Introduction

The Jinping I hydropower station, located in a deep valley of the Yalong River, is the first of the five hydropower stations to be constructed in the middle and lower reaches of the river. The key constructions are a double-curvature arch dam, water diversion tunnels and a major powerhouse. The dam height will be 305 m (to 1,885 m asl) with the retained reservoir level, 1,880 m asl. The total reservoir capacity will be 7.76 billion m3 and the regulated storage 4.91 billion m3. The installed plant capacity is 3,600 MW and the mean annual power generation 16.62 billion KW/h.

In addition to the electricity generation, the scheme will assist in flood prevention and sediment trapping, hence will have a significant economic benefit for this part of China.

The valley was cut through the Triassic Zagunao Formation which consists of sandstone, slate and marble. In view of its location between the Tibetan Plateau and the Sichuan Basin, the area is one of complicated geology and high geostresses, particularly on the left bank of the river (Wang et al. 1998). In addition to superficial unloading of the valley sides, a large number of fractures are found at a significant distance/depth from the valley slope as a consequence of the unique deep unloading (Qi et al. 2004), i.e. an area of antecedent drainage. The deformation and cracking on the left bank (with the highest geostresses) has been exacerbated by the presence of two faults (f5 and f42-9) and a lamprophyre dyke (Huang et al. 2005; Institute of Chengdu Investigation and Design of National Electrical Power 2003). In general, where the rock mass is formed of marble, away from fault f5, the slopes are stable. However, as excavation for the dam foundation has removed some of the marble above fault f5, the stability of this area should be reappraised.

Engineering geological condition

In the area of the dam, the river valley is straight and narrow, trending N25°E. The elevation of the valley floor is about 1,590–1,600 m asl while the slope shoulder is more than 3,500 m asl, with relative relief of about 1,900 m. The right bank is a dip slope with alternating steep and gradual steps. The left bank is a reverse slope, with marble outcropping below 1,820–1,900 m asl with sandstone and slate above. The general slope angle in the marble is 44–70° and in the sandstone and slate, 40–50°.

The dam site is 2 km west of the Jinping fault, within the southeastern limb of the tight Santan syncline. The left slope consists mainly of metasandstone and slate and interbedded marble and greenschist of the Triassic Zagunao Formation, both of which were subsequently intruded by lamprophyre veins. In addition to the four main faults (f5, f8, f42-9 and f2), a number of minor faults are also present in the slope. There are four sets of discontinuities in the abutment slope: (1) N0–30°E/NW∠30–45°, bedding cracks; (2) N40–70°E/SE∠50–70°; (3) N0–30°E, SE∠60–80°; (4) near EW/S∠40–60°. Consequently, the left abutment forms a scarp slope with alternating weak and strong rocks (Fig. 1).

Fig. 1
figure 1

Typical engineering geologic profile of the left abutment slope. a Vertical section along line II-1 of the left abutment slope (selective enlargement). b Horizontal section at elevation 1,930 m

Full size image

A considerable amount of research has been undertaken into the mechanisms responsible for the deep unloading zones (Huang et al. 2005; Huang 2000; Qi et al. 2004). The principal characteristics include:

  1. 1.

    Removal of the horizontal stress in the river valley facilitates the opening of stress-induced fractures in the rock mass.

  2. 2.

    The orientation of the unloading fractures is almost parallel to the river.

  3. 3.

    The deep unloading fractures extend for several metres in length and have apertures of >200 mm.

  4. 4.

    The main fractures occur above 1,680 m asl and generally upstream of the dam.

  5. 5.

    Fractures in the marble below 1,900 m asl extend up to 140 m in from the valley slope.

  6. 6.

    In the metasandstone and silty slate above 1,900 m asl fractures up to 300 mm extend for a horizontal distance of 300 m from the valleyside.

Structural features and deformation mechanism of the left abutment slope

Above 1,975 m asl, the interbedded sandstone/metasandstone and slate dips at about 20°and toppling failures occur. Some fissures have developed along discontinuities which dip steeply out of the slope. Between 1,825 and 1,930 m asl, a number of features are evident in the areas of faults f5 and f42-9:

  1. 1.

