Capturing low-frequency cochlear impedance in time domain models: The role of viscosity.

Abstract

Viscous effects in the cochlear fluid have been shown to be negligible for frequencies greater than 100 Hz [M. Viergever, Ph.D. thesis]. However, they become important at lower frequencies; hence viscous damping is sometimes included in helicotrema models [S. T. Neely and D. O. Kim, J. Acoust. Soc. Am. 79, 1472–1480 (1986)]. Modeling viscosity in time-domain computational cochlear models for wide-band input signals is problematic. Since the boundary layer thickness is inversely proportional to frequency, Navier–Stokes formulations require fine spatial resolutions, and fluid potential formulations are inherently inviscid models. Lumping all viscous effects into the helicotrema is not physiologically accurate as “the helicotrema does not appear to be a dominant constriction” [Lynch et al., J. Acoust. Soc. Am. 72, 108–130 (1982)]. An alternative is proposed whereby a fluid potential formulation is augmented with frequency dependent viscosity corrections that are based on prior analysis of the cochlear geometry and become negligible at higher frequencies. For simplified fluid boundary conditions, the predictions of this fluid model are compared to Navier–Stokes, potential, and lumped viscosity models for frequencies in the range 10 Hz–10 kHz. The simulations are also compared to physiological data for the input impedance of the cochlea.