Research on the deformation mechanisms of red-bed deposits with straight line deposit-bedrock interfaces in Xiangjiaba Reservoir area

Abstract

In the 12 years since the impoundment of Xiangjiaba Reservoir, 28 bank slopes have deformed and necessitated remediation due to reservoir impoundment. This study conducted a statistical analysis of the distribution and types of the 28 deformed bank slopes, revealing that significant proportion of the deformed slopes are red-bed deposits with straight-line deposit-bedrock interfaces. To reveal the deformation mechanism of bank slopes with straight line deposit-bedrock interfaces, two red-bed deposits (D4 deposit and D8 deposit) with significantly different deformation characteristics were selected as study objects. Engineering geological investigations found differences in both the dip angle of the deposit-bedrock interface and the deposit materials between the two deposits. Surface displacement monitoring showed that the displacement curve of D4 deposit exhibited a step-like pattern, while that of D8 deposit exhibited a continuous growth pattern. The numerical simulation results showed that when a high water level is maintained, gentler deposit-bedrock interface dip angles are more likely to result in a safety factor falling below the initial value; when the water level drops, lower permeability coefficients lead to greater decreases in the safety factor. The comprehensive analysis suggested that the core difference in deformation between the two deposits lies in the dip angle of the deposit-bedrock interface. A steeper deposit-bedrock interface leads to deformation primarily controlled by seepage forces during the water level drop, while a gentler interface induces continuous deformation due to the synergistic action of buoyant force and seepage force. The research findings can provide theoretical and technical support for reservoir bank disaster prevention and control, as well as engineering remediation in the Jinsha River basin and similar red-bed reservoir areas.

Introduction

Reservoir impoundment often causes bank slope deformation, and even catastrophic landslides. For instance, in 1961, the instability of the Tangyan landslide at Zhexi Reservoir in China triggered a 21 m high wave, resulting in multiple deaths¹. In 1963, a landslide occurred on the left bank of Vajont Dam in Italy, generating a 175 m wave that resulted in more than 2500 fatalities^2–4. In 2003, after the water level of Three Gorges Reservoir reached 135 m, a massive landslide occurred in Qianjiangping Village, Zigui County, causing 14 deaths and 10 people disappearances^5,6,]7.

To efficiently implement slope management measures, researchers have extensively investigated reservoir bank deformation through experimental^8–10 and numerical approaches^11–13. Research indicates that fluctuations in reservoir water levels alter the seepage field in bank slopes, which is a primary factor contributing to slope deformation. When the water level rises, the buoyant force exerted on the slope increases, reducing its stability, while the seepage force is directed inward, enhancing stability. When the water level drops, the buoyant force decreases, but the seepage force is directed outward. The combined effects of buoyant force and seepage force determine the stability of the bank slope^14,15.

Under water level fluctuations, different types of bank slopes experience significant variations in buoyant force and seepage force magnitudes. To investigate the relationship between bank slope types and deformation, researchers performed statistical analyses of bank slopes in the Three Gorges^16–18, Baihetan¹⁹, Xiluodu reservoir areas^20,21. From a lithological perspective, Li et al^22,23 statistically analyzed the relationship between lithology and landslides in the Three Gorges Reservoir area, finding that red-bed landslides accounted for 66.5% of the total. Red-bed lithology is predominantly composed of interbedded sandstone and mudstone, representing a typical type of easily deformable strata²⁴. Based on the morphology of deposit-bedrock interface, Tang et al.^25–27 found that bank slopes with straight line deposit-bedrock interfaces are prone to form seepage-driven landslides, while those with chair-shaped interfaces are more likely to form buoyancy-driven landslides.

