The near-field dynamics of buoyant helium plumes

Abstract

Buoyant plumes can be found in a variety of naturally occurring phenomena (e.g., volcanoes and hydrothermal vents) and engineering applications (e.g., heat treatment processes), and the underlying hydrodynamic properties are crucial to understanding low-speed reacting flows, such as pool fires and wildland fires. However, despite the prominence of plumes, there are still many outstanding questions pertaining to plumes structure and dynamics that must be addressed to build predictive capabilities in a computationally tractable manner. In this dissertation, I con- ducted a series of high-fidelity numerical simulations using adaptive mesh refinement to model the isothermal injection of helium into a domain filled with air. This configuration is attractive due to the simplicity of the configuration (e.g., two-fluid system, non-reacting flow, and physical re- alizability) while retaining the complexity of the flow dynamics. I first developed the numerical analysis framework necessary to compute relevant statistics on a multi-resolution grid, including high-order azimuthal averaging and cascaded interpolation for gradients. I then conducted a series of simulations varying the Richardson and Reynolds numbers to study the basic kinematics of the flow, uncovering a number of important flow characteristics, including recirculation in very turbu- lent plumes, consequences of transition to turbulence, scaling relationships for streamwise fluxes, and universality through dimensional arguments. Using the same simulations, I performed the first in-depth analysis of the kinetic energy and enstrophy dynamics in buoyant plumes, revealing the sources and sinks for each of these quantities. Two different auxiliary studies were then performed to analyze the puffing frequency for adjacent two-dimensional plumes and for different inflow shapes and levels of confinement. Lastly, I formulated a Galerkin-based reduced-order model for the lam- inar plumes using proper orthogonal decomposition (POD) as the basis functions. To do this, I developed computationally efficient algorithm to compute POD for multi-resolution data.

Michael Meehan

Postdoctoral Research Associate

Mike is a former research associate in the Paul M. Rady Department of Mechanical Engineering at the University of Colorado Boulder and also a former student in the Turbulence and Energy Systems Laboratory, earned his PhD in May 2022.