FACTORS CONTRIBUTING TO THE 2005 TAUM SAUK UPPER RESERVOIR BREACH, REYNOLDS COUNTY, MO

The failure of the AmerenUE Taum Sauk Pumped Storage Powerplant upper reservoir on the morning of Dec 14, 2005 occurred because several factors combined to overwhelm system redundancy. These included: 1) failure of recently-placed anchors for water stage instrumentation conduits; 2) differential settlement of the rockfill embankment that negated intended redundancy of the back-up water stage shutoff sensors; and 3) failure to properly assess the cause of instrumentation problems revealed three months prior to the eventual failure. The unintended deflection of the reservoir's instrumentation ducts and differential settlement of the rockfill dike lead to erroneous stage level readings of as much as 1.8 m.

When the pumps failed to shut down around 5 AM on the morning of the failure the pool began overtopping the parapet wall at a rate of about 83 m3/s, in three locations where differential settlement lowered the original level of the perimeter dike (the facility did not have a spillway).

Forensic assessments revealed that when a new HDPE liner was installed in 2004, the anchors appear to have been loosened by cyclic uplift. This process may have been exacerbated by vortex turbulence, due to the proximity of a 8.3 m diameter glory hole oraface, located about 30 m from the transducer array.

The parapet wall created a thin crested weir overflow condition with the water initially falling 3 m, onto the crest of the rockfill embankment. Post-failure observations of the undermined parapet wall a few hundred meters south of the main breach suggest a plunge pool ~9 m deep was excavated along ~150 m of the parapet wall on the NW side of the reservoir before it gave way via overturning.

Post-failure assessments also revealed that most of the rockfill was end-dumped without benefit of mechanical compaction during construction in 1962-63 and contained a surprising quantity of fines and possibly some highly weathered rock. This likely contributed to hydrocompaction-induced settlement of up to 0.6 m. Up to 2,832 l/s of seepage took place prior to 2004, when the liner was installed. Post-failure observations also revealed that fine grained materials had not been completely removed, as specified in the design, which likely contributed to the abnormally high settlement for a rockfill embankment (2.44 % of embankment height).