From FERC’s Independent Panel Report
by Becky Denney
On December 14, 2005, the Taum Sauk Upper Reservoir on top of Proffit Mountain near Lesterville, Missouri ruptured, sending 1.5 billion gallons of water down Proffit Mountain — severely damaging Johnson’s Shut-Ins State Park.
The water washed away the home of Park Superintendent Toops, his wife and three children washing them across the road into a field. The children were hospitalized but have recovered. Had campers been in the campground, they, too, would have been washed away. The Taum Sauk Reservoir was a pumped-storage hydroelectric plant which employed twelve people.
Some natural features of the park will not recover in our lifetime. The Shut-Ins themselves do not appear to have suffered great damage. Because of danger from debris, swimming is still not allowed after more than six months, although the public can tour the park.
During the week of December 26, 2005, the Federal Energy Regulatory Commission’s (FERC) Director of Dam Safety, Contantine Tjoumas, formed the Independent Panel of Consultants (IPOC) composed of three engineers . IPOC member Dr. Alfred J. Hendron was asked by FERC to visit the Taum Sauk project before the panel was formed. He inspected the breach and remaining embankment on December 15 with FERC staff from Washington and Chicago. The other two members visited the site on December 28, 2005.
The IPOC was asked to write an independent assessment of the technical causes of the failure of the Upper Taum Sauk Reservoir. The results of the report will be used to review other pumped storage projects certified by FERC which have no spillway. With the release of the report, the IPOC has completed three of the four tasks they contracted to do. The uncompleted task is to assist the FERC staff in remedial measures to re-establish the upper reservoir.
The construction of the Taum Sauk Project
#P-2277 was completed in 1962 and the reservoir first began filling in July 1963. Commercial operation began on December 20, 1963. However, the project license was not issued until August 26, 1965. (I see no mention of this in the IPOC report, but Union Electric Company built and operated the plant in 1963 without a federal license, arguing that it would not affect commerce on a navigable stream. The U.S. Supreme Court in May 3, 1965 held that a federal license was required.)
Project Description
The reservoir consists of a continuous hilltop dike which is a concrete-faced dumped rockfill dam (CFRD) from the foundation to elevation 1570 feet. Between Elevations 1570 and 1589 the construction is rolled rockfill. The crest at 1589 is twelve feet wide. A ten feet high, one foot thick reinforced concrete parapet wall extended the crest to elevation 1599 feet as it was originally constructed. But since 1963, the rockfill embankment at various points has settled between one and two feet. On November 6, 2004 the low point on the top of the reservoir at Panel 72 was at elevation 1596.99 feet.
The over-pumping protection systems were changed in November, 2004 when the geomembrane liner was installed to reduce reservoir leakage. The old reservoir control systems were anchored to the concrete face prior to 2004.
“The new system was not anchored to the concrete face because it was decided that the new geomembrane liner should not be penetrated by anchor bolt holes. The HDPE pipe housing the pressure instruments was not positively anchored to the concrete face slab.” (Page 8) There was also an emergency water level protection backup system for the reservoir.
Design Features
The design and construction of the CFRD for Taum Sauk Upper Reservoir was similar to several older CFRDs such as Strawberry Dam and Salt Springs Dam constructed in California. These dams have parapet walls for reflecting waves at normal maximum water storage level, but the maximum water levels are always approximately one to two feet below the crest of the rockfill.
“The design decision made for Taum Sauk Upper Reservoir Dam to routinely store water 6 to 8 feet high on a 10 feet high parapet wall during daily operations made the Taum Sauk dumped rockfill CFRD ‘Unprecedented’ as compared to the previous CFRDs, as summarized by Cooke, 1988 . . .” (Page 10).
Nearly 100 percent of CFRDs prior to 1963 were dumped and many had cracked face slabs and high leakage. This is why no CFRDs were build between 1940 and 1950. Taum Sauk Reservoir was the last newly constructed dumped rockfill CFRD in the U.S.A. Another CFRD, Cabin Creek Upper Reservoir Dam, Colorado was designed at the same time, but it was designed as a compacted rockfill.
The maximum operating level of Cabin Creek CFRD is six feet below the rockfill crest. And, this means the maximum operating level is nine feet below the top of a three feet high parapet wall that is on the crest of the dam.
“The differences in the Taum Sauk and Cabin Creek CFRD designs represent differences in risk tolerances for different engineering firms and individual consultants during the same time frame taking into account the state of the art for CFRD design in the middle 1960’s. It should also be noted that Cabin Creek Dam was overtopped by pumping, but did not fail.” (Page 10).
Standard Operating Procedure
Taum Sauk Project #2277 is called a peaking and emergency reserve facility. In the summer it might generate in the morning, pump from the lower reservoir in the afternoon, generate in the evening and pump again in the early morning. In fall, winter, and spring it would pump less, maybe pumping at night and generating during the day.
Generation, pump-start and duration was determined by system needs and controlled from AmerenUE’s Osage Plant by a microwave system. This was under the direction of a load dispatcher in St. Louis. This type of system is very responsive and can be put on full load in a few minutes.
When the geomembrane liner was installed in 2004 with a new water level monitoring and control system, one pump was set to stop pumping at Elevation 1594 and another at Elevation 1596. But by October 7, 2005 movements of the HDPE pipes housing the pressure transducers in the reservoir had been observed by AmerenUE employees. After October 7, the shutdown elevations were set at Elevation 1592 and Elevation 1594 respectively. An automatic shutdown for both was set at 1594.2 if they weren’t shutdown already.
Primary Causes of Failure of Taum Sauk Dam
The primary root causes are those which caused the overtopping to occur. Any one of five factors might have prevented the overtopping on that particular date, December 14, 2005.
- Water Level Readings Inaccurate
From August 2005, water level plots show erratic behavior that increased until December 14, 2005. The evidence suggests that the pump discharge pattern created substantial forces acting on the protective pipes or the support cables—especially at lower water levels. The cables “as found” after the reservoir collapse show mis-alignment.
On December 14, 2005, Pump Unit #2 was shut down automatically at 4:39 a.m. at a reading of upper reservoir water level of Elevation 1591.6. At 5:15 a.m. Pump Unit #1 was shut down manually from the Bagnell Dam Center just shy of its automatic shutdown. When Unit # 1, the 2nd pump, was shut down the pressure transducer read a reservoir water level of Elevation 1593.7. Since the reservoir did overtop and the lowest elevation of the parapet wall is 1597, the pressure transducer signals were wrong. There is evidence from comparison to the Penstock transducer (used for static measurement in the plant) on December 13 that the pressure transducer might have read as much as 4.2 feet too low. - Emergency Back up
The emergency back up level probes were set at an elevation above the lowest points along the parapet wall. At Panel 58, the Hi-Hi Warrick Probe was set at Elevation 1597.7 where the top of the parapet wall was 1598.0.
“It did not apparently occur to those setting this probe that there were 33 wall panels with their tops lower than the Hi-Hi probe with the lowest one (Panel 72) having a top at Elev. 1597.0.” (Page 35).
So, the way the back-up system was set, there would be a number of panels overtopped before it kicked in. And it didn’t kick in on December 14. - Storage of Water on Parapet Wall
The third factor is that normal operation called for water levels to be as high as one foot below the top of the parapet wall. This required accuracy from the water monitoring system installed in 2004 that was unrealistic.
“The adoption of this 1 foot free board was totally inconsistent with having personnel making key design and installation decisions who were not even aware of the lowest elevation of the parapet wall within the nearest 1 foot.” (Page 36).
The IPOC believes that storing water against a parapet wall on a dumped rockfill dam increases the number of potential modes of failure. For instance, overtopping of a ten feet high parapet wall can cause high velocity of water on the dam crest and increase the erosion to the point that the parapet panel tips over. - No Visual Monitoring
The IPOC states that there was almost no visual monitoring or “ground-proofing” of the Upper Reservoir water levels to verify where the water levels reached on the parapet wall. But managers had every indication that they should be physically checking the water levels. So, during the liner installation that was completed on November 15, 2005, “new visual level indications were painted on the liner reflecting true elevations.”
E-mails indicate the plant manager and workers knew there were problems and did lower the max operating water levels. Between September 25, 2005 and November 23, 2005, workers who witnessed overtopping reported that the gage piping support system failed and was bent. A new tie-down system was ordered and received, and it is noted that a diver may not be available to install it until the end of the year.
The IPOC presents a measured analysis that shows the procedures for safe operation were so inadequate they range from ignorance to criminal. - Omission of Spillway
The IPOC believes that the omission of a spillway from the design was a “most important root cause.” Had there been a spillway with the capacity of the two pumps, there would not have been an overtopping.
Secondary Root Causes
The secondary root causes contributed to making the embankment more vulnerable to overtopping. Overtopping is one of the most frequent causes of embankment dam failure. Certain characteristics cause rockfill dams to be sensitive to failure from overtopping. These are steepness of the downstream slope, compactness of the rockfill, and the percentages of fines and sand. The Cabin Creek Dam in Colorado was built about a year after Taum Sauk with a somewhat flatter slope and well compacted clean granite rockfill. It was overtopped by over pumping but did not fail. The “dirty” rockfill at Taum Sauk had as much as 45 percent sand and was not free-draining for flows imposed by overtopping.
The failure of the Gouhou Concrete Face Sand and Gravel Dam in China occurred as long ago as August 1993. It was constructed of well compacted gravel but contained about 40 percent sand. The Gouhou dam, like Taum Sauk Upper Reservoir, had “dirty” rockfill. Its failure was due to leakage through a concrete face and parapet wall-face joint into an embankment fill that was not free-draining. This may have also happened at Taum Sauk.
Taum Sauk Upper Reservoir had a long history of settlement and leakage with many periods of concern and repair, but the embankment and parapet wall successfully held water for 42 years. The IPOC believes the geomembrane liner installed in 2004 made the dam more stable than it had been. But the steep rockfill embankment was just “marginally stable” because of the dirty, dumped rockfill and the seepage that had occurred. There was no margin for the erosion caused by overtopping and the pore pressures that happened on December 14, 2005.
The report can be found at http://www.ferc.gov.