The 1889 failure of the South Fork Dam: An engineering analysis

The causes of the South Fork Dam failure in 1889 have been the subject of speculation and debate for well over a century. Modifications to the as-built dam including removal of pipes used to control outflow, construction of a bridge over the spillway, addition of a fish guard to the bottom of the bridge, and lowering the dam crest have been hypothesized as contributing factors. A committee established soon after the failure concluded that the dam was doomed from the beginning of its service life. Recent investigations have cast doubt on the historical narrative, though there remains a scarcity of published studies that leverage the full capabilities of current analysis tools. The objective of this work was to quantitatively evaluate the influence of each factor identified as potentially contributory to the failure of the dam. It is motivated by the need to reconsider past events within a modern context, by interpreting and translating the historical record into a contemporary framework of technical analysis, so that the major lessons are remembered and heeded. Current engineering tools including geospatial (ArcGIS), hydrology (HEC-HMS), and hydraulics (HEC-RAS) applications were applied alongside historical publications and testimony to represent the conditions relating to the dam on 31 May 1889, followed by an evaluation of each factor's role with respect to the dam's failure. Results indicate that the dam failed after an estimated average of 6.43 in (163 mm) of rainfall on the contributing watershed, producing a peak inflow to the lake of approximately 6070 ft 3 /s (172 m 3 /s) and overtopping the dam crest at its lowest point by up to 0.6 ft (0.2 m). Neither the bridge nor the fish guard was judged to play a meaningful role in crest overtopping. A proposed auxiliary spillway to the southwest of the dam was considered, but it would have prevented overtopping only if it had been built to mirror the dimensions of the existing spillway. If the pipes had been present and functioning, they could have prevented overtopping of the dam crest by a slim margin, and only if they had been fully opened well in advance of the rainfall. The lowering of the dam crest was the most consequential modification to the as-built dam. The lowest point on the reconstructed dam crest is estimated as 3.1 ft (0.9 m) lower than the original elevation, virtually assuring the dam's failure in response to the 1889 storm. In addition to serving as a case study in technical analysis of historical events, this study reinforces that dams and appurtenant works must be considered and managed as systems. The findings provide an example of how degraded, interdependent components can reduce overall system performance below the threshold of safe operation when linked with specific operating conditions (watershed and rainfall). Of more practical and current relevance, these results reaffirm the value of reliable safety programs and robust oversight of modifications that can affect dam performance, especially for dams having high-hazard potential, and regardless of ownership, location, or age.