The ASEM and the ECAS as Interregional Institutions

Leakage and remedial treatment are summarized in Table 10-4. Dams with the largest leakage rates are listed first.

Previously, it was not common practice to estimate the seepage through the foundation or the concrete face, however, with the use of soft rockfill or dirty gravels as embankment materials it is becoming increasingly important to estimate this seepage and use the estimate as a basis for sizing the internal drainage system.

Seepage through the foundation can be estimated following the usual concepts of flow in porous media, or more complex methods that include the effect of discontinuities in the rock mass, and the effect of the grout curtain (Giesecke et al. 1992).

The flow through the cracks and joints in the slab is difficult to estimate, and often it is preferable to resort to case histories, however there have been several attempts to provide a theoretical basis to the calculations, and these methods may be useful to provide additional support to the estimate (Casinader and Rome 1988). A discussion of the techniques to estimate leakage through cracks and joints in the face slab is presented in Chapter 2.

Leakage is a key parameter concerning the overall performance of the CFRD. Large leakage rates are an indication that opening has occurred to the perimeter and/or face joints and/or that the concrete face has cracked to some extent. Seepage through the foundation may also be a contributing factor to large leakage rates.

The fundamental design concept of the CFRD is that the several embankment zones of the dam including the face support material, filters, transitions, under drainage and the body of the dam remain stable even if extremely large leakage rates were to occur. The ability of rockfill to accept and pass large flows is well known in the literature. Thus, if the embankment zones and the foundation treatment have been designed and constructed appropriately, the large leakage rates are not an indication that safety is a problem, but rather that remedial treatment may be needed to reduce the leakage.

Hydro Tasmania CFRD Experience

Damien Kenneally, 2003, reports the following leakage rates at CFRDs owned by Hydro Tasmania, see Table 10-1

Table 10-1

Hydro Tasmania CFRDs, data current as of April 2003

Dam Year

Completed Height (m) Type of Dam

Current Base

Leakage, l/s Rockfill Type Remedial Action Anthony Dam 1993 42 CFRD 3 Conglomerate and

Sandstone No Bastyan Dam 1983 75 CFRD 3 Rhyolite No Cethana Dam 1971 110 CFRD 5 Quartzite No Crotty Dam 1991 82 CFRD 20 Glacial Gravels No Mackenzie Dam 1972 14 BFRF 1 Dolerite No Mackintosh Dam 1980 75 CFRD 4 Greywacke No Murchison Dam 1982 93 CFRD 6 Rhyolite No Newton Dam 1990 37 CFRD 4

Tuff, Porphyry, Sandstone and

Siltstone

No Paloona Dam 1973 43 CFRD 0.5 Argillaceous Chert No Reece Dam 1987 122 CFRD Not

measured Dolerite No Scotts Peak Dam 1973 43 BFRF 3 Argillite Yes

Serpentine Dam 1971 38 CFRD Not measured Quartzite No Tullabardine Dam 1981 25 CFRD 2 Greywacke No White Spur Dam 1989 44 CFRD 4 Tuff No Wilmot Dam 1970 34 CFRD 0.5 Greywacke No

Legend:

CFRD Concrete Faced Rockfill Dam

BFRD Bitumen Faced Rockfill Dam

In general, excellent performance is reported. Only one dam required remedial treatment. Most CFRDs that have performed well are not widely discussed in the literature.

Experience in China

Dr. Jia Jinsheng, 2003, reports the current leakage rates at several CFRDs in China:

• Shisanling upper reservoir, pumped storage hydro o Concrete lined area 175,000 m²,

o Head 75m, o Leakage 5 l/s.

• Guangzhou upper reservoir, pumped storage hydro o Head 68 m

o Leakage less than 1 l/s

• Tianhuangping lower reservoir, pumped storage hydro o Head 97 m

o Maximum leakage 55 l/s, repaired in 1999 by emptying the reservoir, leakage reduced to less than 5 l/s

• Xikou upper reservoir, pumped storage hydro o Head 37 m

o Maximum leakage 7.7 l/s

• Qinshan CFRD with corrugated rubber waterstop over the top of the perimeter joint o Head 122 m

o Leakage 4 l/s

• Tianshengqiao, see following discussion o Head 182 m

o Leakage as of 2003, 132 l/s

• Chenping

o Head 75 m

o Leakage approximately 70 l/s

• Wan-anxi

o Head 94 m

o Maximum leakage 25 l/s

• Guanmenshan o Head 59 m o Leakage 5 l/s

• Longxi

o Head 59 m o Leakage 3 l/s Two problem dams were reported:

• Gouhou, head 70 m, failed as a result of the opening of the joint between the face slab and the parapet wall following deformation of the zoned gravel embankment. Joint detailing and construction of the dam were poor.

