In-situ shear strength testing of conglomerate formation for a large RCC dam (Koysha, Ethiopia)

Koysha Dam, located in the Southern Nations, Nationalities and Peoples’ Region of Ethiopia, is the fourth plant of the Gibe-Omo cascade comprising Gilgel Gibe (IP=200 MW), Gibe II (IP=420 MW) and Gibe III (1’870 MW) all in operation. The plant includes a 200 m high RCC gravity dam, a gated spillway on the left bank and an open-air powerhouse housing 6 Francis turbines fed by 2 steel penstocks crossing the dam body. The right abutment of the dam is partly founded on conglomerate, composed by a variable proportion of cobble and gravel sized basalt and rhyolite-trachyte sub-rounded elements surrounded by a weak matrix of fine sand and silt particles. The geotechnical characterization of such a complex material resulted in a very challenging task due to the substantial impracticality of collecting high quality, undisturbed and representative samples. Therefore, the mechanical behaviour of the conglomerate has been assessed by means of in-situ large scale shear tests carried out into a 70 m long inspection tunnel. The test results show that in the range of selected confinement pressures (0.3 to 1.0 MPa) the conglomerate exhibits a softening behaviour. Peak friction angle and cohesion are coherent with data and empirical models presented in the literature which indicate a strong correlation between volumetric block proportion and mechanical properties of the Block-In-Matrix (BIM) rock.


Introduction
Koysha dam is located on the Omo river, 130 km downstream of Gibe III dam in the Southern Nations, Nationalities and Peoples' Region of Ethiopia. The project includes a 200 m high RCC gravity dam, a gated spillway on the left bank and an open-air powerhouse housing 6 Francis turbines fed by 2 steel penstocks crossing the dam body. The owner of the project is the Ethiopian Electric Power Company (EEPC) whereas the main contractor is WeBuild SpA (Italy). Koysha is the 4 th plant on the Omo river, downstream of the operating Gilgel Gibe, Gibe II and Gibe III. With its 1620 MW of installed power and 6260 GWh of annual energy production, Koysha is one of the most important projects in the Ethiopian Government's commitment to meet the country's present and future power requirements.
The project presents many geotechnical challenges, one of the most important being the presence of heterogeneous Conglomerate outcropping on the right dam abutment. The Conglomerate is a melange formation composed by unweathered to moderately weathered gravel and pebble clasts into a silty and sandy matrix. From a geotechnical point of view, the Conglomerate represents a typical block-in-matrix rock (bimrock) [1], [2] and [3]. These geological mixtures exhibit a complex behaviour with considerable spatial, lithological and mechanical variability and common fundamental engineering problems related to their geotechnical characterization [4] and [5]. IOP

Geological and geo-structural framework
At broad scale, the project area is located in a physiographical region known as the "western plateau". This region was affected by Cenozoic volcanism triggered by a mantle superplume uplift whose decompression melting generated massive quantities of basaltic magma in the lithosphere and resulted in the formation of a flood basalt province.
The site area consists almost entirely of volcanic rocks composed by Andesitic and Basalt Andesitic flows (A-Formation) that form the lower part of the stratigraphic sequence. This unit belongs to a prerift phase and is constituted by a continuous layered sequence of sub-aerial lava flows some meters thick, dipping towards West at low angle. According to its textural characteristics the A-Formation is divided into a brecciated (B1) and massive member (B2). Both members formed mainly during the quenching of the lava flows. No unconformities in the stratigraphic succession were observed except for decomposed to highly weathered horizons formed after some break during the lava emplacement.
The A-Formation is capped by the Conglomerate Formation (C) on the right bank and by the Columnar basalt Formation (Bc) on the left bank, which constitutes the top of the plateau on both the banks. The plateaux were part of a single tabular structure subsequently dissected by the river erosion that followed the upheaval of the Ethiopian region.
The geo-structural setting of the project area is characterized by a series of tectonic features whose direction is consistent with the one of the main faults at regional scale (20-30°N) that are related to the development of the East African Rift System (EARS) creating a horst-graben tectonic style with system of basins and ranges bounded by faults with a general tilting to the W.

