Using geotextiles to repair earth dams
Geosynthetics | April 2011
By Benjamin C. Doerge, Trent Street, John Chua, Rex Stambaugh, and James McHenry
The Natural Resources Conservation Service (NRCS) has made innovative use of geotextiles to repair several cracked earth dams.
This paper presents how geotextiles were used to repair three dams in Texas, Arizona, and Colorado. The geotextile performs different functions in each of these three dams, all of which are dry structures.
In the Texas dam, the geotextile acts as a chimney filter and was located in the downstream portion of the dam. In the Arizona dam, the geotextile is designed to span existing or future cracks in the earth face on the downstream side of the chimney filter and to prevent loss of drainfill material into the cracks. In the Colorado dam, a structural filter of rock and sand fill sandwiched between two layers of geotextile is used to stop the propagation of cracks from the existing fill and foundation into the newly reconstructed upstream zone of impervious earthfill, as well as to provide a filter layer downstream of this impervious zone.
Earth embankments can become cracked as a result of various mechanisms, including desiccation, differential settlement, foundation collapse, and subsidence.
Cracks can be either transverse or longitudinal to the axis of the dam. Transverse cracks pose a greater threat to dam safety because they provide a direct pathway for uncontrolled flow from the reservoir to the downstream face of the embankment, leading to increased potential for internal erosion and possibly contributing to dam failure.
However, longitudinal cracks cannot be considered benign either, because they can connect otherwise discontinuous cracks and create additional flow pathways through the dam.
The repair of cracked earth embankments typically involves either the installation of:
- an impervious zone or membrane to cut off seepage flow through the dam; or
- a filter layer to prevent the uncontrolled migration of water-borne soil particles from upstream to downstream through cracks in the embankment.
The presence of open cracks in an earth dam creates challenges regarding filter compatibility in the design of any repair system.
The Natural Resources Conservation Service (NRCS) has made innovative use of geotextiles to repair several cracked earth dams. This article describes how geotextiles were used to repair:
- Olmitos-Garcias No. 2 Dam in Texas;
- Florence Dam in Arizona; and
- Cañon C-4 Dam in Colorado.
All three are single-purpose for flood control and are dry structures. The geotextile performs different functions in each. These functions include:
- preventing soil movement through cracks.
- spanning existing or future cracks in the embankment and/or foundation.
- preventing the loss of filter material into existing or future cracks.
- stopping the propagation of cracks from the existing fill and foundation into adjacent uncracked zones.
The geotextile features in these three structures were installed in three different locations within their respective embankments, depending on the intended function, including in the upstream and downstream portions of the embankment and at the embankment centerline. This article presents key features regarding the investigation of the cracking and the design and construction of the repairs, with special emphasis on the geotextile components.
NRCS policy is to not use geosynthetics in embankment dams except in noncritical situations where the safety of the dam would not be adversely affected if the geosynthetic failed to fully perform as intended. A concern with geotextiles is the possibility that they may become excessively clogged over time. However, in these three dams, the functioning of the geotextiles would not be compromised even if they became clogged.
Furthermore, because these dams are dry structures, the chance of the geotextiles becoming clogged is likely much lower than for dams with permanent storage and continuous seepage. For these reasons, the use of geotextiles was allowed on these projects. The geotextile functions described in this paper are presented as applicable only to dams with no permanent storage.
Olmitos-Garcias No. 2 Dam | Starr County, Texas
Numerous cracks and sinkholes were discovered in the embankment during a 1999 inspection. The cracks were both transverse and longitudinal. The sinkholes had formed through erosion of soil into cracks and were found on both the upstream and downstream slopes and on the crest (Figure 2).One 76mm-(3in.-) diameter hole was measured to be more than 7.6m (25ft) deep.
Insects and other animals, including badgers, were using the cracks and holes to facilitate burrowing into the embankment. Numerous mesquite trees were found growing on the embankment and mesquite roots had penetrated into many open cracks to considerable depths.