• Zhushuqiao, head 78 m, leakage tested at 2500 l/s, repaired in 2000 and 2001 after emptying the reservoir. This case history is described later.

CFRDs in China have, in general, performed well and, as a result, are often the selected dam type. The following are under construction in 2003 using the corrugated rubber waterstop:

• Shuibuya, head 233 m

• Hongjiadu, head 180 m

• Jilintai, head 152 m

• Zipingpu, head 156 m

• Yinzidu, head 139 m

• Bajiaohe, head 115 m

Several case histories of dams that included remedial treatment to reduce seepage are summarized below:

Turimiquire (Cooke, 2000)

The 115-m tall Turimiquire Dam in Venezuela was completed in 1980. The design was typical of CFRDs constructed during that time period. The outer slopes of the dam are 1.4H:1V upstream and 1.5H:1V downstream, a conservative design for the excellent limestone rockfill.

The Zone 2 face support material consists of minus 150 mm crusher run rockfill, a material that will segregate into lenses and streaks of fine and coarse material. The dam was well zoned and well constructed. A large capacity limestone rockfill underdrain is located at the base of the dam downstream of the dam centerline.

The reservoir did not begin filling until 1988 because of delay in the completion of the water transfer tunnel. Nearly complete filling of the reservoir occurred during the period, 1988 to 1991; maximum leakage during this period was 300 l/s. In 1994, leakage increased to 5,400 l/s at a rate of about 500 l/s per day over a 10-day period. Tremie placement of silty fine sand was immediately undertaken and the reservoir was drawn down. During the summer of 1995 a second repair by tremie placement of silty fine sand was undertaken; leakage reduced to about 2000 l/s with the reservoir about 5 m from full. During the rising reservoir in 1996, leakage increased to 3000 l/s. A third repair was undertaken; leakage reduced to 1600 l/s with a full reservoir. Because of water supply requirements, leakage at rates up to 3000 l/s is acceptable. In mid-1999, with full reservoir, leakage increased to over 6000 l/s. Leakage reduced to less than 4000 l/s as a result of tremie placement of silty fine sand and gravel, the fourth repair. A repair that included the placement of 7,850 m² of a PVC geomembrane was implemented in the latter half of 2000. Leakage reduced from 6000 l/s to somewhat over 600 l/s.

Leakage was detected using a hydrophone and establishing decibel contours. The leak was concentrated at a location at and above the perimeter joint in an area where the abutment slope is steepest. Divers and a TV camera made detail inspections of the area. A remotely operated vehicle (ROV) was also used in subsequent inspections. The map of the cracked area gives the appearance of an incident that started at the perimeter joint and progressed to a curved major crack above the perimeter joint. The mechanism of failure is visualized to be:

• An initial leak at the perimeter joint, cause unknown,

• Erosion and some of the Zone 2 material into the adjacent Zone 3,

• Nearby cracking as a result of loss of face slab support, and

• Progression of cracking as leakage increased and additional fine-grained material is removed.

One m by 2 m block-outs had been made in the concrete face in order to install piezometers subsequent to the construction of the concrete face. Defects in the later construction of the support for the block-outs and the backfilling of the block-outs are suspected as contributing to the leakage problem.

Aguamilpa

The reservoir began filling in mid-1993. Leakage peaked at about 63 l/s, then reduced to only a few liters per second. In late 1994, leakage increased to 260 l/s, reservoir level at elevation 219, 16 meters below the top of the parapet wall. Flows decreased to below 50 l/s during the summer of 1995 and 1996 with reservoir levels slightly below elevation 200. In 1997, several horizontal and diagonal cracks were detected in the concrete face between elevations 198 and 202. A study of the inclinometer data showed irregularities at several elevations. Diver inspection of the concrete face discovered a horizontal crack at elevation 180 crossing about 10 slabs, 150 m, with maximum opening of 15 mm. The crack was partially sealed with silty sediment; at some locations leakage was evident.

The sudden increase in leakage in 1994 is thought to have been the result of cracks opening as the reservoir was raised to nearly full pool. The reduction is attributed to the sealing of the cracks with sediment and with some reduction of the reservoir level following the rainy season.

The owners engineering staff believe that during the rainy season each year, the cracks open because the pool level increases. As sediment-laden water passes through the cracks, the cracks seal and seepage is reduced. The structure of the sediment is weak and, subsequently, during the next rainy season the cracks re-open and leakage again increases. Peak flows in 1998 and 1999 were 214 and 173 l/s, respectively. Minimum flows in 1998 and 1999 were <50 and <100 l/s, respectively.