Conglomerate formation
The right abutment of the dam is characterized by three main lithological units (figure 1): • Upper Basalt formation (Bc) at the top of the right abutment.
• Conglomerate formation (C) which overlies unconformably the Andesite unit (A) and is capped by the columnar basalt (Bc).   gravel and pebble clast supported conglomerate mainly formed by basalt and rhyolite-trachyte rounded elements. In the boreholes, it occurs as a cemented, strong to moderately strong rock. The matrix is predominantly composed by sand and silt in varying proportions. Locally the conglomerate is interbedded by decimetric plastic clayey levels.
• SILTSTONE AND SANDSTONE (Cs) (8 % of C-Formation) -These two rock types appear as a brown and moderately strong rock. Rock cores have generally a low RQD value (< 45%). Cs levels are placed at the top of the conglomerate formation or as intercalations.
The typical sedimentary succession observed on site is illustrated in figure 2. The Cs lenses outline the following main characteristics: • A lateral downlap geometry in the foreset portion of the fluvial bar.
• A paleocurrent indicating a roughly right to left direction truncated by subsequent channel in a higher fluvial energy period. • An average thickness generally less than 50 cm (locally up to 1 m).
The above characteristics indicate the absence of an upstream to downstream continuity of the Cs levels, due to the braided depositional geometry at time.     [6], [7] and [8].
The tests were carried out inside a 70 m long inspection gallery excavated in the Conglomerate Formation below the ground water level as to ensure near-fully saturated conditions of the material. The testing apparatus is reported in figure 4.  The following testing procedure has been adopted: 1. PREPARATION: each block is carefully cut to the required dimensions (800 × 800 × 400 mm) avoiding disturbance and loosening of the material. The block is then encapsulated into a reinforced concrete pad isolated from the ground by means of a polystyrene panel. 2. CONSOLIDATION: the normal load is gradually increased up to the full value determined for the test by means of a vertical hydraulic jack and a steel section helping to distribute uniformly the load to the test block. In this phase, the pore water pressure in the block dissipates under full normal stress before the shear load is applied. The consequent normal displacement is recorded. The consolidation phase is considered completed when the rate of change of normal displacement recorded at each gauge is less than 0.05 mm in 10 minutes. In all tests, the applied normal stress ranges in the interval n = 0. A typical shear stress-shear displacement curve recorded during the in-situ shear tests is reported in figure 5. The normal displacement measured during testing is also shown. The test blocks show a clear dilatant behaviour with well-identified peak and residual strength. The normal displacement decreases in the first phase of the tests and then increases following the change in volume of the test block. A photograph of the upper portion of the block after testing is reported in figure 6 along with the measured shearing surface. The GE-RCC/Rock shear tests have been carried out about 60 days after the casting of the GE-RCC blocks, applying the same procedure used for Conglomerate blocks.

Test results
Test results have been elaborated according to the procedure reported in [6] and [7]. Shear  and normal n stresses are computed as follows:  Table 2. It is highlighted that the results of 2 tests have been discarded from the determination of the peak strength parameters due to the following reasons: • Irregularities occurred during test execution (tests were interrupted and restarted).
• Odd shape of the failure surface showing a marked downward profile in the direction of shearing.  Regarding the GE-RCC blocks, test results reported in figure 8 show that the cohesion ranges between 110 and 430 kPa, with the highest values observed for the mix with higher cement content. The friction angle appears less dependent on the cement content and varies between 42° and 43.5°, close to that of the Conglomerate blocks. As already mentioned, these values have been measured 60 days after GE-RCC casting. According to [9], an increase of the GE-RCC/Conglomerate contact shear-strength should be expected over time.

Comparison with literature data
A comparison of test results and literature data has been done to confirm the coherence of the measured shear strength parameters. The Conglomerate peak cohesion and friction angle determined via in-situ testing are in good agreement with the empirical methodology proposed by [10] for the determination of the strength parameters of bimrocks based on the volumetric block proportion (VBP) and the strength parameters of the matrix (figure 9). According to [1] and [11], the dimension of the smallest significant block in a bimrock depends on the scale of the problem via the relation Lb = 0.05 Lc (Lb = dimension of the smallest block, Lc = characteristic dimension). Considering the dimensions of the shear-test Conglomerate samples, this corresponds indicatively to blocks of 30÷50 mm. Therefore, based on the granulometric distributions reported in figure 3, one can assume an average VBP of about 30% ÷ 40%.
In the selected range of VBP, the predicted shear strength parameters determined for  = 25° (angle of repose of blocks) and A = 3.0 (model parameter according to [10]) are bimrock = 37°÷43° and cbimrock = 110 kPa ÷ 140 kPa, which are coherent with the in-situ test results. If a smaller VBP is considered, a reduction of the predicted friction angle and an increase of the cohesion are observed with respect to the predicted values (figures 10a and 10b).