, located in the middle of the slope in the downward position. The horizontal soil pressure measured at monitoring point A6113 at the foot of the slope is the second-largest, and the horizontal soil pressure measured at monitoring point A6118 at the top of the slope is the smallest.

After the rainfall, the horizontal soil pressure at monitoring point A6115 increases gradually, from 10.3539 kPa to 11.0619 kPa, between 0 and 2639 s, and then decreases to 10.7079 kPa after 2639 s. By examining the volumetric moisture content distribution in case L2, as shown in Figure 8, it is evident that monitoring point A6115 corresponds to monitoring point 4. The volumetric moisture content at monitoring Point 4 reaches the inflection point first, indicating higher levels of rainwater infiltration. Consequently, the horizontal soil pressure at monitoring point A6115 increases with the prolonged infiltration duration. However, after sustained rainfall, the horizontal soil pressure at monitoring point A6115 gradually decreases due to the influence of the pore water pressure. Monitoring points A6113 and A6122, corresponding to monitoring points 10 and 3, respectively, exhibit increasing trends overall in horizontal soil pressure during rainfall events. These points reach the inflection point later compared to other monitoring points, indicating slower increases in rainfall infiltration. Therefore, the horizontal soil pressures at monitoring points A6113 and A6122 showed increasing trends overall. Monitoring point A6111 corresponds to monitoring point 6, which shows a significant change at 4937 s and subsequently, after 5172 s, experiences a gradual decrease in horizontal soil pressure due to the influence of the pore water. Monitoring point A6116, located at the top of the slope and corresponding to monitoring point 7, experiences a notable increase in horizontal soil pressure, peaking at 5.108 kPa at 2016 s due to considerable rainfall infiltration. Subsequently, the horizontal soil pressure gradually decreased to a minimum of 3.9292 kPa after 2382 s, which is attributed to the decreasing pore water effect seen in the horizontal soil pressure at monitoring point A6116. After 3308 s, it slowly rises again; the horizontal soil pressure at monitoring point A6116 is recorded as 4.7151 kPa at the end of the rainfall. This increase is associated with the decline seen in volumetric moisture content at monitoring point 7 after 3125 s, which resulted in a reduction in volumetric moisture content and a weakened pore water pressure, leading to a gradual increment in horizontal soil pressure. The horizontal soil pressure at monitoring point A6117 gradually rises, from 3.8461 kPa to 4.2307 kPa, between 0 and 3303 s, then gradually decreases to 3.0769 kPa after 3987 s. This is because the soil pressure at monitoring point A6117 increases with the growth of rainfall infiltration time; however, after a period of continuous rainfall, the horizontal soil pressure at measurement point A6117 gradually decreases due to the action of the pore water.

In the comparison of cases L1 and L2, it is observed that before the rainfall, in a stepped red mud dam, the soil pressure is highest in the middle part of the secondary slope, followed by the bottom of the primary slope, and the lowest soil pressure is at the top of the primary slope, regardless of the direction of the soil pressure (vertical or horizontal). For a red mud dam without steps, in the vertical direction, the vertical soil pressure is highest at the middle position of the slope, followed by the foot of the slope, with the lowest value found at the top of the slope. In the horizontal direction, the horizontal soil pressure is highest at the lower-middle position of the slope, followed by the foot of the slope, and it is lowest at the top of the slope. The evolution of the transient saturation zone directly influences the variation in soil pressure when considering the impact of rainfall. In both vertical and horizontal directions, with increasing rainfall infiltration time, there is a corresponding increase in the weight of the red mud itself and a consequent rise in soil pressure. Subsequently, as rainfall continues over time, the soil pressure gradually decreases, due to the influence of the pore water. The evolutionary pattern of the transient saturated zone directly affects the variation in soil pressure when considering the effect of rainfall. The location of the transient saturated zone was obtained from the moisture sensor and pore pressure sensor measurements. The transient saturation zone of the stepped red mud dam body is located at the stepped platform, i.e., at the foot of the primary slope. After rainfall, both vertical and horizontal soil pressures in the transient saturated zone show fluctuations, increasing and then decreasing; vertical soil pressure increased by 9.7% and then decreased by 24.5%, and horizontal soil pressure increased by 9.5% and then decreased by 6.5%. The transient saturated zone of the body of the red mud dam with a slope ratio of 1:3 is located in the middle of the slope, and towards the bottom. After the rainfall, the horizontal soil pressure in the transient saturated zone increases by 6.7% and then decreases by 2.7%, and the vertical soil pressure increases by 18.2% at the end of the rainfall. The increase in soil pressure is due to the increase in the weight of the red mud itself, which increases the soil pressure reading, and when the rainfall infiltration is increasing, the soil pressure box reading decreases due to buoyancy. From the change in soil pressure in the transient saturated zone, it can be seen that the change in soil pressure in the transient saturated zone is greatly reduced at a slope ratio of 1:3, implying a lower likelihood of landslides. This is in agreement with the conclusion of Rahardjo’s research: the gentler the slope, the higher the safety factor [16].

