Kevin B. McGrattan

Large eddy simulations of smoke movement

An approach to field modeling of fire phenomena in enclosures is presented. The conservation equations of mass, momentum and energy are calculated with sufficient temporal and spatial resolution to yield a truly three-dimensional, dynamic picture of the fire plume and its surroundings. The large-scale eddies are simulated directly and sub-grid scale motion is represented by a constant eddy viscosity. Efficient flow solving techniques make it possible to simulate fire scenarios using computational grids in excess of a million cells on modern workstations. Several examples of the methodology are presented.

Numerical simulation of smoke plumes from large oil fires

Abstract A large eddy simulation (LES) model of smoke plumes generated by large outdoor pool fires is presented. The plume is described in terms of steady-state convective transport by a uniform ambient wind of heated gases and particulate matter introduced into a stably stratified atmosphere by a continuously burning fire. The Navier-Stokes equations in the Boussinesq approximation are solved numerically with a constant eddy viscosity representing dissipation on length scales below the resolution limits of the calculation. The effective Reynolds number is high enough to permit direct simulation of the large-scale mixing over two to three orders of magnitude in length scale. Particulate matter, or any non-reacting combustion ;product, is represented by Lagrangian particles which are advected by the fire-induced flow field. Background atmospheric motion is described in terms of the angular fluctuation of the prevailing wind, and represented by random perturbations to the mean particle paths. Results of the model are compared with two sets of field experiments.

Simulating fire whirls

A numerical investigation of swirling fire plumes is pursued to understand how swirl alters the plume dynamics and combustion. One example is the ‘fire whirl’ which is known to arise naturally during forest fires. This buoyancy-driven fire plume entrains ambient fluid as heated gases rise. Vorticity associated with a mechanism such as wind shear can be concentrated by the fire, creating a vortex core along the axis of the plume. The result is a whirling fire. The current approach considers the relationship between buoyancy and swirl using a configuration based on fixing the heat release rate of the fire and imposing circulation. Large-eddy methodologies are used in the numerical analyses. Results indicate that the structure of the fire plume is significantly altered when angular momentum is imparted to the ambient fluid. The vertical acceleration induced by buoyancy generates strain fields which stretch out the flames as they wrap around the nominal plume centreline. The whirling fire constricts radially and stretches the plume vertically.(Some figures in this article are in colour only in the electronic version; see www.iop.org)

Numerical simulation of combustion in fire plumes

Combustion in buoyancy-driven fire plumes is assumed to be diffusion controlled and described by a mixture fraction variable. Experimental data for a 10-cm methanol pool fire was compared to numerical results obtained by assuming the plume was axially symmetric. No turbulence model was used. The influence of imposing axial symmetry independent of the approximations made in the combustion model was tested by comparing simulated and experimental helium plumes. Simulated helium plume results agreed well with the experimental data at sufficiently small heights where most combustion occurs. At larger heights, predictions from the helium simulations were increasingly in error within a volume surrounding the centerline. Prediction error in the pool fire simulations behaved largely in a similar manner, with some error present due to inadequacies in the mixture-fraction-based combustion model. Overall, within the limitations of an axially-symmetric calculation, the behavior of flame flickering and the time-averaged temperature field were reasonably well predicted.

Numerical and experimental gravity currents related to backdrafts

In order to clarify the mixing that occurs in a gravity current which precedes a backdraft, a two-dimensional simulation, a series of salt water experiments, and backdraft experiments were performed. A compartment in a ratio width/height/length of 1/1/2 is used in the experiments and computations. Two different openings were used in the salt water experiments and numerical computations: a fully open end wall and a h1/3 horizontal slot centred in an end wall, where h1 is the compartment height. For the backdraft experiments only the h1/3 horizontal slot was used. The visual observations from the salt water experiments compare well with the numerical simulation. Both show a small mixed region at the gravity current shear interface for the fully open wall and mixing throughout the entire current for the h1/3 horizontal slot. Quantitative comparisons are made in terms of the normalised density differences, β=(ρ0−ρ1)/ρ1, where ρ0 is the higher density and ρ1 is the lower density. The transit time results, i.e., time between opening the compartment and the time the gravity current reaches the wall opposite the opening, for the computations compared favourably with the transit times from the salt water experiments, over the range 0.003<β<0.100. Velocity measurements from the opening of the backdraft compartment, prior to ignition, also are favourable with the numerical simulation.

