Session R24: Biofluids: Physiological VII - Human Voice System Flows

Study of dynamic fluid-structure coupling with application to human phonation

Two-dimensional direct numerical simulations of a compressible, viscous fluid interacting with a non-linear, viscoelastic solid are used to study the generation of the human voice. The vocal fold (VF) tissues are modeled using a finite-strain fractional derivative constitutive model implemented in a quadratic finite element code and coupled to a high-order compressible Navier-Stokes solver through a boundary-fitted fluid-solid interface. The viscoelastic solver is validated through in-house experiments using Agarose Gel, a human tissue simulant, undergoing static and harmonic deformation measured with load cell and optical diagnostics. The phonation simulations highlight the role tissue nonlinearity and viscosity play in the glottal jet dynamics and in the radiated sound.   

Fluid-Structure Interactions as Flow Propagates Tangentially Over a Flexible Plate with Application to Voiced Speech Production

Voiced speech is produced by fluid-structure interactions that drive vocal fold motion. Viscous flow features influence the pressure in the gap between the vocal folds (i.e. glottis), thereby altering vocal fold dynamics and the sound that is produced. During the closing phases of the phonatory cycle, vortices form as a result of flow separation as air passes through the divergent glottis. It is hypothesized that the reduced pressure within a vortex core will alter the pressure distribution along the vocal fold surface, thereby aiding in vocal fold closure. The objective of this study is to determine the impact of intraglottal vortices on the fluid-structure interactions of voiced speech by investigating how the dynamics of a flexible plate are influenced by a vortex ring passing tangentially over it. A flexible plate, which models the medial vocal fold surface, is placed in a water-filled tank and positioned parallel to the exit of a vortex generator. The physical parameters of plate stiffness and vortex circulation are scaled with physiological values. As vortices propagate over the plate, particle image velocimetry measurements are captured to analyze the energy exchange between the fluid and flexible plate. The investigations are performed over a range of vortex formation numbers, and lateral displacements of the plate from the centerline of the vortex trajectory. Observations show plate oscillations with displacements directly correlated with the vortex core location.   

Flow in a Geometrically-Realistic, Vibrating Model of the Human Vocal Tract

Airflow within the human vocal tract is an important component of voice quality. Understanding the nature of the airflow will help better understand voice production, potentially leading towards improved clinical diagnostics and treatments. An up-scaled experimental setup was developed to study three-dimensional flow features in a realistic model of the human larynx. The subglottal and supraglottal sections were made of clear silicone, with geometry derived from CT scan data. A cylindrically-shaped supraglottal section was also fabricated to compare flows with and without anatomically-accurate supraglottal sections. The glottal section consisted of two counter-rotating, mechanically-driven cams, covered by a silicone membrane, to approximate the alternating convergent-divergent profile of vibrating vocal folds. A mixture of water and glycerol was pumped through the system, the index of refraction matching that of the silicone for optical access into the sub- and supraglottal sections. Velocity fields throughout the glottal cycle were acquired using particle image velocimetry (PIV), giving particular attention to differences in flow features (e.g., jet skewing and axis switching) between models with CT-derived and cylindrically-shaped supraglottal geometry. In this presentation, the model design and characteristics will be given, and PIV flow results will be presented and discussed.   

Vocal Fold Pathologies and Three-Dimensional Flow Separation Phenomena

Polyps and nodules are two different pathologies, which are geometric abnormalities that form on the medial surface of the vocal folds, and have been shown to significantly disrupt a person's ability to communicate. Although the mechanism by which the vocal folds self-oscillate and the three-dimensional nature of the glottal jet has been studied, the effect of irregularities caused by pathologies is not fully understood. Examining the formation and evolution of vortical structures created by a geometric protuberance is important, not only for understanding the aerodynamic forces exerted by these structures on the vocal folds, but also in the treatment of the above-mentioned pathological conditions. Using a wall-mounted prolate hemispheroid with a 2:1 aspect ratio in cross flow, the present investigation considers three-dimensional flow separation induced by a model vocal fold polyp. Building on previous work using skin friction line visualization, both the velocity flow field and wall pressure measurements around the model polyp are presented and compared.   

Phonation aeroacoustic source strength estimation from sound pressure measurements

An experimental characterization of monopole and dipole source spectra in a model of the human upper airway is presented. The airway model is a life-scale, vertical, straight duct of square cross section, into which two model vocal folds are placed. Five microphones are positioned in the duct, two below and two above the vocal folds, with a fifth microphone placed at the ``mouth.'' Time-mean subglottal pressure and volume flow rate are measured using a micromanometer and ball-element meter, respectively. In addition, pressure on either side of the model vocal folds are measured using Kulite XCS-093 pressure transducers, and the motion of the model vocal folds is captured using high-speed video. Cross-correlations between the microphone pairs are used to estimate the right- and left-running acoustic wave amplitude spectra above and below the model vocal folds. From these spectra and theoretical matching conditions at the inlet and outlet of the vocal fold constriction, source spectra are constructed. These are compared to independent estimates of source spectra obtained from the difference of the Kulite transducer pressures and the motion of the model vocal folds. Acknowledge support from NIH R01 DC005642 (MK, MM) and ARL E{\&}F program (EC).   

