A history of Karst seepage - Center Hill Dam rehab18 June 2008
Last year the US Army Corps of Engineers classified the Center Hill Dam project as ‘high risk’ due to worsening foundation seepage. The first of three construction contracts to rehabilitate the dam is now underway, and here Linda Adcock and Robert Brimm presents details of the work involved in ensuring the long term stability of the dam
Since construction in the 1940’s, limestone karst seepage has been synonymous with operation of the US Army Corps of Engineers’ Center Hill Dam in central Tennessee. Original designers acknowledged the cave-scattered region would require future grouting to control seepage through the well-developed bedrock joint system. Despite several localised grouting efforts over the years, total seepage has increased to unprecedented levels under normal reservoir conditions. Foundation conditions are deteriorating the clay-filled rock joints within the rims, the abutments and the dam’s foundation. This erosion jeopardises the 246ft (75m) high earthen embankment of a main dam and an adjacent 125ft (38m) high earthen saddle dam. The seepage, however, is most advanced in the rims, where sinkholes align open cavities. The first of three major construction contracts is now underway to increase the long-term safety of the dam. This first contract will place grout in the most critical areas, the main dam embankment foundation and the left rim. The total rehabilitation effort is a seven-year, multi-contract endeavour consisting primarily of grouting and concrete cut-off walls.
Center Hill origin
Center Hill Dam is sited at Mile 26.6 on the Caney Fork River, a major tributary to the Cumberland River. In the early 1900’s, major flooding occurred in the Ohio and Mississippi River basins, causing disastrous loss of life and property, and economic instability. Government agencies began protective measures, which led USACE to develop a comprehensive flood control plan in 1937. The plan proposed construction of 45 reservoirs in the Ohio River Basin, six of which were within the Cumberland River Basin. Four of these reservoir dams were eventually built—Dale Hollow, Center Hill and J Percy Priest in Tennessee, and Wolf Creek Dam in Kentucky.
The Center Hill project was put into service in 1951 to provide important regional benefits of flood control, hydro power, recreation, water supply and water quality. The main dam consists of a 1382ft (421m) long concrete section and a 778ft (237m) long rolled earth embankment. A 770ft (235m) long saddle dam lies within a narrow valley along the right rim, about 1500ft (457m) upstream of the main dam. A three-generator-unit power plant, with a capacity of 135,000kW, and switchyard are located immediately downstream. TN State Highway 96 traverses the top of the dam. Center Hill Lake impounds 2.6Bm3 at its maximum flood control pool elevation of 685 msl. The Caney Fork River below the dam has become one of the most productive put-and-take trout fisheries in the state, complementing the renowned recreation fishery of the reservoir.
Karst geology of the foundation and rims
The site geology is characterised by numerous caves, springs, and sinks. These features are typical of the soluble limestone of the region and indicative of a well-developed karst terrain. Center Hill Dam is founded in nearly flat lying limestone and shale of Ordovician age. In ascending order, the formations present at the dam site are the Carters, Hermitage, Cannon, Catheys, and Leipers of Ordovician age, the Chattanooga of Devonian and Mississippian, and the Fort Payne of Mississippian age.
The contact between the Hermitage and Cannon formation is an unconformity with a significant amount of karst development. The Catheys formation forms the upper portion of the foundation of the concrete dam and earth embankment (see figure below). The Catheys is a finely crystalline to crystalline, hard, fossiliferous limestone. The contact between the Cannon and the Catheys is also an unconformity with localised karst development.
In general, the limestone at the dam site contains well-defined vertical joint sets, caused by a phenomenon called ‘Valley Stress Relief’. Down-cutting of the river through the rock layers has produced near vertical fracturing. The joint patterns are roughly normal to each other and follow the general line of two arms of the bend of the river. The fractures and loose planes create an interconnected system of open features caused by millions of years of groundwater solutioning activity. The fractures and joints are filled or partially filled with soil or residual soil as shown in the middle photograph opposite.
From seepage to piping
Under the pressure of the reservoir head, typically ranging between 125-150ft (38-46m), internal erosion of the cavity infilling material has created a phenomenon known as piping. Typically, piping progresses from the downstream (or outlet end) and propagates upstream until an open conduit is completed. As water erodes material from the karst features, the diameter of the opening conveying the water is ever increasing. This results in an increasing volume of water, a higher water velocity and higher erosive potential. Ultimately, this internal erosion undercuts and erodes the overlying material, resulting in settlement and/or sinkholes in the overlying material. This severe progression of piping is most evident through the left rim.
