Uprating & refurbishment
Weighing the ways to fix a dam25 August 2004
Robert B Jansen takes a look at the different options available when certain rehabilitation work becomes necessary on dams
At some stage in the life of any dam, it may undergo changes that necessitate remedial effort. Each project has characteristics that distinguish it from others, but most will have familiar conditions amenable to analysis and treatment in well-tested ways. In choosing among options for rehabilitation, care must be taken to avoid worsening any situation that is already precarious. If there are indications of incipient failure, the availability of readily operable gates and valves to lower the reservoir may be decisive. In the common case where the disorder is deeply hidden, the assessment of its nature and extent must be conservative. One corrective measure might be more effective or less disruptive than another in certain conditions, but this could be difficult to know if the causes of trouble are not clear. Considerations that may influence the selection of an alternative include:
• Interpretation of information from investigation and instrumentation for defining deficiencies and characterising conditions to be encountered.
• Ability to set and readily adjust procedures in response to developments as the work proceeds.
• Effort required in mobilisation of equipment, materials, and crews.
• Limitations of equipment in coping with site conditions.
• Available working space at the site.
• Potential detriment caused by temporary, partial, or incomplete remedies that could fail and/or conceal deterioration of underlying materials.
• Leakage under a cutoff that is not deep enough.
• Flow through openings in a slurry wall cutoff lacking adequate mixing.
• Difficulties of sealing a cutoff at its contact with irregular foundation rock boundaries.
• Potential problems with seating a cutoff in presumed impervious soil foundation strata that might not be permanent under imposed higher water pressures.
• Difficulties in achieving a lasting, tight closure with sheet piling.
• Remedial work causing local changes in hydraulic gradient that might redirect and accelerate underflow.
• Limitations of embankment overlays for augmentation of spill capacity.
• Disadvantage of long-term dependence on operation of permanent remedial facilities that could be interrupted, such as pumps and power lines.
• Damage caused by drain maintenance procedures.
• Need for periodic renewal of remedial action, as in grouting of rock cavities partly filled with natural erodible materials or in repair of an earth blanket on a reservoir floor subject to sinkholes.
• Need for future access to drains and other remedial works, where feasible, for observation, maintenance, repair, or extension.
An embankment dam may be tested rapidly and most severely during the initial reservoir filling, when water enters defective zones of the fill or its foundation, as occurred at Teton dam in Idaho, US, Fontenelle dam in Wyoming, US, Bradfield dam in England, Jatiluhur dam in Indonesia, and Nahal Oz dam in Israel. However, in some projects, significant problems have not become apparent for long times after first storage, extending eight years at Walter Bouldin dam in Alabama, US, nine years at Mill River dam in Massachusetts, US, 12 years at Baldwin Hills dam in California, US, 28 years at W.A.C. Bennett dam in British Columbia, Canada, 44 years at Whiteman’s dam in Alberta, Canada, 54 years at Hardy dam in Michigan, and 85 years at Willow Creek dam in Montana (Jansen, 1983, 1994, and 2000). Latent defects may remain undiscovered until deterioration exceeds a tolerable limit or until loading reaches an unprecedented level, as in extreme floods or earthquakes.
Whether problems develop quickly or slowly, readiness for remedial action requires that potential weaknesses be assessed and addressed while they are still manageable. Soil and rock obtained from borrow areas and quarries may vary widely from design objectives. Some embankment zones could be subject to weathering and contain disproportionate amounts of erodible fines. The integrity of earthfill depends on its density, gradation, plasticity, water content, pore pressures, shear strength, and soil fabric, the representative values of which are variable and not easily determined. The possible consequences of nonuniform conditions include:
• Collapse of loose soils.
• Internal erosion.
• Erosion of embankment soils into bedrock fractures or into coarse alluvial or glacial foundations.
• Erosion of embankment soils into joints or cracks in adjoining structures or by inflowing leakage through such openings or from unsealed foundations.
• Differential settlement at zone interfaces.
• Settlement and cracking at irregular boundaries or at contacts with foundations containing soluble or erodible materials or weak rock susceptible to sliding or subsidence.
• Susceptibility to liquefaction under seismic loading.
