Abstract:
Bear Creek Lake Dam in Lee County, Arkansas, is located on a deposit of loess which is approximately 70 feet in depth below the crest of the existing spillway. Loess is a silt sized material deposited by wind. Due to the nature of deposition, the loess is very loose in its natural state and prone to erosion and collapse when wet. The existing reinforced concrete spillway was built in 1938 and significant repairs were performed in 1980 and 1993 due to foundation issues and then completely failed in 2019, likely due to the erosion of the loess subgrade. The site is located within the New Madrid Seismic Zone and the loess is susceptible to cyclic failure during the design earthquake to a depth of about 70 feet below the crest of the spillway. Some of the problems encountered are, there is a relatively small database of earthquake ground motions for the Eastern US as well as a gap in the study of cyclic failure of transitional material/silts in general and, more specifically loess as it relates to geotechnical earthquake engineering. A Site-Specific Seismic Response Analysis was performed, after which the results of three different methods to analyze cyclic failure of soils were compared, which include the liquefaction of sands, the cyclic softening of silts and clays using the simplified method, and finally a method for transitional silts suggested by Dadashiserej et. al (2023). The soils are susceptible to cyclic failure in each method, but to different degrees. A Deep Soil Mixing (DSM) grid was considered the most efficient solution to mitigate the consequences of cyclic failure. The engineering team modeled the design earthquake with the improved ground using QUAKE/W and SIGMA/W and compared this with the numerical model stress reduction approach outlined in Nguyen et al (2013). While the soils onsite are problematic and susceptible to cyclic failure, it was not economically feasible to bring the factor of safety against cyclic failure to above 1.2 using a uniform layout and area replacement ratio. Therefore, considerations were given to allowing the soil inside the cells to liquefy, if the consequences could be mitigated, or increasing the area replacement ratio only in the most problematic zones. Based on our findings, more study is needed on cyclic failure of loess due to seismic activity in the New Madrid Seismic Zone, and a successful design should not exclude the possibility of cyclic failure if the consequences of cyclic failure can be mitigated.