Acoustic emission induced by seepage through granular soils

The relations among acoustic emission (AE) activities, flow parameters, and soil properties were studied to develop a database and/or methodology of using AE technique for detection of excessive seepage. Laboratory experiments and data analyses were performed using specially designed data acquisition instruments and computer based data analysis devices. The overall experimental system was composed of a seepage permeameter, a pressurized water supply reservoir, a hydrophone, signal conditioner components, a data acquisition system, and a computer based data analysis system. In addition, an isolation chamber with steel-wired shielding screen was used during the tests to reduce the background noise and electromagnetic effect.; Three limestone gravel samples with very uniform gradations were investigated. For each soil sample, four specimens with different densities were prepared; and six levels of hydraulic gradients were applied to each test specimen. Thus, a total of 72 tests were performed. For each test, the FFT analysis was performed for a frequency span of 800 to 13,600 Hz. A total of 264 records were obtained in every measurement.; The analysis results showed that the seepage induced AE in the test soils were broadband Gaussian signals which were zero-mean, normally distributed, and leptokurtic. A minimum seepage velocity ranging from 0.020 to 0.023 m/s was required to trigger meaningful AE in these granular soils. All of the measured autospectral density functions showed that the most prominent AE activities occurred within a frequency range of about 0.8 to 10 kHz. All results of variance, standard deviation, absolute peak, 95th percentile peak, mean square value, ring-down count analyses showed that the AE activities increased with increasing seepage velocity. Most trend lines obtained from these analyses for each test specimen resembled more exponential than parabolic curves. However, these trend lines did not yield a well-defined relationship between acoustic activities and soil properties. This is possibly caused by the effect of system modulation in the measured data and the signal conversion problems in the amplitude and time domain analyses. (Abstract shortened by UMI.)...