Soil investigations upstream and downstream of the embankment discovered that the alluvium in the foundation was low in density (1.4 g/cm3) and contained pinholes ranging in diameter from 1.6–6.4mm (1/16–1/4in.). Screening procedures based on density and other index properties indicated that the top 3.7–4.3m (12–14ft) of the foundation alluvium contained collapsible materials. Test pit excavations revealed that the cracks tended to increase with depth, up to a width of 51mm (2in.).
This finding indicated that the top of the embankment was in compression while the bottom was in tension, suggesting that the embankment foundation had experienced subsidence. Based on all the evidence, it was concluded that the cause of the cracking was collapse of the foundation upon wetting during periodic storage of water in the reservoir following runoff events.
Several large-scale infiltration tests were performed from the top of the embankment to determine the extent of the cracking. Three 4.6m- (15ft-) deep test pits were excavated into the top of the embankment and filled with gravel. Both transverse and longitudinal cracks were observed in the walls of the pits. Water was pumped into each pit at a rate of 0.05 m3/s (775gpm) until the pit was full or reached a constant level.
In all cases, the pits emptied quickly, and generally little or no flow emerged from either the upstream or downstream slopes adjacent to the pit. Following the infiltration tests from the top of the dam, two test pits were excavated near the downstream toe of the dam to a depth of 4.9m (16ft) to determine if seepage waters had migrated downstream and below the surface of the dam. No sign of seepage water or the dye introduced into the upper test pits was found at either location.
Based on the results of the infiltration tests and other observations at the site, it was clear that an extensive system of cracks had developed in both the embankment and the foundation alluvium.
Geologic investigations indicated that the top 0.6–0.9m (2–3ft) of the claystone in the foundation was highly fractured. Therefore, the design feature to control the possible migration of soil particles through the embankment and foundation needed to extend from the top of the embankment down to unfractured claystone and from abutment to abutment. Three alternatives were considered (NRCS, 2002):
- chimney filter on the downstream slope of the embankment and into the foundation.
- impermeable zone or membrane in the upstream slope of the embankment and into the foundation.
- geotextile placed in the downstream slope and into the foundation.
For reasons of economy and constructability, the geotextile option was chosen. A nonwoven, needlepunched geotextile was selected for its filtering and elongation properties. A 542 g/m2 (16oz.) fabric was specified to provide adequate tensile strength to span existing cracks or cracks that may develop throughout the life of the structure. Good resistance to root penetration was also desired. At the same time, operation and maintenance guidelines were developed to include control of mesquite and other woody vegetation on the embankment.
The geotextile was placed deep enough within the embankment to ensure stability against internal hydraulic uplift pressures even if the geotextile became fully sealed and was able to maintain full reservoir head on its upstream side. An additional stability berm was installed to the approximate mid-height of the embankment at the downstream toe (Figure 3).The geotextile did not need to function as a drain, so no outlets were required. Its primary purpose was to stop the migration of any soil particles through the dam by its filtering function.
Florence Dam | Pinal County, Arizona
In 1977, cracking in Florence Dam was investigated by means of three test pits (SCS, 1978). In 1982, additional trenching with water testing was performed (SCS, 1982). Cracking of the homogeneous earthfill embankment consists largely of transverse cracks, but several longitudinal cracks have also been observed. The transverse cracks range in width from hairline to 51mm (2in.) and are generally widest at the surface and decrease in width with increasing depth. The transverse cracks are visible at the crest and are seen for nearly the entire length of the dam, with four large concentrations covering 25% of the total length. This cracking pattern is consistent with tensile cracking due to desiccation.
The maximum depth of cracking observed at any SCS dam in Arizona is about 6.7m (22ft), and the maximum depth at which cracks become hairline is 3.4m (11ft) (NRCS, 2004). The cracking in Florence Dam is thought to be primarily due to desiccation, but some cracking due to collapse has also been suspected by some investigators (NRCS, 2004). Collapsible materials have been found in the foundation outside the footprint of the dam, but engineering judgment determined that its effect upon the existing structure was not significant enough to require mitigation.