3.2. Effects of Different Slope Ratios on Dam Stability under Rainfall Conditions

3.2.1. Distribution of Moisture Content in Red Mud Dams under Rainfall Conditions

The locations of the soil moisture sensors for case L3 are shown in Figure 18. The rainfall duration for case L3 is 74 min and 52 s. As can be seen from the distribution of volumetric moisture content for case L3, as depicted in Figure 19, the order of monitoring points relative to the volumetric moisture content and the arrival of the inflection point is as follows: monitoring points 9, 10, 3, 1, 4, 5, and 7. Case L3 has a steeper slope, with a slope ratio of 1:1.

In the comparison of cases L2 and L3, in which the initial water content of the red mud remains consistent but the gradient of the slopes vary, the paths of rainfall infiltration and the progression of transient saturation zones differ under identical rainfall conditions. In case L2, with a relatively gentle slope, the first area to reach saturation is located in the middle-lower section of the slope. Conversely, in case L3, with steeper slope, the first area to reach saturation is proximate to the summit of the slope. In steeper slopes, the transient saturation zone is first formed in the upper part of the slope, while in slopes with gentler gradients, the transient saturation zone is first formed in the middle-to-lower part of the slope. Slope gradient has a substantial influence on slope runoff. Under the same rainfall conditions, the steeper the slope is, the faster the runoff rate is, and the more serious the erosion of the slope’s surface is. Huang found by experimentation that the runoff rate is linearly related to the slope, a relation which conforms to the equation $v=hx+l$, where v stands for the average speed of runoff, and x stands for the slope. For steeper slopes, the faster the runoff rate is at the lower part of the slope, the smaller the actual infiltration amount is. The same conclusion was obtained in this test, i.e., the transient saturated zone was first formed in the upper part of the slope when the slope was steeper, while the transient saturated zone was first formed in the middle-to-lower part of the slope in slopes with a lower gradient [35].

3.2.2. Distribution of Infiltration Peaks in the Body of the Dam

The changing trend of volumetric moisture content in case L3 is analyzed to derive the infiltration line trend of case L3. Figure 20a,b illustrate the varying trends of infiltration lines for cases L2 and L3, respectively. From Figure 20b, it is evident that volumetric moisture content at monitoring point 9 reaches the turning point first, followed sequentially by monitoring points 10, 3, 1, 5, and 7. The infiltration patterns on the slope surfaces for cases L2 and L3 exhibit dissimilarities. Initial moisture content is identical for cases L2 and L3. L3 has a steeper slope. The region of transient saturation initially manifests in the upper-middle part of the slope in case L3. In contrast, the slope of case L2 is more gradual. The region of transient saturation initially emerges in the lower-middle part of the slope in case L2.

3.2.3. Distribution Patterns for Pore Water Pressure

The locations of the pore water pressure sensors for case L3 are shown in Figure 21. Case L3 exhibits a sharper slope compared to case L2. Analysis of the pore water pressure variation trend in case L3, as depicted in Figure 22, reveals that monitoring point A6126, which is situated near the slope’s peak, experiences a rapid increase in pore water pressure at 2118 s. Similarly, monitoring point A6123, which is located at the slope’s midpoint, also displays a swift rise in pore water pressure at 2687 s. The volumetric moisture content distribution for case L3, illustrated in Figure 19, establishes that monitoring point A6123 corresponds to monitoring point 10, which reaches the inflection point at 1414 s, while monitoring point A6126 corresponds to monitoring point 9, which reaches the inflection point at 1136 s, with monitoring point 9 reaching it first, followed by monitoring point 10. The red mud structures at monitoring points A6123 and A6126 are the first ones to incur damage, in contrast to the structures at monitoring points A6124, A6125, and A6127, which remain intact during the rainfall period.

