Wolf Creek Dam is located on the Cumberland river in Kentucky, US. It is operated and maintained by the Nashville District of the US Army Corps of Engineers, providing flood control, power, recreation, water supply and water quality benefits for the Cumberland River system. Lake Cumberland, impounded by the dam, is the ninth largest reservoir in the US and the Corps’ largest reservoir east of the Mississippi river (see Figure 1).

Construction at Wolf Creek began in 1941 and was completed in 1950. The 1748m long dam is a combination earthfill and concrete gravity section. However, the original design and construction techniques of the 1930s and 1940s were inadequate to control seepage beneath the dam. Design considerations of the day did not fully account for the impact of the underlying geological karst features on the dam’s performance. The problematic formations beneath the dam, which have been pertinent to its seepage problems, are the Catheys Formation and overlying Leipers Formation. Both are hard, thin to massive bedded, argillaceous limestone interbedded with thin, well cemented, calcareous shale.

Furthermore, during construction the alluvium was left in place under the majority of the embankment and did not allow designers the opportunity to inspect the condition of the rock. Apart from the cutoff trench, no foundation treatment occurred beneath the embankment.

The design depended on a narrow, steep-sided cutoff trench with a single line grout curtain to block seepage in the foundation. The cutoff trench was designed to be under the upstream face of the embankment and parallel to the dam axis, except at its left terminus where it turned and tied into the last concrete monolith number 37. It was designed to be 3m wide at the base with steep 1V on 1-1/2H side slopes. A schematic of the cutoff trench is shown in Figure 2.

Early during construction of the trench a solution channel was intercepted running along the planned trench alignment. The decision was made to clean out and use this feature as the trench. Several large caves and numerous other solution features of varying size intercepted the trench at right angles.

Figure 3 shows a view of the solution channel/cutoff trench just upstream of its tie-in to Monolith 37. Note the large caves in the trench face. The man standing in the trench bottom gives an idea of the size of the openings.

Obtaining compaction of impervious fill against the side walls and plugging intercepting solution features was not considered important, as long as a 3m width of compacted material in the centre of the trench was achieved. The sidewalls of the trench were therefore left steep, irregularly shaped, and with overhangs that prevented tight contact and good compaction between the fill and rock. Placement and compaction was often by hand in solution features and under rock overhangs.

As water moving through the rock intercepts the trench, it is likely to have moved along the poorly compacted contact between the fill and rock, crossing the trench at weak spots. Bridging of the embankment across the narrow, steep sided trench means that cracking and hydro-fracturing of the fill became a possibility.

The trench stepped down from the right abutment to its tie-in at monolith 37 in several high vertical steps or benches. These are potential locations of differential settlement in the trench fill that could cause cracking. Typically, these steps also coincided with solution features crossing the trench. Thus, concentrated flow in solution features may occur at cracks in the trench fill which provide an avenue for seepage. As there are no filters, it is possible that trench material has been piped into open features.

Distressing signs

Wolf Creek Dam appeared to function normally, without any visible signs of distress, until 1967. This period was prior to the current dam safety programme, before any performance monitoring instrumentation was installed in the dam.

The earliest anomalies were observed in 1962 in the form of wet areas near the downstream toe towards the right abutment. Then in 1967 a small sinkhole was found near the embankment toe in the general vicinity of the wet areas. This was followed two years later by two more sinkholes near the downstream toe, above the switchyard in the wraparound area.

Immediately following discovery of the first sinkhole, the District embarked on an emergency exploration, instrumentation and grouting programme. This lasted from 1968-70 and resulted in about 8212m3 of grout solids being placed in the rock foundation, with the majority in the highly solutioned rock in the wraparound area. This work was generally recognised as saving the dam.

The investigations indicated that seepage was occurring through and/or under the cutoff trench and through the system of solution features. The seepage consisted of piping material filling these features, and subsequently overlying embankment material that collapsed into the voids. Eventually this progression of piping and collapse worked its way to the surface resulting in the sinkholes. Dye tests showed that the system of solution features went through the sinkhole and muddy flow areas.

A concrete cutoff wall was needed for the long term reliability of the dam. Two walls were recommended and subsequently installed. One was located downstream between the switchyard and river to protect the switchyard foundation from the surging and eroding action during power generation. The second was located along the crest of the embankment. A board of consultants recommended that the second wall should extend the full length of the embankment into the right abutment, extending to a depth of at least 1.5m below the Catheys–Leipers contact.

