Tailings Dam Performance Monitoring by Combining Coda Wave Interferometry with Distributed Acoustic Sensing

Abstract

Advances in distributed fiber optic sensing technologies are enabling new methods to monitor changes in tailings dam performance. Distributed acoustic sensing (DAS), a distributed fiber optic sensing technology relying on Rayleigh light backscattering, can provide continuous spatial and temporal coverage along the length of a fiber optic cable extending tens of kilometers. In 2019, nearly six kilometers of fiber optic cable were installed at ∼1 m depth along an active upstream tailings dam in northern Canada. DAS seismic data were acquired at 400 Hz over a four-month period, from April to August 2021. We applied coda wave interferometry to a 120 m cable segment to obtain relative changes in seismic velocities (dv/v). Such coda waves are typically dominated by Rayleigh surface waves and dv/v can be used as a proxy for shear wave velocity changes. The dv/v estimates decrease by up to ∼1.9% over an initial two-month period of spring thaw and rainfall. Subsequently, dv/v recover by ∼1%, and generally show an inverse correlation with tailings pond levels up until the end of data acquisition. This correlation is supported by a known power-law relationship between shear wave velocity and effective stress. Rayleigh surface wave sensitivity kernels incorporating nearby seismic cone penetration testing data are used to estimate the approximate depths of dv/v sensitivity at ∼10 m. Despite active construction causing noise contamination, we obtain stable cross-correlation waveforms with as little as one hour of data per day. Overall, our results demonstrate how DAS can be used to augment geotechnical monitoring networks by providing in situ estimates of dv/v to inform changes in tailings dam performance over time. In addition to design and ongoing maintenance, tailings dams require robust geotechnical monitoring to reduce the risk of a potential failure. Here, we demonstrate a novel method for tailings dam monitoring by combining a type of fiber optic sensing known as distributed acoustic sensing with methods from passive seismology. We rely on energy from the ambient seismic wave field to infer changes in shear wave velocities of up to ∼1.9% at depth from ∼7 to ∼14 m, corresponding to springtime thaw and rainfall. Following this, we observe an inverse correlation between the inferred shear wave velocity changes and the nearby tailings pond levels. This method has important implications for monitoring changes in tailings dam performance, considering that shear wave velocities enable direct inferences of soil stiffness which can be used to help inform the stress state, liquefaction susceptibility and degree of cementation of tailings materials. Furthermore, distributed acoustic sensing is capable of monitoring over tens of kilometers, advantageous for improving the spatial coverage at large tailings storage facilities.