Timely detection of such deficiencies requires investigation and frequent close inspection in concert with carefully placed and monitored instrumentation. Flow paths through or under embankment dams have been investigated by dye tracing and by borehole thermometers, piezometers, weirs, and meters equipped for measurement of constituents in the discharges. Plotting of piezometric and thermal contours can be useful in defining any leaks. Visible evidence may be deceptive if seepage is passing into pervious foundation that intercepts and retains eroded soils. Critical flaws might also be hidden by superimposed structures or by overhanging fill that roofs a cavity. The degree of urgency may depend on the dam’s operating experience. If it is in questionable condition at normal reservoir level and has not yet been tested by flood or earthquake, repair should warrant a high priority.
In examining a dam suffering from any of the cited problems, the shortcomings of design and construction may become readily apparent. However, remedial action may be much more difficult than the necessary work would have been originally. The presence of the operating dam and reservoir denies easy access to faulty components of the embankment and the foundation. Innovative features may have to be introduced to overcome the obstacles and to eliminate defects from outmoded practices and inferior workmanship.
Drain defects may be attributable to design mistakes, including failure to anticipate the detrimental changes that can occur in operation. Clogging by migrating materials can be expected where gradations of adjoining zones are incompatible and filters are lacking or insufficient. Drainpipes have been displaced by deformation of surrounding zones or have been obstructed by organic growths or by corrosion and breakage. Cleaning by chemical treatment and/or pressure washing has caused damage in some cases, especially where filters and screens were unable to retain loose particles. Access to old drains for maintenance or repair has been limited or practically impossible in many projects. Remediation may require taking them out of service by grouting or removal, and by replacement with accessible facilities of proper design. Such works may be vital to dam survival if defective drains are actually contributing to internal deterioration. Very careful attention to design and construction details is needed to ensure that the added features are stable and of adequate capacity, without causing undesirable changes in the seepage regime.
Typical clay soils are generally resistant to erosion. However, dispersive clays are susceptible to piping due to deflocculation where water passes through cracks, as along conduit and structural interfaces or in dessicated embankment. These soils consist of particles that have little electro-chemical adhesion and are distinguished by high sodium cation content of the pore water. Where earthfills composed of such materials have been damaged, repairs have been made in several ways, including placement of superior fill, filter protection, and treatment with lime.
Wister dam was completed in 1949 on the Poteau River in Oklahoma (Erwin and Glenn, 1991). As originally designed, it was an unzoned 30m high earth embankment with upstream and downstream toe berms composed of waste shale. The dam was constructed on about 9.1m of alluvium overlying sandstone and shale bedrock. To combat the tendency of the fill to erode internally, work in 1949 included injection of cement grout into the embankment and the foundation, driving of steel sheet piling along the upstream berm, and placement of a drain system in a limited reach along the downstream toe. Also, mud grouting of the embankment was attempted, but this was aborted after cracks appeared in the dam crest. From the time of initial operation, the dam passed seepage at large rates under high piezometric pressures. Service was continued for forty years under these conditions. Then a thorough investigation disclosed extensive cracks and weak strata in the embankment and the alluvium. The mud grouting in 1949 had caused hydrofracturing. In addition, a large part of the fill was found to comprise dispersive soils. The sheet piling had many openings. Remedial action had to be taken. This included a slurry wall of plastic concrete through the embankment and alluvium and extending several feet into rock, a downstream sand filter blanket, an enlarged drain system, revision of the downstream slope, and lime treatment of the exposed surface soils to increase their resistance to dispersive erosion. The rehabilitation was completed in 1991.
Problems have been found at certain sluiced fills, which in the basic design known as hydraulic fill are constructed of soils conveyed and placed by water. A dam type called a semihydraulic fill is also placed by water but the soils are brought to the site by other methods of transport. In some projects, dredge pumps were used and sluicing was done directly from the borrow pit to the dam. In other cases, the borrow was excavated by dragline and hauled and dumped into bins at the dam site where hydraulic ‘giants’ (high-pressure nozzles) washed it to dredge pumps and pipes delivering to the dam. The basic differences in procedures for handling the materials have usually resulted in poorer performance for the semihydraulic type, in which the outer parts are the remainder after materials were sluiced from the inner edge of the dumped fill toward the core by water jets. The partially washed or unwashed zones extending to the dam face retained substantial amounts of the original fines and therefore constituted a barrier confining water in the central and intermediate parts of the embankment. In contrast, the outer section of the true hydraulic fill is more pervious so that it allows outward drainage to enhance stability. Ideally as envisioned in its design concept, which recognized the need for zoning and filter protection, the hydraulic fill grades from coarsest material at the face to sand at the edge of the core pool to impervious fines in the central zone. However, control of zone limits was difficult. The actual cross section was usually only a crude approximation of the ideal.
