An Engineering Geological Appraisal of the Leakage Problem in Dora-1 Earthen Dam, Tigray: Implications for its Stability

Leakage is one of the major problems facing the functionality and sustainability of dams. It occurs through the embankment body, reservoir, foundation, and abutments. This study was conducted to identify the main causes of the leakage problem at the Dora-1 dam, located in the northern part of Ethiopia. It is an earthfill dam with a height of 43.5 m, crest length of 454 m, and reservoir capacity of 4.67 million cubic meters. Part of the embankment body was wet and swampy up to 20 m high from the ground due to leaking water. Geological investigation, laboratory test of the construction materials (including grain size analysis, specific gravity and water absorption, Atterberg limit, free swell, dispersion, permeability, and shear strength), and electrical resistivity investigation were used to identify and pinpoint the possible causes of the leakage problem. Results of the study show that the favorable geological features responsible for the occurrence of leakage include: (a) geological contact between sandstone and moderately to highly weathered basalt unit at the left abutment, (b) the gravelly sand deposit at the central foundation and (c) dyke outcrop at the river course within the reservoir running in the upstream-downstream direction. Results of laboratory tests for clay core show medium to high compressibility, good to poor workability, and semi-pervious to impervious permeability when compacted. The water absorption and the percentage finer of the filter material don’t satisfy the filter criteria and the shell material was found to be semi-previous. The anomalous in the resistivity survey result confirms the situation. Slope stability analysis of the embankment showed instability conditions at full reservoir level. Close follow-up and a downstream stabilization structure, including rock and gravel support, were recommended.


INTRODUCTION
The Federal Government of Ethiopia initiated a rural development strategy, with the main objective of attaining food security at household level, and bringing about an overall socioeconomic development (Haregeweyn et al., 2006).Thus, a number of dams have been constructed to overcome the recurrent drought, which is caused by highly erratic and variable (spatial and temporal) rainfall.However, most of the dams didn't bring continuous economic growth due to technical challenges faced after the construction of the porjects.This is due to, among others, the reason that most dams are constructed without detailed engineering geological investigations.It was reported by Berhane et al. (2016a); and Abdulkadir (2009) 1.1.Regional Geological Setting The regional geology of the Dora area is part of the northern province of Ethiopia which is characterized by rocks of varied composition, which include Mesozoic sedimentary units, volcanics and Quaternary deposits.The northern Ethiopia is mainly underlain by weakly metamorphosed rocks capped by sedimentary and volcanic rocks.After this work, several authors start to investigate the geology of Ethiopia in general and northern Ethiopia in particular (Alene et al., 2000;Alene et al., 2006;Beyth, 1972;Bossellini et al., 1997;Dow et al., 1971;Gebreyohannes et al., 2010;Kazmin, 1972;Tadesse, 1996;Tadesse et al., 1999;Teklay et al., 1998).According to these authors, northern Ethiopia is covered by variety of rocks, varying in age and genesis.Based on stratigraphic sequence, the northern Ethiopian stratigraphy is classified (from the youngest to the oldest) as flood basalt volcanics, Amba-Aradam Formation, Antalo Supersequence, Transitional Unit, Adigrat Sandstone Formation, Enticho Sandstone and Edaga Arbi Tillites, and Upper Complex Metamorphic (Basement) rocks.

Description of the Site
Dora dam is located in the southeastern administrative zone of Tigray Regional State, particularly in Hintalo-Wajerat Woreda (district) near Adi-Keyh and Debub towns (Fig 1).It is found at 82 km south of Mekelle City across Weynat River.Adi-Keyh town (at 80 km from Mekelle City) is accessible through the Mekelle−Addis Ababa asphalt road.The specific dam site is accessible through a 2 km gravel road.The physiography of the dam site and its vicinity varies from an elevation of 2100 to 2800 m above mean sea level.It is the result of both volcano−tectonic and surface (erosion) processes.The area drains toward Raya valley and Afar basin.The salient features of the dam are presented in table 1.
Table 1.Salient features of the Dora-1 dam.

METHODOLOGY
In general, field descriptions of soil, rock and discontinuities, and laboratory tests were carried out according to the methods suggested by the American Society for Testing and Materials (ASTM) and the International Society for Rock Mechanics (ISRM, 1981(ISRM, , 2007)).