In 2004, 15 test pits were excavated along the upstream toe of the embankment to the depth corresponding to the anticipated depth of the proposed centerline filter trench. The purpose of these test pits was to verify that the proposed trench would intercept all cracks through the embankment and foundation and to assess the collapse potential of foundation soils. In all 15 test pits, no cracks were found in the foundation soils. This finding supports the conclusion that the transverse cracking is confined to the embankment.
The repair consisted of a centerline trench filled with filter material to intercept any suspended soil particles being transported through cracks in the embankment.
The 0.6m- (2ft)-wide trench would be excavated to a maximum depth of 9.4m (31ft), or 2.7m (9ft) deeper than the deepest cracks observed in any Arizona dam investigated by the NRCS/SCS. At this depth, the trench penetrated a minimum of 0.3m (1ft) below the stripped ground line along the centerline of the dam. The filter material was designed to be a broadly graded mixture of sand and gravel to give it excellent self-healing properties. It was also less prone to bulking than more uniform sand material.
A nonwoven geotextile was designed to be placed against the upper 3.4m (11ft) of the downstream trench wall prior to filling the trench with filter material, with an additional 1.2m (4ft) secured to the top of the dam with securing pins (Figure 5).The top of the filter was terminated at a depth of 0.3m (1ft) below the top of the dam to allow for 0.3m (1ft) of fine-grained soil cover over the filter.
The primary purpose of the geotextile was to mitigate downstream movement of filter material into any cracks in the trench wall with a width that exceeded the filter material’s D50 size (1.19–4.76mm). The filter material was not designed to be coarse enough to bridge across cracks as wide as those observed at Florence Dam (up to 51mm) wide.
Since cracks narrow to hairline below a depth of 3.4m (11ft), the cracks below this depth are expected to be thin enough to be bridged over by the filter material’s D50 size. Cracks wider than this below a depth of 3.4m are unlikely based on historical observations of cracks. Coverage with geotextile to a depth of 3.4 m over the entire length of the embankment was considered reasonable due to the distribution of transverse cracking identified along the length of the crest (NRCS, 2004).
For the geotextile to perform its intended function, it needed to have adequate tensile strength to withstand the forces that might act on it.
No proven method for analyzing a geotextile’s ability to span open cracks has appeared in technical literature. Therefore, the designer developed a method to check the adequacy of the geotextile (NRCS, 2004). This method is based on consideration of the stresses acting on a failure plane in the filter material at the time of formation of a new crack.
It is assumed that a failure plane must develop in the filter material for a crack to form. The stresses acting on the failure plane in the filter are conservatively assumed to be completely transferred to the geotextile (i.e., with no slippage). The resulting tensile stresses acting on the geotextile are compared to its ultimate tensile strength. The tensile capacity of the geotextile in the vicinity of the crack is estimated to be equal to the ultimate tensile load from an ASTM D 4632 Grab Tensile test (ASTM, 2010).
The design calculations showed that a 203 g/m2 nonwoven geotextile would be adequate to survive the formation of a 51mm- (2in.-) wide crack. The maximum tensile load was calculated to be at a depth of 3m (10ft).
Cañon C-4 Dam | Fremont County, Colorado
In 1991, additional cracks and holes in the dam were noted. In the following years, the condition of the dam continued to worsen. On March 31, 2004, the Colorado Division of Water Resources issued an order to the dam owner to breach the dam because of its apparent unsafe condition. This order prompted the owner (sponsors) to request an immediate investigation by the NRCS to determine the cause of the cracks and holes and to evaluate the feasibility of repairing the structure.
The investigation of C-4 Dam was conducted in April 2004 (NRCS, 2004). Numerous cracks and holes were identified, including both transverse and longitudinal cracks. The holes were discovered to be surface expressions of cracks into which eroded soil had been lost. Lines of holes were found to be connected by cracks that had been obscured at the surface by deposition of eroded or windblown material.