Some sluiced fills have low densities and may be susceptible to excessive settlement. At a dam of this type in California, subsidence was so severe that reservoir water was passing over the top of the core. Where the soils in these dams are comparatively loose and cohesionless, they may be subject to internal erosion, or to liquefaction under seismic loading. Some sluiced cores have been found to contain interbedded coarse strata that may invite leakage. The extent of such intrusion into the central zone depended on inevitable irregularities in the core shoreline, nonuniform distribution of flow from the sluicing nozzles, and variations in gradation of the materials transported to the site. Testing has shown shear strengths of sluiced embankments to range considerably, depending partly on the histories of deposit and transport of the material sources and the consequent angular or rounded shapes of the particles.
A 52m high semihydraulic fill constructed in southeastern US in 1930 provides an illustrative example of a project with multiple deficiencies. In the mid-1990s, inadequate spill capacity was remedied by raising the dam 2.1m and constructing a second spillway. The principal problems at the embankment stemmed from the indefinite zoning and consequent elevated internal water pressures and marginal stability resulting from the semihydraulic method, as well as placement of the penstock under the dam rather than in a secure abutment tunnel.
Under the core, a foundation trench was excavated 15m wide and about 4.5m deep at the maximum section. Some of this core trench reached into sound rock. Excavation was bottomed mostly in overburden and/or decomposed rock under the rest of the core, the base of which has a total width of roughly 60m at the highest dam section. Under the remainder of the dam, foundation preparation consisted of stripping to remove organic matter.
A very large riveted-steel penstock with concrete encasement was located in a trench under the dam. This was originally the only low-level means of lowering the reservoir. Because of nonuniform support along its alignment, the penstock suffered differential settlement. An internal longitudinal girder was installed in the penstock to span the weak foundation zone. The steel liner bulged inward from external water pressure during its first unwatering. Drain holes were drilled in the liner in the reach under the upstream shell of the dam.
The embankment consists of three kinds of fill: (1) puddled core ranging from sandy silt to silty sand, with dry unit weight of about 70 pounds per cubic foot, (2) intermediate sluiced zones adjoining the core, composed of silty sand and gravel with average dry unit weight of about 92 pcf, and (3) dumped fill forming the outside slopes, comprising a wide-ranging mix of silt, sand, and gravel with average dry unit weight of approximately 86 pcf. There is a small rockfill drain zone forming the downstream toe. Because of the high fines content in the shells – roughly 25% in the sluiced fill and 50% in the dumped fill – and the consequent poor drainage capability, the phreatic surface was excessively elevated. The quality of materials tested throughout the dam was variable and generally low, with measured standard penetration resistance ranging approximately between 3 and 12 blow counts per 0.3m. The foundation overburden left in place was indicated to be in the upper part of this range. Given all these conditions, the ability of the dam to withstand earthquake was analysed and found to be marginal. Another factor calling for attention was leakage from the penstock into the surrounding embankment, due mainly to the drain holes drilled through the steel liner to relieve external water pressure during penstock emptying, and then left unsealed.
In weighing remedial alternatives, the improvement of embankment drainage was a primary consideration, along with the planned repair of the penstock, which has been enabled by completion in 2004 of a new low-level tunnel outlet in an abutment. Installation of a new drain system for the embankment has been scheduled in the period of lake drawdown for the penstock work, when dewatering of the affected downstream fill can be done safely. The ongoing remedial programme also gives attention to (1) repair of areas of the upstream slope damaged by periodic drawdowns, (2) a filtered drain feature to control seepage along the penstock, (3) strengthening of the riser pipe of the elevated surge tank and the structural members supporting the tank, and (4) correction of inadequacies at the penstock headgate. At some other sluiced fills where drainage was not as complicated, enhanced stability has been ensured by addition of berms. At this dam, working space for such earthwork is limited by the surge tank, power house, and related facilities at the downstream toe of the embankment. The natural channel carrying discharges from the spillways and from the new outlet is close to the power house.