However, significant differences in deformation mechanisms exist even among bank slopes with the same type of deposit-bedrock interface^28–30. Some researchers have divided bank slopes with a chair-shaped deposit-bedrock interface into gentle and steep slope sections (Fig. 1), and analyzed the effects of immersion position, the length ratio of gentle to steep sections (L2/L1), and the permeability coefficient on slope deformation. For instance, Luo et al²⁸ found through their research on water immersion position that when the immersion position is located in the steep section, seepage-driven landslides are prone to occur; when located in the gentle section, buoyancy-driven landslides are more likely. Yang et al²⁹ investigated the dip angle β of the steep section and found that with a larger dip angle, bank slopes are more prone to deformation during reservoir water level rise. Although previous studies have systematically investigated the deformation mechanisms of chair-shaped deposit-bedrock interface bank slopes, a comprehensive understanding of the deformation mechanisms for straight-line deposit-bedrock interface slopes remains limited. Unlike chair-shaped structures, the dip angle of straight line deposit-bedrock interfaces is largely consistent from the slope crest to the toe, lacking distinct anti-slide sections. Consequently, variations in the deposit-bedrock interface dip angle will lead to differences in bank slope deformation mechanisms. Furthermore, other factors such as rock and soil mass permeability and water level fluctuation rates may also influence the deformation mechanisms of straight line deposit-bedrock interface bank slopes, necessitating in-depth investigation.

Fig. 1.

The slope with a chair-shaped deposit-bedrock interface²⁹.

Over the 12 years of Xiangjiaba Reservoir’s operation, many bank slopes experienced deformation^31,32, requiring control measures to ensure the safety of nearby residents and their property. Statistical analysis of the treated slopes revealed that red-bed deposits with straight line deposit-bedrock interfaces were the primary subject affected by Xiangjiaba Reservoir impoundment. To analyze the effects of deposit-bedrock interface angle and permeability coefficient on the deformation of straight line deposit-bedrock interface slopes, this study selected two typical straight line deposit-bedrock interface red-bed deposits (D4 and D8 deposits) as research subjects. The deformation characteristics of the two deposits were analyzed through engineering geological surveys and surface displacement monitoring. Numerical simulations were conducted to analyze the effects of varying deposit-bedrock interface angles and permeability coefficients on the safety factor of the deposits. Based on the combined monitoring and numerical simulation results, the deformation mechanisms of the D4 and D8 deposits were investigated.

Study area

Engineering geological conditions

Xiangjiaba Reservoir is located at the border of Sichuan and Yunnan provinces, with its upstream connected to the Xiluodu hydropower station (Fig. 2). The lithological are shown in Table 1. The mainstream shoreline of Xiangjiaba Reservoir is 156.6 km long, with a length of 44 km for the first-level tributaries. This area has a subtropical humid climate, with the rainy season occurring from May to October, and an average annual precipitation of 908.1 mm. Xiangjiaba Reservoir started impounding water on October 10, 2012, with water levels reaching 370 m in October 2012 and 380 m in September 2013. Subsequently, the reservoir operated between 370 and 380 m following the regulation schedule, with a drop before the flood season and impoundment at its end (Fig. 3).

Fig. 2.

Location of Xiangjiaba Reservoir area. (a) Map of China; (b) Xiangjiaba Reservoir area.

Table 1.

Lithology of the strata in Xiangjiaba Reservoir area.

Strata	Lithology
Cretaceous (K)	Quartz sandstone, mudstone, siltstone
Jurassic (J)	Purple mudstone, fine quartz sandstone and mudstone interbed
Triassic (T)	Quartz sandstone, siltstone and mudstone interbedding, limestone with shale
Permian (P)	Limestone with shale, basalt
Silurian (S)	Shale with limestone
Ordovician (O)	Sandstone, siltstone, limestone and grayish green shale
Cambrian (∈)	Silty dolomite, siltstone, mudstone, limestone
Sinian (Z)	Clastic dolomite and limestone; Purple red sandstone, shale with limestone

Fig. 3.

Variation curve of water level and daily rainfall in Xiangjiaba Reservoir area.

Strata from the Sinian to the Cretaceous are exposed in the reservoir area, except for the Carboniferous strata. The upstream section of Xinshi Town is primarily composed of Triassic and older strata, with lithologies dominated by limestone, basalt, and sandstone. The downstream section beyond Xinshi Town is primarily composed of Jurassic strata (Fig. 4), with red-bed mudstone and sandstone as the dominant lithologies. The reservoir area contains structures such as the Yaziba Fault and Loudong Fault, which are spreading predominantly in SN.

Fig. 4.

Location of deformed slopes in Xiangjiaba Reservoir area.