    With the release of stress due to erosion, tension cracks have developed in the area of f5 which have shattered and hence weakened the integrity of the rock mass, although it is still tightly compressed.

  2. 2.

    Fault f42-9 has an EW orientation and dips to the south at 40–60°. The fractured zone consists of heavily weathered breccia, rock debris and fault gouge and is commonly 0.2–0.4 m wide but sometimes reaches 1 m. In the upper part of the fault zone the breccia is loose. Whilst the footwall is only slightly deformed and cracked, the hanging wall is more extensively fractured.

Fault f42-9 acts as the lower, downstream boundary of the deformed and cracked rock mass on the left abutment. Stream erosion and trenching have resulted in intensive unloading of the left slope, which is cut by fault f5. The deformed and cracked rock mass is generally stable as it is confined by the surrounding rock mass. However, this rock will be partly excavated in order to form the dam abutment and foundations.

Stability analysis of the deformed and cracked rock mass

In the left abutment slope, definite boundaries of the deformed and cracked rock mass include downstream fault f42-9 and the outside fault f5. A uniform boundary has not been formed in the upstream side. According to the above analysis, the rock mass outside of tension fissures zone SL44-1 and lamprophyre veins is a key rock mass which controls deformation and stability of the left abutment slope. Parametric analysis on boundaries of the key rock mass is as follows (Table 1).

Table 1 Parameters of rock mass and key discontinuities
Full size table
  1. 1.

    Fault f 42-9 is a main slip surface in the downstream side; its attitude is N80–88°E/SE∠39–51° as measured in all the adits above 1,885 m asl, consistent with the general attitude of N84°E/SE∠40° obtained by spatial analysis. The strength parameters of the shattered fault zone are: angle of internal friction 15.5°, cohesion 16 kPa.

  2. 2.

    Tension fissures zone SL 44-1 is the upstream cut surface, with a general attitude is SN–N10°W/NE∠52–65°. It is believed to be a significant feature in over 50% of its plan area.

  3. 3.

    Lamprophyre veins form the rear boundary; the general attitude is N50–75°E/SE∠60–75°, and the average attitude (from spatial analysis) N67°E/SE∠73°.

  4. 4.

    Fault f 5 has a general attitude of N35°E/SE∠84° and forms much of the excavated surface. In some areas it passes into the marble face itself such that marble outside of the fault f5 supports the deformed and cracked rock mass like a “rock wall”.

Orientation of all discontinuities In the Cartesian three-dimensional coordinate system, X axis trends east, Y axis trends north, and Z axis trends upwards. Using this system, equations were developed for each boundary based on the positions of the discontinuity uncovered in the adits, such that the key blocks could be defined (Zong 1996).

Stability analysis of key blocks

Computation model

Fault f42-9 does not cross the fault f5 hence two models were adopted to compute the stability of the deformed and cracked rock mass within f5.

  1. 1.

    Integral model Fault f42-9 or SL44-1 is extrapolated through the rock mass outside of f5 to the free face, and forms a movable block.

  2. 2.

    Blocked model Fault f5 acts as a boundary between the deformed and cracked rock mass and the marble which prevents the broken rock mass from sliding.

Two schemes were designed for the foundation excavation: Scheme I, a scheme with a large excavated volume and Scheme II with a small excavated volume. Computations were undertaken for both schemes for comparison and it was decided to proceed with Scheme I. This involved excavating part of rock mass outside f5 such that fault f42-9 protrudes through the excavated surface. Discontinuities striking NW or NNW in the slope act as the upstream boundary and form the deformed and cracked rock mass with other boundaries of f42-9 and lamprophyre veins. Specific connectivity of discontinuities striking NW or NNW is less than SL44-1, hence 50% is recommended.

Four working conditions were considered: natural, heavy-rain, earthquake and heavy-rain plus earthquake. The heavy-rain condition involved a water table at a quarter of the height of the rear surface of the block. For the earthquake condition, a horizontal peak acceleration of 0.1g and vertical peak acceleration of 0.05g were considered.