In 1985, sinkholes were observed in the top of 40m high Hardy dam in Michigan, US, which was built in 1931 by sluicing fine sand dumped from railroad cars operating on wood trestles. Ground penetrating radar was used to examine the embankment, locating 266 anomalies, of which 162 were identified as voids caused by rotting of trestle timbers left in place after construction (Silver et al., 1986). The sand of the embankment was found to be generally loose. Therefore, alternative remedial methods that would induce disturbing densification were rejected. In 1986, the cavities were grouted by tremie using a lean mix of 6% cement, 54% fly ash, 3% bentonite, and 37% sand injected at low pressure (between zero and 15 psi) measured at a gage at the top of the pipe.
While sluiced fills have experienced notorious deficiencies, other earthfills where design and construction practices would not meet modern standards have displayed similar inadequacies in zoning, densities, drains, and filters, as well as interbedding of pervious and impervious soils. The security of an embankment can be threatened by sliding along a single weak layer of fill or by leakage through a free-draining layer that was intended to be tight.
Embankment defects at instrument installations
Substandard earthfill also may result from poor compaction of fill around instrument columns during construction. This may be evidenced first as depressions in the crest. Repair by shallow backfill could cause arching that conceals deeper defects, including subsidence of the core and consequent inward and downward deformation of filters. Leakage could develop through the narrow, weakened zones in the upper part of an embankment.
In 1996, two sinkholes were discovered in the top of W.A.C. Bennett dam in British Columbia (Stewart et al., 1997). This 183m high dam, completed in 1967, is a sand-and-gravel fill with a central core of nonplastic silty sand. The holes were at survey benchmarks made of steel pipes based on the bedrock and extended upward above the fill as construction progressed. During investigative drilling at the initial 2m3 sinkhole a few days after its appearance, an abrupt collapse resulted in a cylindrical cavity 7m deep and about 2.5m across. Testing disclosed that disturbed core material at the site of that sinkhole extended as far as 5m from the benchmark tube and close to the rock foundation at a depth of 114m. Seismic crosshole shear wave tomography showed a pattern consistent with direct CPT measurements, revealing significant differences between the damaged zone and the undisturbed fill. A high content of fines was apparent in the blanket drain. Action was taken to restore the weak fill so that its properties were similar to those of the unaffected core. This was done by compaction grouting to unprecedented depth, using a mix of silt, sand, and pea gravel with a gradation approximating that of the core, without cement. This method was chosen instead of jet grouting or deep soil mixing because of their depth limitations and use of cement grout. Crosshole tomography and reduced seepage measurements have indicated that the grouting was effective.
At another Canadian dam in the 1980s, collapsed backfill was found at instruments (piezometers and inclinometers) in the thin central till core of the embankment. Because of inadequate compaction around the instrument risers, low-density material comprised as much as half the core width in the upper part of the dam at those locations. Thermal profiles in the weak zones showed large variations, suggesting the possibility of concentrated leakage. Any protective plan had to take into account the reduced effective dimensions and density of the core and the short seepage path at the top, as well as the masking effect of bridging of the fill and concealment of eroded material by interception in coarse zones or foundation recesses. In an extreme scenario, the soil bridge could collapse and water passing through disturbed core and filter zones could carry fines into the downstream shell. If this could happen under normal static conditions, there would be little need for seismic analysis.
In considering the course of action for this dam, the described circumstances were weighed against the lack of turbidity or sharply rising leakage measured at the weirs. Also, some reassurance was found in the high quality of construction in the rest of the embankment away from the instrument columns and in its lack of sinkholes. The decision on the basis of this more favorable assessment was to defer remedial treatment and to continue to scrutinise the performance of the dam to ensure early detection of migrating materials that would trigger corrective measures. This approach depends on the ability to react quickly enough to avoid an uncontrollable washout. It can be compared with the cautious remedial program adopted for W.A.C. Bennett dam in coping with sinkholes at instrument columns.
In such cases where the location and cause of a defect are well identified, and the integrity of the rest of the dam is unquestioned, attention obviously must be focused on that defect to determine the damage and its possible future limits if deterioration were to continue at the present rate – or to accelerate, as is common with internal erosion. The unknowns that may remain even after thorough investigation and analysis, including the paths and places of deposit of lost materials, must be recognised in drawing a credible worst scenario. In weighing options, the location of the dam and the availability of emergency resources could be decisive. At a project in a remote area where access might be restricted by severe weather, the rapid mobilisation of equipment, materials, and personnel stationed locally would be imperative to stem outflow through a suddenly increasing breach. Conditions could become unmanageable at a defective dam struck by earthquake while the reservoir was high.