A plan has been developed to seal the crack at elevation 180. Recent inspections of the crack have indicated a lengthening on the order of 40 m; total length is now about 190 meters.

Generally, the performance of Aguamilpa has been satisfactory. Face slab cracking was attributed to differential settlements caused by the use of dissimilar embankment materials. At Aguamilpa, low compressibility alluvial material was placed within the upstream shell of the dam while relatively high compressible rockfill was placed in the downstream shell.

Xingo

During fill placement, cracks in the surface of Zone 2B were observed, close to the left abutment. Cracks had an average width of 20 mm but some were as wide as 56 mm and were essentially vertical. Offsets of the order of 15 mm were reported. Initially the cracks were sealed on the surface with mastic and fill placement resumed. New openings within the same cracks occurred as well as new cracks at higher elevations in the same zone. Prior to the

placement of the face slab, cracks were filled with sand; the surface was re-graded, then compacted with a vibratory roller.

Settlements and deformations at Xingo continued within the rockfill zones after reservoir filling.

Normal behavior was reported during the first 1.5 years with leakage on the order of 110 l/s.

Subsequently, the rate of settlement of several instruments increased significantly over a period of about six weeks, then returned to similar rates recorded prior to the increase. Leakage increased to rates ranging from 180 to 200 l/s. Diver inspections indicated major cracks in the same areas where cracks had occurred in Zone 2B during construction. At one location, an 8-m long crack, 15 mm wide, was detected and an offset of about 300 mm was observed between two slabs.

The on-going settlement, face cracking, and the increase in leakage were closely related (Sousa, 1999). Leakage penetrated the dam fill reaching layers of less pervious material where rockfill with higher fines content was placed. It is believed that the increased rates of settlement were caused by wetting and saturation as a result of the increased leakage in this area. The increased settlement caused cracks to open further. Most probably, cracks in the Zone 2B material opened again allowing increased leakage. Re-opening of cracks also explains why the complete sealing after dumping of dirty sand was not obtained.

The increase in the rate of leakage was attributed to the opening of the joints and cracks in the concrete slabs of the left abutment (openings of 36 mm were observed). Hair-line cracks were detected in the face slab before the filling of the reservoir but they were not treated since they were superficial. Cracks appeared on the left abutment during the construction period and appear to have been produced by the topography of the rocky foundation in the left abutment, strongly inclined in the downstream direction.

Remedial treatments were carried out because of the slow stabilization of displacements in the left abutment and to preserve the integrity of the material supporting the face slab. The observed rates of leakage are considered acceptable, Eigenheer, et al, 1999.

Ita

Reservoir filling began in late February 2000; a full reservoir was reached in late April 2000, two months later. Leakage increased from 160 l/s in late February to 1700 l/s in mid-May, a period of about 2 ½ months. Inspections of the face slab using divers and a remotely operated vehicle (ROV) revealed a series of horizontal and sub-horizontal cracks in 15 panels, 10 to 15 meters above the perimeter joint. The cracks occurred toward the right abutment under a reservoir head of 65 to 85 meters. Cracks were found to be open as much as 7 mm. Dumping of clay and sand were successful in reducing the leakage from 1700 l/s to 380 l/s. The underlying cause of the face cracks is not known. Analyses of data collected from the instrumentation system indicate no abnormal behavior.

Additional foundation grouting was also underway at the same time as the remedial clay-sand dumping. The grout curtain was deepened in the area between the dam and the spillway gate structure to treat pervious interbeds between basalt flows. The extent to which the abutment is a

source of leakage is not known and the effect of the additional foundation grouting in reducing leakage is not known.

Modern CFRDs have emphasized the need for filter protection at the perimeter joint. At that location, the zone 2A is well compacted as is the adjacent zone 2B. The result is a dense, high modulus material located within three to six meters of the perimeter joint. At distances away from the joint the face slab is supported by a less dense material, thus creating the possibility of bending stresses and face slab cracking on the order of eight to ten meters above the perimeter joint. Some engineers have suggested that this condition may have contributed to the face slab cracking at Ita.

Ita was the first CFRD to use the curb method for face protection. No bond break material was used on the surface of the curb to break the bond between the face slab and the curb. Some engineers have suggested that stress transfer between the curb and the face slab may have contributed to the face slab cracking at Ita. Chapter 10 includes a description of the curb method for surface protection.

