The Xiaolangdi Multipurpose Dam Project is located in central China on the Yellow River. It comprises a 154m high rockfill dam, a large underground powerhouse with six turbines and a total installed capacity of 1800MW, nine flood and sluice tunnels and six power tunnels. All underground structures are on the left bank (Fig 2). The power cavern, which is 26.2m wide, 61.4m high and 251.5m long, was originally designed to be supported with systematic bolting and shotcrete. Downstream of the power cavern are a transformer chamber with a cross section of 15.2m by 18.3m and a gate chamber with a span of 10.6m.

The Ertan Hydropower Project is on the Yalong River in western China. The project comprises two spillway tunnels; two diversion tunnels; a log passing tunnel; an underground powerhouse with six turbine units and a total installed capacity of 3000MW; and a 240m high concrete arch dam (Fig 3). The powerhouse and all the tunnels, except one diversion tunnel and the spillway tunnels, are in the mountain on the left bank. The powerhouse complex consists of the powerhouse, which is 25.5-30.7m x 65m x 280m; the transformer chamber, 18.3m x 25m x 215m; and the surge chamber, 19.8m x 60-70m x 203m.

Geology and rock properties

Xiaolangdi site

The powerhouse cavern at Xiaolangdi has a rock cover of 75-95m, overlain by a 10-20m thick layer of soil. The rock is mainly unweathered sedimentary rock and there are no significant faults or crushed zones near the powerhouse. The rock has a dip of 10°. The powerhouse is situated in thickly bedded fine sandstone and siltstone. Several clay intercalations parallel to the bedding planes were identified during investigation (Fig 2).These are 1-20mm thick and filled with clay and other soft materials. In addition to the bedding planes, four sets of joints are mapped in the powerhouse area, dipping at 75-85° and persisting less than 500mm. Joint planes are mainly rough and planar11,12.

The in-situ stresses in the powerhouse area were measured by the overcoring and hydraulic fracturing and measurements indicate that (i) the major principal stress is vertical and approximately identical to the gravitational stress; (ii) the minor principal stress is horizontal and about 0.8 times the major principal stress1. The vertical stress at the crown of the powerhouse is approximately 3MPa. Rock quality is estimated at 52-64 (RMR system) and 8.3-12.6 (Q system). The groundwater table is approximately half way up the power cavern. The measured uniaxial compressive strength of the rock is in the order of 60-150 MPa, and the friction angle is estimated as 41-46° for the intact rocks and 27-33° for the joints11.

Ertan site

The Ertan project is located in a steep, narrow valley whose sides rise to 400m above the river bed. Average rock cover above the powerhouse complex is 250-300m. The rocks consist mainly of granodiorite, gabbro and metabasalt. Contact zones between different rock types are, in general, characterised by fluent transitions without distinct interfaces.

During the site investigation, six joint sets were identified in the powerhouse area. Three sets are well developed in granodiorite and gabbro, while the other three sets are in basalt. The friction angle of joints ranges from 30-40°. The wet compressive strength of the intact rocks is approximately 180MPa. Encountered weakness zones vary in thickness and type of filling. Maximum observed thickness is 1500mm; most zones are 200-500mm thick. Frequently, they are accompanied by jointed zones with decreasing intensity of jointing and increased distance to the weakness zone. Most weakness zones comprise rock fragments and crushed rock material.

Nine 3D in-situ stress measurements were performed in the adits in the powerhouse area. They confirm that the stress situation is dominated by the topographic effects, which is often the case in a narrow valley with steep sides. The major principal stress dips to the river bed and increases with lower elevation; the maximum stress was found at the river bed. It was concluded that the major principal stress1 dips at 22° on average and is 20-30MPa in granodiorite and 30-40MPa in basalt, while 2 and 3 are in an average relation to 1 of 0.5-0.7 and 0.27 to 0.363.

Using the Q system, the cavern complex rock mass was classified as being ‘good to very good’ in general, with limited zones of ‘fair to exceptionally poor’ rock (jointed zones and weakness zones).