Comparing cases L2 and L3, the initial volumetric moisture content of the red mud is identical, while the slope gradients differ. Under the same rainfall conditions, the slope of case L3 is steeper, and the sensor locations of case L3 are situated in the middle and upper regions of the slope, where the rapid increase of the pore water pressure curve occurs. Therefore, a slope with a steeper gradient may experience damage in the red mud structure on the slope surface in a shorter time, and the range of the red mud structure damage would be in the middle and upper portions of the slope. The location of the maximum pore water pressure is the same as the location of the transient saturated zone predicted by the moisture sensor, proving the correctness of the predicted trajectory of the infiltration peak.

3.2.4. The Variation Patterns of Vertical and Horizontal Soil Pressure over Time

The Variation in Vertical Soil Pressure with Respect to Time in Case L3

The locations of the soil pressure boxes in case L3 are shown in Figure 23. As can be seen from the vertical soil pressure distribution depicted in Figure 24 for case L3, before the rainfall the vertical soil pressure at monitoring point A9104 is the highest, and it is located in the middle of the slope at the downward position; the vertical soil pressure at monitoring point A9110 near the top of the slope is the second highest; and the vertical soil pressure at monitoring point A9112 at the top of the slope is the smallest.

After rainfall, the vertical soil pressure at monitoring point A9104 exhibits a gradual increase from 0 to 3978 s. Figure 19 illustrates the distribution of volumetric moisture content in case L3, indicating that the rise in vertical soil pressure at monitoring point A9104 is attributable to the augmentation of red mud weight caused by rainwater infiltration. Monitoring point A9107, which corresponds to monitoring point 10, experienced a rapid surge in pressure between 1021 s and 1839 s due to the increased weight of the mud itself caused by rainwater infiltration, followed by a notable decline starting at 1839 s due to the buoyancy effect of pore water on the soil pressure box. Monitoring point A9110 corresponds to monitoring point 9, which first reaches the inflection point of volumetric moisture content. Nevertheless, as monitoring point 9 resides near the slope’s apex, rainwater swiftly permeates downwards along the slope post-infiltration, resulting in a stable–decrease–increase–stable pattern in the vertical soil pressure at monitoring point A9110. At the two corresponding points of monitoring point A9106 and monitoring point 7, a minimal volumetric moisture content alteration is observed before a rising-then-falling curve in vertical soil pressure after 3484 s, attributed to rainwater accumulation at the slope’s base. Monitoring point A9109 corresponds to monitoring point 1, and shows a slight increase in vertical soil pressure from 0 to 1829 s, escalating from the initial 4.4207 kPa to 4.7488 kPa, then rapidly declining between 1038 s and 2452 s to 3.2961 kPa, which is followed by a gradual rise after 2452 s, culminating in a pressure of 4.0927 kPa by the end of the rainfall. Monitoring point A9112 corresponds to monitoring point 3, and is where a slight elevation in vertical soil pressure occurred from 0 to 897 s due to increased rainfall infiltration at the slope’s peak. However, subsequent to the continued rainfall, the pressure notably decreases due to the buoyancy effect of the pore water. Given the higher elevation at the top of the slope, some rainwater will cascade downhill post-infiltration. As the buoyancy effect of the red mud pore water on the soil pressure box diminishes, the vertical soil pressure experiences a slight increase once more.

The Variation in Horizontal Soil Pressure with Respect to Time in Case L3

As can be seen from the distribution of horizontal soil pressure depicted in Figure 25 for case L3, before the rainfall, the horizontal soil pressure measured at monitoring point A9105 is the highest; this value was recorded in the middle of the slope, while the values recorded for horizontal soil pressure at the top of the slope and the foot of the slope were the smallest.