The problems caused by poorly compacted backfill at instrument risers have raised serious questions about the advisability of such installations. In many cases that have been investigated, the defective earth columns have functioned as seepage sumps and leakage conduits. In effect the instruments are intrusive foreign elements that do not provide readings reliably representative of conditions in the normal embankment. Therefore, the basic intent of monitoring has not been served, while the dam suffers from an introduced defect that is potentially dangerous.
Many reviews of older dams have found that flood discharge capacity was inadequate. Among the remedial options that have been examined is modification of an embankment by armoring to allow overtopping. In the past, this was viewed with disfavour because of failures, which could be attributed in some cases to inferior design and/or construction, as well as to the overlay’s veiling of deteriorating embankment conditions. However, the evolution of roller compacted concrete technology has brought new attention to the concept. The list of rehabilitations by concrete or soil-cement overlays has grown, although most are on low embankments and have been subjected to only limited discharge. Designers must still be mindful of the well-established cautionary considerations, including:
• Forces imposed by flow velocities.
• Impact of overflowing debris.
• Underseepage and uplift and need for drains.
• Erosion of underlying fill.
• Tailwater pressures at toe of slope.
• Effects of voids and roughness of overlay surface.
• Effects of steps in slope.
• Shrinkage and thermal cracks.
• Sliding resistance.
• Energy impact at toe.
• Embankment settlement.
• Approach conditions.
• Peripheral cutoffs.
• Training walls.
• Quality of the concrete mix and its placement and compaction, particularly at geometric changes.
Recognition of these factors has led to improvements in design and construction procedures (US Society on Dams, 2003). Concrete overlays on embankments are generally used only to provide incremental capacity to pass infrequent floods where the costs of conventional alternatives would be prohibitive. In any case, underdrainage is essential. A slab extending over the downstream slope will restrict normal seepage in the fill so that uplift tends to develop. Drains for pressure relief must also intercept spillway discharge leakage through cracks in the overlay. Contamination of the underdrain system can result from infiltration of dirty floodwater, as well as from muddy runoff during construction or movement of underlying dam materials during normal operation. Filter protection is needed to keep the drain clean. The concrete and its underdrain have to be placed on firm, dry embankment. The section at the spillway crest must have secure lift joints and be heavy enough to resist delamination from uplift or damage by debris-laden overpour. Here and elsewhere on the edge of the overlay, safeguards are needed to control hydraulic forces from underflow. The adjoining upstream embankment face needs to be protected from erosion. At the toe of the chute, the forces from energy dissipation and tailwater must be considered. Depending on the slope and the expected flood impact, special precautions may be required to control creep or sliding of the slab, including possibly shear keys or anchorage.
The design for an overlay on Arthur R. Bowman dam in 1994 was based on a thorough comparison of RCC and a concrete slab continuously reinforced through and between panels (Hensley, 1993). The latter alternative was favoured because of its superior control of cracking and offsets. Underdrainage was provided by a thick sand-and-gravel layer. This 75m high zoned embankment in Oregon, completed in 1961, has a broad central earth core and relatively thin rockfill shells on cobble and sand-and-gravel intermediate zones. An overlay on a downstream face composed of such coarse materials could pose less concern about uplift and erosion than one placed on soil.
There is a need to emphasise that an overlay is a substitute for a more acceptable spillway that is separate from the embankment. Even with the most careful engineering and surveillance, it has the serious drawback of blocking the underlying fill from visual inspection, thus preventing earliest warning of a worsening condition. The sound principle of preferred avoidance of fluid conduits in, or on, or under an earth dam remains valid. It should be circumvented only when the dam site environment precludes a safer alternative that is structurally and financially feasible. A permissible exception might be an auxiliary spillway created by a slab over an embankment on rock in a topographic saddle on the reservoir rim. Such a facility could fail as a fuse plug without exceeding acceptable limits on the extent of reservoir loss.
