Sinkholes and Seepage: Embankment Repair at Hat Creek 1

When a sinkhole caused a seepage issue at the Hat Creek No. 1 forebay, a geotechnical investigation and a unique engineering solution were required to design and construct a solution to the problem that would prevent future sinkholes and seepage.

By Robert Bowers, Kevin Burlingham and Joseph Sun

Pacific Gas and Electric Company’s Hat Creek No. 1 and Hat Creek No. 2 Hydroelectric Project was constructed in 1921. In 2002, a new Federal Energy Regulatory Commission operating license for the project allowed for a 1-foot increase in the forebay water elevation. The water surface elevation was raised to this height in 2005, and within a year seepage was evident at the toe of the northern embankment. PG&E made several attempts between 2006 and 2009 to eliminate the seepage.

On March 14, 2012, a sinkhole was discovered on the crest of the embankment that led PG&E to remove the powerhouse from service and drain the forebay to determine the source of the problem and craft a long-term solution. An in-depth geotechnical investigation by PG&E and AMEC Environmental & Infrastructure was followed by a unique engineering solution to address the restrictive site and environmental constraints while correcting the larger issue.

sinkhole at hat creek

Background of the issue

The northern embankment in the forebay of the Hat Creek No. 1 facility is 310 feet long, with the maximum section having a height of 14 feet. Before 2005, the normal maximum water surface elevation was 3,206.7 feet. FERC issued a new license for the Hat Creek project on Nov. 4, 2002. In 2005, PG&E raised the normal maximum water surface elevation to 3207.7 feet, as required by the new FERC license.

Soon after the increase in water surface elevation, PG&E personnel discovered new and consistent seepage along the eastern portion of the landside toe of the northern embankment. Most of the waterside forebay embankment slopes were lined with original construction rock and mortared rock covered by several layers of gunite (or shotcrete) applied over the life of the project. A combination of many animal burrows on both the landside and waterside of the berm and construction practices of the time were thought to have contributed to the seepage problems.

After a site reconnaissance by the PG&E Geosciences Department in 2006, visual observation led to the conclusion that the seepage likely occurred through the lining, which was cracked. To remedy this problem, the forebay elevation was lowered below the level where the observed seepage stopped, and a shotcrete lining was applied by PG&E General Civil Construction crews over an approximately 12-foot-wide by 300-foot-long section of the northern embankment’s waterside slope. The repair was keyed into the existing lining starting at about 1 foot below the water surface elevation. The shotcrete was applied up to the crest of the embankment. This repair appeared to slow, but not completely stop, the observed seepage once the water level returned to normal maximum operating elevation.

In February 2009, seepage had once again increased as measured by a Parshall Flume installed at the site. A site reconnaissance and evaluation was conducted by the PG&E Geosciences Department to develop recommendations for mitigating the seepage. The reconnaissance revealed that the discharge occurred along the landside interface between the embankment fill materials and native bedrock elevation, providing evidence of minor piping and/or erosion of materials from the dam. Additionally, plant operations personnel had observed since 2005 that the seepage noticeably decreased when the forebay reservoir water level was lowered by as little as 1 foot. Appropriate notifications to FERC and fish and game agencies were made shortly after the site reconniassance, and the forebay elevation was lowered by 1 foot, which caused the seepage to cease completely. The importance of the sensitivity of this 1-foot forebay level change affecting the seepage was not fully recognized. An extensive investigation to find the root cause of this unusual sensitivity was not performed.

In response to the increased seepage, the PG&E Geosciences Department designed a landside earthen berm and seepage collection system in July 2009 to capture seepage coming through the northern forebay embankment and stop the migration of fines. This design assumed a dominant seepage path transverse to the dam axis. The design did not take into account the possibility of undocumented elements buried in the dam that could facilitate seepage longitudinal to the dam axis.

In July 2009, PG&E General Civil Construction constructed a 40-foot-long berm near the embankment’s maximum section. Once the berm construction and drainage work were completed, the water level in the forebay was returned to the normal maximum water surface elevation of 3,207.7 feet.

Seepage was not visible at the landside toe of the embankment but instead was being captured in the new embankment drain system. Observed seepage was less than 1 gallon per minute.

However, in January 2011, seepage reappeared, this time at the toe of the embankment near the intake structure where the berm and drainage measures were not constructed. After consultation with the agencies and a FERC notification, the water surface elevation was again lowered 1foot in January, and again the seepage ceased.

