CFRD under pressure29 February 2008
The discovery of displacement and cracks at Campos Novos CFRD in Brazil shocked the professional dams community. Suzanne Pritchard reports what happened to the structure from the investigation findings that were given to delegates at a recent symposium
Extensive rupturing of the concrete face at Campos Novos dam in Brazil has been the focus of intense study. Similar events occurred at other CFRDs shortly after, during the period 2005-6, and drew attention to a phenomenon that had previously gone undetected. It was discovered that high compressive strains could be imposed on concrete face slabs due to an adverse combination of dam height, low rockfill deformation modulus and an unfavourable valley shape.
Understanding this phenomenon, the agents involved and evaluating the design of very high CFRD dams were among the objectives behind the 3rd Symposium on CFRDs in Brazil in October 2007. Several of the papers focused specifically on the experience gained at Campos Novos.
The discovery of longitudinal displacement and cracks on the concrete slabs at the dam in 2005 had taken the international field of CFRD professionals by surprise. Campos Novos was designed and built with the most up-to-date technology and strictly followed empirical design criteria used in hundreds of other CFRDs worldwide. The “inventor” of CFRD himself, J Barry Cooke, was consultant to the project in the early stages and was involved until his death in 2005.
Construction of the 202m high, 12M m3 dam began in October 2001 and was completed in February 2005. The rockfill embankment, originated from basalt rock, was divided into zones so that the upstream third and the central area of the dam material - denominated as 3B and 3D - were compacted in layers 1m thick, wetted at the rate of 200 l/m³. The voids factor of those zones was equal to 0.22 on average and the compacted unit weight is 2.14 tonnes/m³. The uniaxial compression tests showed that the resistance of rockfill material was over 75MPa. The downstream third, consisting of materials of the 3C/3D type, was compacted in layers of 1.6m without wetting; the average unit weight of this zone being 2.02 tonnes/m³. The transitions under the slab – materials 2B and 3A – were compacted in layers of 0.5m reaching void factors of about 0.20.
For protection of the processed transition and regulation for concreting of the upstream face, an extruded concrete face was implemented concurrently with the transitions, with an average cement rate of 75 kg/m³.
The rockfill embankment was implemented in three stages and the concreting of the slabs of the face in two stages. The rockfill embankment was first raised upstream and the slabs concreted up to El.568m, which corresponded to approximately 52% of the final height of the dam and flood recurrence period of 1:500 years.
The downstream rockfill embankment was then raised to El. 570m and later to El. 660m. Slab concreting was carried on after the completion of the rockfill embankment. The average production of compacted rockfill was 700,000m³/month, while the 16m wide slabs were implemented at an average speed of 2.9m/hr.
Initial filling of the reservoir began in October 2005 and the level rose 160m in 20 days. Instrumentation at the dam indicated that transversal deformations of the rockfill embankment were as expected. But on 24 October 2005 the dam experienced a rupture along joint 16-17, from El. 639m-El. 656m. The dam crest is at El. 666m.
The following day, a crack was observed on slab 15 in the horizontal construction joint of the second and third stages of construction. On 29 October, new displacements occurred on joints 17 and 18, increasing damage on the central joints. Subsequent repair works, mainly rebuilding of the damaged joints, were carried out above the water level.
However, during January and February 2006, further cracks were observed in slabs 27 and 28, along with a sudden and noticeable seepage through the dam. On 26 February, silt and fine sand were poured over the slabs and, as a result, the seepage was reduced substantially from 1350 l/s to 950 l/s. But when the reservoir was raised from El. 642m to El. 652m, the flow increased rapidly up to 1500 l/s. The reservoir was lowered immediately.
Unrelated problems were being experienced at one of the diversion tunnels, which necessitated drawing down the reservoir completely in June 2006. This revealed a 300m long transverse crack in the dam structure across the valley. Intensive spalling and deformation of rebars along the crack led to the conclusion that the rupture was due to high compressive stresses, now acting along the slab in a longitudinal direction.
The vertical joint rupture, which was first identified above the reservoir level, did not extend to the bottom of the valley in one single joint but migrated across the slabs to three successive neighbouring joints as it progressed downwards.