Experience at 105m high Jatiluhur dam in Indonesia showed problems that can develop with structures and conduits embedded in embankments. The upstream shell of this earth-cored rockfill was built around a massive concrete structure combining a shaft spillway and a power house. The discharge conduits from those facilities were placed under the dam, along with an access gallery. Reservoir impoundment started in 1965. Associated with embankment deformation, cracks and leakage developed in the gallery and the spillway/powerhouse structure tilted upstream. Extensive rehabilitation has been needed to arrest the adverse movement.
Earthfills placed on soil foundations (alluvium, colluvium, eolian deposit, glacial drift, landslide, residual soil, volcanic ejecta, or talus) may present exceptional challenges in remediation where the dam base has been destabilised by piping, which might originate from incompatible gradations of dam and foundation materials or through open leakage passages at the contact of overburden and bedrock. The natural materials under the embankment are not likely to be of uniform permeability. Escaping water will seek the more pervious strata that cross the site. Without an effective cutoff, particles eroded from the fill and the overburden may thus have ready transport channels. In the common case where reservoir service cannot be interrupted to allow time for extensive modifications, remedial action might include grouting of rock and/or overburden, or construction of a cutoff wall through the dam to reach a solid, nonerodible base. Since these walls are costly, other options are usually considered first. An alternative with a long history is a sheet pile cutoff, which in early times might have been composed of wood planks but which is now commonly made of steel. Although sheet piling is comparatively inexpensive and may be installed rather quickly, its benefits have been questionable in many cases because of deterioration and ineffective interlock or lack of penetration into a tight base.
Soil-bentonite slurry cutoff walls
Soil-bentonite slurry cutoff walls have been used for many years, and in most cases they have been effective. However, with the advance of technology, some competitive methods have been adopted with more confidence. The tightness of a soil-bentonite water barrier is governed by the quality and gradation of the natural materials in the mix, as well as the bentonite content. Thorough blending is essential. Head differentials across the trench could be high enough to move the backfill into adjoining coarse zones. In practice, such problems have been avoided or overcome in various ways. This type of cutoff is still an appropriate choice for projects where it may have advantages of economy, adaptability, and fast construction.
A soil-bentonite wall was constructed as part of the 2002-2004 remediation of 39m high Keechelus dam, an earthfill completed in 1917 on the Yakima river in Washington. The original embankment was built to enlarge a natural lake created by a moraine. The fill was dumped from rail cars on timber trestles and then was spread by Fresno scrapers (Pabst, 2002). Compaction, by a steel wheeled roller, was done only in the upstream half of the dam because the trestle restricted access to the downstream half. From the outset of operation, abnormal seepage manifested by sinkholes and sand boils was experienced at the site. Although there was general leakage through the glacial deposits, a principal problem was caused by water loss through an alluvial fan at the right abutment. Ground penetrating radar and test pits disclosed voids left by the abandoned, decaying wood members in the dam. Fine material eroded from the earthfill had entered and clogged the drain system. Tests showed the foundation and embankment soils to be deficient in internal filtering capability. To eliminate the extensive defects, most of the dam was removed and replaced. The new design incorporated filter zones and drains, as well as a soil-bentonite cutoff wall in the alluvial fan.
Compaction grouting is a widely accepted method for increasing the density of embankment and foundation soils by injection of thick mortar grout under high pressure. In contrast to conventional cement grouting, densification is achieved by compaction rather than penetration, thus limiting grout movement and preventing hydrofracturing. The extent of the treated zone can be established closely by control of the introduced bulb of low-slump grout. The procedure is particularly effective in filling voids and compacting loose soils susceptible to erosion and liquefaction. In such cases where the embankment core at and near the crest has been reduced by damage, compaction grouting would be clearly preferable to options that could cause further disturbance, particularly those involving vibration. A modified version of this method was an appropriate choice for remedying effects of sinkholes at W.A.C. Bennett dam, where a mix without cement was used to reproduce closely the characteristics of the normal fill. This would avoid any differential deformation or locally elevated seepage pressures that might result from inclusion of a rigid cemented mass.