In July 2011, PG&E Geosciences contracted with AMEC to develop repair alternatives. Based on the construction, repair, and seepage history, several contributing conditions were identified: possible leakage through gunite construction joints and/or interfaces with mortared rock or through features beneath the existing lining and animal burrows. AMEC suggested repair alternatives that could be implemented using a phased approach, from least intrusive methods to test pits within the berm section. By early fall 2011, with the forebay water surface elevation still at reduced, a plan to implement the repairs was developed. The repair alternatives had yet to be implemented by the time a sinkhole was discovered.

location of the seepage at the Hat Creek No. 1 forebay embankment, as well as where the new berm was built to prevent seepage
This photo shows the location of the seepage at the Hat Creek No. 1 forebay embankment, as well as where the new berm was built to prevent seepage.

Studies of the site

Upon discovery of the sinkhole on March 14, 2012, and subsequent draining of the forebay, PG&E and AMEC engineers visited the site to observe the location and condition of the sinkhole. Once the forebay was drained, PG&E and AMEC engineers visited the site to observe the location and condition of the sinkhole. The sinkhole (about 3.5 feet long by 2 feet wide x 5 feet deep) was located near the interface of the original berm and the 2009 repair berm. During the site visit, the sinkhole was photographed, measured and probed with a hand soil probe to evaluate the character of the soil. Numerous animal burrows were observed on the landside slope of the embankment. An interview with a PG&E crew member also revealed that while the forebay was being drained, water was seen on the waterside face of the berm at several locations along the joint between the new and old shotcrete.

To determine both the cause of the seepage and the final repair alternatives, PG&E and AMEC performed an in-depth comparison of the original design drawings with existing conditions, which showed a couple important factors to be considered. First, the original forebay lining had been overlaid with various generations of gunite or shotcrete, indicating that the embankments had a history of seepage. Second, as much as 6 to 8 feet of soil/sediment were located on the landside slope of the embankment over the years as a result of sediment collecting in the forebay.

To support the development and design of the repair measures, PG&E initiated a physical site investigation to develop an understanding of the cause(s) and extent of the sinkhole as well as the character and composition of the northern embankment and the underlying foundation materials.

Exploration of the sinkhole was accomplished using a phased approach. In Phase I, California Push Technologies, performed 21 cone penetrometer test (CPT) soundings to help develop an understanding of the extent of the sinkhole and the character and composition of the northern embankement and underlying foundation materials. Shallow test pits were also excavated and logged in the vicinity of the historically observed seepage. The CPTs showed that the embankment material in the vicinity of the sinkhole and intakee was indicative of very low-strength soil or voids.

Excavation of the test pits demonstrated that the original embankment was constructed without significant foundation preparation, resulting in seepage coming from thin lenses of sand between old fill and recent fill.

Phase I also included visual inspection of the slopes of the embankment, where seepage was found to still be occuring near the sinkhole, even though the forebay reservoir had been drained for more than a week. No evidence of piping of the foundation or embankment materials was observed around the northern embankment. The outlet pipe from the 2009 berm seepage collection system was dry.

Phase II consisted of backfilling the sinkhole with low-strength slurry concrete material, excavating and logging an exploratory pit (about 40 feet long by 20 feet wide by 8 feet deep) centered on the sinkhole, and excavating a “slot trench” (about 20 feet long by 4 feet wide by 14 feet deep) at the bottom of the pit to the embankment foundation. Additional test pits were excavated along the landside toe, west of the sinkhole. A FERC representative was present during the investigation.

This portion of the investigation showed dozens of animal burrows in the vicinity of the sinkhole. Dredged sediment was also identified on the landside of the northern embankment. All fill observed was relatively soft and loose, silty sand that would be susceptible to animal burrowing. The Phase II investigation also revealed that the waste soil/sediment on the landside slope of the northern embankment was placed directly on a rock facing slope protection that was likely a continuation of the existing exposed rock facing nearest the intake structure. The rock facing was encountered in the bottom of the exploratory pit.

The test pits excavated along the landside toe of the original northern embankment encountered loose/soft fill that was prone to caving and susceptible to animal burrowing. Rock facing similar to that revealed by the exploratory pit and slot trench was encountered in all but the westernmost test pit.

Based on the northern embankment’s performance history and the Phase I and Phase II field investigations, it was determined that the sinkhole was caused by a combination of factors, as listed below.

Deterioration of the original northern embankment caused by animal burrowing played a role because once the forebay elevation was raised by 1 foot, the dens were flooded and the unobstructed pipe from the lake to the animal burrows provided a seepage path through the embankment. Further lateral seepage paths were seen in the buried rock facing that was present on the landside slope of the original embankment. This rock facing likely allowed the seepage to eventually bypass the 2009 berm seepage collection system. Erosion of the embankment fill through the voids in the waterside shotcrete/gunite lining likely shortened the seepage path through the embankment fill. Additional weight of the 2009 berm acting on the waste soil and sediment that had been placed on the landside embankment slope could have contributed to the collapse of the loose and weakened embankment fill and the formation of the sinkhole.