Integral model

Figures 2 and 3 show the discontinuities and faces in the integral model; the calculated parameters are given in Table 2. The results indicate:

Fig. 2
figure 2

Positive axonometric project map of integral model. a Scheme I (volume: 3,862,543m3). b Scheme II (volume: 3,360,316m3)

Full size image
Fig. 3
figure 3

Positive axonometric project map of blocked model. Scheme II with small excavation: a cracked rock mass, b anti-slide rock wall. Scheme I with large excavation: c cracked rock mass, d anti-slide rock wall

Full size image
Table 2 Calculation results of two excavation schemes in the integral model
Full size table
  1. 1.

    In Scheme I, the volume of the block is about 3.36 million m3. Stability coefficients before and after accumulating water are more than 1.57 except in the heavy-rain condition, and the block has two sliding planes. Under the heavy-rain condition, the block has a single sliding plane and its stability coefficient is less than 0.6. In Scheme II, the volume of the block is 3.86 million m3. Stability coefficients are more than 1.45 except in the heavy-rain condition, and the block has two sliding planes; in the heavy-rain condition its stability coefficient is less than 1.2.

  2. 2.

    Under all working conditions, stability coefficients for Scheme I are greater than for Scheme II, indicating that Scheme I has a load reduction effect. As the unloading position is not considered in the load reduction effect, the stability of the rock mass is appraised according to the block model.

  3. 3.

    Specific connectivity of SL44-1 influences the stability of the rock mass. When calculated at 10% intervals, the stability coefficient increased 0.068–0.078% under the same conditions. This was monitored during the construction period and it was concluded that taking the specific connectivity of the upstream boundary as 50% would be conservative.

  4. 4.

    If the groundwater table is high, the block appears to slide on a single rather than a double face and its stability coefficient is greatly reduced. Draining water from the slope and reducing the groundwater level was therefore considered an important element in the construction sequence.

Block model

In the block model, the deformed and cracked rock mass is considered separately and its surplus sliding force applied on the rock wall outside of f5, so the stability of the rock wall was used to appraise the stability of the left abutment slope. There are two probable failure modes for the rock wall outside fault f5:

  1. 1.

    Punching shear mode, i.e. the deformed and cracked rock mass wedges and damages the rock wall;

  2. 2.

    Pushing shear mode, i.e. the rock wall slides along the plane where there is least resistance to it and the toe of the shear is at the same elevation as the valley floor. The computational analysis shows that the stability coefficient of the rock wall under the pushing shear mode is small (Lin et al. 2007).

The results are shown in the Table 3 and indicate that:

Table 3 Stability calculation results of block model under two excavation schemes
Full size table
  1. 1.

    Slope stability reflected by the rock wall blocking the movement satisfies the safety requirement except in the heavy-rain condition. The stability coefficient for Scheme II is less than for Scheme I, i.e. Scheme I is more beneficial in reducing the load of the cracked rock mass.

  2. 2.

    Drainage is important for the stability of the slope.

Stability analysis of local blocks

There are two key sets of discontinuities, which strike NW and NNW with average orientations of 75° at 55° and 210° at 70°, respectively. In Scheme I with major excavation, the computed stability of the potential largest local block is controlled by these two sets of discontinuities (Fig. 4). The results (Table 4) indicate that:

Fig. 4
figure 4

Positive axonometric project map of local blocks a Discontinuities with attitude of 75°∠55°; volume: 1,239,819 m3 b Discontinuities with attitude of 210°∠70°; volume: 219,932 m3

Full size image
Table 4 Stability calculation results of local blocks
Full size table
  1. 1.

    A 1.239 million m3 block is controlled by discontinuities with an attitude of 75° at 55°. The stability coefficient is 1.164 under the natural condition, 1.069 under the earthquake condition and less than 0.61 under the heavy-rain condition when its specific connectivity is 50%. Its stability coefficient reduces slightly after accumulating water.

  2. 2.