The slabs had moved horizontally, one against the other, and stopped after interpenetrating about 12cm. Spalling resulting from the shearing planes in the upper surface alternated with intense crushing of the concrete at the base of the slabs over the curb elements at different sections of the slabs.
Remedial action involved cleaning the concrete face and removing the ruptured concrete stretches and damaged joints. Detailed mapping of the affected zones was made to understand the movements of the face and the stress under which it was submitted. Even after the reservoir had been drawn down, excess stresses were still detected in the concrete where cracking had occurred, indicating it was still under residual stress.
To repair the damaged compression joints, the joints between slabs 16 and 20, located in the central area, were left open with a 5cm space between them. This freed up a space of 20cm which would enable the compressed slabs to move during the second filling of the reservoir. These open joints were filled with asphalt mastic. The joints were covered with an 8mm thick blanket of an elastomer (EPDM) and were tied to the slabs with a continuous galvanised foil and anchor.
After lowering the reservoir and refurbishing the central region of the dam face, new monitoring instrumentation was installed in the concrete to help assess performance in the areas damaged during the first filling. More than 30 instruments were installed in the dam, including: thirteen joint meters to monitor the closing of the central joints between slabs 16 to 20; seven strain gauges to check the level of stress at the zones affected during the first filling; Ten electro levels placed in the right abutment of the dam to assess the face deflection.
Following the repairs, reservoir impoundment started again in November 2006
Avoiding high leakage
Seepage flows during the second filling were still of a similar value to those recorded in the first impoundment. Consequently, sub-aquatic investigations were undertaken using divers and robots. While the checks verified that, in general, the condition of slabs was acceptable, there was still a small amount of damage in some stretches of the upper blanket that protects the open compression joints. These suction points allowed seepage through the open joints.
In the second filling, deformation measurements taken in the rockfill embankment and the concrete face were an average of 35%, which was an improvement on the readings at the first impoundment. Longitudinal deformation, which was deemed to be the main cause for cracks in the first filling, behaved differently this time in an upstream-to-downstream direction. Strain gauges installed where cracks and ruptures were evident in the first filling did not record compression above 1.5MPa.
Maximum deflection of the concrete slabs was much less than during the first filling and the rockfill was four times stiffer on reloading. The maximum operating level of the reservoir was achieved a year ago and the project has operated normally since. Leakage was stable at 1200 l/s. Although values of leakage in the order of 600 l/s to 2000 l/s do not represent a risk to the stability of the CFRD, Cruz and Pinto believe that important issues should be understanding the cause of leakage and continuous observation of how flow versus time proceeds during the life of the dam.
Flows through rock foundation are expected to be small and the measurement of flow through the foundation of the concrete structure, usually founded in the same rock mass, can be a good indication of the flow that seeps through the dam foundation.
Furthermore, Cruz and Pinto believe that the empirical solution related to the repairs carried out were not effective in reducing the flow to a low value with time. They question what is lacking in the design and construction of such CFRDs to avoid high leakage at the dams, and ask whether continuous observations should be recommended over the next few years.
Instrumentation at the Campos Novos dam was installed inside the rockfill embankment. It was only capable of evaluating deformation in horizontal and vertical directions, and these revealed ‘normal’ behaviour similar to CFRDs of the same size.
However, the cracks and ruptures that occurred in the face slab at the dam were caused by displacement in abutments to the centre of the valley. Such displacements cannot be measured inside the rockfill embankment with the usual sets of instrumentation installed at CFRDs. The valley shape in Campos Novos, with a ratio of 2:92 (L/H) directly influenced this type of behaviour.
Cruz and Pereira believe that a new type of instrument to measure longitudinal displacement at the axis of the dam would be of great value to evaluate the behaviour of the rockfill as a whole. This is a challenge for instrumentation specialists. They also believe that displacement forecasting using numerical methods should be developed in three dimensions to anticipate displacement to the centre of valley, which may or may not affect the performance of the slab.
This report was drawn from papers at the 3rd Symposium on CFRDs, in Brazil, in October 2007. For further information and conference proceedings, go to: www.cbdb.org.br
|Campos Novos dam - facts and figures|
The dam is located on the Canoas river in the south of Brazil. Its main characteristics are: