Surface Waves and Flow-Induced Oscillations along an Underground Elliptic Cylinder Filled with a Viscous Fluid

Abstract:

I made a linear analysis of flow-induced oscillations along an underground cylindrical conduit with an elliptical cross section on the basis of the hypothesis that volcanic tremor is a result of magma movement through a conduit. As a first step to understand how the self oscillation occurs because of magma flow, I investigated surface wave propagation and attenuation along an infinitely long fluid-filled elliptic cylinder in an elastic medium. The boundary element method is used to obtain the two-dimensional wave field around the ellipse in the frequency-wavenumber domain. When the major axis is much greater than the minor axis of the ellipse, we obtain the analytic form of the dispersion relation of both the crack-wave mode (Korneev 2008, Lipovsky & Dunham 2015) and the Rayleigh-wave mode with flexural deformation. The crack-wave mode generally has a slower phase speed and a higher attenuation than the Rayleigh-wave mode. In the long-wavelength limit, the crack-wave mode disappears because of fluid viscosity, but the Rayleigh-wave mode exists with a constant Q-value that depends on viscosity. When the aspect ratio of the ellipse is finite, the surface waves can basically be understood as those propagating along a fluid pipe. The flexural mode does exist even when the wavelength is much longer than the major axis, but its phase speed coincides with that of the surrounding S-wave (Randall 1991). As its attenuation is zero in the long-wavelength limit, the flexural mode differs in nature from surface wave. I also obtain a result on linear stability of viscous flow through an elliptic cylinder. In this analysis, I made an assumption that the fluid inertia is so small that the Stokes equation can be used. As suggested by the author’s previous study (Sakuraba & Yamauchi 2014), the flexural (Rayleigh-wave) mode is destabilized at a critical flow speed that decreases with the wavelength. However, when the wavelength is much greater than the major axis of the ellipse, the unstable solution does exist, but its linear growth rate in amplitude becomes almost zero. Therefore, the unstable solution effectively disappears in the long-wavelength limit, suggesting that the aspect ratio of the conduit is needed to be sufficiently large if the flow-induced oscillation caused by a moderate magma speed is an origin of volcanic tremor.