THE 188m high Miel I dam – the highest roller compacted concrete (RCC) dam in the word built to date – was put into operation in April 2002. Located on La Miel river in Colombia, the dam is part of the Miel I hydroelectric project. Approximately 1.76Mm3 of RCC was used during the construction of the 340m long structure.

Geotechnical instrumentation

Planning of the geotechnical instrumentation for the dam followed a logical and simple process, first defining the purpose of the instrumentation as well as the important parameters that must be measured, and then selecting the types and characteristics of the instruments required to achieve such an objective. Basic steps followed in this process included:

• Evaluation of the geological and geotechnical conditions at the site: after removal of surface residual soils and colluvium, the dam foundation is composed mainly of hard and sound micaceous gneiss with excellent bearing capacity.

• Type of structure, construction methods, and special requirements during operation: RCC gravity dam, placed in 0.3m thick layers. This step, in combination with the one above, allowed for the identification of the mechanisms that will control the structure’s behaviour, and the possibility of evaluating RCC conditions, such as stresses and strains, as well as strength and thermal properties that could have an effect on dam behaviour.

• Analytical evaluation of structure behaviour: this involved a series of two and three-dimensional analysis by means of the Finite Element Method (FEM). Several types of analysis were performed including incremental, thermal-gravitational, static and dynamic analysis.

• Analytical evaluation of failure mechanisms to control sliding, overturning, and bearing capacity failure at the dam foundation, and excessive seepage. The analysis performed allowed the dam behaviour to be predicted and determined under different load conditions such as static, thermal and dynamic, as well as allowing the identification of geotechnical and structural parameters that require a follow-up by means of the instrumentation. At the same time they were used to establish the expected variation rates of each of these parameters, and the type and characteristics of the most adequate instruments required to monitor and evaluate the dam behaviour.

• Considering the individual nature of the dam, a very important step in the process included the research of past experiences in projects with similar characteristics. This step allowed for a comparison between the analytical results obtained with the observed behaviour in other similar structures which had instrumentation. In particular, it allowed for the optimisation regarding types and quantity of instruments, and instrumentation locations.

Throughout this process several important questions were raised, regarding both geotechnical and structural behaviour of the dam. To provide adequate answers to these questions requires an instrumentation program specifically designed to monitor and evaluate the different parameters that control behaviour of an RCC dam of this magnitude during the different project stages, in particular, during construction, reservoir filling and operation. A basic step taken during the development of this program was to identify possible cautionary measures or remedial actions to be taken in case instrumentation results deviate from the preset allowable ranges – see Table 1.

Monitoring behaviour

Based on the results of the process described above, the following were identified as important parameters to be monitored: total displacements of the structure, and relative displacements within the structure itself; temperature, stress and strain (cracking) status; water pressure and seepage; and earthquake related ground motions.

Structure displacement

Results of the analytical modelling performed for Miel I dam, for the several failure mechanisms evaluated, showed that the following needed to be monitored: total dam displacement during earthquakes, along the contact RCC rock; relative displacements between RCC blocks during earthquakes – design properties between layers used for these analysis are angle of friction between layers of 45º, cohesion C=10kg/cm2, and unit weight of 2.5 ton/m3.

For both cases, allowable displacement ranges were established based on the results of the analysis for the design base earthquake (DBE) and maximum credible earthquake (MCE).

Temperature

The dam is zoned with four mix types, with a variable cement content between 85 and 150kg per m3 of concrete. In an RCC dam, the adiabatic temperature change is one of the most important parameters to be monitored, considering its influence on the behaviour of the dam. The thermal decrease and deformations or volumetric changes induced in the dam body are associated to two different phenomena: 1) The initial cooling phase, where the concrete is still fresh, affects most severely the dam faces, crest, foundation and abutments, where the generation of high and sudden tensile stresses could induce unallowable deformations; 2) The slow cooling of the RCC over the years also induces tensile stresses which can affect the dam body. However, because this phenomenon is occurring while the concrete is gaining strength and has a greater modulus of deformation, its effect is expected to be less significant.

Based on laboratory tests with the different RCC mixes to be used, Table 1 shows the criteria established to control and reduce cracking in the dam to allowable levels.

As described later a series of strategically located thermocouples were installed across the dam for temperature monitoring.

Stresses and strains

A combination from the results of the stress, strain and temperature analysis showed several zones of high stress concentration, both compressive and tensile. Although high, the magnitude of the compressive stresses expected was judged not critical for the dam. On the contrary, development of tensile stresses were considered critical to the security and behaviour of the structure. This parameter will be monitored by means of the installation of pressure cells and strain gages located across the dam body. From laboratory testing, the following strength parameters were established for the different mix types to be used: 85, 100, 125, and 150 kg/m3.