Distribution of deformed slopes

The statistical analysis was performed on slopes deformed due to reservoir impoundment and requiring control measures, based on field surveys and defined data of deformation slopes over the last 12 years. The analysis indicates that 28 slopes experienced deformation (Fig. 4), with 4 located in limestone and sandstone regions and 24 in red-bed regions. Among the 24 slopes in red-bed regions, only 4 are rock masses, while the remaining 20 are deposits. This indicates that the primary subjects affected by reservoir impoundment in Xiangjiaba Reservoir area are red-bed deposits.

Structural types of red-bed deposits

To investigate the deformation mechanisms of red-bed deposit slopes, the structures of 20 red-bed deposits were classified. Previous studies on slope classification in the Three Gorges Reservoir area found a strong correlation between slope deformation characteristics and deposit-bedrock interface morphology, categorizing the latter into four types: straight-line, fold-line, arc shape, and chair shape (Fig. 5). Except for the straight line deposit-bedrock interface, the other types can be divided into gentle and steep sections^29,30. Therefore, based on an analysis of drilling data, this study statistically analyzed the length ratio (L2/L1), total length, and dip angles of the gentle and steep sections of deposit-bedrock interfaces.

Fig. 5.

Four types of the deposit-bedrock interface. (a) Straight line interface; (b) Fold line interface; (c) Are shaped interface; (d) Chair shaped interface.

In the statistical analysis, the division standard of 5° is employed²⁹. When the dip angles of the front and rear deposit-bedrock interfaces exceeded 5°, the distinction of gentle and steep sections was made; otherwise, no distinction was made. All straight line deposit-bedrock interfaces were treated as gentle slope sections (Fig. 5a).

The locations and numbers of the 20 deposits are shown in Fig. 6, while the statistical results are presented in Table 2 and Fig. 6. The lengths of the deposit-bedrock interfaces are mainly concentrated in the range of 200–600 m (Fig. 7a). The length ratio of the gentle to steep slope sections of deposit-bedrock interfaces (L2/L1) is mainly within the 0:10–1:9 range, accounting for 50% of the total (Fig. 7b), making it the predominant type of deformed deposits in Xiangjiaba Reservoir area. The deposit-bedrock interfaces of deposits with a length ratio in the 0:10–1:9 range are relatively straight and are often defined as the straight-line type²⁵. Combined with Table 1, the dip angle of the gentle section of the deposit-bedrock interface for deposits with the ratio (L2/L1) in the 0:10–1:9 range is primarily between 10° and 25° (Fig. 7d). Therefore, this study focuses on deposits with deposit-bedrock interface dip angles ranging from 10° to 25° to investigate their deformation mechanisms.

Fig. 6.

Location of deformed deposits in the red-bed region of Xiangjiaba Reservoir area.

Table 2.

Statistical data of the deposits.

Deposit number	Volume (× 104m3)	Total length(m)	Length ratio of gentle to steep sections	Dip angles of steep section (°)	Dip angles of gentle section (°)
D1	56	180	0:10		23
D2	180	260	0:10		17
D3	390	300	4:6	38	3
D4	97	300	0:10		22
D5	140	530	0:10		15
D6	28	130	1:9	58	7
D7	170	245	5:5	34	0
D8	180	490	0:10		15
D9	119	260	4:6	33	7
D10	5250	1070	3:7	20	8
D11	49	230	1:9	44	13
D12	36	170	2:8	48	14
D13	126	520	5:5	19	13
D14	210	515	1:9	36	13
D15	96	350	4:6	30	3
D16	138	580	0:10		11
D17	415	420	1:9	50	7
D18	60	245	5:5	31	6
D19	62	315	7:3	24	8
D20	72	200	6:4	20	0

Fig. 7.

Statistical data of the deposit-bedrock interface. (a) Total length; (b) Length ratios; (c) Dip angle of steep section; (d) Dip angle of gentle section.

Case study

Based on the above statistical results, this study selects two deposits with straight line deposit-bedrock interfaces (D4 and D8) as typical cases. The dip angles of the gentle sections of deposit-bedrock interfaces in the D4 and D8 deposits are 22° and 15°, respectively. The surface displacement curves of the two deposits show distinct differences during water level fluctuations. The curve of D4 deposit exhibits a step-like pattern, whereas that of D8 deposit shows a continuous growth.

