After significant leakage was observed at Morocco’s Hassan Addakhil dam, research work was undertaken to discover the source of the problem
The Hassan Addakhil dam is situated at Foum Rhour on the Ziz river, about 10km from Errachidia city in Morocco’s Tafilalet province. The purpose of the dam is to regularise the wadi in order to carry out the Ziz valley agricultural and Tafilalet lands development, as well as to provide flood protection.
Construction work on the dam was conducted from May 1968 to March 1971, and filling of the reservoir took place during November 1970. After filling began, significant leakage flow was observed downstream, in particular at the auxiliary dam (saddle dam) ‘C’. This paper chronicles the treatment of this leakage problem.
Works and foundations
The project (Figure 1) consists of the main dam, the auxiliary dams A and C, the spillway, the outlet bottom and the irrigation galleries.
The main dam is an impervious clay core embankment. Tightness is provided by the sealing, on banks at the valley bottom of the core to the impervious red marls layers which ensure the reservoir tightness. The filters ensure the necessary transition between the core and the upstream and downstream shells.
Drainage is ensured by an interceptor drain connected to the core protection filters downstream. The two filter-drain layers are turned over to the downstream drainage blanket, of which the higher part is protected by another filter layer. This blanket allows the evacuation of the infiltrations through the core. The dam crest width is 800m and its crest length is 740m. Maximum height is 85m.
West of the main dam, two auxiliary dams were constructed (A and C), whose bodies with roughly the same constitution as the main dam.
The auxiliary dam C has a crest length of 1270m and a maximum height of 22m. The auxiliary dam A was developed with a crest of 135m and a height of 17m.
The spillway is located west of the main dam between the auxiliary dams A and C. This spillway constitutes a 43m long weir, levelled to 1122.20NGM, and continuing with a chute ending in a flip bucket. A channel excavated in the rock assures the out-flow of water toward the Ziz natural bed.
An update of hydrologic studies based on years of observations showed the necessity of a supplementary spillway. This spillway was executed after the first dam filling.
Geologically, the region of the dam and the reservoir is situated in the limit of the High Atlas. At the Foum Rhiour on this side of the mountain, the High Atlas Jurassic folds yield cretaceous layers. The site chosen for the dam is situated to the entry of a gorge dug by the Ziz wadi, through an asymmetric anticlinal of limestones and Jurassic marls. The same succession of limestone and marls surrounds the reservoir and appears on the surface to a level where continental sandstone and marls cut it up while enveloping the extremity of the plunging.
The most impervious section of this succession of marls and limestones, which interstratifies presenting an upstream slope of about 20º, has been chosen to constitute the support of the impervious core of the dam.
During the first filling in 1970, after that the Hassan Addakhil dam reservoir level exceeded 1088NGM, water resurgences appeared in the Chaâba downstream of the spillway (the Chaâba is a small Ziz right bank tributary, which drains infiltration water at the auxiliary dams downstream). In 1972, the reservoir water level elevation reached 1101NGM and new resurgences were observed in the Chaâba, but in the high part, to the auxiliary dam C downstream.
The different resurgences appeared in horizons situated stratigraphitely below the ‘impervious layer’, which is connected to the grout curtain of all works except that of dam C (apart from its left bank end).
In October 1976, the reservoir water level elevation reached 1113.32NGM for the first time. The leakages, for the whole of the collected and measured sources in a spillway downstream, then reached 342 l/sec.
To discover the origin of these significant leakages, twelve double piezometers were implemented in 1977, partly in the left bank abutment and partly at the downstream toe of auxiliary dam C. These piezometers highlighted the presence of a nappe in the sandstone formations surmounting limestones. This rather homogeneous nappe was established to about 7m under the reservoir level and was directly in relation with it.
Recently the reservoir level exceeded 1110NGM again and some significant leakages were observed mainly at the downstream toe of auxiliary dam C, where was a small lake formed.
This situation preoccupied the owner and particular vigilance by the operating staff was requested in the event of a rise of the water level. In particular, new potential source observation near the auxiliary dam C, resurgences on the downstream slope and water turbidity visual control were recommended. The owner also executed different studies in order to appreciate the consequences of leakage on dam C’s stability and the measures for avoiding any incident.
A widened study of various existing documents was carried out. This study gave a better idea of the geology and complex hydrogeologic structures of the Foum Rhour site.
The site geology is presented as a succession of sedimentary rocks layers forming the southern side of a synclinal. The layers are inclined toward the upstream and right bank under dam C. The alternation of the layers is as follows (from the base to the top):
• 60m of karstic limestones known as the Aalénien, based on a green marl bench of Toarcien.
• 110m of alternations of the red brown marls, the limestones and the limestones-clay (often clearly green). It is in these alternations that the layer known as impervious layers is located, in the higher half. Its roof is 25m under the limestones with a thickness of 13 to 14m.
• About 50m of limestones (Dogger).