The District undertook a pre-installation exploration and grouting programme from 1970-75 along the alignment of the embankment cutoff wall consisting of borings on 0.76m centres. This programme served two purposes. It grouted openings along the wall alignment to prevent potential problems with the wall installation. It also provided information on the condition of the rock that the District could use to select the founding depths and lateral extent of the cutoff wall.

Based on these explorations, the bottom of the wall varied in its termination depth. In contrast to the original recommendation, the wall was only carried below the Catheys–Leipers contact at two locations, and the majority of the wall terminated in the upper Leipers formation. Laterally the wall tied into the end of concrete monolith 37 and was carried about two-thirds of the distance towards the right abutment.

These decisions have been mischaracterised at times as being cost driven, but were based on sound technical grounds at the time. Borings on 0.76m centres along the wall alignment provided a wealth of information upon which the wall extent was based. The designers held the view that grouting would seal the relatively minor openings in the rock indicated in the pre-installation exploratory holes below the selected founding depths and beyond the ends of the wall. A profile showing the limits of the wall is shown in Figure 4.

In retrospect, the decisions made concerning wall depth and length contributes to the reoccurrence of problems seen today. The original wall has worked well in cutting off features it intercepted but it simply did not go deep enough, or extend laterally far enough, to intercept all the significant features. Subsurface investigations and other indicators of distress confirm features still exist that have not been cut off. Over time, seepage has found these new paths under and around the ends of the wall and is once again increasing.

Since completion of the wall in 1979, the District has been monitoring various indicators of performance. A variety of instrumentation has been installed over the years. These consist of piezometers, displacement monuments, uplift cells, weirs, inclinometers, and alignment plugs. In addition, observations of the physical manifestations of the foundation seepage problems in the embankment and downstream areas are done routinely. Project personnel inspect the embankment and downstream areas daily for signs of problems. A brief discussion of some of the performance indicators follows.

Wet areas downstream of the dam

After the grouting programme and wall installation, most of the wet areas disappeared. However, over time, persistent wet areas redeveloped primarily near the right end of the dam along the downstream toe. Since 1990 the extent of the wet areas has steadily increased reaching the maximum extent in March 2004 after a two and a half year interval of sustained high lake levels.

Piezometric levels

Because of its history, Wolf Creek dam is highly instrumented with piezometers. Since 1968 over 300 have been installed. Currently 150 piezometers are monitored monthly and more frequently as conditions dictate. A select group of 25 in critical locations are read weekly.

Water levels in piezometers immediately upstream of the wall, with screened intervals set in the rock foundation, are equal to and react with the lake level. These have shown the lack of head loss across the cutoff trench and its ineffectiveness as a seepage barrier.

Downstream of the wall it was expected foundation pressures would drop to a small percentage of head water levels. However, immediately after installation only a slight reduction occurred and levels remained higher than anticipated. It was concluded the pressures would dissipate over time. However, this has not occurred and in fact several critically located piezometers have risen, with two piezometers in the wraparound section reflecting a 4m rise since 1984. Two embankment piezometers downstream of the wall have high levels and respond to headwater changes. Additionally, five piezometers generally located downstream of the embankment show that flow amounts are slight – approaching a trickle but illustrate the increasing seepage.

Settlement monuments

Subsidence of the embankment crest as measured by surface monumentation is occurring in the wraparound area; represented by both continuing settlement as well as an increase in the rate.

Embankment Investigation

In 2002 and 2003, 12 borings were drilled in the embankment using the resonant sonic drilling method. These holes were at various locations downstream of the wall. Six of the borings encountered soft zones within the embankment. One hole, located 1.2m downstream of the wall, encountered 2.1m of very soft, saturated clay at the top of rock. Additionally, two other borings over 30m downstream of the wall in the wraparound section also encountered soft material at the top of rock.

Temperature survey

In September 2004, a temperature survey of the screened interval of project piezometers was performed. Two cold spots identified in the survey were attributed to foundation seepage. Cold Spot 1 was present at the interface between the embankment and concrete, reinforcing the suspicion that seepage is occurring beneath the masonry section. About 37m downstream of this is Cold Spot 2, which registered cold temperatures in two piezometers with their tips set at different elevations. Overall, the temperature survey confirmed the seepage of cooler reservoir water past the wall and grout curtain.