Along both the left and right abutments, piping is likely along both vertical joints and weak, horizontal bedding planes. Seepage through the main dam and saddle dam foundations is implicit from a combination of historic foundation data and current distress indicators such as abnormal piezometer levels, downstream wet spots and springs, cold zones at depth, and abnormal settlement.
Embankment dam construction
Both the main dam embankment and the saddle dam are constructed of high quality, well-compacted clay. The main dam embankment lies on alluvial soils for much of the footprint. The designers believed conventional cutoff trench foundation treatment, supplemented with a single line grout curtain, would provide adequate seepage protection. The 10ft (3m) wide trench, however, was not deep or wide enough to sufficiently act as a long-term seepage barrier. The trench was designed with 1V:1H side slopes and is within primarily the Catheys and Cannon formations. The rock in these formations is soluble limestone riddled with fractures and vertical joints ranging in size from fractions of an inch to more than 50ft (15m). Within most of the trench, 6-10ft (2-3m) of the rock/soil matrix was removed. Stepped into the left abutment, slopes along the axis of the cutoff trench were near vertical and irregular (see photo, below).
Embankment seepage history
For the first 18 years after construction, no problems were detected with the embankment. In 1967-68 at Wolf Creek Dam, however, muddy flow in the tailrace was observed, followed by large active sinkholes on the downstream embankment. These projects were designed and built in similar geology with similar design philosophy. The Wolf Creek piping crisis heightened concern with the similarly constructed Center Hill embankment. Prior to the Wolf Creek sinkhole, no instrumentation existed at either project. Permanent piezometers were installed throughout the Center Hill embankment and switchyard in 1969.
In the mid 1970’s wet areas downstream of the embankment were first observed, leading to a series of studies. One study was temperature profiling from the bottom of the top of rock piezometers. Several zones of colder temperatures were found. Exploratory drilling in the late 1970’s indicated cavities throughout the project area. Many solution channels crossed the core trench within the rock.
District personnel concluded that a number of potentially serious conditions existed and an embankment grouting programme was performed along the axis of the embankment from 1982-84. Drilling was single line on 2.5ft (0.6m) centers. Soft zones were found with drill water loss and connections between drill holes.
Reservoir water was found to be travelling between the embankment/rock contact and within the rock. Drill fluid losses and large grout takes required a contract modification to fill cavities. A total take of about 30,000ft3 (849m3) of solids were placed in this programme.
Left rim seepage
Early in the field investigations of the dam site, designers realised the thin reservoir rims on the left abutment might develop seepage through solution channels and joints in the Cannon and Catheys limestone formations. Field investigations were performed on several springs. These investigations did not reveal immediate concerns; however, designers considered remedial work would be required in the future after impoundment. Cavities encountered in the exploratory borings were primarily in the purer limestone of the Cannon and Catheys formations, but were too intricate to be effectively treated. In January 1949, seepage developed through the left abutment and rim as the initial reservoir was filled. A small spring located 3000ft (914m) downstream of the left rim (labelled ‘Picnic Spring’) flowed 1ft3/sec (0.03m3/sec) prior to impoundment. As the reservoir reached elevation 629 the flow increased to over 38ft3/sec (1.08m3/sec). Seepage paths were located using exploratory drilling. The reservoir was lowered about 50ft (15m) to elevation 583 and a US$2.2M (1949 dollars) grouting programme was performed along the left rim.
A single line 3500ft (1067m) long grout curtain extended from within the left abutment through an adjacent ridge in a southwesterly direction. An approximate 400ft (122m) section within the ‘saddle’ section of the ridge was not grouted. This section is capped by the Chattanooga Shale formation, which misled designers about the extent of solutioning. They mistakenly concluded the Chattanooga Shale had served as an impervious cap or roof that protected the underlying limestone. The 1949 grouting programme reduced seepage to 1ft3/sec (0.03m3/sec), yet over time the seepage slowly increased.