Field Study
Before the start of fieldwork, literature review was conducted from existing published and A total of 10 VES, five along the dam crest and the other five at the berm parallel to the dam axis, were conducted with Schlumberger array.In addition, two electrical resistivity profiling, one at the dam crest and one at the berm, were conducted with Wenner array.For the VES "AB/2" ranges from 45 to 100m, while for the profiling survey "a" varies from 10 to 45m and the suvey length was 330m for profile one and 260m for profile two.Direction of survey or AB direction was in the northeast-sourthwest direction along the direction of the dam axis.The VES data were analyzed by IPI2win software (Bobachev, 2002).Primary data were collected from direct geological investigation of the dam site, reservoir and from laboratory tests.Close observation was conducted at the foundation, abutments, reservoir, embankment body, reservoir rim and the spillway parts.Since the dam was already constructed data were collected from downstream and upstream toe of the dam and its surroundings and interpretation of VES and resistivity profiling survey data.Five test pits with a depth that varies from 2 to 3m were excavated at the downsteam toe of the dam for visiul inspection.Schmidt hammer or rebound test (L-type) was carried out to estimate the strength of intact rocks found near the reservoir area.The Schmidt hammer rebound data collection and interpretation were conducted as per the ISRM suggested methods (Aydin, 2008) and ASTM D5873 procedures.Moreover, weathering description and classification of rock mass was executed as per the guidelines and suggestions of BS 5930:1981 (Anon, 1981).

Laboratory Tests and Procedures
The construction materials for the dam were obtained from borrow areas near the dam site.
The purpose of the sampling was to evaluate the borrow materials and samples were disturbed to simulate the specific purposes.Earth and MS Excel).

Geology
The geology of the Dora-1 dam catchment area comprises of Quaternary Deposit, Basalt and Sandstone rock units (Fig 2 ).

Quaternary Deposit
It covers flat to gently sloping areas, found at downstream part of the dam, at the reservoir area and at upstream area near the river.It consists of alluvial and colluvial sediments.The colluvial sediment consists of angular pebble to cobble size basaltic rock fragments, and sandstone fragments of boulder size.It shows variable thickness, degree of compactness, texture and color.The maximum thickness reaches up to 7m at the site.The alluvial deposit is exposed along the stream courses, and it consists of fluvial origin of rounded to sub-rounded pebbles and boulders, and poorly sorted sediments of clay, silt, sand, and pebbles.

Basalt
This unit is exposed almost in all directions and most parts of the catchment.The dominant varieties are aphanitic and porphyritic basalt.The porphyritic basalt is characterized by porphyritic texture.It has mineral phenocrysts of olivine, pyroxene, and plagioclase.

Sandstone
This unit is found in the elevated areas forming rugged terrain.In areas where the sandstone is exposed at the elevated areas the capping volcanic rocks are eroded away and subjected to faulting and folding..This shows existence tectonic activity and surficial geological processes in the past.The sandstone unit is characterized by medium to coarse grained, white in color, rounded to sub-rounded, and commonly composed of quartz conglomerate.
Figure 2. Geological map of Dora-1 dam and its catchment area.

Engineering Geology of the Dam Site and the Reservoir
Dora-1 dam is one of the dams located in Tigray Regional State which was constructed without an adequate geological and geotechnical investigations.In this research, the rock masses found at the abutment, reservoir rim and the spillway parts of the dam site were characterized based on their origin, degree of weathering, spacing and orientations of the discontinuities and the intact rock strength.

Central Dam Foundation
The dam is founded at a river valley on alluvial deposits and residual soils which are underlain by slightly weathered basalt.The central foundation of the dam site is covered by river channel fill deposits of gravelly sand (TWRB, 2013).As it is already described in

Dam Abutments
The right abutment is covered by a moderately to highly weathered and fractured basalt unit In left abutment basalt and sandstone units are exposed.The contact between the two units runs across the dam axis.Leaking water was observed during field observation along this area.This geologic feature is found to be favorable for water leakage from the reservoir.
According to TWRB (2013), the recommended depth of excavation for the foundation was shallow, only 5m at the abutments and 9m at the central foundation.There was no concrete data regarding the actual construction or excavation depth at different part of the foundation.
It was observed that the depth of weathering is beyond a depth of 5 m at the abutments and their surrounding.The strike of the dykes at the spillway and the geologic contact at the left abutment is presented in figure 5.