Glottal aerodynamics in compliant, life-sized vocal fold models

This talk presents high-speed PIV measurements in compliant, life-sized models of the vocal folds. A clearer understanding of the fluid-structure interaction of voiced speech, how it produces sound, and how it varies with pathology is required to improve clinical diagnosis and treatment of vocal disorders. Physical models of the vocal folds can answer questions regarding the fundamental physics of speech, as well as the ability of clinical measures to detect the presence and extent of disorder. Flow fields were recorded in the supraglottal region of the models to estimate terms in the equations of fluid motion, and their relative importance. Experiments were conducted over a range of driving pressures with flow rates, given by a ball flowmeter, and subglottal pressures, given by a micro-manometer, reported for each case. Imaging of vocal fold motion, vector fields showing glottal jet behavior, and terms estimated by control volume analysis will be presented. The use of these results for a comparison with clinical measures, and for the estimation of aeroacoustic source strengths will be discussed. Acknowledge support from NIH R01 DC005642.   

Control volume analyses of glottal flow using a fully-coupled numerical fluid-structure interaction model

Vocal fold vibrations and the glottal jet are successfully simulated using the modified Immersed Finite Element method (mIFEM), a fully coupled dynamics approach to model fluid-structure interactions. A self-sustained and steady vocal fold vibration is captured given a constant pressure input at the glottal entrance. The flow rates at different axial locations in the glottis are calculated, showing small variations among them due to the vocal fold motion and deformation. To further facilitate the understanding of the phonation process, two control volume analyses, specifically with Bernoulli's equation and Newton's 2nd law, are carried out for the glottal flow based on the simulation results. A generalized Bernoulli's equation is derived to interpret the correlations between the velocity and pressure temporally and spatially along the center line which is a streamline using a half-space model with symmetry boundary condition. A specialized Newton's 2nd law equation is developed and divided into terms to help understand the driving mechanism of the glottal flow.   

Viscous Flow Structures Downstream of a Model Tracheoesophageal Prosthesis

In tracheoesophageal speech (TES), the glottis is replaced by the tissue of the pharyngeoesophageal segment (PES) as the vibrating element of speech production. During TES air is forced from the lungs into the esophagus via a prosthetic tube that connects the trachea with the esophagus. Air moving up the esophagus incites self-sustained oscillations of the surgically created PES, generating sound analogous to voiced speech. Despite the ubiquity with which TES is employed as a method for restoring speech to laryngectomees, the effect of viscous flow structures on voice production in TES is not well understood. Of particular interest is the flow exiting the prosthetic connection between the trachea and esophagus, because of its influence on the total pressure loss (i.e. effort required to produce speech), and the fluid-structure energy exchange that drives the PES. Understanding this flow behavior can inform prosthesis design to enhance beneficial flow structures and mitigate the need for adjustment of prosthesis placement. This study employs a physical model of the tracheoesophageal geometry to investigate the flow structures that arise in TES. The geometry of this region is modeled at three times physiological scale using water as the working fluid to obtain nondimensional numbers matching flow in TES. Modulation of the flow is achieved with a computer controlled gate valve at a scaled frequency of 0.22 Hz to mimic the oscillations of the PES. Particle image velocimetry is used to resolve flow characteristics at the tracheoesophageal prosthesis. Data are acquired for three cases of prosthesis insertion angle.   

Evaluation of Synthetic Self-Oscillating Models of the Vocal Folds

Approximately 30{\%} of people will suffer from a voice disorder at some point in their lives. The probability doubles for those who rely heavily on their voice, such as teachers and singers. Synthetic vocal fold (VF) models are fabricated and evaluated experimentally in a vocal tract simulator to replicate physiological conditions. Pressure measurements are acquired along the vocal tract and high-speed images are captured at varying flow rates during VF oscillation to facilitate understanding of the characteristics of healthy and damaged VFs. The images are analyzed using a videokymography line-scan technique that has been used to examine VF motion and mucosal wave dynamics \textit{in vivo}. Clinically relevant parameters calculated from the volume-velocity output of a circumferentially-vented mask (Rothenberg mask) are compared to patient data. This study integrates speech science with engineering and flow physics to overcome current limitations of synthetic VF models to properly replicate normal phonation in order to advance the understanding of resulting flow features, progression of pathological conditions, and medical techniques.   

Patient-Specific Computational Modeling of Human Phonation

Phonation is a common biological process resulted from the complex nonlinear coupling between glottal aerodynamics and vocal fold vibrations. In the past, the simplified symmetric straight geometric models were commonly employed for experimental and computational studies. The shape of larynx lumen and vocal folds are highly three-dimensional indeed and the complex realistic geometry produces profound impacts on both glottal flow and vocal fold vibrations. To elucidate the effect of geometric complexity on voice production and improve the fundamental understanding of human phonation, a full flow-structure interaction simulation is carried out on a patient-specific larynx model. To the best of our knowledge, this is the first patient-specific flow-structure interaction study of human phonation. The simulation results are well compared to the established human data. The effects of realistic geometry on glottal flow and vocal fold dynamics are investigated. It is found that both glottal flow and vocal fold dynamics present a high level of difference from the previous simplified model. This study also paved the important step toward the development of computer model for voice disease diagnosis and surgical planning.