Taking the empirical approach

29 February 2008



Practical experience is favoured over design theories in CFRD structures. At a recent symposium in Brazil, delegates heard of the latest experiences worldwide following problems in some dams. Suzanne Pritchard reports


Guided by experience and not by theory, the design of concrete faced rockfill dams (CFRD) has evolved empirically over the past 40 years. As many members of the international industry agree, successful and unsuccessful experiences should be known, understood and thoroughly discussed so that the progress and development of CFRDs are assured.

It was the experience gained by engineers who worked on the ruptured concrete face at Campos Novos dam in Brazil which provided the motivation for the 3rd Symposium on CFRDs held in Brazil in October 2007. (See p27 for details on Campos Novos dam).

“Due to the fact that such configuration and cracks [at Campos Novos dam] were observed for the first time, a great interest was raised internationally in the technical community and also in other non technical groups,” explains Lailton Vieira Xavier from the symposium’s organising committee.

He adds: “Another relevant point was that many CFRDs with heights in the range of 180m-230m were under construction at that time. This was a concern for professionals involved and forced the implementation of additional measures on other projects based on the partial understanding of what was happening at Campos Novos dam. For these reasons, we found it an appropriate time to create the opportunity to bring together Brazilian and other international experts. We aimed to share and further develop the knowledge on CFRD, particularly related to very high dams.”

The previous two CFRD meetings, held in 1985 and 1999, were inspired and organised under the leadership of J Barry Cooke who was described as the great promoter and developer of CFRD dam technology throughout the world. Furthermore, Edilberto Maurer, chairman of the Brazilian Committee on Large Dams, said he could be called “the father of this type of dam in modern times”. Cooke died in 2005.

“We recognise the fundamental contribution of J Barry Cooke to the development of CFRD dam technology,” Xavier said, “not only for his technical qualification but especially for his never ending disposition of sharing his knowledge.

“The symposium was held in his honour.”

Promoted by the Brazilian Committee on Large Dams (CBDB) and organised by Engevix Engenharia and the Santa Catarina branch section of CBDB, the symposium was held in Florianopolis. Technical papers on CFRDs were divided into three themes covering: prediction of rockfill deformation; construction methods; and instrumentation and performance of the dams. Thirty four papers were given, covering the experiences of CFRDs in Brazil, China, Ecuador, Iceland, Mexico and the US.

Technical visits were arranged to the 202m high Campos Novos dam on the Canoas River in Santa Catarina and the 187m high Barra Grande on the Pelotas River between Rio Grande do Sul and Santa Catarina.

Natural limits

As Cooke stated in 1984, the basic principle behind the design of CFRDs is that it is empirical – based on previous practical and successful experiences instead of design theories. This has been considered an acceptable approach due to the intrinsically stable nature of the structure. Even when the rockfill is subjected to high shear resistance, the dam remains stable. Furthermore, although water flow can pass through the embankment, the dam remains stable because the rockfill is draining freely and so pore pressure does not build up.

In his work, Cooke stated that the principle of precedents must be followed but should not restrict progress, either for the implementation of higher dams or for the adoption of more cost-effective solutions.

Various papers presented at the symposium drew upon the recent experience of several CFRDs. Extensive rupturing of the concrete face under compression occurred at Campos Novos, in October 2005, and Barra Grande dam almost a year later, in September 2006. A similar situation was also reported at Mohale dam, in Lesotho, in February 2006.

One consequence of the compression problems commented on was some questioning of the empirical procedure and the emphasis placed on Finite Element Modelling (FEM) as an alternative method for the design of the dam. “However,” said Nelson L de S. Pinto[1], “the complexity of the physical problem to be tackled analytically sets a natural limit to what can be accomplished by mathematical modelling as yet. Well understood experience continues to be the best and essentially the only approach to the design of CFRD dams, including the very high ones.”

Pinto added that although numerical modelling was a promising design tool, it was still under development. He said it was helpful for dam design through the easy parametric analyses and comparison of alternative solutions but had not yet become the main instrument for CFRD dam design.

Filho, Saboya Jr and Maia[3] discussed in their paper the disparity between acquired and applied knowledge of CFRD.

“Difficulties in carrying out geotechnical laboratory tests in representative samples of rockfill have encouraged great efforts towards a good instrumentation programme,” the authors explained. “However, the results from instrumentation have not been matched with improvements in numerical models applied to CFRDs.” He added that aspects involving numerical analysis of rockfill dams are full of classical symbolisms representing the huge gap that exists between the state of the art and its practice.

