Meeting challenges in Ethiopia10 August 2010
The Beles II and Gilgel Gibe II hydro projects in Ethiopia have brought varied tunnelling challenges that were successfully overcome by TBM manufacturer and subcontractor Seli. Report by Patrick Reynolds
In the last few years, four TBMs supplied by Italian manufacturer and tunnelling subcontractor Seli were used to traverse varied, and often complex, geology in Ethiopia to construct the long tunnels for two major hydropower schemes – Beles Multipurpose (‘Beles II’) and Gilgel Gibe II. The projects called for a range of equipment, including one of the most advanced shields being used for weak ground, as anticipated at Beles II, but it was a Gilgel Gibe II that the most difficult challenges were faced when a double shield TBM unexpectedly encountered an extremely difficult fault zone.
Seli has worked on a number of hydro projects among the many sectors it has served over its 60 years of operations, which it celebrates this year. In Ethiopia, as often elsewhere, it was able to draw upon its experience as a tunnel subcontractor as well as innovative equipment provider to have its crews and engineers plan ahead and address the challenges to be faced, such as the known weak stretch of ground at Beles II, but also work to cope with and overcome unexpected difficulties when met, such as on Gilgel Gibe II.
The developer on each project is the Ethiopian Electric Power Corp (EEPCo), the main contractor was Salini Costruttori and the tunnelling subcontractor – and equipment supplier – was Seli.
Located in Amhara region in north west Ethiopia, the Beles II scheme takes water from Lake Tana along a 12km long headrace tunnel to an underground powerhouse of 460MW (4 x 115MW, Francis units) installed capacity.
The plant discharges to a 7.2km long tailrace which conveys the flow to the river Jehana, a tributary of the Beles. The underground works on the project also include a 270m long penstock shaft and a 90m high surge shaft.
Beles II - Headrace
Seli selected a dual mode TBM to operate either as a double shield universal (DSU) or an earth pressure balance (EPB) machine to deal with the varied geology along the alignment of the headrace. The majority of the drive – approximately 10km – was in basalt up to UCS 350MPa with some local faults, but the 8.1m diameter TBM also had to bore through more than 1.8km of loose soils, described as lake deposits.
The cutterhead was equipped with 52 x 17” backloading and recessed discs for the hard rock drive, and switchover to EPB excavation mode took a few weeks to perform on the machine, one of Seli’s most advanced. The TBM, equipped to deal with squeezing ground and to undertake face treatment, cost about 15% more than a standard DSU TBM.
The tunnel is lined with 7.2m i.d concrete rings, each 1.5m long, formed of 300mm thick segmental concrete segments (6+key).
The TBM was launched in late 2006 and made reasonable progress, having bored about two-thirds of the tunnel after two years. Before then, and just after the midway point in the basalt drive, there was a large, blocky face collapse and recovery work called for polymer resins from Innotek to fill the void and consolidate the rock mass, restoring the ground for the drive to continue.
By early 2009 the TBM had switched excavation mode and was driving through the lake deposits on the final stretch of its journey. Boring through the lake deposits, the EPB-DSU holed through in November last year.
Beles II - Tailrace
Geology along the tailrace was reasonably good basalt with some stretches of agglomerates and tuffites, and overburden was up to 120m in areas, and 250m-370m elsewhere. About 17 major discontinuities, mostly vertical and obtuse to the alignment, were expected and would present locally poorer ground conditions.
Seli used an 8.07m diameter double shield TBM for the tailrace drive, which was bored from the outlet end of the tunnel towards the powerhouse. Like the headrace tunnel TBM, the tailrace machine’s cutterhead was equipped with 52 x 17” backloading and recessed discs. The cutterhead drive had a maximum torque of 5,250kNm, and the maximum design advance rate was 6m/h.
To test the ground ahead of the TBM, the machine is also fitted for probe drilling and 18 holes through which to extend the drills.
The tunnel lining for the tailrace was the same as for the headrace tunnel, and the annular gap filled with 8mm-12mm peagravel and grout.
The TBM was launched in mid-2007 and completed its drive just less than a year later, in May 2008, and had advanced at an average of approximately 20m/day, achieving a best day of 36m (24 rings) and best week of 189m (126 rings).
Tunnelling operations on the tailrace were performed in 3 x 8hr shifts, 7 days per week with scheduled daily maintenance in the mornings, service extensions and probe drilling when required. In total, the efficient tunnelling system had 51.9% of the time in excavation and 23.4% in maintenance. Where there was time lost a prime reason was the blocking effect of the pea gravel from the quarry-derived, angular aggregate because no local fluvial sources of gravel were available.