• Continentals Red sandstones and clays (superior Jurassic lower Cretaceous) forming the reservoir bottom.
Under dam C and on the impervious layer exist a few limestones meters, which are probably very pervious.
Three types of leakage seem to be able to develop under dam C:
• Type one: At the core/foundation contact. On account of the dip on the left bank-right bank, the foundation was excavated in ‘teeth of saw’. In addition, it has been kept at about 500m of the left bank, an abrupt discontinuity zone at a superficial rocky bar. These discontinuities are likely to generate the tension zones in the core, especially as the setting up and the compaction of the materials are made more difficult.
• Type two: Above the impervious layer through or under the grout curtain, towards the synclinal SW, by the continental red layers or Dogger limestones.
• Type three: Below or in the impervious layer, on account of tightness defects of this layer or in karstic limestones, either directly or via the reservoir faults.
As regards to the works security, only types one and two have been considered of a practical interest. The deep losses do not have an influence on the security of the works themselves.
The monitoring device used at the dam includes piezometers, piezometric cells, tassometers, leakage measurements points and a topographic measurement network.
To study the problem, only the auxiliary dam C piezometers (Figure 2) and the measured leakages, which make it possible to provide information on the state of the critical zone, were used.
The follow-up of the monitoring was mainly centred on the permeability of the sandstone – when realising P1 to P and P13 piezometers, the water tests revealed total water losses almost everywhere.
The established piezometry, at about 80m on the downstream of the grout curtain at the level of auxiliary dam C, confirms the presence of a free flow nappe (the deep piezometers of the horizon 1080 indicate appreciably the same levels as the superficial piezometers of the horizon 1100). The synthesis diagram (Figure 3) shows that the head losses at the auxiliary dam downstream toe reach only 4 to 6m in the central third of the auxiliary dam. Towards the right bank (about 7m to P13), head losses are more significant. It is worth noting that for an average increase in the reservoir of 19m, the nappe raising is rigorously equivalent (piezometers P1B, P2B, P3B and P13B) and as soon as it is established, the nappe oscillates in a synchronous way with the reservoir level.
All of these results lead to doubts about the real efficiency of the grout curtain which probably constitutes only a minor obstacle in the flow upstream-downstream. This flow is established in the sandstone formations, constituting the three quarters right bank of the auxiliary dam C foundations.
In the left bank, the two piezometric profiles upstream-downstream (Figure 4) show that the ‘impervious layer’ on which the curtain is connected delimits in alternations two nappes, one under the ‘impervious layer’ and the other above.
Above the ‘impervious layer’, the natural ground upstream of the grout curtain already causes a head loss of 10 to 12m in comparison to the reservoir level when this level exceeds 1112NGM. The grout curtain presence causes a supplementary head loss of about 7m.
However it can be noted that the high levels of piezometer P10 (levels to 1105), are comparable to those of the P6 piezometer located at the grout curtain upstream. Under ‘the impervious layer’ (P7 to P9 piezometer at the left bank abutment), the nappe appears slightly fed. And its maximum level did not exceed 1091.5 NGM (the ‘impervious layer’ base).
The only plausible explanation of these lower levels towards left bank lies in the foundation configuration and the grout curtain under the auxiliary dam core. Indeed, in the first 200m from the left bank, the grout curtain is founded on the ‘impervious layer’, then on the supplementary 150m the curtain is connected to alternations and to the Dogger limestones. Finally, only a curtain 25m deep is found in the sandstone layers.
It should be noted that all formations described above are bathed by a reservoir. The auxiliary dam C is the only work of the project where different formations pass from the upstream to the downstream without being interrupted by a pervious curtain, natural (impervious layer) or artificial (grout curtain).
It can be said that the leakages were foreseeable, but their importance was not, or could not be, evaluated at the time of the design. The various leakages that appeared in the small valley of the auxiliary dam C are measured locally near the points of resurgence, the ‘higher source of the Chaâba’ at about 200m downstream of the auxiliary dam, the S2 to S4 sources correspondent to low neckline points or the rupture of ‘impervious layer’. Finally, the total of these leakages is controlled with a rectangular reservoir which bars the small valley of Chaâba at the spillway downstream, where the first leakages appeared (in May 1971), coming from karstic limestones subjacent to the alternations.
The leakage flow follows an exponential law that corresponds to the forecasting since headwater, saturated foundation surface and flow path sections, all increase together. The extrapolation to Normal Level (RN) 1122,20 (Figure 5) revealed a probable flow from 500 to 600l/s, that is a flow almost 50% higher than that recorded at the beginning of February 1990.
The increase in leakage and in the foundation uplifts can compromise stability in three ways:
• The stability of the auxiliary dam’s downstream slope can be endangered by the increase of the pore pressure existing in the downstream shell, or in a foundation layer sufficiently close to the auxiliary dam to create a potential rupture area.
• The development of significant uplift in layers near the surface, at the auxiliary dam downstream or near to the abatements of the main dam.