Water pressure and seepage

The design of the waterproofing system for the dam called for 20m deep grout curtains, complemented with a series of drainage holes both along the dam’s foundation and its abutments, and within the dam’s body. The target is to reduce permeability beneath the dam to values of 10-7m/sec and water seepage pressures to less than 1/3 of its original value in the vicinity of the drainage galleries and zero at the dam toe. Water pressures will be monitored using piezometers installed along the foundation and within the dam body.

To control water seepage, a novel system was designed for Miel I in order to protect the dam’s upstream face, consisting of: a 0.4 to 0.5m thick strip of grout-enriched RCC (GE-RCC) across the full depth of the upstream face; a PVC waterproofing membrane placed directly over the GE-RCC strip.

In addition to the above, a reduction in water seepage through the dam was achieved through the use of four types of RCC mixes used (each with different cement contents), as well as with a series of drainage galleries to be constructed at different levels within the dam, as shown in figure 2.

Water seepage will be monitored with strategically located weirs and Parshall flumes.

Seismicity

The project is located in a medium-intensity seismic zone, as defined by the NSR-98 (Colombian Seismic Code). Dynamic analysis was performed both for the DBE, with a maximum acceleration of 0.22g, and the MCE, with a maximum acceleration of 0.30g. Results showed that the most unfavourable condition associated with dynamic loading, which can induce important tensile stresses and strains within the dam’s body, occurs during a seismic event with the reservoir to its full capacity. The need to monitor the dam in such a case is therefore of outmost importance.

Instrument selection and location

The following aspects were taken into consideration for the selection of the instruments:

• Parameters to monitor

• Range, accuracy and sensor type. Operating range and accuracy of each instrument were selected based on the evaluation of the results of the different analysis performed, including sensitivity analysis carried out to determine the significance that changes in each of these parameters could have in dam behaviour.

• Instrument technology. To allow for automatic data acquisition, make it more timely, and at the same time optimise the number of reading units required, it was decided to unify instrument technology, selecting, whenever possible, vibrating wire (VW) sensors, which can provide the accuracy and cover the ranges desired for the different parameters to be monitored.

Location of each of the instruments was based on the critical control areas shown by the stress distribution, deformation and thermal analysis. However, another very important aspect considered was the need to ease wire handling and diminish the interference with the construction process. As shown in the typical cross-section in Figure 2, this was accomplished by establishing six instrumentation levels, all of these coinciding with the preset levels of the drainage galleries located within the dam body, through which cables were installed to connect to the dataloggers.

Instruments were installed on five levels (except the lower) along three cross-sectional areas, one near each abutment, and one near the dam’s axis. At some specifically selected stations, clusters of instruments were installed, typically including pressure cells and strain gages, allowing for a very precise control of the stress-strain relationship within the RCC blocks. Also, some of the pressure cells, strain gages and VW piezometers included a thermistor, which will not only serve for additional temperature control and monitoring, but can be used as a backup system, to validate information obtained from the thermocouples.

Crackmeters

A total of 127 crackmeters were proposed to monitor relative displacements between RCC blocks. As mentioned above, joint opening must be monitored to control the effects of RCC cooling, both short and long term, as well as relative movements induced by the more critical dynamic earthquake loads. About 120 crackmeters were located over the contraction joints, along each of the six vertical instrumentation levels established, near the upstream face of the dam which, in this regard, is more critical than the downstream face. The remaining crackmeters were installed at the longitudinal joint in the centre of the dam. At most of the control points, a ‘rosette’ of three orthogonally placed crackmeters (X-river direction, Y-dam direction, Z-vertical direction) was installed.

Multiple Point Extensometers (MPE)

Possible vertical and lateral movements within the rock along the abutments and below the foundation must be monitored. For this purpose the instrumentation programme designed calls for a total of four MPEs, each with six sensors. Two of the MPEs are to be located at the dam foundation, following the spillway’s axis, and one along each of the abutments.

Thermocouples

The instrumentation programme designed for the dam needed 334 thermocouples. As mentioned, some of the other instruments installed have thermistors, as a complement to the information obtained from these thermocouples. Two main aspects were considered in order to determine the location of these instruments. Firstly, the need to monitor and control the temperature gradient at a specific point within the dam, and its variations from point to point at a specific elevation across the dam was considered. The other aspect was the fact that the dam was designed to be constructed with four different RCC mixes, each with a different thermal behaviour, and therefore requiring separate temperature monitoring. The areas close to the faces, particularly the upstream face and around the galleries, have the highest density of thermocouples. According to the thermal analysis performed, these areas will be exposed to the largest thermal gradients, and consequently have a greater potential to crack. Other points where temperature development was considered critical, and therefore required specific monitoring with strategically located thermocouples, include the conventional concrete block at the spillway ogee, and its contact with the RCC.