In 1968 weirs were installed to monitor increasing flows through the left rim. Drilling operations were focused on the reservoir rim along the original grout line and, in 1980, cavities were encountered. Although seepage was confirmed in the left rim, the safety of the main embankment was a higher priority and funds were directed to the 1983-84 embankment grouting. In the late 1980’s a number of shallow sinkholes were noted along the ‘Hidden Springs Hollow’ both upstream and downstream of the grout curtain. In the spring of 1991, during record high pool, approximately 3823m3 of chert and clay discharged from the Picnic Spring. In the subsequent months, several new sinkholes appeared along the ‘Hidden Springs Hollow’ above the Picnic Spring. The largest of the sinkholes (#11) measured approximately 25ft (8m) across and 20ft (6m) deep. The uphill side of the sinkhole exposed a near vertical weathered limestone. The material within the sinkhole was comprised of similar chert and clay seen at the Picnic Spring.
In 1993, a second 2100ft (640m) single-line grout curtain was installed along the same alignment as the original curtain. The 1993 curtain was designed with the intent to close the 400ft (122m) gap. Several large cavities were intercepted at depth with one hole consuming over 20,000ft3 (566m3) of solids without effect. The gap was reduced, however; approximately 150ft (46m) could not be grouted within the scope of the contract due to the size of the cavities. The cavity was 200ft (61m) below the top of ground and measured roughly 80ft (24m) high and 15ft (5m) wide. A dye trace confirmed that the cavity connected with the Picnic Spring. The water level in the cavity fluctuates with headwater. The increase is disproportional, suggesting that a natural, internal weir restricts the water through the hill at approximate elevation 635.
Prior to grouting, the Picnic Spring flow often exceeded the 16ft3/sec (0.45m3/sec) weir capacity. After completion of the grouting programme, normal flows were reduced to 3ft3/sec (0.08m3/sec). Since the 1993 supplemental grout curtain, seepage has again increased and collapse of sinkholes have occurred downstream of the grout curtain along Hidden Springs Hollow. Exploration has revealed that the cave system feeding this seepage path is large and complex. Openings have been encountered that extend from the river base elevation to the top of the flood control pool. Flow out of the Quarry Springs and Picnic Spring has increased to approximately 20ft3/sec (0.57m3/sec) since the 1993 grouting programme.
Following heavy rains and a high pool elevation in the spring of 2003, sinkhole #11 in the left rim collapsed and for several days flow could be seen at the bottom. Dye placed into sinkhole #11 appeared unobstructed at the Picnic Springs location and significant head cutting continues to be observed.
Right rim seepage
Original designers also documented concerns about seepage through the thin right rim. Two uncontrolled seepage paths are currently in the right abutment. Shortly after impoundment, a slight seep of less than 1ft3/sec (0.03m3/sec) appeared immediately adjacent to the concrete dam at elevation 580, the Cannon/Catheys geologic contact. A weir was installed in 1956 to monitor the flow at this upper leak. By 1959, seepage had increased to over 5ft3/sec (0.14m3/sec) and a hot bitumen grouting programme successfully intercepted and sealed the leak. The intent of the programme was to back up the bitumen with cement grout, but was not possible due to complete void filling with the bitumen. The bitumen mix was initially 100% effective, however, within a year the seepage began yet again.
The seepage reoccurrence is likely due to the bitumen deteriorating and to continued erosion of clay filling adjacent to the bitumen.
In 1965, 15 years after impoundment, another seepage outlet appeared at a cave feature 200ft (61m) downstream from the main dam along the Hermitage/Cannon contact. By 1968, flow totalled 4ft3/sec (0.11m3/sec) and a large weir box was constructed. By 1971 flows increased to over 6ft3/sec (0.17m3/sec) and in 1976, the 9ft3/sec (0.25m3/sec) flow became muddy, discolouring much of the tailwater. In 1987, muddy flow was again noted and by this time, flow was averaging 13ft3/sec (0.37m3/sec). The 1992 flow averaged 10ft3/sec (0.28m3/sec) and, by 1996, frequently overtopped the lower leak weir. The weir box was severely damaged by overtopping and high tailwater in 2003.
