TO enhance its hydroelectric production capacity for the next century, the Azienda Energetica Metropolitana (AEM) of Torino, Italy, began building the 150MW Pont Ventoux-Susa hydroelectric plant in Piemonte in 1996 – a scheme designed to provide the area with an additional 457GWh/year.

For environmental reasons and to achieve the best functional scheme, most of the plant was built underground, inside the mountain between the two valleys of Monginevro and Moncenisio passes, with only the water intake, the reservoir and the delivery to the river outside.

The intake is located near Sauze d’Oulx village, in Pont Ventoux , on the Dora Riparia river at an altitude of 1046m asl. The diversion canal for the project is a 14km tunnel, with a gradient of about 1/1000, which supplies daily the Val Clarea reservoir.

This reservoir, with a 560,000m3 capacity, supplies a pressurised 4km tunnel, the upper surge tank and the 1.3km long penstock, which has an average diameter of 3.2m. At the foot of the penstock, at an altitude of 491m asl and a head of 515m, lies the power station, equipped with two Francis turbines, with a speed of 750rpm each, coupled to an 85MVA synchronous generator. After the downstream surge tank, the 1.6km long outlet tunnel returns the water into the Susa demodulation reservoir, which has a capacity of 420000m3.

One of the turbines in the power station is also coupled with a 13.5m3/sec pump, which transfers water from the Susa lower lake to the Clarea upper reservoir during the night.

The design and construction of the hydroelectric scheme was awarded to to Pont Ventoux Scrl, an association between Astaldi of Rome, as a main contractor, and Eiffage of Paris, France. Milan, Italy-based Alpina was appointed as engineering consultant for the design and underground works

Geology

The geological knowledge of the site came from extensive studies on the basis of deep drillings, surface surveys and analysis of geostructural surveys performed in the tunnels during excavation. The excavation geological data collection confirmed that the lithotypes found at the site consist essentially of phylladic schistose crystalline micaschists composed of calcite and mica or with decimetric to metric intercalations of marbles and silicates or, sometimes, with centrimetric to pluricentimetric intercalations of more or less quarzitic and chloritic gneisses. This meant that there are often gradual variations in composition, which are difficult to foresee in the rock mass.

The plant consists of about 22km of tunnels for hydraulic purpose and about 2.7km for transit.

Most of the hydraulic tunnels (about 15km of them) were bored by two 4.75m diameter Robbins TBMs (type 148-212-3, 35 cutters, total power of cutter head drive 895 kW, gripper system). The good quality rock mass allowed a successful TBM performance, which gained ground on a high average (15m/day) peaking at 40m/day.

Because of the rock quality, bolts were used as provisional supports, with steel arches set when passing through fault zones with bad geomechanical characteristics.

A careful monitoring of the TBM typical progress parameters allowed, after the very first tuning of their values, to achieve both a successful back analysis of the passed zones and a reliable prediction of the forthcoming ones.

The remaining 6.5km of hydraulic tunnels were excavated by drill and blast method due to:

• Geometrical and functional properties such as: high gradient, narrow bending radiuses, shortness of single sections; changing of internal geometry.

• Remarkable fault zones and bad quality of the rock mass. All hydraulic tunnels of round section had a concrete lining 25-35cm thick, cast by an iron framework for lengths of 36m.

In all pressurised hydraulic tunnels, liquid mortar injections structurally joined the concrete lining to the rock allowing them to work together to support hydraulic loads inside the tunnels.

A detailed modelling of structural interaction between rock and concrete showed that the traction working stresses inside the lining were allowable with no need for steel reinforcement. This achieved a large economical saving in comparison with the preliminary budget. However steel fibers were added to the concrete to avoid microfissures to the lining.

Where a large water head was found inside the rock mass, a drainage system was set between the concrete and the rock to lower the hydrostatic pressure to the load bearing capacity of the lining.

The vertical section of the penstock was blocked by concrete inside a vertical shaft 220m high, drilled by a raise borer. This excavation system allowed quick working progress.

The upper surge tank of the plant consists of a vertical cylinder 80m high, with an internal diameter of 12m. It was not possible to excavate the surge tank with a previous axial hole drilled by a raise borer; instead, the cylinder, with a diameter of 13.4m, was excavated by steps, 2.4m deep each, by drill and blast system, followed by casting the 0.7m thick reinforced concrete.

The transit tunnels have a U-shaped section (variable between 25m2 and 45m2) and they were all excavated with drill and blast system. The good quality of the rock mass and no other finishing requirements led, in most cases, to a single lining made of shotcrete reinforced by steel mesh.

Further underground works include:

• Two gradient surge tanks, one of which is 450m long.

• Three small caverns, two of which are for underground transformer seating, 2400m3 in volume each, and one for the butterfly valve seating, 2800m3 in volume.

• Four steep sloped tunnels, developing through variable routes connecting the turbines and the pump to the penstock.

