Abstract
On 12 May 2008, an earthquake of magnitude 8.0 on the Richter scale occurred at Wenchuan in China. Zipingpu dam also experienced the earthquake. The maximum ground acceleration near the top of the 156 m high rockfill dam exceeded 2 g. The right dam abutment slope, which accommodates all the discharge facilities, remained intact after this severe earthquake. However, the natural slopes on the opposite side of the reservoir were severely damaged. It has been found that the reinforcement provided by more than 1000 prestressed cables in the right abutment slope helped to stabilize it. Measurements from multipoint extensometers and load cells installed in the slope showed that there were sudden changes in displacement and stress during the earthquake. This is clear evidence that reinforcement with anchored cables can be successfully used to stabilize a slope.
Zipingpu hydroproject is located in the upper reaches of Minjiang River, about 60 km from Chengdu, the capital city of Sichuan Province, China, and only 10 km from Dujiangyan City, which has a population of 300 000 (Fig. 1). The water-retaining structure of the project is a 156 m high concrete face rock-fill dam. The total reservoir storage capacity is 1.112 billion m3. Owing to the constraints of the topography, all the water discharge structures are built within a bar-type ridge on the right bank, and their construction has involved large excavations and high slopes.
Location of Zipingpu dam.
The Zipingpu project area is only 17 km from the epicentre of the 12 May 2008 Wenchuan earthquake of magnitude 8.0 on the Richter scale. The site seismic intensity was estimated to be IX–X degree (Zheng et al. 2008; Zhao et al. 2011). The original design adopted a peak acceleration of 0.26 g with an annual exceedance probability of 2% over 100 years (Gao & Yang 2002). A macro-seismograph installed at No. 2 monitoring station on the dam crest recorded a seismic peak acceleration of 1.635 g along the dam axis, 2.063 g horizontally and 2.064 g vertically (Fig. 2; Guan 2009). It is suggested that the recorded peak acceleration of the dam bedrock may have exceeded 0.500 g (Chen et al. 2008), which is much greater than the dam's design seismic resistance. The right abutment of the dam accommodates all the water discharge facilities including the spillway, powerhouse tunnel and flood discharge tunnel. The high slopes of this abutment are therefore of vital importance to the safety of the rockfill dam. In other words, if the abutment failed, the natural inflow of the Mingjiang River, which is 500 m3 s−1 on average, would have no water release path and would eventually overtop the crest of the embankment dam. All the slopes were heavily reinforced with anchor cables, which helped the right abutment to remain intact. In contrast, the natural slopes on the opposite side of the reservoir collapsed extensively because of the earthquake. This demonstrates that the slope reinforcement design of the right ridge is robust.
Ground movement measured at the crest of the dam (Guan 2009).
This paper presents a case history that includes the information obtained from the field investigations and data measured by the multipoint extensometers and load cells in the powerhouse intake structure slope. Further information has been given by Li et al. (2011) and Ma et al. (2011).
Geology
The right bank of the Zipingpu dam area is located within a narrow ridge of rocks of the Triassic Xujiahe formation. The water discharge facilities, such as the flood discharge and water supply tunnels and the spillway, are located within this ridge. The ridge extends along a syncline whose axis trends 050°–060°, with bedding dip angles of 25–35° on both flanks. Therefore, both of the inlet and outlet slopes are composed of formations of in-dipping rock strata, as shown in Figure 3.
Geological cross-section of the narrow ridge of the right abutment; values in metres. Legend: 1, interbedded disturbed belt; 2, medium fine sandstone; 3, siltstone mingled with fine sandstone and argillaceous siltstone; 4, siltstone mingled with argillaceous siltstone.
The lithology of the slope includes Triassic hard and soft sandy shale. Each alternating layer is composed of sandstone, argillaceous siltstone and an upper coal shale. Hard, medium and thick sandstone accounts for 49%, siltstone for 37% and interbedded coal shale for 14% of the total layer. The syncline contains an interbedded shear zone of coal shale, which lies near the outlet of the flood discharge tunnel, as shown in Figure 3. The 55 – 87 m wide fault zone comprises mylonite that originated from the coal shale, fault gouges and sandstone tectonic lens; this zone has a complex composition and the rocks are soft. Such an area is prone to disintegrate when soaked with water. The rock mass mainly consists of two sets of joints with dip direction/dip of 280°–315°/10 – 35° and 230°–260°/60 – 80° respectively. The hydropower project required excavation of slopes with heights of over 150 m with complex geological conditions and poor rock quality. The rock masses of these slopes are highly fractured and classified as Grade 4 according to the Chinese standards (Ministry of Water Resources (MWR) 1995, 1999), which represents rather poor rock quality. The entire project is dependent on the stability of these slopes.
Slope reinforcement design
Figures 3 and 4 show the longitudinal profile and cross-section of the intake structure slope. The 123 m high slope has a width of 90 m and an average grade of 45° with a vertical cut c. 50 m deep. The upper part of the cross-section consists of moderately weathered medium fine sandstone designated Type II in Table 1, whereas the lower part is of moderately weathered argillaceous siltstone designated Type III. An interbedded shear zone lies between the two layers of rock and is designated Type I in Table 1.