After the rainfall, the horizontal soil pressure at monitoring point A9105 remains relatively stable, between 6.3688 kPa and 6.4101 kPa, from 0 to 2065 s, gradually decreases after 2065 s, and decreases to 5.2108 kPa by the end of rainfall. Monitoring point A9111 basically remains between 4.146 kPa and 4.2941 kPa from 0 to 2688 s, and then gradually decreases after 2688 s, decreasing to 3.208 kPa by the end of the rainfall. Monitoring points A9105 and A9111 are in the middle of the slope and reach their water content inflection points earlier during the rainfall, and the decrease of horizontal soil pressure is caused by the pore water effects. The horizontal soil pressure levels at monitoring points A9108 and A9102 basically do not change during rainfall events. Given the distribution of volumetric moisture content for case L3 depicted in Figure 19, it can be seen that the volumetric moisture content levels at monitoring points A9108 and A9102, which correspond to monitoring points 5 and 7, are almost unchanged until the end of the rainfall. Both monitoring point A9103 and monitoring point A9113, considered overall, show decreasing trends until the end of the rainfall, and by the end of rainfall the soil pressure at monitoring point A9103 decreases from 3.0023 kPa to 2.5404 kPa, while the soil pressure at monitoring point A9113 decreases from 3.1041 kPa to 2.3281 kPa. Monitoring points A9103 and A9113 are located at the top of the slope and near the top of the slope, respectively, where rainfall infiltration is high, and the reduction in horizontal soil pressure observed is due to the action of pore water.

The analysis of the horizontal soil pressure distribution in cases L2 and L3 indicates that, although the initial volumetric moisture content of the red mud in both cases is the same, the slope gradient ratios differ. Before the rainfall, there are discrepancies in the vertical and horizontal soil pressure distributions on the red mud slopes in these two cases. It is evident from the soil pressure distribution in cases L2 and L3 that, as the slope incline increases, the location of the maximum vertical soil pressure shifts nearer to the slope’s base, while the maximum horizontal soil pressure tends to be closer to the upper segment of the slope. After rainfall, the soil pressure distribution is notably influenced by the infiltration trajectory of the rainwater. When the slope gradient ratio is 1:1, the transient saturated zone is located near the top of the slope and the formation time of the transient saturated zone is shorter. In contrast to the 1:3 zone, where the soil pressure increases and then decreases, the 1:1 zone shows a decrease in soil pressure during the rainfall period, with the vertical soil pressure decreasing by 14.8% and the horizontal soil pressure decreasing by 29%. The decrease in soil pressure is due to the effect of the increased pore water on the buoyancy of the soil pressure box.

3.3. Discussion

The formation of the transient saturated zone is mainly due to the combined effects of soil properties, rainfall intensity, time, slope type, slope gradient, and slope runoff, as shown in Figure 26. The poor permeability of red mud makes formation of a transient saturated zone on the slope easier, and the formation of a transient saturated zone is the main cause of landslides.

Under the same rainfall conditions, where the slope ratio is 1:3, the transient saturation zone of a stepped red mud dam appears at the stepped platform, while the transient saturation zone of a red mud dam without steps appears in the middle of the slope, towards the lower end. The stepped red mud dam reduces the runoff path of the rainwater, and the runoff velocity then slows down, the height between platforms decreases, the hydraulic gradient decreases, and the erosion of the slope surface by the runoff from the slope surface is lessened. Under the same rainfall conditions, when the slope ratio is different, the transient saturated zone for a slope ratio of 1:1 appears on the slope surface near the top of the slope, and the higher the transient saturated zone, the greater the damage caused by landslides. From the sequence of rainfall infiltration times, it can be seen that the steeper the slope, the shorter the formation time of the transient saturated zone; accordingly, the steeper the slope of the red mud dam body is, the greater the risk of landslide damage. The locations of the transient saturation zones are shown in Figure 27.