D4 deposit

Engineering geological investigation

D4 deposit is located on the left bank of Fuyan river, a primary tributary of the Jinsha River (Fig. 6). Its front edge lies below the water level at an elevation of 340 m, while the rear edge is at 470 m. The deposit is bounded by a gully on the left and a ridge on the right (Fig. 8). The underlying bedrock of the deposit consists of interbedded sandstone and mudstone of varying thickness from the Middle Jurassic Shaximiao Formation (J_2s). The bedrock attitude in the gully is 310°∠23°, and the deposit forms a dip slope.

Fig. 8.

D4 deposit. (a) Satellite image; (b) Topographic map; (c)–(f) Crack.

In September 2014, when the water level dropped, timber-framed buildings on the deposit experienced deformation and cracking (Figs. 8c–e). Over the following years, the deformation of the buildings intensified, and cracks appeared on the northern roadway surface (Fig. 8f). To ensure the safety of residents, property, and traffic, anti-slide piles were installed outside the roadway in January 2020.

To determine the structure of the deposit, two profiles were set along the escarpment, and eight boreholes were drilled (Fig. 8b). Borehole results, depicted in the profile diagram (Fig. 9), reveal a relatively flat deposit-bedrock interface. The dip angle of the interface along profile A’-A is 22°, whereas it is more gentle along profile B’-B, at 17°. The D4 deposit has a thickness of 15 m on average and a total volume of 97 × 10⁴ m³. The material composition of the deposit consists of clay containing gravel, with a gravel content of 30% (Fig. 9c).

Fig. 9.

Profile and deposited materials of the D4 deposit. (a)–(b) Profile; (c) Photograph of the borehole.

Deformation monitoring

In July 2016, two surface displacement monitoring points were set up at locations with significant deformation in the front and middle sections, numbered GPS1 and GPS2 (Fig. 8b). In September 2017, an additional surface displacement monitoring point, numbered GPS3, was added at the rear edge. The surface displacement monitoring data were periodically collected manually using a total station. The horizontal displacement vectors as of January 2020 are shown by the blue arrows in Fig. 8b, indicating that the horizontal displacement direction of the landslide is perpendicular to the contour lines and points towards the free face. The maximum horizontal displacement of 494 mm was observed at point GPS1 in the front. Point GPS2 in the middle recorded the second largest displacement at 434 mm, while point GPS3 at the rear showed the minimum at 126 mm. This characteristic of decreasing displacement magnitude from the front to the rear, combined with the distribution of surface cracks (longer and wider cracks at the front), indicates that D4 deposit exhibits retrogressive deformation characteristics.

Surface displacement monitoring data, as shown in Fig. 10, reveal that the deformation of the D4 deposit primarily occurs during water level drops. For instance, from June 22 to September 22, 2016, as the water level dropped from 378 to 371 m at a rate of 0.6 m/d and then remained stable, the displacements at points GPS1 and GPS2 sharply increased to 213 mm and 190 mm, respectively, with corresponding displacement rates of 2.32 mm/d and 2.06 mm/d. During periods of reservoir water level rise and high water level maintenance, the increment in surface displacement was small. For example, from September 22, 2016, to April 19, 2017, displacements at points GPS1 and GPS2 increased by 27 mm and 18 mm, respectively, with corresponding displacement rates of 0.105 mm/d and 0.07 mm/d, significantly lower than those during the water level drop period. The influence of rainfall on deformation was limited. For instance, from May 8 to June 21, 2018 (when reservoir water level changes were small but rainfall was significant), displacements at points GPS1 and GPS2 increased by 4 mm and 7.3 mm, respectively, with displacement rates of 0.09 mm/d and 0.16 mm/d.

Fig. 10.

Surface displacement monitoring data of D4 deposit.

In summary, the D4 deposit features a relatively steep deposit-bedrock interface with a dip angle of 22°, representing a retrogressive landslide, and its deformation is primarily controlled by reservoir water level drop.