Rock support

Xiaolangdi powerhouse

The crown of the powerhouse was initially designed to be supported by 6m and 8m rockbolts at 1.5m spacing and a 200mm thick reinforced shotcrete layer. During the tender period, several clay intercalations were found above and in the roof of the powerhouse in an exploratory shaft. It was feared that these intercalations might open during or after construction, so eight rows of 25m long, 1500kN tendon anchors were introduced with a pattern of 4.5m by 6m. The sidewalls of the powerhouse are supported by 6m and 10m rockbolts at 1.5m spacing.

Ertan powerhouse

For rock of ‘very good to fair’ quality, the following rock support was proposed: the crowns are supported by 6m and 8m rockbolts at 1.5m spacing and the sidewalls by 5-8m rockbolts at 1.5m spacing, combined with 15-20m tendon anchors, tensioned to 1750kN, spaced at 4.5m in the powerhouse and 6m in the surge chamber. Steel fibre or wire mesh reinforced shotcrete with a thickness of 70-150mm is applied in most places.

When the top heading of the powerhouse arch was completed in April 1994, the rock seemed to be of good quality and it was decided to reduce the tendon anchors in the powerhouse by 50% and by 60% in the surge chamber. However, from June 1994, rockbursts and stress related problems started to occur in the caverns with increasing frequency. More tendon anchors had to be installed in the lower parts of the sidewalls and the pillars between the caverns. Chain linked mesh and cable bolts with steel fibre shotcrete were applied at the upstream springline of the powerhouse and the surge chamber affected by rockbursts.

Instrumentation

Instrumentation at Xiaolangdi

No instruments were installed immediately after crown excavation. During bench excavation, convergence measuring instruments were installed instead of the designed multi-point extensometers. In spite of late installation, the results still give useful information about the deformation pattern. Maximum displacements at the crown occurred after heading excavation, while displacements at the sidewalls developed with benching (Fig 4). These observations conform well with the predictions from the UDEC analyses carried out in 19965. Maximum displacements in the powerhouse are 3mm at the crown and 18mm at the sidewalls. When the top heading is excavated, displacement at the crown reaches its maximum value and remains the same during the benching operation. However, displacement at the downstream crane beam increases with bench excavation. Contrary to the deformation, the stress concentration at the crown increases with the benching, up to approximately 4MPa after excavation.

&#8220In-situ stresses have a strong impact on underground cavern stability either in a positive or a negative way”

Instrumentation at Ertan

The instrumentation system consists of 20-28m long multi-point extensometers and rockbolt stress meters. Most instruments were installed immediately after excavation. Displacements at the crowns of the three caverns are very small, normally less than 2mm with a maximum of 13.3mm. However, those at the sidewalls are much larger. Larger displacements occurred at the upper and central parts of the upstream walls and at the lowest part of the downstream wall in the powerhouse (Table 1). Deformation development at the powerhouse sidewalls can be correlated clearly with the bench excavations. When a 6m bench is excavated, displacements usually increase about 10-20mm then stop after a period of creeping6. Stresses in the rockbolts are clearly related to deformation. Rockbolt meter readings show that induced rockbolt stresses are normally less than 15 % of yield capacity at the crowns, while those at the sidewalls may reach 30-60 %.

The authors believe that, for a powerhouse cavern with a height greater than the span, this is a normal deformation pattern, i.e., displacements at sidewalls are larger than at the crown, provided that the horizontal in-situ stress is greater than half of the vertical stress, which is normally so for powerhouse caverns.

Rockbursts at Ertan

When the top headings of the caverns were excavated, no rockbursts were reported. However, when benching proceeded and the height of the caverns increased, rockbursts and stress related problems started to occur with increasing frequency. The major burst events could be related to large bench blasts or earthquakes far away, and they normally happened in massive and competent granodiorite.