Reservoir Geology
During the research work, the reservoir area was found to be partially filled by water (Fig 6a) and by sediment derived from the catchment.From the TWRB (2013) report, the reservoir area is mainly covered by alluvial deposit and residual soil.From the field observations, the alluvial deposit found at the downstream of the embankment is a mixture of sand and gravel with minor cobble to boulder sizes.In addition, a massive and late coming basaltic dyke is

Geophysical Investigation on the Embankment Body
In embankment dams, resistivity surveying is important to detect potential weaknesses zones, anomalous seepage and to verify the safety of the dam (Sjodahl, 2006).The resistivity survey of Dora-1 dam was carried out by TWRB (2013).The leakage problem was experienced during construciotn phase when reservoir was started to impound water while the height of the dam reached 37m.Electrical resistivity raw data containing the profiling and VES measurements were collected from the bureau.This data were analyzed and interpreted by integrating with the new engineering geological map and other data collected during the present research.
Figure 7. Engineering geological map of Dora-1 dam reservoir area and its environs with cross-section along the direction of the dam axis.

Vertical Electrical Sounding (VES)
VES 1 to 5 were conducted at the crest of the dam and VES 6 to 10 at the berm of the dam.

Electrical Resistivity Profiling (ERP)
The two profiles were aligned in the NE-SW direction (left abutment to right abutment).The first profile (ERP1) which was conducted at the crest of the dam with a=15, 25, 35, and 45 m, while the second profile (ERP2) was executed at the berm of the dam with a=10 and 20 m.The field photograph in (Fig 9b) shows the contact zone.The contact between the rock units crosses the dam axis.Discontinuities that intersect the dam axis, dipping toward downstream are the most promising features for leakage of water from reservoirs.So, the contact at left abutment of Dora-1 dam could be one contributing cause for the occurrence of leakage.

Geotechnical Condition of the Embankment Body
Leakage is severed problem at this project to the extent that water is leaking directly through the embankment body.Portion of the embankment at the downstream face up to a height of 20 m was found to be saturated and wet (swampy) when the reservoir water level was at around 25 m.When the reservoir level was dropped the extent of the wet area was also dropped.Generally, as observed in the field the height of wet portion of the embankment varies with the fluctuation of the reservoir water level.Such phenomenon and observations are clear indications that the safety of the embankment is under question.The danger signs of

S.
No.

Grain Size Analysis
The grain size analysis results for the filter, shell and core materials are presented in (Fig 12).
The primary purpose of a filter material is to provide a safe drainage and prevent migration of fine particles from the protected zone.To achieve these purposes filter criteria is checked and fulfilled.The Terzaghi criterion ( < 4) addresses the retention requirement.For the present research, the test result was 0.55, which satisfies the criteria.For the second Terzaghi criterion ( > 4) the result was 56, which satisfies the criteria.Other requirements in the Terzaghi criterion for a good filter are: ✓ The gradation curve of the filter material should be approximately parallel to the gradation curve of the protected soil (core material), especially in the finer range.This criterion is satisfied as can be seen in figure 12.
✓ Filters should not contain more than 5% fines and fines should be cohesionless to ensure that the filter is pervious and prevent cracking.This criterion is not satisfied.Because, the percentage of fines of the filter material ranges from 5.8 to 9.3%.
✓ The filter should not contain particles larger than 75 mm to minimize segregation.This criterion is satisfied.

B) Sherard's Criterion (Sherard and Dunnigan, 1989)
According to this criterion, the filter is considered as good filter if it fulfills the following.