Geological profile of the intake structure slope; values in metres. Legend: 1, interbedded disturbed belt; 2, medium fine sandstone; 3, siltstone mingled with fine sandstone and argillaceous siltstone; 4, siltstone mingled with argillaceous siltstone.
Shear strength parameters
By analysing the relationship between the main structural surfaces and the designed excavation surfaces of the intake structure slope, possible failure modes of the slopes were reviewed on the basis of a simplified kinematic analysis (Chen et al. 2001). It was reported that potential failure modes such as planar or wedge slide and toppling were not possible (Hoek & Bray 1977). The instability of the slope of this highly fractured slope may develop along a circular slip surface, and the method of Sarma (1979; see also Hoek 1987) may be suitable for assessment of the factor of safety for design purposes. As an equivalent to this method, the computer program EMU (Chen & Wang 2000), which virtually employs an upper bound approach developed by Donald & Chen (1997), was used for the analysis. It avoids the simplification related to the use of limit equilibrium methods concerning static and kinematic admissibility. The program starts from an initial trial slip surface and ends at a critical one. Its input parameters include the designers' recommendation based on the Mohr–Coulomb criterion of cohesion and friction angle as shown in Table 1. In addition, the Hoek–Brown failure criterion (Hoek & Brown 1997), which offers an empirical non-linear strength of jointed rock mass, was adopted for checking purposes. This criterion involves the input of rock mass rating index RMR (Bieniawski 1989), uniaxial compressive strength of the intact rock σc, and m, a coefficient that accounts for the lithology of the rock, the values of which are also shown in Table 1.
The analysis considers a 10 m deep tensile crack and a pore water pressure coefficient of 0.1. The calculated factors of safety associated with the design strength parameters and the Hoek–Brown failure criterion are 1.34 and 1.10 respectively as shown in Figures 5a and 6a, which are diagrams produced by EMU (Chen et al. 2001). The Chinese slope design code specifies that for extremely important hydraulic structures, the allowable factor of safety should range between 1.30 and 1.50 (Ministry of Water Resources (MWR) 2007). It was therefore considered that systematic reinforcement of the intake slope by prestressed cables was necessary.
Slope stability analysis using the designers’ recommended parameters. (a) Without reinforcement, F = 1.34; (b) with reinforcement, F = 1.41. I, II and III are rock types (see Table 1). The shaded area indicates the sliding mass above the slip surface.
Slope stability analysis using Hoek–Brown parameters. (a) Without reinforcement, F = 1.10; (b) with reinforcement, F = 1.323. I, II and III are rock types (see Table 1). The shaded area indicates the sliding mass above the slip surface.
The front slope was reinforced with 77 pieces of 3000 kN anchor cable with a length of 45 – 60 m for elevation between 841.00 and 865.00 m, 123 pieces of 1500 kN anchor cable with a length of 40 – 67 m for elevation between 828.00 and 837.00 m, and 88 pieces of 2000 kN anchor cable for elevation between 813.00 and 825.00 m. The combined anchoring force of all these cables is 591 500 kN. These cables were installed over the front face, which is 90 m wide. Figure 7 show the layout of the anchor cables on the reinforced slope. The reinforcing load applied on the slope is then 6572 kN m−1. Based on this load, the factors of safety after slope reinforcement associated with the design strength parameters and the Hoek–Brown failure criterion are 1.41 and 1.323 respectively as shown in Figures 5b and 6b (Chen et al. 2001).
Front view of the intake structure slope and the instrumentation; MIP, multipoint extensometer; AIP, load cell (values in metres, scale: 1:1000).
The anchor cable used in the reinforced slopes is in accordance with ASTM A416-90a standard (AASHTO 1995) and was 1860 MPa, 15.24 mm diameter steel strand with high strength under relaxation and unbonding. Two types of prestressed anchor cables were used: (1) bearing load 2000 kN with 12 strands; (2) bearing load 1500 kN with nine strands. Each strand consists of seven steel wires, 5 mm in diameter. These cables have decentralized tension and compression. The design lengths of the cables are 45 m for the 2000 kN and 40 m for the 1500 kN cables. Figure 8 shows the intake structure slope under construction.
The intake structure slope under construction.
Field observations after the earthquake
Witnesses reported that during the earthquake, the whole reservoir could not be seen because of dust and smoke. When this cleared, they saw that the mountains on the opposite side of the reservoir were seriously damaged by landslides (Figs 9 and 10), and dozens of people were killed by the surge wave of the reservoir water. One span of the bridge that crosses the reservoir also collapsed (Fig. 11). In contrast to this severe damage, the right abutment, which contains all the water discharge facilities, basically remained intact (Fig. 12). The only damage within the right abutment was to the gatehouse, but it did not collapse (Fig. 13). It is widely acknowledged that the heavy cable reinforcement helped to stabilize the slope. The field investigations showed further evidence, as one of the anchors showed serious ‘pull-out’ (Fig. 14).