From the change trend of pore water pressure determined in the test, we can verify the location of the transient saturated zone predicted by the moisture sensor, since the location of maximum pore pressure is the location of transient saturated zone formation. The larger the pore pressure is, the more the shear strength of red mud decreases, and the more prone it is to landslides on the slope. The findings of the earth-pressure test in the experiment reflect the pore water content of the red mud. At the beginning of the rainfall period, when the rainfall caused the water content of the red mud on the surface to rise, the earth-pressure box readings increased due to the increased pressure of the overlying soil, and then decreased due to buoyancy when the rainwater infiltration gradually increased. The change in the earth-pressure sensor values is related to the amount of rainwater infiltration in the red mud, and the values of the earth-pressure sensor are consistent with the change in the moisture content sensor.

Through the analysis of the law of the evolution of transient saturated zones, as applied to a red mud dam body under strong rainfall conditions, it can be seen that when designing the red mud dam body, the slope type and slope design should be fully considered. As for red mud dams in operation, artificial intervention should be carried out in the possible transient saturated area to discharge the water in time to avoid the formation of a transient saturated area.

A red mud tailings pond has a large area, and under the influence of periodic rainfall events over many years, the overall moisture content of the red mud tailings pond will continue to increase, while the water level in the tailings pond keeps increasing due to long-term infiltration [36]. The infiltration line of a red mud tailings pond is relatively high compared to that of a coarse-grained tailings pond, which can easily cause high water-head pressures on the tailings dam’s slopes and lead to infiltration failure. In subsequent research, research on seepage failure in the red mud tailings pond will be conducted, which will continue the focus on the sustainable development of tailings ponds, and contribute to the safety protection and environmental protection of tailings ponds.

4. Conclusions

This paper systematically explores the evolutionary laws of the transient saturated zones of red mud dams with different slope types and slope gradient ratios, reveals the damage mechanisms of red mud dams encountering continuous heavy rainfall under different working conditions, and provides references for preventing landslides and guaranteeing the quality of dam construction during the period of red mud damming. The following conclusions are drawn:

(1) The saturation turning point and the rapid achievement of 100% soil volumetric moisture content vary under similar rainfall conditions between stepped and non-stepped red mud dams. In a stepped red mud dam, the soil volumetric moisture content turning point occurs earliest at the step platform, whereas in a non-stepped red mud dam, it is initially observed in the middle, and slightly lower on the slope, at the same slope ratio. The step platform can act as a buffer platform, altering the water infiltration path in the red mud dam and reducing the water’s potential energy within the dam’s body due to the decreased infiltration height. In reality, the height of red mud dam structures often exceeds 100 m, resulting in an increased risk of landslides. Therefore, creating steps within the red mud dam’s body is crucial.

(2) Under identical rainfall conditions, the bodies of red mud dams with equivalent moisture content levels but varying slope ratios exhibit distinct trajectories of rainfall infiltration and transient saturated area evolution. The steeper the slope is, the higher the position of the transient saturated area first occurring is, the faster the infiltration peak infiltration speed is, and the easier it is for the slope’s surface to become saturated. When encountering heavy rainfall, the steeper the slope of the body of the red mud dam, the higher the landslide occurrence position, and the greater the risk. Therefore, selecting a suitable slope ratio for the red mud dam body is necessary.

(3) Different slope types and rates correspond to different internal force distributions. The evolutionary pattern of the transient saturated zone formed after rainfall directly affects the change in soil pressure. The transient saturated zone of the stepped red mud dam body is located at the stepped platform, and the vertical earth pressure at the stepped platform increases by 9.7% and then decreases by 24.5%, while the horizontal earth pressure increases by 9.5% and then decreases by 6.5% after rainfall occurs. Horizontal earth pressure in the transient saturated zone of a red mud dam body with a slope ratio of 1:3 first increased by 6.7% and then decreased by 2.7%, and the vertical earth pressure had increased by 18.2% by the end of the rainfall. The soil pressure in the transient saturated zone of a red mud dam body with a slope ratio of 1:1 showed a decreasing change throughout the rainfall period, with the vertical soil pressure decreasing by 14.8% and the horizontal soil pressure decreasing by 29%. The decrease in earth pressure is due to the effect of the increased pore water pressure on the buoyancy of the earth-pressure box. Obviously, the soil pressure in a transient saturated zone with a slope ratio of 1:1 is more affected by rainfall.