As part of seepage exploration of the entire project, drilling for the lower leak was performed in 1979. Of 20 drill holes only four angle holes intercepted the seepage paths, and connected to the lower leak by dye test. In the upstream right rim reservoir, both dye and an underwater television camera were unsuccessful in locating the seepage source(s). The District monitored the flow through the 1980’s, noting both upper and lower leaks continued to increase. Ten observation wells were placed into the right bluff in 1992 to intercept the lower leak. The holes were drilled from the downstream right bluff and half the holes intercepted the lower leak.
In 1993, a remedial grouting programme for the left and right rim was funded. A fan shape triple line grout curtain consisted of two cement grout lines and one chemical grout line. The curtain was placed parallel to and upstream of the axis of the dam. The grouting failed to intercept the lower leak but was successful in reducing the upper leak to 1ft3/sec (0.03m3/sec). However, in 1996, a new seep emerged about 50ft (15m) downstream and at the same elevation as the upper leak. Since 1996, this seep has doubled to 2ft3/sec (0.06m3/sec). The cave features of the lower leak and a second dry cave approximately 500ft (152.4m) downstream of the lower leak have similar entrance elevations within the lower Cannon formation. The cave floors terminate along the horizontally bedded Hermitage formation. In 2001, a team mapped the two cave systems to a distance of about 140ft (42.7m) each. Several clay-filled joints were observed roughly perpendicular to the cave and parallel to the right rim bluff. Although these right abutment seepage paths are not currently a dam safety concern, this uncontrolled seepage continues to increase. If left unchecked, the right rim could become a safety issue; consequently, right rim grouting is an integral part of the remediation plan.
Beginning in 2003, the Nashville District began studies to support a major seepage rehabilitation project. In 2005, the Center Hill Dam, along with the similarly constructed Wolf Creek Dam in Kentucky, was designated in USACE’s Class I Action Category for dam safety remediation, the highest priority for all USACE projects. Approved in 2006, the Center Hill Major Rehabilitation Evaluation Report recommends permanent barrier walls and supplemental grouting into the main embankment and saddle dam foundations as long-term remediation measures.
Grouting is also proposed to arrest seepage along both abutments, rims, and the concrete dam. Drilling into the dam foundation for rock information began in November 2006. Haul roads were improved in the left rim in the summer of 2007. The left rim and main embankment grouting is deemed the most critical work to increase the safety of the dam. This US$88M grouting contract was awarded to a joint venture of Kiewit Southern Company, Inc. and Advanced Construction Techniques, Ltd. The joint venture team includes subcontractors Hayward Baker, Inc., Gannett Fleming, Inc. and AMEC.
A second grouting contract is planned to be advertised in late 2009 for the right rim, right abutment and saddle dam foundation grouting. Barrier wall construction will follow the grouting and is expected to begin in 2011. The total plan cost is US$260M (fully funded costs, which includes inflation) and the project is estimated to be completed by the end of 2014. Technological advancements in grouting and cut-off wall construction have significantly improved over the past decade, especially with the advent of computer monitoring and controls. The Nashville District expects the rehabilitation work to make a safer dam for many years to come.
The project will be funded initially through annual appropriations from Congress. The hydro power produced at Center Hill is marketed by the Southeastern Power Administration (SEPA) to preferred area customers. SEPA is required to repay a share of the costs, based on hydro power usage, to the treasury at the end of construction. Several lake area water supply users are also required to repay a percentage of the rehabilitation costs.
Interim risk reduction measures
Since 2005, the District has modified the lake level operation with the aim of reducing high pools typically experienced in the winter and spring months to help take some stress off the foundation. The target lake elevations are 630 msl in the summer and no lower than 618 msl in the winter. This plan will be reevaluated as construction progresses. Determining the appropriate levels has meant addressing the risk to those downstream, as well as the impacts to those upstream who depend on the lake. Also, environmental impacts play a role in determining lake levels. The process is complex and continual and public safety is always USACE’s highest concern. It has an aggressive dam safety programme, including 24/7 visual monitoring of the dam, increased frequency of instrumentation readings, dam safety training and emergency planning.
The article was written by Linda Adcock, Center Hill Seepage Rehabilitation, Project Manager and Robert Brimm, Professional Geologist, Nashville District of the US Army Corps of Engineers.
For the most up-to-date information on construction, lake information and project news visit: www.lrn.usace.army.mil/CenterHill