Pinopolis West dam in South Carolina, US, is a homogeneous rolled embankment built in the 1940s. Investigation in 1983 disclosed three areas of its foundation with thin strata of loose silty sands potentially susceptible to liquefaction under the postulated seismic loading. Remedial work in those areas in 1989-1990 included bolsters placed against the downstream face, on foundation prepared by excavation of soils beyond the original toe. Compaction grouting was applied to densify loose foundation sands under the dam to improve their resistance to liquefaction. Standard penetration N-values were increased from a pre-treatment average of about 4 to a range of 11 to 38 blows per 0.3m. Closely monitored ground heave was on the order of one inch at the surface (Baez and Henry, 1993).
Another method for improving soil foundations is jet grouting, which is accomplished by injecting grout under high pressures (generally 4000 to 6000 psi) through small (2mm to 4mm diameter) rotating jetting nozzles to effect a soil-grout mix. When the hole has been drilled to the required depth, the rod is raised and/or lowered as the grout mixes with the soil. The excess mixture of soil and grout flows to the surface. Jet grouting can be done within definite boundaries, making it especially suitable, for example, in isolated loose soil strata or in zones immediately under or adjacent to concrete structures. Working space requirements are modest. Because of its ease of control and its applicability to a wide range of materials, it has been used on numerous projects with an outstanding record of success. The method does require special equipment and well-experienced crews. Procedures have evolved through several variations and refinements. Single-rod and double-rod versions involve high-pressure pumping of cement slurry to mix with the soil. The double-rod system adds air injected as a concentric shield around the grout jet. In a triple-rod process, the air shield surrounds a high-pressure water jet that moves soil to the surface while cement slurry is introduced by tremie.
A notable example of jet grouting on a large scale was in construction of the deep cutoffs for the cofferdams of 245m high Ertan dam, an arch on the Yalong river in China (Sembenelli and Sembenelli, 1999). This method was chosen over conventional grouting or concrete cutoff walls because of the required fast completion, the presence of materials ranging from silt to boulders, and the need for ability to increase dimensions where the encountered conditions dictated. The multiple-row cutoffs extended as deep as 40m through alluvium ranging from sands, gravel, and cobbles to coarse angular blocks of bouldery debris. The upstream cutoff penetrated fully to bedrock by jet grouting and was extended deeper into the rock by cement grouting. Similar procedures were followed for the downstream cutoff except that its middle part was bottomed in a thick silt stratum. At abutment rock contacts, cutoff broadening was done by adding two short rows to the three-row geometry adopted for the project. Spacing of the jet-grouted columns was 1m in each row and 0.75m between rows. Innovative procedures applied in this job included use of fragmenting (transversely slotted) PVC casing inside the steel casing of the drill holes, preventing caving after the steel was withdrawn. The designed, closely spaced, circumferential cuts in the temporary plastic pipe ensured that it disintegrated readily under the 400-bar (6000-psi) pressure of the grout jets. Injection was done with a double-fluid system comprising rods delivering grout and compressed air separately through coaxial lines, the grout flowing through the central conduit and the air through the outer one. The air enveloping the grout jet served to maximise the effect of the grout impact on the soil and to facilitate the removal of fines and their replacement by grout. A higher bentonite content was used in the downstream cutoff where there was a need for accommodation of possible differential settlement at the contact with the steep rock abutments. The jet grouting in the Ertan project, which was accomplished on a fast schedule under water head differences exceeding 10m, set high standards with adoption of a multiple-row layout and an advanced injection technology.
Wickiup dam, a 30m high roller-compacted embankment completed in 1949 on the Deschutes river in Oregon, was modified in 2001-2003 to improve its resistance to earthquake, focusing on jet grouting of low-density foundation materials susceptible to liquefaction (Bliss, 2003). A principal weakness was in the silts and fine sands underlying that reach of the dam designated as the left wing dike, which has a maximum height of about 12m. Standard penetration N-values in the most critical strata were very low, ranging from zero to three blow counts per 0.3m. The remedial work also included new filter-protected drains under a downstream berm constructed mainly of a mixture of soils and jet grout waste.