The most prominent features observed on the embankment (Figure 7) included the following defects:
- a line of several dozen large holes—up to 0.9m (3ft) in diameter—along the upstream edge of the crest of the dam (Figure 8);
- transverse cracks over both abutments; and
- a line of small holes—25–152mm (1-6in.) in diameter—parallel to the main axis of the dam, approximately one-third of the way down on the downstream slope of the dam.
Subsequent trenching in May 2004 revealed several additional transverse cracks in the central part of the dam (i.e., not over the abutments).
The primary cause of the cracking was determined to be differential settlement due to collapse of the foundation materials upon wetting. During the original construction of the dam, a surface layer of highly collapsible aeolian/colluvial soil had been removed, but a lower layer of moderately collapsible alluvium was left in place (Figure 9).
Underseepage from periodic impoundments of runoff water had triggered some degree of collapse of the foundation, but the centerline cutoff trench to bedrock largely prevented the wetting front from reaching downstream of the dam centerline. However, a possible seepage window at the base of the cutoff trench was suspected, based on additional drilling. The large settlement under the upstream portion of the dam, coupled with the negligible settlement of the crest of the dam, caused a large, continuous longitudinal crack to form along the upstream side of the crest (Figure 10).This crack gave rise to the line of large holes observed on the crest of the dam.
Other longitudinal cracks were formed in numerous locations as the periodic wetting fronts advanced through the foundation and triggered episodes of foundation collapse. The transverse cracks were judged to be the result of differential settlement over the steep bedrock abutments or between zones of variable compressibility and/or permeability in the main valley.
The line of small sinkholes on the downstream slope was found to be located immediately above the coarse gravel zone of the chimney drain (Figure 11).These holes were attributable to piping of soil into the gravel zone during episodes of downward percolation of precipitation or snow melt water. These holes were judged to be a cosmetic problem rather than a dam safety issue.
Three alternatives were considered for the filter layer to prevent migration of soil through cracks in the dam. All three consisted of a continuous filter extending from the sandstone bedrock to the top of dam and from abutment to abutment. The main variable was the location of the filter layer.
The alternatives considered included (NRCS, 2004):
- a sloping filter blanket near the downstream slope of the dam.
- a vertical filter trench on the dam centerline.
- a structural filter within the upstream slope of the dam.
The downstream location was eliminated because it had the potential to induce wetting of the downstream portion of the foundation, resulting in additional collapse and cracking. This alternative also did nothing to address the questionable integrity of the centerline cutoff trench.
The downstream alternative also involved greater excavation depths because the bedrock dips in the downstream direction. Finally, the downstream alternative would have interfered with the existing toe drain and the principal spillway outlet structure. The centerline trench was not chosen because it would have involved an excavation depth of approximately 21m (70ft) from the top of dam to the top of bedrock. In addition, excavating around the principal spillway conduit would be problematic.
The upstream alternative was selected because it allowed for the placement of a filter layer as well as an upstream impervious zone extending all the way to bedrock (Figure 11). This upstream cutoff would deter any future wetting of the foundation materials, thus minimizing further collapse.
Another advantage of this alternative was that it provided multiple lines of defense against migration of soil particles through the dam, including:
- the recompacted upstream impervious zone limits overall seepage through the dam; and
- the structural filter prevents the loss of any material from the upstream impervious zone in case it should develop cracks in the future.
The existing inlet structure for the principal spillway was far enough away from the embankment that the excavation required for the upstream alternative did not affect it. The excavation would expose the principal spillway conduit, but because the conduit is founded on the bedrock in the right abutment, no serious complications were anticipated.
The structural filter consisted of a 600mm- (2ft-) thick “sandwich” of a sand and cobble mixture between two layers of a 475g/m2 (16oz.) nonwoven geotextile (Figure 12 & Table 1).The filter was designed to perform several functions:
- prevent the migration of soil particles out of the upstream impervious zone and into existing or future cracks in the downstream portion of the dam.
- bridge over existing or future cracks in the original fill and foundation.
- stop existing or future cracks in the original fill or foundation from propagating up into the newly compacted upstream impervious zone.