• Finally, the regressive erosion can take place in fine soils, when a high hydraulic gradient causes flow velocities able to involve the fine particles (piping).
The first risk type was quantified by the control stability analysis, which showed that the auxiliary dam stability remains insufficient. For the two other instability types, particular interest was paid to the piezometer values, to the water path, and to the soil quality to appreciate the risk level.
To acknowledge and guard against the incurred risks, a series of operating instructions were established for the operating staff and a series of measures (immediate and long-term) were considered.
The drainage of the work downstream
In principle, this is the most effective solution since it removes the instability issues. However it does not solve the problem of the leakage, and may even increase it.
For the drains to work at a maximum efficiency, it is necessary to have a sufficient quantity, and close enough to permit a real decompression of the constituted superficial foundation in part of fine materials with low permeability. But even though these drains are drilled in great number, they will always present the risk of not being completely efficient. Moreover, those already in place didn’t have much effect on the hydraulic grade. Drainage wells are more expensive and need a relatively long time to be executed. The best solution would be the execution of draining trenches perpendicularly to the auxiliary dam axis.
These trenches also work better at the foundation ground. A geotextile would be placed like a filter and the trenches filled with clean gravel, with draining properties. Some trenches would be completed by a sump in order to increase their efficiency and to measure the leakage flows. The trenches not provided with a sump can be connected to those with one by horizontal collectors. At the time of the trench’s realisation, it is necessary to be able to intervene quickly if met with a permeable layer with an important water flow. It is also necessary to hold in reserve the geotextiles and the gravel, which can be used when needed.
It will also be necessary to avoid leaving an open trench and each trench will be closed again before opening the following. The best method of excavation consists of opening sections from the downstream to upstream, by closing and protecting each section before the attack of the following. This work must be completed by the piezometer installation, intended to control the trenches efficiency.
Creation of an embankment at the auxiliary dam downstream
This operation can be undertaken alone or as a complement to the solution of the drainage if adopted, and after the end of its realisation the measured uplifts would not be lower than the ground level. An adequate weighting material will be retained, having the resistance properties of auto-drainage, guaranteeing the compacted material stability and permeability. The compacted material will be separated from the ground by a geotextile on which it will pose a sand and gravel protective layer 30cm thick.
The embankment volume must be able to generate compressive stresses in the foundation, even though the uplifts reach the normal water level.
Lowering of the reservoir water level
It is important to determine, according to the stored possible flood in the reservoir, the safe water elevation level in the reservoir at which it is necessary to decrease. This value can be reached in a few days, if the bottom outlet is entirely opened without creating problems downstream.
A good drainage of the foundation permits lowering of the pressures downstream and an increase of the auxiliary dam’s safety level. However, the drainage risk may not be sufficient in the case of piping.
Before any treatment, there is a necessity to undertake the most complete possible hydrogeologic study to define the volume of the losses and the origin of the leakages in the reservoir.
In the case of grouting curtain reinforcement, it will be necessary to drill through the upstream core.
Before the realisation of reinforcement work, the following instructions were indicated to the dam operators. The auxiliary dam C security is not assured anymore if the measured pressure in the downstream piezometers in the high measuring chambers exceeds 1110NGM. It is necessary, therefore, to lower the reservoir level if the uplift reaches this value.
A second situation deserves an attentive examination: the water turbidity. To be immediately informed of a change in this area, it was suggested to pass a part of spring water of the higher Chaâba through a geotextile pilot filter. This filter replaced at the regular period would be weighed before and after use (after drying) to detect any solid deposits.
Finally, it was desirable to limit the total volume of the acceptable losses. It was obviously difficult to fix a rational leakages limit. The value of 500l/sec was retained like the limit for the measured total flow at the main spillway.
It was necessary to analyse the evolution of every resurgence according to the reservoir level and to react quickly if a rupture was observed (in the sense of an abrupt increase).
A solution that consists of draining the foundation at the auxiliary dam downstream and laying out an auto-draining weighting material at its downstream toe, will:
• Lower the nappe level and decompress the downstream superficial foundation.
• Generate compressive stresses in the foundation with the stabilising weighting material, even though the uplifts reached the normal reservoir level.
It is clear that this solution will not solve the significant flow problem of the leakages, but it nevertheless increase the auxiliary dam’s stability. The confrontment works finally executed, relative to this solution, are:
• Setting up of a stabilising embankment auto-draining whose crest is fixed at 1114.00NGM. The crest width of this berm is 35m and its downstream slope is 1 V/2 H.
• Raising the existing piezometers.
• Realisation of two draining trenches, one skirting the embankment downstream toe LB-RB and the other being transverse at the auxiliary dam axis.
• Execution of a drainage line at the trench LB-RB, formed by the vertical drains 6m deep.
Hamza RouguiI – Hydraulic engineer, Rabat, Morocco
Mohamed Arjouane – Geologist engineer, Rabat, Morocco