Pressure cells

Thirty-six pressure cells are to be installed in rosettes composed of three cells each (in the X, Y, and Z directions) in order to determine the state of stress at the upstream and downstream boundaries, as a result of temperature changes, static loads (weight of concrete, water), hydraulic pressures, and eventually dynamic loads. The pressure cells were located at points where critical compressive or tensile stress concentrations are expected, as shown by the different models and analysis performed, thus allowing for a comparison between the predicted and actual stresses at these critical locations.

Strain gages

At total of 36 strain gages were installed, most of them grouped at the locations selected for the pressure cells, thus allowing close monitoring of the stress-strain ratio. Strain gages were distributed across the upstream and downstream boundaries, which, based on the results of the analytical models, were considered as the most critical surfaces to be controlled. Similarly, these instruments were installed in rosettes along the orthogonal preset X, Y, and Z directions.

Vibrating wire and hydraulic (manometric) piezometers (VWPs)

These instruments were used to control the development of water pressure and the flow net. VWPs were installed both within the foundation and the abutment rock, as well as across the dam body. Hydraulic piezometers were installed only along the foundation and the abutments, to provide a better coverage of water seepage and pressure within the rock mass. As shown in Figure 2, in the foundation and the abutments, a total of 25 VWP sensors were installed and distributed along seven drill holes, with two to five sensors per hole. Additional coverage was provided through the installation of 46 hydraulic piezometers, placed along holes to be drilled from the galleries located at each of the six instrumentation levels. Within the dam body, 36 VWP sensors were installed, also distributed along each of the six instrumentation levels.

Frequency of measurements

During the construction phase, a weekly reading was established as the basic monitoring frequency for most instruments, except for thermocouples. Taking into account thermal gradient development, thermocouple readings will be made every 12 hours during the three days after their installation, every 24 hours for the following three days, and then on a weekly basis. As discussed previously, such monitoring frequency shall be modified in accordance with the rock-structure behaviour, based on actual instrumentation results, analysis and observations taken.

Automatic data acquisition system (adas)

The instrumentation programme designed for the Miel I dam involves a complete monitoring system with more than 620 sensors. The quantity of instruments and amount of information expected is such that automation of the process is a must in order to obtain timely, useful and precise knowledge of the entire instrumentation system, and consequently dam status. It is important to note that the use of an ADAS does not eliminate the need for qualified and experienced personnel to manage and analyse the information obtained from the instrumentation, as well as for the visual inspection and the maintenance of the entire instrumentation system.

One of the important aspects in the selection of the instruments to the Miel I dam was to unify the operation technology as much as possible, in order to incorporate them to the data acquisition system. Vibrating wire technology adapts well to such a requirement. The ADAS selected for Miel I is made of the following components:

• Multiplexers: These units enable many sensors to be connected rapidly and easily to the readout box. Taking into account the amount of instruments to be read, and that a maximum of 32 two-conductor sensors can be attached to each unit, a total of 27 multiplexers are required.

• Dataloggers: Actual readout of the instruments will be made through dataloggers. Being connected directly to the multiplexers, these units will allow continuous processing, storage, validation and transfer of data obtained from each instrument. Dataloggers will be provided with a real-time control system, to control date and time of readout, data transfer, and printout. Each datalogger can handle a maximum of six multiplexers. Considering the specific distribution of the instrumentation levels at Miel I, a total of six dataloggers will be required, each connected to two to five multiplexers. Data transfer between the dataloggers and the computer will be made via fiber optic cables, which provides maximum speed and signal quality, in what is an electronically noisy environment.

• Data Processing Unit: This processing unit corresponds to a computer, provided with the necessary capacity and features to allow an optimum and permanent performance of the automate system.

Results

Miel I dam is instrumented to enable performance monitoring. The dam is being extensively monitored with instruments installed at six different levels inside the body of the dam. Continuous information during construction has been recorded and analysed.

Electric piezometer readings are reported below 50 kPa; pressure cell readings are below 3.6MPa; and crackmeter readings are within the range of ± 2mm.

RCC placement temperatures ranged between 28°C and 31°C, with minimum and maximum values directly associated to ambient temperatures. Maximum recorded temperature inside the dam body has been 50°C for the 150kg/m3 mix, 44°C for 125kg/m3 mix and 39°C for the 100kg/m3; all of them were developed approximately at the third day, after RCC placement.

After one year of operation on the reservoir, total seepage through the dam body has been recorded below 2.5 litres/sec.


Author Info:

Pedro Leguízamo. Civil Engineer, PML, Geotechnical Engineering & Instrumentation Consultant, Cll 22 D 58-39 Int 4 – 203, Bogotá, Colombia. Tel: (157)-4846588. Email: pedro.leguizamo@gmail.com

Tables

Table 1