I.
< 9, is satisfied because the value for this ratio is found to be 0.55.

II.
The coefficient of permeability (cm/s) of dense filter should be in the range of k = 0.2 to 0.6 (D15) 2 (mm) with average value of 0.35 (D15) 2 mm.
For this criterion the permeability test results of the filter material are summarized in table 3.
Table 3. Evaluation of the filter material as per the Sherard's permeability criterion.The permeability of filter material is categorized in to pervious as per USBR (1998).
However, all of the samples were not with in the recommended range of Sherard's permeability criterion (Table 3).The filter material doesn't fulfill all the quality requirements.
The percentage of fines (particles passing #200 sieve 0.075mm) obtained in the laboratory test ranges from 5.8−9.3% which is above the permissible value.According to ASTM C 33 (1999), the fines content of fine aggregate should be less than 3 percent.The presence of higher amount of fines can result clogging of the filters and drains.In addition, the filter doesn't fulfill the second (II) criteria of Sherard.by fair to poor in their shear strength when saturated, highly compressible, and poor workability as a construction material.So, these soils are considered unsuitable material as impervious core.

Specific Gravity and Water Absorption
Water absorption gives an idea about the strength/durability of aggregates.Aggregates having high water absorption are more porous in nature and are considered unsuitable.According to ASTM C128 (2012); and Kosmata et al. (2003), the absorption value for fine aggregates should not be more than 2%.The water absorption of the filter materials of the present research resulted in the range of 3.72 to 4.25%, while the specific gravity was about 2.9 (Table 5).

Free Swell and Linear Shrinkage
According to Holtz and Gibbs (1956), soils are grouped in to non-expansive, intermediate and expansive soils based on their degree of expansiveness.Swell value of less than 50% are non-expansive, 50−100% are considered as intermediate and greater than 100% are grouped in to expansive soils.The free swell value of the core material ranges between 65 &78.4% (Table 5), and for the shell material it ranges from 50 to 60%.All the results fall in the intermediate class.Intermediate soils are not considered as sever, but such soils could be potential for some problems.

Atterberg Limit
Atterberg limit tests were conducted for the core and shell materials show low liquid limit.
Test results are summerised in table 5 and in the plasticity chart (Fig 13).
Figure 13.Classification of the core and shell materials using plasticity chart.

Dispersivity of Core
Bell and Maud (1994) cited in Fell et al. (2014) indicated that, soils with > 50% dispersion are regarded as highly dispersive, 30-50% moderately dispersive, 30 to 15% slightly dispersive, and < 15% non-dispersive.For the present research all the core materials fall under the non-dispersive soil type (Table 4).The permeability results from moulded samples are summarized in table 5.According to USBR (1998), soils with permeability value <110 -6 cm/s are impervious, 110 -6 cm/s to 110 -4 cm/s) are semi pervious and > (110 -4 cm/s) are pervious.The permeability results for the filter material fall in the previous and both the core and shell materials fall in the semi previous.Generally, the shell material is supposed to be pervious or at least it permeability should be different from that of core material.Otherwise, the embankment shall be homogenous types.This is not the case for the present dam.

Shear Strength Test Results
In this research, unconfined compression strength test (UCS) and direct shear test were conducted for the core and shell materials.The shear parameters were used in the slope stability analysis (Table 5).

Stability Analysis of the Embankment Slope
The stability analysis was conducted for three selected reservoir water levels: at a full level of the reservoir (40.5 m), at 25 m (the maximum reserve water level that the dam was reached in the years of 2017 and 2018) and at 20 m.For the analysis the embankment fill material's shear strength parameters were used as input.
From the analysis results, the slip surfaces and the calculated factor of safety for the reservoir level of 25 m and 20 m are presented in figure 15, and for the full reservoir water The dam is therefore, unstable at a full reservoir water level.

CONCLUSION AND RECOMMENDATION
The engineering characteristics of the rocks/soils and geological structures at the dam foundation, reservoir area, and the abutments favor for the occurrences of leakage.Water is leaking along the contact between sandstone and moderately to highly weathered basalt units at the left abutment.The foundation and reservoir of the dam site is covered by gravelly sand and loose residual soils.Field observations and analysed data showed that part of the embankment in the downstream side is wet and saturated.With respect to the quality of the construction materials, some of the embankment materials were found unsuitable (filter with high percentage of fines, permeability of core and shell materials).The permeability of the core material falls under semi-previous category.The water absorption (above 2%) and the percentage of fines of the filter material don't satisfy the filter criteria.The presence of higher amount of fines can result clogging of the filters and drains and can result pore water pressure.The resistivity results showed strong variations for the same embankment material.
This variation was interpreted as the presence of leakage triggering features, such as defects, layering of the embankment fill material due to compaction differences during construction.
From the stability analysis conducted the factor of safety at a full reservoir level was 0.835.
The dam is, therefore, unstable at a full reservoir water level.A layer of clean compacted gravelly sand/rock which can act as a drain is recommended at the downstream toe of the ), the deposit is about 6 to 7m thick, pervious, non-plastic and very loose.Subsurface water flow was observed at the test pits in the central foundation (Fig3).