Golillas

The following description of leakage at Golillas is taken from Amaya and Marulanda, (2000):

“At Golillas the situation was more complex, because leakage took place through these joints, but also through the plinth foundation. Since the problem was not completely solved, and leakage is still occurring, the development is summarized:

• During the first filling of the reservoir, a rapid erosion of the clayey fills in the main joints in some sectors of the foundation was evident, producing total seepage surpassing 500 l/s, with a tendency to be higher. This condition forced the emptying of the reservoir, when it was at 50% of its height.

• After repair works at elevation 2915 m in the right abutment, at the contact between the plinth and the foundation, reservoir filling was completed, almost to its maximum level (El.

2995 m). Still important leakage was registered, but more controlled, on the order of 1080 l/s.

• The operators of the project lowered the reservoir to elevation 2965 m. This allowed treatment of the area close to the abutments. Upstream of the plinth, above this elevation, loose material was removed, cleaned and the main joints were filled and finally the surface was reinforced with shotcrete. The next filling of the reservoir only registered a total seepage of 650 l/s, half of what was measured before. This is attributed to two main facts: with the superficial treatment, the seepage path for the foundation was doubled (the plinth was incremented with the reinforcing of the slope) and the fine materials that fell in the perimeter joint during cleaning.

• During the next 15 years of operation, a period during which the dam has not experienced any more important deformation, seepage reduced in a natural way. At the maximum

reservoir level seepage was around 270 l/s. About mid 1999, after the reservoir remained at its maximum level (El. 2997.5 m) during 10 months, a time much longer than usual, seepage increased suddenly by more than 200 l/s to approximately 470 l/s. This could have been generated by the partial washing of the material deposited in the perimeter joint.”

The leakage did not affect the stability of the dam, however, a pumping system to return the water to the reservoir was implemented. An analysis indicates that this measure is economical because the water that is returned to the reservoir can generate more energy than is required for pumping.

“It is clear that the design of the triple seal used at Golillas was not adequate. The intermediate PVC seal shears and the mastic did not penetrate the upper seal when movement is essentially vertical. Some mastic can lose plasticity with time or low temperature, as occurred at Golillas.”

Minase

The Minase Dam was completed in 1964 by the Ministry of Construction, Japan. The following is a summary of pertinent data:

• The 66.5-m-high dam was constructed in the period 1958-1963.

• The foundation consists of fine-grained sedimentary rocks, shale, tuff, and mudstone.

• Rockfill slopes, downstream 1.4H:1V between 5 m wide berms every 20 m (1.65H:1V overall average slope) and 1.35H:1V upstream. The angle of repose of quarry rock was 1.3H:1V and shaking table results indicated that a 1.4H:1V slope was stable when subjected to 0.2g acceleration.

• The rockfill forming the body of the dam was liparite. The rockfill was placed in lifts (thickness not presented in the reference) and compacted by sluicing with a volume of water about four times the fill volume. The void ratio of the sluiced rockfill was 0.41.

• The concrete face was placed on rockfill described as “packed large rock”.

• The concrete face was placed in slabs mostly measuring 10 m by 10 m. The area of the steel reinforcement was 0.5% of the area of the concrete section.

First filling of the reservoir resulted in about 10 cm of crest settlement and 10 cm of horizontal displacement at the crest. The maximum values of settlement and horizontal displacement of the concrete face at about mid-height were 33 cm and 28 cm respectively. Leakage after reservoir filling, as measured by a weir at the downstream toe, measured 220 l/s at the highest water level.

This leakage was not considered to be excessive. However, upon inspection of the face, the horizontal joints at about mid-height of the dam and the perimeter joint at the left abutment were repaired. As a result, leakage reduced to about 100 l/s.

In June, 1964, Minase Dam was shaken by the Niigata Earthquake (M 7.5, 147 km epicentral distance from the dam, 75 cm/s² estimated peak ground acceleration at the dam). As a result of this earthquake, the crest settled about 15 cm and displaced horizontally about 10 cm. The

earthquake temporarily increased leakage from about 100 l/s to somewhat over 200 l/s. Within a few days, leakage returned to pre-earthquake levels.

Long term settlement over the period, 1963 to 1975, added another 15 cm for a total of 40 cm over 12 years. Total horizontal crest displacement was about 30 cm over the period, 1963 to 1975. Leakage gradually increased over the years as a result of the long-term settlement. By mid-1978, 15 years after the first reservoir filling in 1963, the leakage had increased to 400 l/s at full pool.

Repair of the concrete face, consisting of the placement of “gravel asphaltic mastics” over the

Repair of the concrete face, consisting of the placement of “gravel asphaltic mastics” over the

In document ZHANG CHAO (página 148-151)