,- .. LARGE EDDY SIMULATIONS OF BUOYANT PLUMES - — —— William E. Meil, Art Johnson, Kevin B. McGrattan and Howard R. Baum Building and Fire Research Laboratory National Institute of Standards and Technology Gaithersburg, MD 20899 ABSTRACT . An approach to the study of gas phase combustion and convection processes in fires using a combination of mathematical analysis and computer simulation is presented. It seeks to solve the governing equations directly (if approximately) by decomposing the fire into a large scale convective and radiative transport problem coupled to a small scale model of combustion and radiative emission. The combustion model assumes that all of the thermodynamic properties of the fluid are tied to the local mixture fraction, which is convected by the large scale motion, which in turn is driven by the heat released by the combustion processes. The large scale flow is studied using finhe difference techniques to solve large eddy simulations of the Navier-Stokes equations, As a first test of the numerical approach a buoyant helium plume k simulated and results compared to a companion laboratory experiment. INTRODUCTION The difficulties associated with analyses of tire phenomena originate with the fact that the active combustion zone of a fire play<’two distinct roles which encompass widely different length and time scales. The combustion zone is the region where the local mixing of gasified fuel and air produces the chemical energy release and radiant energy emission that sustains the fire. These processes occur on length scales ranging from a fraction of a millimeter to a few centimeters. At the same time, the combustion zone is a source of buoyancy which induces large scale mixing of air and combustion products, forming a plume which can persist as an organized structure over length scales ranging from a few meters to tens of kilometers, depending on the scenario of interest. The plume in turn, acts as a giant pump that induces a flow pattern throughout the entire structure enclosing an indoor fire. Equally important, the interaction of the large scale mixing and small scale combustion processes creates a combustion zone that is not necessarily small compared with the plume that it generates. Finally, the radiative transport from the entire combustion zone back to the condensed phase fuel surface provides the feedback Heeded to supply the fire with the fuel required to maintain itself. Much of the work in fire research concerns the movement of smoke and hot gases in an enclosure. Depending on the scope of the particular scenario, the fire itself is described in relatively simple terms as a source of heat and combustion products [1]. The reason for this is that the fire itself usually occupies a very small fraction of the flow domain; and there is simply not enough spatial resolution to describe fluid motion on length scales at which combustion takes place. Conventional field models using k-&representations of turbulence often include an empirical description of the combustion processes, but this description relies heavily on the level of turbulence prescribed by the user through the choice of parameters. The approach outlined below, by contrast, seeks approximate solutions to the governing equations directly, by considering combustion, convection, and thermal radiation in parallel, allowing each to evolve separately on its own length and time scale. The complexity of the combustion model which is implemented depends on the spatial and temporal resolution which can be provided by the solution of the Navier-Stokes equations without any empirical turbulence model. Present computational resources allow for two-dimensional simulations with Reynolds numbers approaching 105, and three- dimensional calculations approaching 104. In the present work, a few 2D, axisymmetic plume simulations will be presented. Simulations in 3D are presently underway, as is the implementation of a soot and radiation model. Before these more elaborate additions are made, however, we seek confirmation that the hydrodynamic model adequately describes the flow field. As a first test of the hydrodynamic model a buoyant helium plume was simulated and the results compared to those from laboratory experiments. The approach taken for both the combustion and isothermal scenarios is presented next, followed by helium plume results. NUMERICAL MODEL Following the analysis of Rehm and Baum [2], we assume that the flow velocity is much less than the sound speed; the temperature and density variations are large, but the pressure variations are small. The gases are assumed to be ideal and variable transport properties are allowed. Under these assumptions the equation of state and conservation equations for momentum and total mass are (1) 187 Proceedings: Combustion Institute/Eastern States Section, Chemical and Physical Processes in Combustion, Fall Technical Meeting, October 16-18, 1995, 187–190 pp, Worcester, MA