Figure 3
Figure 3. a) Subsurface water inflow into the test pit and pumping out.b) Foundaton materials are very pervious and loose.Photographs are taken from TWRB (2013).
overlaid by residual soil.The spillway is located at the right abutment side.At the right side of the spillway excavation, moderately to highly weathered and fractured basalt rock unit is exposed.This unit is cross-cut by a dyke dipping at an angle of 54° to the upstream direction(Figs 4a & b).The geology at the left side of the excavation is not the same with the right side of the excavation.The left side comprises shale, moderately to highly weathered basalt and pyroclastic rock units stratified one over the other, and is dipping at an angle of 36° to the downstream direction(Figs 4c & d).Two dykes which have width of 0.8m each and spaced at 46m are exposed at the left side of the excavation.The dykes are massive and dipping to the upstream direction(Figs 4a & b), so they can act as barriers.

Figure 4 .
Figure 4. Rock units and structures exposed at the spillway, a) Dyke at the left of the spillway excavation, b) Highly to completely weathered basalt overlaying pyroclastic bed, c) Dyke at the right of the spillway excavation, d) Shale rock unit at left of the spillway excavation.All photographs taken by the second author in 2018.

Figure 5 .
Figure 5. Rose diagram showing the strike of the 3 dykes (shaded plot) found at the spillway area and the geologic contact (solid line) at the left abutment in relation to the direction of the dam axis (broken line arrows pointing towards the center color of the circle. observed at the river course in the reservoir area(Fig 6b)  which extends towards the right abutment(TWRB, 2013).The basaltic unit trends toward the downstream part.This geological feature (contact between the dyke and surrounding basaltic rock) could be a possible leakage path.The basalt rock unit is moderately to highly weathered and the discontinuities are closely−spaced (Fig6c).The rim of the reservoir is comprised of basalt and sandstone rock units.At the upstream side, the basalt is covered with residual soils and the sandstone unit is found exposed at the steep slope.The sandstone unit at the left abutment side was excavated at steep angle (nearly vertical).This can lead to sliding of rock blocks in to the reservoir during full impoundment.The sandstone unit is slightly weathered and is characterized by the presence of moderately−spaced discontinuities (Fig6d).Simplified engineering geological map of the dam site and reservoir area is depicted in figure7.

Figure 6
Figure 6.a) Over view of the rim of the reservoir, b) Basaltic dyke at the upstream river course of the reservoir (photograph taken and modified from TWRB (2013), c) moderately to highly weathered basalt at the rim of the reservoir, and d) Fractured sandstone unit at the rim of the reservoir.All photographs except b taken by the second author in 2018.
The individual VES data gives information about the vertical resistivity variation of the subsurface beneath the VES point.To interpolate the data throughout the section between the VES points, the resistivity data were analyzed and presented in terms of apparent resistivity pseudo-sections constructed along selected VES points.Interpretations were made based on dam profile, field observation and resistivity pseudo-sections.The low resistivity values within the embankment body and natural ground were interpreted as the presence of defects, compaction difference during construction and weak zone.The survey for the pseudo-section (Fig8) was conducted along the center of the dam crest.The center of the dam crest is the core material.The resistivity variation detected within the single material (core material) could be interpreted as the result of presence of defects, compaction difference, and moisture variations.The relatively low resistivity value (<19.31m) found at a horizontal distance of 25 m infers to be the contact zone between the embankment body and the natural ground and the presence of geological defects.In the section starting from the horizontal distance of 150 m the resistivity decreases gradually from the top up to a depth of penetration by the current electrode spacing (AB/2) of 10 m, and again increases from AB/2 of 10 m up to AB/2 of 45 m.This variation in resistivity for a single zone (core material) indicates the presence of layering or stratification of the core fill material as a result of construction deficiencies (poor bondage between successive lift) and/or the presence of defects that could enhance leakage through the embankment body.