- cap the existing chimney drain to prevent continued loss of earthfill material into the coarse gravel zone of the drain.
The cobbles in the middle of the filter zone had a D100 of 381mm (15in.) and a D50 of 127–178mm (5-7in.). The cobbles were graded to be large enough to bridge across cracks up to 127mm (5in.) wide in the existing embankment fill. Filter compatibility between the cobbles and open cracks was based on criteria for drainfill placed against perforated or slotted pipe (USACE, 1986). The D50 size of the filter must be at least as large as the crack width.
The heavy, nonwoven geotextile was designed to provide both strength and ductility (elongation) in the event that future settlement occurred in the foundation. The geotextiles also provided filter compatibility between adjacent materials and would divert any seepage from existing cracks within the remaining portion of the original embankment. The sand in the structural filter served to cushion the geotextiles from the angular cobbles and to fill the voids between the individual rocks.
The construction consisted of six main steps:
- excavation of the upstream cutoff trench and level area above existing chimney drain;
- cleaning, inspection, and dental grouting of any cracks found in the bedrock;
- placement of the bottom geotextile on the downstream face of the cutoff trench, extending to the top of the existing fill left in place;
- placement of the sand/cobble material on the bottom geotextile;
- placement of the top geotextile; and
- backfill of cutoff trench and reconstruction of upstream slope of embankment (see Figures 13 and 14).
A mock-up of the structural filter was constructed prior to actual placement to test construction techniques and to provide a template from which the quality of the construction could be measured. The single geotextile and filter sand layer for the top 16ft in the embankment were constructed concurrently with the earthfill placement.
- Geotextiles have been used to construct robust, cost-effective repairs for cracked NRCS earth dams with no permanent water storage.
- Geotextiles can be used for a variety of functions in repairing cracked earth dams, including:
- filtering to prevent migration of soil particles through cracks in the embankment and foundation.
- reinforcement to span cracks and retain filter material.
- preventing the propagation of cracks from cracked zones into uncracked zones.
Ben Doerge, P.E., G.E., USDA-NRCS, 501 W. Felix Street, Bldg. 23, Fort Worth, TX 76115; +1 817 509 3759, email@example.com
Trent Street, P.E., USDA-NRCS, 101 South Main, Temple, TX 76501; +1 254 742 9892, firstname.lastname@example.org
John Chua, P.E., USDA-NRCS, 230 N. First Avenue, Suite 509, Phoenix, AZ 85003; +1 602 280 8838, email@example.com
Rex Stambaugh, P.E., USDA-NRCS, 655 Parfet Street, Room E200C, Lakewood, CO 80215; +1 720 544 2811, firstname.lastname@example.org
Jim McHenry, P.E., USACE, 1645 S. 101st E Avenue, Tulsa, OK 74128; Ph. +1 918 669 7670, email@example.com
ASTM International. (2010). “Test Method for Grab Breaking Load and Elongation of Geotextiles (D 4632).” Annual Book of Standards, Section Four, Volume 4.13. ASTM International. West Conshohocken, Pa.
Natural Resources Conservation Service. (2004). Design Folder, C-4 Embankment Restoration, Cañon Watershed.
Natural Resources Conservation Service. (2004). Design Folder, Florence Dam Repair.
Natural Resources Conservation Service. (2002). Design Folder, Olmitos-Garcias Watershed, Site No. 2 Repair.
Natural Resources Conservation Service. (2004). Engineering Investigation Report, Cañon Watershed, C-4 Dam.
Natural Resources Conservation Service. (2004). Technical Assistance Report, Florence Dam Repair.
Soil Conservation Service. (1982). “Crack Investigation, Florence Retarding Structure, Florence Watershed,” Internal Memorandum.
Soil Conservation Service. (1978). “Cracking of Earth Dams in Arizona,” Internal Memorandum.
U. S. Army Corps of Engineers. (1986). “EM 1110-2-1902, Seepage Analysis and Control for Dams.” Department of the Army. Washington, D. C.