D8 deposit

Engineering geological investigation

D8 deposit is located on the right bank of the Jinsha River (Fig. 6). Its front edge is below the water level at an elevation of 360 m, while the rear edge is at 491 m. The deposit is bordered by gullies on both sides (Fig. 11). The underlying bedrock of the deposit consists of mudstone with a small amount of sandstone from the Jurassic Ziliujing Formation (J_1-2z). The bedrock attitude at the rear is 310°∠12°, while the bedrock attitude at the front is 300°∠15°. The deposit forms a dip slope.

Fig. 11.

D8 deposit. (a) Satellite image; (b) Topographic map.

On September 12, 2013, when the water level first reached 380 m, cracks appeared on the surface of the deposit’s front edge (Fig. 11b). To ensure the safety of residents and their property, the houses on the deposit were relocated. Over the following years, with fluctuating water levels, small-scale bank collapses occurred, and the front-edge cracks expanded further.

To investigate the structure of the deposit, a profile line was established along the escarpment direction, with three boreholes arranged (Fig. 11b). The borehole results are shown in the profile diagram (Fig. 12). The A’-A profile shows that the deposit-bedrock interface is relatively flat, with a dip angle of 15°. The average thickness of the deposit is 15 m, with a total volume of 140 × 10⁴ m³. The deposit consists mainly of clay, with only 10% gravel content (Fig. 12b).

Fig. 12.

profile and deposited materials of the D8 deposit. (a) Profile; (b) Photograph of the borehole; (c) Deep monitoring data.

Deformation monitoring

In April 2014, one deep monitoring point and three surface displacement monitoring points were arranged along the A-A’ profile (Fig. 11b). Deep monitoring data were manually collected at irregular intervals using an inclinometer. The inclinometer was numbered INC1, and displacement towards the direction of the escarpment was considered positive. Surface displacement monitoring data were manually collected at irregular intervals using a total station. The surface displacement monitoring points were numbered GPS1 to GPS3. The horizontal displacement vectors as of October 2020 are shown by the blue arrows in Fig. 11b. The maximum horizontal displacement of 102 mm was observed at point GPS1 in the front, while the horizontal displacements at the middle and rear monitoring points were 47 mm and 36 mm, respectively, indicating that D8 deposit exhibits retrogressive deformation characteristics.

The curve of deep monitoring data shows a clear inflection point, as illustrated in Fig. 12c. The deposit has formed a sliding surface, with the middle of the sliding surface located at the deposit-bedrock interface. At the front edge, soils from ZK1 at the depth of 20 m show a soft-plastic state, with significantly higher clay content than surrounding soils (Fig. 12b). Based on the morphology of the deposit-bedrock interface and borehole results, it is inferred that the front part of the sliding surface is located within the deposit (Fig. 12a).

Surface displacement monitoring data, as shown in Fig. 13, reveal a continuous growth trend in displacement at all monitoring points. Taking point GPS1 as an example to analyze the influence of water level fluctuations on the deformation of D8 deposit, when the reservoir water level drops, the surface displacement rate is relatively fast. For instance, from April 16 to September 23, 2016, as the water level dropped from 377 to 371 m at a rate of 0.5 m/d, the cumulative horizontal displacement at point GPS1 increased from 34.3 mm to 49.7 mm, corresponding to a displacement rate of 0.09 mm/d. When the reservoir water level rises, there is a small initial decrease in displacement. For example, from August 13 to September 14, 2017, the cumulative horizontal displacement at point GPS1 decreased from 59.7 mm to 53.6 mm. When the water level remains high, displacement shows a continuous increase. For instance, from October 16, 2017, to January 13, 2018, the cumulative horizontal displacement at point GPS1 increased from 55.5 mm to 61.4 mm, corresponding to a displacement rate of 0.07 mm/d. The influence of rainfall on the deposit’s front edge was limited, primarily affecting displacement in the middle section. For example, from June 15 to August 24, 2019, point GPS1 experienced almost no deformation, while the displacement at point GPS2 increased from 33.1 mm to 45.1 mm.

Fig. 13.

Surface displacement monitoring data of D8 deposit.

In summary, the D8 deposit features a relatively gentle deposit-bedrock interface with a dip angle of 15°, representing a retrogressive landslide, and its displacement shows continuous increase during both reservoir water level rise and drop.