The final repair included drivings sheet piles down to the bedrock
The final repair included drivings sheet piles down to the bedrock, removing the 2009 berm, constructing a sub-drainage system, building a wider, more stable embankment, and more.

Choosing a solution

Both near-term and long-term repair scenarios were explored to address the deteriorated condition of the northern embankment. The goal was to mitigate the leakage, refill the forebay and return the powerhouse to service as quickly as possible while a long-term solution was designed for future implementation.

Because the exploratory test pits in Phase II of the site investigation revealed favorable foundation conditions, a sheet pile wall was considered as a short-term means to provide a seepage cutoff barrier within the embankment. Because it appeared that the seepage was occurring through the uppermost embankment fill, the sheet piles would not need to extend significantly into the underlying foundation materials to reduce the observed leakage. A system of subsurface drains and instrumentation would also have to be installed to monitor seepage and the phreatic surface within the embankment. It was also recommended that the forebay water level continue to be restricted by 1 foot below the maximum under this repair.

The long-term repair scenarios centered on either a complete rebuild of the northern embankment or construction of a setback embankment on the landside slope of the original northern embankment. The setback embankment would also require internal seepage control and monitoring measures.

Several factors contributed to the selection of the final repair method. Design alternatives that concentrated on the landside of the embankment were desirable from a schedule standpoint because this would reduce work within the waterway and therefore the need for long lead permits and agency approvals. Any short-term repair options, expected to be in place for no more than one to two years, would likely require forebay water level restrictions and careful performance monitoring. From an operating perspective, it was desirable to implement a long-term remediation to limit future outages, forebay drains and subsequent fish rescues in this popular camping and fishing location.

To eliminate future outages for long-term repairs and limit the total area of disturbance, the project teamdecided to implement a combination of the near-term and long-term repair recommendations while the forebay and canal were already drained.

The final repair method consisted of driving sheet piles through the center of the original embankment down to the bedrock foundation; removing all of the 2009 berm and a portion of the original embankment; exposing the bedrock foundation in the excavated area; constructing a sub-drainage system and placing a sand filter on the exposed original embankment slope; placing and compacting imported embankment fill to create a wider, more stable embankment; and installing instrumentation along the crest of the embankment.

Once a design scheme was chosen, AMEC performed analyses to support the design of the northern embankment repairs. These analyses included compatibility of filter/drain materials with the existing embankment fill and foundation soils, and stability analyses to establish the size and configuration of the new landside berm and assess performance during earthquake shaking.

Early in the design process, the length of the sheet pile wall and new landside berm had to be determined, based on the past performance, findings from the site investigation, and topography of the landside slope. For the eastern extent of the sheet pile wall, it was decided that in order to minimize the risk of damaging the existing intake structure foundation, the sheet pile wall would start about 2 feet from the intake structure wing wall. The seepage that would travel through this gap would be intercepted by the filter and drain in the new landside berm. To minimize any deformations during a seismic event and help mitigate any future issues with the embankment, the new landside berm was designed to widen the crest and have a 2(H):1(V) landside slope.

Following the preliminary design of the repairs, the stability of the repaired northern embankment was evaluated for representative and potentially critical loading conditions, including static loading during construction, at end of construction, during steady-state seepage, and immediately following rapid drawdown, as well as seismic loading during steady-state seepage. The benefit the sheet pile cutoff wall might have on the stability of the temporary and permanent waterside and landside slopes of the analyzed sections was conservatively not considered in the stability analyses.

Embankment cross-sections at two locations were developed for the stability analyses. One section represented the northern embankment’s maximum section. The other section was located near the intake structure where the waterside slope steepens to about 0.7(H):1(V) and represents a small portion of the northern embankment, roughly 20 to 30 feet.

The minimum static factor of safety for the embankment sections was for the long-term case and corresponds to a potential slip surface located on the waterside slope where steepened slopes exist. This computed factor of safety is less than that required for acceptable performance, although the long-term performance of this steep slope has been satisfactory. The forebay was drawn down several times in its history of operation, most recently when the sinkhole was first discovered, and the performance of this waterside slope was satisfactory. The satisfactory performance reflects the conservative strength parameters and assumptions used in the analysis.

The stability of the embankment immediately after excavation of the waste soil/sediment that exists on the landside slope of the northern embankment was also evaluated. Because the factor of safety was marginal for this case, it was recommended that heavy equipment should not be allowed access to the top of the temporary cut slope during construction of the repairs.

A quality control and inspection program was developed to ensure the materials and construction activities were inspected and tested to the correct specification and frequency. Additionally, it was decided to have the engineer-of-record (or representative) onsite during construction to monitor quality control, answer questions regarding the design and observe material conditions as they were encountered. Piezometers were also specified to be installed through the crest of the embankment, with readings required on a monthly basis for the first two years after embankment repairs are implemented. Regular monitoring of the piezometers can be used to evaluate the effectiveness of the sheet pile cutoff wall and whether or not adverse seepage conditions are developing.