Pinto, taking into account the nature of CFRDs, doubted that numerical modelling would ever become the main instrument for designing the structures unless the interpretation of the physical phenomena involved was much improved. He believed that the main design decisions were bound to continue for a good time, basically dependent on the empirical approach.

Indeed, empirical design decisions have now been taken following the rupturing incidents at Campos Novos, Barra Grande and Mohale dams. Pinto said that the analysis of recent experience had resulted in much greater emphasis on rockfill compaction, stronger central face slabs and the use of soft joints as practical and cost-effective provisions to expand the domain of successful CFRDs. Such design improvements had been made on the very high CFRDs presently under design and/or construction.

“Their behaviour will ratify experimentally those design decisions and will constitute basic reference for future dams,” Pinto said.

Rockfill:

The most effective provision to reduce compressive strains in the concrete face of CFRDs is to improve the rockfill modulus – improving the compaction and extending the section of better compacted upstream Zone 3B to at least two-thirds of the cross-section. The significantly more deformable downstream shoulder, classically accepted for lower dams on economic considerations, should be avoided in very high dams.

One basic design principle for the central portion of the face slab under compression is to eliminate or minimise all features that contribute to stress concentration along the vertical joints: These include:

• Encroachment of the mortar pad into the slab design section.

• V-notch detail at the top of the joint.

• Vertical loop of the copper water stop in the bottom of the slab.

• Concrete against concrete contact at the joints. This can be overcome by inserting a wood pad between slabs.

The use of anti-spalling reinforcement is also recommended to strengthen the borders of the slabs along the joints. After the recent incidents, thickening of the slabs in the central portion of the dam has been adopted in some dams under construction. The upper portion of the central slabs remaining to be cast was thickened to t = 0.4 + 0.002 H in Kárahnjúkar dam in Iceland and to 0.5 + 0.003 H in Bakun dam in Malaysia. Thickening the slab improves resistance under compression and provides room for a better distribution of the steel bars. Double reinforcing instead of the usual single matt at the centre of the slab is advisable for the slabs under compression. Heavier reinforcing, above the usual 0.3% – 0.4%, is not deemed justifiable.

Curb elements

Extruded curb elements were used as the finishing surface for the upstream face of the rockfill on the dams at Campos Novos, Barra Grande and Mohale. The curb elements, first devised for the construction of Itá dam in Brazil in 1999, are a most practical way to: define the upstream face and confine the upstream transition 2B for compaction; provide a stable surface for protection against erosion from rainfall runoff; and, set up a firm surface for construction of the concrete face.

Use of curb elements has become a way of life in practically all CFRDs built since Itá. The possibility that these could contribute to compression accidents has been raised but the arguments have not been convincing to the point of inducing the abandonment of the practice.

Bond-breaker products between the curb surface and the concrete slab have been suggested to possibly reduce the shearing forces transmitted at the interface. At Kárahnjúkar dam, a 3mm thick bitumen bond breaker (3.5 l/m² of a hot asphalt emulsion) was placed under the upper portion of the concrete of the nine central slabs. On the Bakun project, a layer of sand-asphalt was placed over the curbs before placement of the slab at the dam. At El Cajón dam in Mexico, no bond-breaker element was used.

A thin bond-breaker layer can be effective in eliminating or reducing adhesion but is not effective in cutting down friction. The roughness and natural irregularities of the final curb element face is such that it is essentially impossible to reduce the angle of friction by a thin layer of any smooth product. The reduction of adhesion between the curb and the concrete slab can only have a limited influence.

In Campos Novos and Barra Grande dams, plastic sheets had been placed over the curb elements to produce a bond-breaker effect and, although adhesion was essentially eliminated, no noticeable protection was gained.

Soft vertical joints

A system of soft vertical joints in the central region of the face to absorb part of the horizontal strains is effective in reducing the compressive strains in the face slab.

Several dams under design and at the beginning of construction are being provided with soft joints. At Bakun dam, six central joints were reopened in the concrete slab cast in Phase 1 and will be continued to the top of the dam, with an opening of about 50mm, and a soft wood board filler capable of deforming more than 50% under a 10MPa load. In the reconstruction works at Campos Novos, four 50mm wide joints were sawed along the old cold joints and filled with mastic and covered by fibre reinforced rubber membrane. The joints closed smoothly a total of about 60mm and no signs of high compressive stresses were registered.