With the headrace drive by far the more difficult compared to the tailrace on Beles II, the Seli tunnellers found the unexpectedly greater technical challenge came on Gilgel Gibe II, underway at the same time.
Gilgel Gibe II
Gilgel Gibe II is a 428MW (4 x 107MW, Pelton units) scheme. The headrace tunnel is smaller diameter but, at 26km, is far longer than that of Beles II.
Geology anticipated along the alignment included five main rocks – tertiary volcanics, rhyolite, trachyte, basalt and some dykes. As hard basalt was anticipated for the bores, the TBMs selected for the headrace drives were two 6.98m diameter DSUs that would bore from either end of the tunnel, on Inlet and Outlet drives.
The tunnel lining is 1.6m long concrete rings of 6.8m outside diameter and formed of four hexagonal, 250mm thick honeycomb segments.
In the end, the TBMs encountered 24 geological difficulties, or classified, ‘events’, of varying scale and nature on the headrace excavations – 15 in the Inlet drive and nine in the Outlet drive. By far the most difficult to overcome was Event 19, on the Inlet drive.
Gilgel Gibe II – Inlet Drive
The drives started well and advanced 10m-20m per day despite the almost continuous presence of weaker rock and ravelling faces. Probes were drilled up to 50m ahead of the TBMs, From early on, though, it was the Inlet drive that was being more hampered by geological problems – more than three times as many events to deal with even before Event 19 struck.
Chemical grouting was employed for local face stabilisation and some bypass digs were required to handle the early events. Overcutting helped to counter squeezing ground. The geological events caused hold-ups ranging from a few days to about a month. But when Event 19 occurred, in October 2006, about 14 months and just over 4km into the Inlet drive, it was to take more than 20 times as long to resolve.
The TBM had already stopped when the problem emerged that would be classified as Event 19. There was a sudden extrusion and collapse of the face against the TBM, the progressive creep of the rock mass and crushing force against the shield bent and broke parts of the TBM, pushing it back more than 600mm and displaced it laterally by more than 400mm. Probing ahead, with some difficulty, hot and high pressure mud was found – 40 degrees Centrigrade and up to 35 bar- 40 bar pressure.
Efforts were made first to open a small tunnel over the TBM to start to release the shield but high rock pressures prevented success. Next, a Back Chamber was to be opened up fully around the TBM and also an exploratory adit was excavated to the left from which boreholes would be drilled to probe the weak rock mass nearer the fault. However, the mud broke into the adit and overtopped bulkhead to reach and partly bury the TBM in the main tunnel. Two further surges of mud soon followed, and measurements showed they caused rapid reduction in acting pressure.
The tunnellers proceeded by opening up an adit to the right from which exploratory galleries were formed and much longer boreholes drilled to reach into and beyond the fault. A programme of drainage helped to lower the pressure in the rock mass around the TBM head and the partly constructed Back Chamber around the shield. The adit was then turned towards the front of the TBM to create space for the insitu recovery and repair work but could not continue due to loads and deformation rising, as a consequence, in the Back Chamber.
Eventually, without further excavation and the Back Chamber completed, the TBM could be fully accessed and was dismantled by early 2008. Removed to the surface, it was refurbished, rebuilt and re-assembled, and the cutterhead diameter was increased, to 7.07m, to help the TBM reduce the squeezing effects from rock mass with plastic behaviour. The shield was assembled again underground in a new chamber, located about 400m back from the site of Event 19. The intervening distance had been backfilled with concrete.
The TBM was relaunched in August 2008 on a bypass route around the fault, through which resin injection was used to help consolidate the clayey basalt and to limit squeezing. Beyond the fault the TBM rejoined the headrace alignment.
The time and difficulty in overcoming the problem of Event 19 meant that the Outlet drive did much more of the overall excavation than originally planned, and the Inlet drive would only bore the same distance again beyond the fault zone. The TBMs completed the headrace excavation in the middle of last year.
With the four TBM bores finished on Beles II and Gilgel Gibe II, and the machines long gone, there is one remaining challenge at the latter project. Some months after the operations began at the hydro scheme a blockage was found in a short section of the headrace, on what was the Outlet drive. The investigations have revealed the cause to be a rockfall at a zone of undetected, weak ground above and well behind the tunnel lining. It is believed the weak rock mass, comprising some loose large blocks in weak soil, was agitated by hydraulic pressure variations induced by the active headrace. Recovery work is well underway and the repairs to the tunnel should be completed soon.