Differences between D4 and D8 deposits

Differences in the surface displacement curves of D4 and D8 reveal that D4 deposit shows a noticeable increase in displacement only when the water level drops, with no significant deformation during other periods (Fig. 10). In contrast, the surface displacement curve of the D8 deposit shows a continuous increase both when the water level drops and when the high water level is maintained, except for a slight decrease when the water level rises (Fig. 13). It is evident that the D4 deposit is more affected by the drop in the water level, while the D8 deposit is more influenced by the drop in water level and the maintenance of high water level.

The differences in the deformation characteristics of the deposits are primarily related to their structural features. Comparing the structural differences between the D4 and D8 deposits, it is evident that both have relatively flat deposit-bedrock interfaces, but the dip angles differ significantly. The dip angle of the deposit-bedrock interface in the D4 deposit is 22°, while in the D8 deposit it is only 15°. In addition, the two deposits differ significantly in material composition. The D4 deposit consists of clay containing gravel, with 30% gravel content, whereas the D8 deposit is primarily clay with only 10% gravel content. The difference in gravel content leads to variations in material permeability, which in turn causes differences in the magnitude of permeation force.

Deformation mechanisms of the D4 and D8 deposits

Based on the above analysis of the differences between the D4 and D8 deposits, this paper considers the dip angle of the deposit-bedrock interface and the permeability coefficient as influencing factors. A numerical model is used to compute the safety factor of the deposit at various times. Then, the deformation mechanism of the straight line deposit-bedrock interface deposit is revealed based on the variation of the safety factor.

Numerical model

Based on the A-A' profile of the D4 deposit, deposit models with deposit-bedrock interface dip angles of 22°, 15°, and 10° were constructed using GeoStudio software and numbered as S22, S15, and S10, respectively (Fig. 14a). Additionally, research³³ on the permeability coefficient of deposits within the reservoir area indicates that permeability coefficients primarily fall within the range of 0.1 to 1. To analyze the impact of different material permeabilities on deformation, four saturated permeability coefficients were assigned to the deposit: 0.01 m/d, 0.1 m/d, 1 m/d, and 10 m/d.

Fig. 14.

Calculation model and water level (a) Calculation model of the deposits; (b) Curve of water level fluctuations and the calculation curve.

To simulate water level fluctuations in Xiangjiaba reservoir area, the water level regulation data from 2015 to 2018 were used as a reference (Fig. 14b). As shown in Fig. 14b, the X-axis starts on September 1 each year. The water level generally rose from 370 to 380 m at a rate of 1 m/d, being maintained a high level for approximately nine months. Then, it dropped back to 370 m at the same rate and remained low for around three months. Based on this regulation pattern, a water level fluctuation rate of 1 m/d was used in the numerical simulation. In the S22 model, the initial water level was set at 370 m, rising to 380 m at a rate of 1 m/d and remaining at that level for 260 days. Then, it dropped back to 370 m at the same rate and was maintained for 90 days (Fig. 14b). To eliminate the influence of immersion positions on the calculation results, the water levels in the S15 and S10 models were lowered by 7.5 m and 13.2 m, respectively, compared to the S22 model. Borehole samples revealed saturated densities of soil ranging from 1.9 to 2.1 g/cm³. Consequently, the density of the deposit material is taken as 2 g/cm³. Based on the inversion results for the D8 deposit at the water level of 370 m, the cohesion of the deposit was taken as 20 kPa, and the internal friction angle as 15°.

The numerical modeling process is as follows²⁹: First, transient analysis of the seepage field at different times was conducted using the SEEP/W module. The safety factor was then calculated using the Morgenstern-Price method³⁴.

Calculation results and analysis

The calculation results are shown in Fig. 15. It can be seen that the safety factor increases significantly as the water level rises, and the lower the permeability coefficient, the greater the increase in the safety factor. When the high water level is maintained, the safety factor continuously decreases. When the water level drops, the safety factor decreases significantly. When the low water level is maintained, the safety factor gradually increases. The safety factor decreases are observed both when the high water level is maintained and when the water level drops.

Fig. 15.

Calculation results of the safety factor. (a) S22 model; (b) S15 model; (c) S10 model.

