Advancing internal erosion monitoring using seismic methods in field and laboratory studies

Abstract

This dissertation presents research involving laboratory and field investigation of passive and active methods for monitoring and assessing earthen embankment infrastructure such as dams and levees. Internal erosion occurs as soil particles in an earthen structure migrate to an exit point under seepage forces. This process is a primary failure mode for dams and levees. Current dam and levee monitoring practices are not able to identify early stages of internal erosion, and often the result is loss of structure utility and costly repairs. This research contributes to innovations for detection and monitoring by studying internal erosion and monitoring through field experiments, laboratory experiments, and social and political framing The field research in this dissertation included two studies (2009 and 2012) of a full-scale earthen embankment at the IJkdijk in the Netherlands. In both of these tests, internal erosion occurred as evidenced by seepage followed by sand traces and boils, and in 2009, eventual failure. With the benefit of arrays of closely spaced piezometers, pore pressure trends indicated internal erosion near the initiation time. Temporally and spatially dense pore water pressure measurements detected two pore water pressure transitions characteristic to the development of internal erosion, even in piezometers located away from the backward erosion activity. At the first transition, the backward erosion caused anomalous pressure decrease in piezometers, even under constant or increasing upstream water level. At the second transition, measurements stabilized as backward erosion extended further upstream of the piezometers, as shown in the 2009 test. The transitions provide an indication of the temporal development and the spatial extent of backward erosion. The 2012 IJkdijk test also included passive acoustic emissions (AE) monitoring. This study analyzed AE activity over the course of the 7-day test using a grid of geophones installed on the embankment surface. Analysis of root mean squared amplitude and AE threshold counts indicated activity focused at the toe in locations matching the sand boils. This analysis also compared the various detection methods employed at the 2012 test to discuss a timeline of detection related to observable behaviors of the structure. The second area of research included designing and fabricating an instrumented laboratory apparatus for investigating active seismic wave propagation through soil samples. This dissertation includes a description of the rigid wall permeameter, instrumentation, control, and acquisitions systems along with descriptions of the custom-fabricated seismic sensors. A series of experiments (saturated sand, saturated sand with a known static anomaly placed near the center of the sample, and saturated sand with a diminishing anomaly near the center of the sample) indicated that shear wave velocity changes reflected changes in the state of stress of the soil. The mean effective stress was influenced by the applied vertical axial load, the frictional interaction between the soil and permeameter wall, and the degree of preloading. The frictional resistance was sizeable at the sidewall of the permeameter and decreased the mean effective stress with depth. This study also included flow tests to monitor changes in shear wave velocities as the internal erosion process started and developed. Shear wave velocity decreased at voids or lower density zones in the sample and increased as arching redistributes loads, though the two conditions compete. Finally, the social and political contexts surrounding nondestructive inspection were considered. An analogous approach utilized by the aerospace industry was introduced: a case study comparing the path toward adopting nondestructive tools as standard practices in monitoring aircraft safety. Additional lessons for dam and levee safety management were discussed from a Science, Technology, Engineering, and Policy (STEP) perspective.