Figure 8 .
Figure 8. Apparent resistivity pseudo-sections (Ohm-meters) along VES1, VES2, VES3, VES4 and VES5 points at the dam crest, and VES6, VES7, VES8, VES9 and VES10 points at the dam berm starting from the left abutment to the direction of the right abutment.

Figure 9
Figure 9. a) Field photo illustrates the wet portion of the embankment body (double arrow); b) Field photograph showing the contact between the basalt and sandstone units at the left abutment.Solid line shows the trend of the contact.Part of the reservoir (water) is visible in upper left side of the photograph.Photographs were taken by the second author in 2018.

Figure 10 .
Figure 10.Graphical plot of the electrical resistivity profiling a) along the dam crest (profile one) and b) along the dam berm (profile two).The broken lines indicate interpreted as geological defects and variation in degree of compaction during construction.
Figure 11.Photographs showing leaking water (a) and erosion features (b) at the downstream face of the embankment.Arrow indicates leaking (flowing) water from the embankment body.Photographs were taken by the second author in 2018.

Figure 12 .
Figure 12.Grain size distribution curve of the embankment fill material.
Shell1 and Shell2 of the shell material are classified as SM (Silty sand with gravel) and sample Shell3 is classified as GM (Silty gravel with sand).As it is portrayed in the grain size distribution curve(Fig 12), the shell material is consist of higher amount particles finer than #200 sieves (0.075mm) and it ranges from 43.5−49.3%.Obviously, the presence of higher amount of fine particles in the shell material can also lead to migration of fines from the shell material in to the filter material when water seeps in to the filter.This can result clogging of the filter.The purpose of shell material is to support the embankment (stability), and it should be pervious.But, actually the shell material used in constrcting the dam has excess fines content, this can result clogging of the filter which in turn can result pore pressure and triggers leakage via the embankment and detorirate its stabilty.The soil classification indicates that sample Core1 of the core material classified as CL and samples Core2 and Core3 as MH.As per USBR (1998), CL (Inorganic clays of low to medium plasticity) soils are ranked as 3 rd based on the relative desirability of the soil as core material for embankment (suitable material).But, the MH (Sandy elastic silt) soils are characterized

(
40.5m) in figure14.The slope stability analysis was conducted using SLOPE/W model, one of the GeoStudio-2007 software products.The calculated factor of safety for the reservoir levels of 20 m and 25 m are 2.162 and 1.909 respectively.This shows stable condition of the embankment.But, the most critical condition of embankment occurs if the reservoir is fully impounded.The minimum factor of safety obtained at a full reservoir level is 0.835(Fig 14).

Figure 14 .
Figure 14.Stability analysis results for a reservoir level of (a) 20 m, (b) 25 m; and (c) for a full reservoir water level.
unpublished documents, journals and books related to dam and leakage problem.Secondary data, including engineering geological reports of the dam, electrical resistivity data (profiling and vertical electrical sounding, VES) and dam design documents were collected from TWRB.The geophysical survey of the Dora-1 dam was carried out when the leakage problem was appeared after construction completion in 2014.The location for the starting and end points for Profile one are 0572126mE; 1440560mN and 0571858mE; 1440364mN respectively.Similarly for Profile two the points are 0572188mE; 1440509mN and 0571979mE; 1440386mN.The leakage problem was experienced starting from the first reservoir filling.In this research, geophysics was used to detect leakage causing geological features.The electrical resistivity survey was carried out at the crest and the berm.The height of the berm from the lowest river is about 24m.

Table 2 .
Dora-1 based on the U.S. Society on Dams (2011) guidelines are listed in table 2. In addition, field photographs that show the danger signs are presented in figures 9 and 11.List of danger signs for an embankment dam (U.S. Society on Dams, 2011) and their occurrences in Dora-1 dam.

Table 4 .
Summary of the double (dispersion) hydrometer test results.

Table 5 .
Classification of the soils used as construction material based on their index properties as per the Unified Soil Classification System (USCS).