A large-scale landslide in the opposite side of the reservoir.
The opposite river bank, showing severe damage.
The cross-reservoir bridge, with one span that fell down during the earthquake.
The reinforced right abutment in close vicinity to the dam and powerhouse after the earthquake.
The gatehouse of the intake structures after the earthquake.
The cable with cracks around the head (Courtesy of Yueping Yin).
Displacement measurements on the slope of the intake structure
To measure the movement of the slope, four multipoint extensometers were installed on the slope of the intake structure. MIP1 and MIP3 each have four points installed at 4, 10, 18 and 30 m into the slope respectively. MIP2 and MIP4 each have six points installed at 3, 7, 12, 18, 26 and 38 m into the slope respectively (Table 2).
Layout of the multipoint extensometers
Figure 15 shows the measurements during the 12 May 2008 Wenchuan earthquake. It can be seen that all four multipoint extensometers show a sudden change in displacement readings. Taking MIP1 (Fig. 15a) as an example, the measuring points located at 4, 10 and 18 m depth displayed sudden jumps with outward displacements of 6.03, 6.43 and 5.14 mm respectively at the moment when the earthquake occurred, whereas the reading taken from the point located at 30 m depth showed virtually no change. One may presume that the potential slip surface was somewhere between 18 and 30 m depth from the slope surface. Similar interpretations can be drawn from the other three multipoint extensometers. Tables 3 and 4 give the magnitude of movement of all the measuring points. The maximum displacement was found at MIP3, where the measuring point at 4 m depth showed a displacement of 10.7 mm.
Measurement of the multipoint extensometers along section A–A' (section location shown in Fig. 7). (a) MIP1. (b) MIP2. (c) MIP3. (d) MIP4.
Displacements recorded by multipoint extensometers MIP1 and MIP3
Displacements recorded by multipoint extensometers MIP2 and MIP4
The installation of these multipoint extensometers was completed on 29 December 2002. After remaining static for 5.5 years, they were all ‘awakened’ by the earthquake, offering excellent information on the internal movement of the rock mass. This is good evidence demonstrating the serviceability of our geotechnical field monitoring facilities.
Load cell measurements of prestressed cables
The layout of the anchor load cells on the slope is shown in Figure 7. Figure 16 shows the measurements of these cells, from which the following findings may be drawn.
AIP1 and AIP7 showed virtually no change during the earthquake.
AIP4 and AIP12 showed small decreases in cable load with magnitudes of −12 and −16.9 kN respectively (the minus sign indicates a decrease of load). This indicates slight loosening of the two cables.
All the remaining nine load cells displayed a sudden increase of the cable loads, showing that these cables provided stabilizing forces at the moment when the earthquake occurred. The magnitudes of the increased loads range from 12.4 to 86.1 kN.
Anchoring force–time graph when anchor cable dynamometers were installed. (a) AIP1; (b) AIP2; (c) AIP3; (d) AIP4; (e) AIP5; (f) AIP6; (g) AIP7; (h) AIP8; (i) AIP9; (j) AIP10; (k) AIP11; (l) AIP12; (m) AIP13. It should be noted that data before 8 January 2008 are truncated. They are all virtually the same as those recorded after that date until the earthquake occurred on 12 May 2008.
Table 5 summarizes the load increase at the moment the earthquake took place.
Increase of load cell reading
Again, the load cell measurement shows that after a long period of sustaining readings of virtually zero load increment, the majority of the load cells accurately recorded the sudden changes at the moment of the earthquake, demonstrating the long-term serviceability of these instruments.
Conclusions
This paper describes the excellent performance of the heavily reinforced right abutment of Zipingpu dam during the 12 May 2008 Wenchuan earthquake, in contrast to the severe damage of the natural slopes on the opposite side of the reservoir. It demonstrates the effectiveness of the prestressed cabless, which are a common measure in geotechnical engineering for slope stabilization.
Visual inspections showed that the slopes excavated for the powerhouse, flood discharge tunnel and spillway slopes and reinforced by prestressed cables remained virtually intact during the severe ground movement. The excellent performance of the right abutment allowed all the water discharge facilities to work properly after the earthquake and helped the damaged dam to survive by allowing the reservoir to be emptied for repair. On the other hand, the opposite side of the reservoir was severely damaged by landslides, causing a widespread water surge that killed dozens of people.
Readings of the four multipoint extensometers installed at the powerhouse intake slope invariably showed sudden movements of the relevant measuring points, detecting internal movements of the slope up to 10.7 mm.
Of the 13 load cells installed at the powerhouse intake slope, nine showed a sudden increase of cable load that ranged from 12.4 to 86.1 kN. This demonstrates that the cables successfully stopped a possible disastrous movement of the slope.
Acknowledgements and Funding
This research work was supported by the National Basic Research Program of China (Grant No. 2011CB013502) and the Chinese National Science Foundation (Grant No. 51479214).
- © 2016 The Author(s)
Performance of the reinforced right abutment slope of Zipingpu Dam during magnitude 8.0 earthquake, Wenchuan, China