Soil mix wall
At 20m high Jackson Lake dam in Wyoming, constructed in stages in 1905, 1911, and 1916, the principal embankment was founded on liquefiable interbedded soils. In the late 1980s, those sediments were strengthened by the soil mix wall (SMW) method developed by Seiko Kogyo in Japan, in which grout is injected through hollow-stemmed earth augers as they are advanced. The foundation work done by SMW consisted of a cutoff wall at the upstream toe and a complex of containment cells in a honeycomb pattern. The equipment comprised track-mounted rigs containing two augers of 914mm diameter or three augers of 864mm diameter with fixed spacing. The cutoff was constructed with the three-auger system and the honeycomb was built with both kinds of rigs. Blades between the auger flights mixed the cement grout with the soil. A cutting head at the bottom of each auger facilitated the advance through harder materials. Because the SMW process involves return flow to the surface of a mixture of water, soil, and cement, provision must be made for disposal of this waste. In rating the available remedial measures for this case, the soil mix wall was assumed to have a possible advantage over jet grouting because of its procedure to ensure positive overlapping of the soilcrete columns. Experience on this job did involve some problems in plugging of augers and in achieving an adequate mix of silts and clays, as well as difficulties in penetrating cobble deposits in the construction of the cutoff wall. That part of the work was finished by conventional drilling and grouting (Luebke et al., 1991).
Microfine cement grouting
Remediation at the Logan Martin project in Alabama has included a cutoff through the floodplain alluvium by microfine cement grouting. In resolving the uncertain condition of overburden left in place in the original construction, the curtain grouting in the floodplain set a new precedent. It was selected instead of a more conventional cutoff wall because of limited working space and the benefit from the special talents and experience of the owner’s project team. An advantage of the adopted approach was in targeting only the permeable alluvial zone, which could also be done by jet grouting but without its requirements for handling of waste overflow. The grouting extended over a reach of about 200m in alluvium with a total thickness ranging up to 6m. This comprises a relatively coarse lower layer of sand, silty sand, and gravel as thick as 4.5m between the bedrock and an upper layer with very high fines content (as much as 75%). Based on the results of experiments, a selection was made from several sources of microfine cement. The grouting programme was focused on the coarse layer, establishing a single-row curtain with holes spaced at 0.9m and extra holes placed where water tests showed that seepage was still in excess of strictly set objectives. At some stations, the initial grouting had reduced the test water loss more than 90% (e.g., in one pair of 10-minute tests made at 25psi, the loss was 2139 litres before grouting and 49 litres after grouting. Coring and water testing confirmed that the special cement injected at low pressure had penetrated materials with substantial percentages of fines. Some holes had takes as high as one hundred 24.9kg bags of microfine cement. Water testing and piezometric measurement have indicated that the curtain provides effective seepage control.
Dam on Karst
Construction of Logan Martin dam on the Coosa river in Alabama, US, was completed in 1964 in an area noted for karst geology. It is 30m high and 1890m long, comprising concrete gravity and earthfill sections. Its 1417m long east embankment is founded on dolomite in the river channel, on shallow alluvium overlying dolomite in the floodplain, and on residual soils from there to the east end.
During the first reservoir impoundment in 1964, muddy boils were seen in the river downstream, initiating a long history of leakage accompanied by numerous sinkholes in the reservoir floor. In April 1968, a large sinkhole opened at the top of the east embankment while there was muddy discharge in the river. Then several major depressions were found in the upstream slope. Foundation rock sealing to establish a curtain has been continued through several programs over the years, comprising nine rows of cement grout holes extending over various lengths and depths, culminating in an intensive effort since 1990 in which grouting has been as deep as 180m in the rock (Robinson and Williams, 1997). This has produced steep declines in piezometric levels and in discharge from a weir at the toe of the embankment in the river section, from a peak of 55 liters/sec to 14 liters/sec. Total leakage past the dam site is presently about 20m3/sec. This is believed to be largely very deep discharge not reachable by grouting or other alternative such as a wall installed by trenching. It does not affect the primary objective of the remedial program, which is to ensure continuing protection of the underside of the embankment from attack by rising foundation leakage.
In weighing the options for remedial action, many factors must be taken into account, the effect of which will vary from project to project. The most appropriate alternative method at one site might not serve as well at another site with a different complex of conditions. The dam, its foundation, and the environs may have aspects that are not readily determined or analyzed. Principal variables include the adaptability of remedial methods to local conditions and resources, site accessibility, weather, project operational requirements, the time available for the job, and the possible benefits and detriments from the introduced features during the corrective work and after its completion. There is a need to keep abreast of continuing advances in techniques and equipment that may tip the scales in favor of one method or another.
Robert B Jansen, Consulting Civil Engineer, 509 Briar Road, Bellingham, WA 98225, US.