In addition to the engineering analysis completed as a basis for the design, a temporary construction potential failure mode analysis workshop was conducted jointly by PG&E, AMEC and FERC. The workshop participants postulated a number of potential failure modes for the various loading conditions for the northern embankment during construction.

For each failure mode that was deemed credible, the participants listed the factors that made it more likely to occur (adverse factors) and those that made it less likely to occur (favorable factors). Each adverse condition was then mitigated with risk reduction measures that were carried forward into the design process.

A more stable embankment was constructed at the Hat Creek No. 1 facility
A new, wider, and more stable embankment was constructed at the Hat Creek No. 1 facility once the sub-drainage system was installed.

Construction process

The investigation, design and agency consultation phases of this project were extensive. With appropriate permits and approvals in place, construction commenced on Nov. 5, 2012. PG&E’s Hydro General Civil Construction was the construction lead for the work. Beginning an earthwork project late in the year proved to be challenging due to wet and frozen conditions typical for the region. Many adjustments to schedule, construction methods and materials were necessary to maintain quality control while achieving the desired embankment remediation.

After several weeks of weather delays, and following removal of a significant portion of the original embankment along with any rock fill/facing that was buried, the bedrock foundation for the new landside berm was exposed and prepared for placement of embankment fill. The bedrock foundation was proof-rolled (rolled over with heavy equipment to ensure it was sound) and any frozen material was excavated before placement of any embankment fill. The original embankment slope that remained after the completion of the excavation was compacted using an HED shaker vibratory plate compactor mounted on an excavator arm.

The sub-drainage system was installed after the foundation was exposed. The collection pipe consisted of 6-inch-diameter corrugated high-density polyethylene (HDPE) pipe. A new seepage monitoring weir was placed at the east end of the system that matched the existing weir at the site. The sub-drainage system was split into two sections (east and west) so as to be able to determine in the future through which section of the embankment the seepage was occurring.

After completion of the sub-drainage system, embankment fill was placed and compacted on the landside of the northern embankment to form the new landside berm. The imported embankment fill consisted, at first, of a red clayey sand at the base of the new landside berm followed by a brown silty sand with gravel for the upper portion of the berm.

Overnight temperatures during the embankment construction were below 0 degrees F, while daytime temperatures rarely reached above 10 F. These temperatures caused the exposed embankment to freeze to a depth of 6 to 9 inches, which necessitated that any frozen material on the original embankment had to be removed before the granular filter material was placed on the slope.

Three piezometers were installed along the crest of the repaired embankment, on the landside of the sheet pile wall, to observe any seepage through the cutoff wall. All three were installed down into the weathered bedrock to a tip elevation of 3,198 feet. Readings of the water levels in the piezometers were taken before and during the refilling of the forebay. Work was completed in March 2013.


This work successfully repaired the conditions that led to the creation of the sinkhole and mitigated future seepage.

Owners of aged earthen canals and forebays are accustomed to finding areas of seepage and leakage. Recognizing changes in seepage conditions after an operational change is vital to avoiding a potential failure.

In the case of Hat 1, fluctuations in the forebay elevation showed the condition to be elevation-dependent but did not give indication of where the seepage was entering the berm. Recognizing when to lower the forebay elevation or call for a complete draw-down of the canal due to changes in seepage locations or increases in seepage volume greatly reduced the risk of a potential failure and subsequent environmental damage.

The investigation to determine the cause of the Hat 1 sinkhole was critical to determining the appropriate repair and its limits. Previous efforts to compare as-built conditions with original design drawings were not thorough enough to reveal potential sources of the seepage and cause of the subsequent sinkhole.

Additionally, the first two attempts to repair the seepage were completed with limited knowledge of subsurface conditions. The need to “dig deeper” to find out what is leading to the observed conditions is critical. Although the previous repair attempts may have been appropriate for some applications, it was eventually apparent that an “off-the-shelf” design would not work for this site without revealing the cause of the seepage and its extent.

Having a representative of the engineer-of-record onsite during construction allowed for decisions regarding materials, methods and design alterations to be made in a timely manner. Assembling an experienced and knowledgeable team familiar with all aspects of dam construction and remediation was essential. As with any project, a technically diverse team will provide the best solutions to persistent and difficult issues.

Robert Bowers is a civil engineer with Pacific Gas and Electric Company. Kevin Burlingham is a project engineer at AMEC Environmental & Infrastructure. Joseph Sun is principal geotechnical engineer at PG&E.

Previous articleHydro Currents
Next articleCost-Effective Passage Design for High-Head Hydro Facilities

No posts to display