By comparing differences in safety factor among the models, it can be seen that the safety factor of the S22 model first drops below the initial safety factor when the water level drops. For the S15 and S10 models, there is a difference in the time when the safety factor first drops below the initial value. Models with higher permeability coefficients (> 0.1 m/d) experience the drop when the high water level is maintained, while models with lower permeability coefficients (≤ 0.1 m/d) experience it when the water level drops. Based on the above analysis, it is evident that due to differences in permeability, some of the S15 and S10 models are more significantly affected by high water level maintenance.

Analysis of the calculation results for high water level maintenance

To analyze the differences between the S15 and S10 models and other models at high water levels, a comparison of the groundwater levels on Day 0 (initial state) and Day 159 (high water level) is made for three models (Fig. 16). As shown in Fig. 16a, the deposit in the S22 model can be divided into the saturated region, the hydro-fluctuation region, and the unsaturated region. The submerged volume in the hydro-fluctuation region corresponds to the volume below the groundwater level line. The submerged volume reflects the change in buoyant force. The larger the submerged volume, the greater the buoyant force, and the greater the reduction in the safety factor of the deposit when the high water level is maintained. By comparing the submerged volume in the hydro-fluctuation region of the S22 model with the S15 and S10 models, it can be observed that the lower the dip angle of the deposit-bedrock interface, the larger the submerged volume, which leads to a greater impact of high water level maintenance on the S15 and S10 models. Moreover, the submerged volume is also influenced by permeability. The higher the permeability coefficient, the larger the submerged volume, which leads to the safety factors of the S15 and S10 models with higher permeability coefficients being lower than the initial safety factor when the high water level is maintained.

Fig. 16.

Calculation results of groundwater level line when the high water level is maintained. (a) S22 model; (b) S15 model; (c) S10 model.

Analysis of the calculation results for water level drop

Due to the adverse effects of the water level drop on all types of deposits, this study calculates the change in safety factor (ΔF) when the water level drops, using Eq. (1):

1

In Eq. (1), F₂₇₀ represents the safety factor of the deposit on Day 270, and F₂₈₀ represents the safety factor of the deposit on Day 280. As shown in Fig. 17, the amount of decrease in safety factor for each model increases with the reduction in permeability coefficient when the water level drops.

Fig. 17.

The decrease in safety factor during water level drop. (a) S22 model; (b) S15 model; (c) S10 model.

To analyze the effect of permeability coefficient on the amount of decrease in safety factor when the water level drops, the S22 model is taken as an example. The groundwater level line when the water level drops to 370 m (Day 280) is shown in Fig. 18. As the permeability coefficient increases, the hydraulic head at the deposit-bedrock interface gradually decreases, but the model with a permeability coefficient of 0.01 m/d does not follow this trend. Because of the low permeability coefficient, the seepage rate is slow. When the high water level is maintained, the hydraulic head at the deposit-bedrock interface rises slowly.

Fig. 18.

Calculation results of groundwater level line in the S22 model when the water level drops.

The above process does not affect the variation pattern of the head at the front edge when the water level drops. As the permeability coefficient increases, the hydraulic gradient at the front edge decreases, leading to lower dynamic water pressure. When the water level drops, the primary cause of deposit deformation or reduction in safety factor is the outward dynamic water pressure. The larger the permeability coefficient, the smaller the dynamic water pressure and the lesser the amount of decrease in safety factor.

Analysis of deformation mechanism

Based on the above calculation results, the deformation mechanisms of the D4 and D8 deposits are analyzed.

The deposit-bedrock interface of D4 deposit is relatively steep, with a dip angle of 22°, resulting in a smaller volume of the hydro-fluctuation region (Fig. 19a). Due to the small buoyant force generated when the water level rises and maintains a high level, the safety factor of D4 deposit remains higher than the initial value. The deposit experiences minimal deformation in this phase (Fig. 19b). The material composition of D4 deposit consists of clay containing gravel, with a gravel content of 30%, resulting in a high permeability coefficient (> 1 m/d)³³. But when the water level drops, the safety factor remains lower than the initial one, causing deformation. Moreover, when the rate of water level drop is faster, it generates higher dynamic water pressure, leading to greater deformation. For example, when the water level dropped at a rate of 1 m/d in July 2018, the displacement increase at GPS1 on D4 deposit was 164 mm (Fig. 19b). In contrast, when the water level dropped at a rate of 0.13 m/d in July 2017, the displacement increase at GPS1 on D4 deposit was only 28 mm (Fig. 10).