    A 2.2 million m3 block is controlled by discontinuities with an attitude of 210° at 70°. It is stable under all working conditions when its specific connectivity is 50%.

  3. 3.

    The stability of localised blocks has a close relationship with the specific connectivity of the upstream discontinuities and a rise in the groundwater table as a consequence of heavy rain.

Slope deformation analysis in construction period

Excavation of the left abutment slope began in September 2005 from elevation 2,110 m. Excavation above elevation 1,960 m was completed in December 2006, excavation above the dam crest (1,805 m asl) was completed in July 2007 and excavation above elevation 1,710 m was completed in September 2008.

Support of the abutment slope included pre-stressed cables and anti-shear replacement tunnels (tunnels excavated and then backfilled with reinforced concrete). Numerous monitoring measures were adopted at different positions on the slope throughout the excavation. The results for two of the adits (see Fig. 1) are discussed briefly below.

  1. 1.

    Adit PD44 is 203 m long at elevation 1,930 m. Three kinds of monitoring measures were adopted.

    1. a.

      The development of fractures in the tunnel wall at two main locations: section 1 from 74 to 86 m behind the rock face, between SL44-1 and lamprophyre veins; section 2 from 114 to 134 m where deep fractures (SL44-4) occur.

    2. b.

      There were five monitoring points for the valley span. The section from 92 to 47 m behind the rock face showed the largest deformation, reaching 0.02 mm/day from 10th June 2008 to 9th July 2008.

    3. c.

      There are seven measuring points for tunnel convergence. As seen in Fig. 5, two showed deformation, i.e. section ID-3 (70–16 m) and section ID-4 (116 to 0 + 146 m). Between June and July 2008, the average rate of deformation was 0.038 mm/day at ID-3 and 0.029 mm/day at ID-4.

Fig. 5
figure 5

Deformation curve of PD44 by plumbago-rod convergence indicator

Full size image

Combining the results of the valley span monitoring and the convergence indicated that the section from 98 to 152 m in the PD44, between lamprophyre veins and f44-9, has apparent deformation, and its mean strain gradually reduces from outside to inside (Fig. 6).

Fig. 6
figure 6

Average deformation rate of each section in the PD44

Full size image
  1. 2.

    Adit PD 42 is located downstream of PD 44 and also at elevation 1,930 m asl. The convergence results are shown in Fig. 7. In section ID-4 (64.9–77.2 m and inside of fault f42-9) the accumulative displacement at 3rd July 2008 was 3.045 mm but there was no apparent deformation in the other monitoring points. The average deformation rate was 0.027 mm/day, slightly less than that in PD44.

    Fig. 7
    figure 7

    Accumulative deformation of each section in PD42 by plumbago-rod convergence indicators

    Full size image

The monitoring results from two typical adit positions show that sections with apparent deformation are all located within the range of influence range of the pre-stressed cables and yet continued to deform at a low rate. They indicate that whilst the left abutment slope is generally stable with the current support, further monitoring would be beneficial.

Recommendations and conclusions

The left abutment slope of the Jinping hydropower station has complicated engineering geological conditions, including some large faults and lamprophyre veins which have a signification effect on the stability of the slope.

In the process of valley formation, many deep stress-release fractures developed in the sandstone and slate related to the lateral unloading such that a deformed and cracked rock mass is present.

Before excavation, the slope was generally stable due to the mass of marble between fault f5 and the valley slide which prevented the slope from sliding. However, to construct the dam required removal of part of the marble out of fault f5. Based on engineering geological analysis and block theory, the key blocks which control stability of the slope were analysed and the following conclusions drawn:

  1. 1.

    The block which is controlled by fault f5, f42-9 and the deep fractures zone SL44-1, is generally stable under present excavation/support scheme except in the heavy-rain condition. However, safety requirements are satisfied during the construction period and after the reservoir has been impounded.

  2. 2.

    The local block which is controlled by discontinuities striking NNW has a low stability and greater support is required.

  3. 3.

    Monitoring results in adits PD44 and PD42 show that in some sections within the range of influence of the pre-stressed cable anchors, deformation is continuing, albeit at a slow rate.