Soft horizontal joints

Horizontal soft joints in the most critical central area should also be a beneficial provision. However, horizontal joints are not yet a clear designers’ choice. For one thing, the longitudinal compressive stresses seem to be less than the horizontal ones as they are not directly affected by the narrow valley shape most responsible for the mishaps, although the height of the dam may be a determinant factor in this case.

The three reported incidents give evidence of having started by the rupture of the central vertical joint, the transverse crack most probably being a consequence of the weakening of the concrete at the ruptured central zone. Also, horizontal joints interfere more with the construction of the face by interrupting the slip-forming process and require more attention.

The joints should not close completely during construction under the action of gravity to preserve the deformability needed to release the strains due to the reservoir load. They should be restricted to the central region of the face where face deflections are essentially uniform to avoid differential displacements due to the proximity of steep abutments.

Along the slope, movement of the slabs are bound to produce extra shear on vertical joints in the border with continuous slabs. Horizontal soft joints should help to reduce compressive stresses in the slope direction and deserve consideration in cases where the high compressive stresses are deemed critical.

Impervious fill upstream

The preventive measures against large seepage in Kárahnjúkar, Bakun, and Campos Novos dams include raising the impervious fill upstream of the face to about 40%-50% of the maximum head. It covers the most critical zone as far as the transverse crack is concerned and is considered a worthwhile provision.

Even after the reported accidents at Campos Novos, Barra Grande and Mohale, the CFRD structures remained inherently safe, Pinto told delegates. The operation of the dams was not affected during the episodes (the Campos Novos reservoir was emptied for unrelated reasons) and there was no concern at any time about the safety of the dams.

Such confidence in the structures explains why, according to Jose Antunes Sobrinho, vice president of Engevix Engenharia, CFRDs have become popular and are built worldwide. Their many advantages include:

• Stability.

• No piping.

• Fast construction.

• No effects of rain on construction.

• Safer river diversion.

• Easy to find equipment.

• Use of almost all types of rocks.

Contribution of China

Since the 1980s, Sobrinho acknowledges, China has played a major role in the development of CFRDs. Chinese engineers have been responsible for the majority of structures built subsequently.

Representatives from the China Institute of Water Resources and Hydropower Research, and the Kunming Hydroelectric Investigation Design and Research Institute[2] explained at the symposium that CFRD constructio in China started in 1985, and engineers have learned much from international experience but they have also developed their own technology. The government and the National Foundation of Natural Science have sponsored a series of fundamental research programmes into CFRD engineering.

Many advances have been made in the design and construction of CFRD structures in the country. Records show that, up to 2005, there were 160 CFRDs more than 30m in height, and amongst that total there were 50 higher than 100m. The 178m high Tianshengqiao No. 1 (TSQ-1) CFRD was completed in 2000. Its reservoir capacity, rockfill volume, area of concrete face slab and installed capacity currently rank it as the largest in the world although the 233m high Shuibuya CFRD will be the highest in the world upon completion.

At present, CFRD is often the preferred choice for dam engineers in China. With the rapid development of hydropower in the country, the authors Zeping Xu, Ahengkun Du and Lichun Zou said that they believed that more and higher CFRDs would be built in the coming years.

In the southern part of China, there are several proposed hydropower projects that are proposed to be build with CFRDs that would be more than 300m high. The dams are currently at the feasibility study stage.

The authors acknowledge in their paper that as CFRD technology develops even further, more challenges will be encountered in design and construction. Formal experience and design criteria may need some further reconsideration for future CFRDs as planned structures continue to increase in size and height. This, of course, is not only applicable to China, but also to the rest of the world.

IWP&DC would like to thank the Brazilian Committee on Large Dams (CBDB) and Engevix for their permission to draw upon papers presented at the 3rd Symposium on CFRDs. Additional copies of the proceedings may be requested through CBDB at www.cbdb.org.br


Barra Grande slab rupture Barra Grande slab rupture
Mohale dam Mohale dam
Campos Novos CFRD Campos Novos CFRD
Engevix Engevix
CFRD symposium discussions CFRD symposium discussions
Mohale CFRD slab joint Mohale CFRD slab joint
Barra Grande dam Barra Grande dam


Privacy Policy
We have updated our privacy policy. In the latest update it explains what cookies are and how we use them on our site. To learn more about cookies and their benefits, please view our privacy policy. Please be aware that parts of this site will not function correctly if you disable cookies. By continuing to use this site, you consent to our use of cookies in accordance with our privacy policy unless you have disabled them.