Fig. 19.

Profile and surface displacement curve of D4 deposit. (a) Profile; (b) Surface displacement monitoring data.

D8 deposit has a relatively gentle deposit-bedrock interface with a dip angle of 15° (Fig. 20a). The material of D8 deposit is primarily clay, with 10% gravel content, and a permeability coefficient between 0.1 m/d and 1 m/d. Due to the lower permeability, a larger inward seepage force is generated when the water level rises, leading to a slight reduction in displacement. When the high water level is maintained, the gentler deposit-bedrock interface results in a larger volume in the hydro-fluctuation region, generating greater buoyant force that leads to a continuous increase in displacement. When the water level drops, the low-permeability material generates a larger outward seepage force, causing a sharp increase in the displacement of the deposit (Fig. 20b).

Fig. 20.

Profile and surface displacement curve of D8 deposit. (a) Profile; (b) Surface displacement monitoring data.

From the above analysis, it is evident that the core difference in deformation between the D4 and D8 deposits lies in the dip angle of the deposit-bedrock interface. D4 deposit has a relatively steep deposit-bedrock interface. When the reservoir water level rises, the growth of buoyant force is limited, and its deformation is controlled by seepage forces during the water level drawdown period. In contrast, D8 deposit, due to its gentler interface, experiences continuous deformation triggered by the synergistic action of buoyant force and seepage force.

Limitation

This study simplified the deposit as a homogeneous material and employed a two-dimensional model for analysis, which did not fully account for the heterogeneity of actual materials and three-dimensional spatial effects. The two-dimensional computational profile adopted in this study aligns with the deformation direction of the deposit. However, it does not account for the constraining effect of surrounding materials on the deformation, which leads to an underestimation of the safety factor derived from the two-dimensional simplified calculation.

Conclusions

Based on field surveys in Xiangjiaba Reservoir area, a statistical analysis of the distribution and types of deformation slopes in the area was conducted. The deformation characteristics of red-bed deposits with straight line deposit-bedrock interfaces (D4 and D8 deposits) were analyzed through borehole and monitoring data. The deformation mechanisms of red-bed deposits with straight line deposit-bedrock interfaces were analyzed through numerical calculations. The conclusions are as follows:

1. Statistical analysis indicates that the primary subjects affected by reservoir impoundment in Xiangjiaba Reservoir area are red bed deposits, with those possessing straight line deposit-bedrock interfaces accounting for 50% of the total.

2. Analysis of the structural and deformation characteristics of two straight line deposit-bedrock interface deposits showed that the D4 deposit has a relatively steep deposit-bedrock interface with a stepped deformation curve, while the D8 deposit has a gentler deposit-bedrock interface with a continuously growing deformation curve.

3. The analysis of the numerical calculation results reveals that the smaller the dip angle of the deposit-bedrock interface, the more pronounced the effect of maintaining high water levels. The smaller the permeability coefficient of the deposit, the greater the outward dynamic water pressure during water level drop, and the larger the reduction in the safety factor.

4. Based on the analysis of monitoring data and numerical calculation results, it is concluded that the difference in deformation between the two deposits in the dip angle of the deposit-bedrock interface: D4 deposit has a relatively steep deposit-bedrock interface, causing its deformation to be controlled by seepage forces during the water level drop period. In contrast, D8 deposit, due to its gentler interface, experiences continuous deformation triggered by buoyant force and seepage force.

Author contributions

Yijun Xiong: Validation, Writing—original draft. Fuchu Dai: Conceptualization, Funding acquisition, Writing—review and editing. Yunpeng Qi: Data curation, Formal analysis, Writing—review and editing. Shouwen Li: Software, Visualization, Writing—review and editing.

Funding

This work was supported by the Second Tibetan Plateau Scientific Expedition and Research (STEP) Program (No. 2019QZKK0905).

Data availability

All data generated or analysed during this study are included in this published article.

Declarations

Competing interests

The authors declare no competing interests.