Several serious rockbursts were reported from 1994 to 1996. In September 1995, there was a major rockburst, the most severe so far, at the rock pillar between the transformer chamber and the powerhouse, causing great damage to reinforcement already installed. Rockburts continued for over nine hours, starting in the transformer chamber upstream wall and two busducts and continuing in the downstream wall of the powerhouse, ending with serious rockbursts in two draft tubes. Several vertical cracks in the concrete lining were found in busducts 1 and 2, with a maximum of 30-50mm, 5-7m from the downstream sidewall of the powerhouse9.

The areas affected have been strengthened by bolting and shotcreting: plain shotcrete or mesh reinforced shotcrete was replaced by steel fibre reinforced shotcrete. In the upstream springline of both the powerhouse and the surge chamber, more rockbolts were installed with big plates. Eleven metre long cablebolts, combined with chainlink mesh, were installed in parts of the crowns. At the lower downstream wall of the powerhouse, 14 tendon anchors, 3500kN and 25m long, were added.

Since the pillars between the caverns seemed to be highly stressed and fractured, more tendon anchors have been installed there: six in the pillar between powerhouse and transformer chamber and 21 in the pillar between the transformer chamber and the surge chamber, where large displacements were measured. Later inspections found that six tendon anchors at the sidewalls of the powerhouse and three at the upstream wall of the surge chamber failed due to large deformations. These anchors have been replaced.

Several numerical modelling analyses were carried out independently before and during construction. A 3D elastic FEM analysis shows that: stresses are concentrated up to approximately 50MPa or more at the upstream springline and downstream corners of the caverns; parts of the pillar in the surge chamber are stressed to approximately 70MPa; parts of the busducts and draft tubes are stressed to 60MPa; and bends in the draft tubes near the powerhouse are highly stressed to 110MPa5.

It can be seen that the modelling results correlate well with the burst occurrence in terms of which areas are prone to rockbursts. Stress concentration at the upstream springline increased as bench excavation proceeded, and affected areas also increased. The measurements indicate that the stress concentration at the upstream springline of the powerhouse is up to 52.7MPa.

The pillars between the caverns are already overstressed: up to 64.4MPa was measured in the pillar between transformer chamber and surge chamber. Furthermore, they are weakened by the busducts and the draft tubes, resulting in even higher stress concentration, causing rock cracks, rockbursts and large deformations. More attention should have been paid to the impact of the busducts and the draft tubes on pillar stability. For these areas and also for the pillar in the surge chamber, tendon anchors through the pillars with anchor plates on both sides may be the best solution.

Other studies show that the deformation at the sidewalls of the caverns also increased when benching was under way, implying that the tendon anchors at the upper part of the sidewalls should not be fully tensioned because subsequent deformation will bring the anchors to their full tension. If this had been taken into consideration, probably no anchors would have failed.

It was demonstrated in Ertan that steel fibre reinforced shotcrete combined with end anchored rockbolts with large plates were very effective in preventing rockbursts or spalling. End anchored rockbolts have been used as protection against rockbursts in tunnels and caverns in Norway since the 1960s. The concrete lining of the busducts at Ertan was cracked during benching and was very difficult to repair. It would have been better if a flexible support system of shotcrete and rockbolts had been used during benching and the busduct lining delayed until the situation had stabilised.

It is believed that the potential for rockbursts was built up by stress concentrations formed by redistribution of stress during excavation and triggered by large bench blasts or earthquakes. The most severe rockburst – in September 1995 – occurred after one stroke penetration of a 20m bench in the surge chamber and a distant earthquake. Large blasts should be limited to minimise the shock to the caverns.

Conclusions

  • In-situ stresses have a strong impact on underground cavern stability in a positive or a negative way

  • The amount of rock support should not be determined solely by rock quality but also, to a large extent, by the stress conditions

  • Monitoring data from Ertan demonstrate that both the deformation and the stress of rockbolts at the crowns are much less than those at the sidewalls

  • It is implied that the crowns are self-supporting arches

  • Numerical analyses are useful for assessing the stability of an underground cavern – e.g., which areas are prone to rockbursting or spalling and how deformation patterns develop during excavation.



Related Files
Location Map
Dam Project Layout
Perspective View of the Ertan Hydropower Project
General Layout
Deformation diagram