• All the underground works including the overflow of the upper Clarea reservoir, its flow off drainage and off take to the pressure tunnel, consisting of several tunnels and caverns.

• A flow off drainage tunnel of the lower Susa reservoir, 49m2 in section, 340m long and designed for a maximum flow rate of 180m3/sec.

Power Cavern

Among all the underground works that make up the Pont Ventoux hydro plant, one of the most important structures is the power cavern, which is 50m long, 21m wide and 21-49m high, with the vault and walls self supported by pretensioned rebars.

The design of the cavern had to take into consideration the problems arising from its location and orientation inside the rock mass between the Cenischia Valley and Dora Riparia. In the site chosen, the foliation dip is between a nearly horizontal direction (10°) and a slightly sloping one (25°), in clockwise direction S-SW.

The joints and the faults detected are few and narrow. These discontinuities do not affect the geomechanical rockmass classification. Joints and faults are closed and without any water flow.

The rockmass appears generally closed, and its geomechanical average is between 2nd and 3rd Bieniawski class, in accordance with the found lythotype and the structural and mineralogical characteristics. As a result, it was decided to slightly move the location of the plant into a zone with better mechanical characteristics.

In July 1996 a feasibility study was undertaken, allowing official approval of the location of the power plant, and the start of the final design. The study for the final orientation decision was led by the construction of spectral curves representing the behaviour of the volume of the wedges as a function of the angle expressed in degrees with respect to the NS axis. From the analysis of the spectrum of the key blocks, a range of favourable orientations of the power plant was obtained and identified between 10° and 30° in a clockwise direction from the NS axis. The final value selected was 17°.

The preliminary design was a mushroom-shaped cavern supported by a heavy reinforced concrete arch. However, the new cavern location and the new assessments on geomechanical parameters lead the project engineers to re-design the cavern shape to minimise the rock supports. Possible cavern shapes, including egg and U-shape, were studied using hybrid FEM/BEM numerical analysis. It was found that the new shapes generated less decompressed zones and considerably reduced the tensile failure on the walls.

A U-shape was finally chosen for the cavern, due to ease of excavation and better use of space.

Support system

Before starting work, the length and spacing of rockbolts and rebars needed to be decided. Analyses of the wedge failures and the length of overstressed zones – including potential wedge identification and study of the database of wedges according to location and physical characteristics – were useful in designing the extension required for the supports.

The Alpina engineering team decided to create a double support system with regards to length and installation methods, to differentiate the action of the longer bars from the shorter ones.

The short bars, chemical resins injected with short setting times and low prestressing values, limit loosening of the rock during excavation, creating a cortical reinforced zone.

Tensioned bars, anticorrosion double coated and cement grout injected, are the final support system of the cavern. In both cases, Dywidag bars with high strength steel are used.

After the untensioned bolts installation, a 10cm shotcrete lining was applied to limit the small wedge failures. The support system is then completed by another 15cm lining of fiber-reinforced shotcrete.

The cavern excavation was carried out in subsequent stages by blasting. The excavation of the roof cavern took five stages with maximum opening of 8m length and 5m height. The benches cavern excavation followed the same method; a deepening slip road links the upper bench to the lower, for rock removal.

A monitoring system was used to compare computed and read values, and to monitor rock mass performance over time, to ensure expected values were not exceeded during construction.

The monitoring instrumentation included two sections of seven multiple-point borehole extensometers, with readings taken by remote data acquisition.

Back analysis

During the first excavation, the logical flow chart for the design under construction was outlined by deformation monitoring compared with expected values, adjustment to the design procedure and an under construction design logical flow chart.

The underevaluation of the displacement of the boundary computed by FEM and BEM models meant the design team had to use distinct elements models. The numerical method DEM was chosen as the best way to simulate the rockmass performance, if compared to the one computed by finite elements methods. The distinct element model, performed by UDEC and 3DEC numerical codes, allowed the study of the displacements at each excavation stage, together with the bolts and the shotcrete lining stresses. The spacing foliation is the joint generator (method) parameter used for six distinct element models. An automatic generator to create joint patterns based on statistical parameters is included in DEM codes. The main discontinuities were matched on the statistical joint distribution.

The back analysis models were developed in different zones for joint generation, strength, deformability and boundary condition. Both untensioned bolts and tensioned rebars were included in the numerical models. The standard Coulomb slip model was used to simulate the joint and main discontinuities.

All civil works on the cavern are now complete, as is excavation of the tunnels with the exception of the 14km long diversion tunnel. This had been split into two 7km long TBM drives with both Robbins machines going to face in 1997.

The first 7km section ran smoothly with a record rate of 40m recorded in one day. Three kilometres into the other 7km, the second TBM hit a fault zone and boring was suspended. Completion of the remaining 4km will utilise drill and blast from two faces and is scheduled to take one year.