Proposing a fix

As instrumentation installed at the dam had demonstrated that seepage was still a problem, the Nashville District conducted an extensive rehabilitation evaluation study. This concluded that the best course of action to take to remedy the problem would be the construction of a new concrete barrier cutoff wall. This new wall will start immediately upstream of the right concrete monoliths and run the length of the embankment, into the right abutment, for about 1280m (see figure 5). It will be constructed to a depth which is deeper than the deepest sections of the original wall, and as much as 23m deeper than the majority of the original wall.

The founding depth will be at least 7.6m into the Catheys formation, well below the zone of solutioning. With a minimum 0.6m thickness, a depth extending up to 84m and a total surface area of the face of approximately 27,750m3, the Wolf Creek barrier wall is unlike any other in the world.

In an effort to reduce the risk associated to dam safety, and simultaneously allow for further evaluation of the new barrier wall, the District elected to proceed with the rehabilitation efforts in two phases.

The first phase began in September 2006 and was completed by August 2008. It involved the installation of a double line grout curtain on each side of the alignment of the new barrier wall. This phase sealed openings in the rock to both improve the short term reliability of the foundation, and reduce barrier wall construction problems due to slurry loss. It also provided foundation information used in setting the final wall limits.

In December 2007, during the execution of this foundation grouting, the District issued the solicitation for phase II of the foundation remediation for the wall. The ‘best value’ request for proposal (RFP) method was used for procuring the contractor for this work.

Extensive data from historical records and soil information, together with guide specifications for the cutoff wall construction, were the base for the preparation of the RFP that was divided into a technical and price proposal. Unlike a traditional low-bid procurement, selection was made considering the soundness of the approach and experience of the firm, along with price.

The District awarded the barrier wall contract to Treviicos-Soletanche JV, a Treviicos led joint venture between Treviicos South, (the North America subsidiary of Trevi headquartered in Italy) and Soletanche Construction (a subsidiary of Soletanche-Bachy of France). The contract amount was US$341.4M over a four-year period.

Dam safety will be paramount throughout all the different phases of construction. This approach resulted in a series of additional steps in the construction process prior to installation of the barrier wall. The JV elected to perform a supplemental probing and grouting programme to detect and treat soft contact zones at the interface between the embankment and rock foundation. This will allow safer installation of a second major dam safety inspired step; the installation of a protective concrete embankment wall (PCEW) – designed to safeguard the embankment against the effect of the construction activities as depicted in Figure 6.

The PCEW will be seated into the top of rock, with a thickness of 183cm and depths averaging 43m. Following completion of this, the barrier wall installation will be a combination of secant piles and rectangular panels to depths of 84m and into the underlying rock.

All the major specialised equipment for the execution of the barrier wall is manufactured and customised for the particular requirements of the project. The equipment manufacturing units of the JV partners’ parent company, Soilmec (part of the Trevi Group) and Soletanche-Bachy itself, have been involved from the early stages of the project.

Project update

The notice to proceed for the 48-month contract was given to the JV in October 2008. The mobilisation and preparatory works are almost completed and the next phase of operations have started with the execution of the supplemental probing and grouting programme, and completion of the foundation grouting.

The installation of the PCEW started in April 2009 and is anticipated to continue for just over 14 months. Construction of the barrier wall is expected to start in July 2009 with completion scheduled for the first half of 2012.

The authors are: Michael F. Zoccola, Chief, Civil Design Branch, Nashville District, Corps of Engineers, Estes Kefauver Federal Building, 801 Broadway Street, Nashville, TN 37203, US. Email: Michael.F.Zoccola@usace.army.mil

Stefano Valagussa, Vice President, Special Projects, Treviicos Corporation, 38 Third Avenue, 3rd Floor, Charlestown, MA 02129, US. Email: svalagussa@treviicos.com



Dam features

The concrete portion of Wolf Creek dam consists of 37 gravity monoliths that extend 547m across the old river channel. With the top of the dam at elevation 773, it has a maximum height of 79m above the founding level. The spillway section has ten 15m x 11m tainter gates and six 1.2m x 1.8m low level sluice gates. To the right of the spillway section, the power intake section has penstocks feeding six turbines rated at 45MW each in the powerhouse downstream. Non-overflow sections on either end complete the concrete portion of the dam.
The embankment section extends from the end of the concrete gravity portion 1200m across the valley to the right abutment. It has a maximum height of 65m above the top of rock. The non-zoned embankment is composed of well-compacted, low plasticity clays, from the valley alluvium. Where the embankment and concrete sections tie together is a critical section that is referred to in this paper as the ‘wraparound’ section. Here the embankment wraps around the end concrete monolith number 37.