Effects of slow wind on localied radiative ignition and transition to flame spread in microgravity

An experimental and numrical investigation of ignition and the subsequent transition to flame spread over a thermally thin cellulosic sample is described. The experiments were conducted using a lamp as an external radiant source in a 50% oxygen atmosphere at three diffeirent wind velocities of 0.2, and 5 cm/s in a 10 s drop tower. The results show that there are no significants effects of the slow wind on the ignition-delay time. Photographic sequences of both the experiments and the calculations show that the wind increases the flame propagation speed in the upwind direction. while decreasing it in the downstream direction. The downstream clame fails the transition to flame spread and becomes a tail of the upstream flame. The downstream char front propagates much slower than that for the upstream direction. Three-dimensional, time-dependent numerical solutions to the Navier-Stokes equations are used to simulate the experiments. Three global degradation reactions describe the pyrolysis of the sample paper, and one gasphase reaction describes the combustion of the fuel gases. The model results reflect the qualitative features of the experiments and also are in reasonable quantitative agreement, give the uncertainty of the gasphase reaction mechanism.

Numerical Modeling of Pool Fires Using LES and Finite Volume Method for Radiation

The thermal environment in small and moderate-scale pool flames is studied by Large Eddy Simulation and the Finite Volume Method for radiative transport. The spectral dependence of the local absorption coefficient is represented using a simple wide band model. The predicted radiative heat fluxes from methane/natural gas flames as well as methanol pool burning rates and flame temperatures are compared with measurements. The model can qualitatively predict the pool size dependence of the burning rate, but the accuracy of the radiation predictions is strongly affected by even small errors in prediction of the gas phase temperature.

Improved Radiation And Combustion Routines For A Large Eddy Simulation Fire Model

Improvements have been made to the combustion and radiation routines of a large eddy simulation fire model maintained by the National Institute of Standards and Technology. The combustion is based on a single transport equation for the mixture fraction with state relations that reflect the basic stoichiometry of the reaction. The radiation transport equation is solved using the Finite Volume Method, usually with the gray gas assumption for large scale simulations for which soot is the dominant emitter and absorber. To make the model work for practical fire protection engineering problems, some approximations were made within the new algorithms. These approximations will be discussed and sample calculations presented.

Simulation of smoke plumes from large pool fires

A large eddy simulation model of smoke plumes generated by large outdoor pool fires is presented. Theplume is described in terms of steady-state convective transport by a uniform ambient wind of heated gases and particulate matter introduced into a stably stratified atmosphere by a continuously burning fire. The Navier-Stokes equations in the Boussinesq approximation are solved numerically with a constant eddy viscosity representing dissipation on length scales below the resolution limits of the calculation. The effective Reynolds number is high enough to permit direct simulation of the large scale mixing over two to three orders of magnitude in length scale. Particulate matter, or any nonreacting combustion product, is represented by Lagrangian particles that are advected by the fire-induced flow field. Background atmospheric motion is described in terms of the angular fluctuation of the prevailing wind and represented by random perturbations to the mean particle paths. Sample computations are presented and compared with plumes generated by large crude oil pool fires. Also presented is an assessment of the potential environmental hazard posed by burning marine oil spills.

Ignition and transition to flame spread over a thermally thin cellulosic sheet in a microgravity environment

Abstract An axisymmetric, time-dependent model is developed describing auto-ignition and subsequent transition to flame spread over a thermally-thin cellulosic sheet heated by external radiation in a quiescent microgravity environment. Due to the unique combination of a microgravity environment and low Reynolds number associated with the slow, thermally induced flow, the resulting velocity is taken as a potential flow. A one-step global gas phase oxidation reaction and three global degradation reactions for the condensed phase are used in the model. A maximum external radiant flux of 5 W/cm 2 (Gaussian distribution) with 21%, 30%, and 50% oxygen concentrations is used in the calculations. The results indicate that autoignition is observed for 30% oxygen concentrations but the transition to the flame spread does not occur. For 50% oxygen the transition is achieved. A detailed discussion of the transition from ignition to flame spread is given as an aid to understanding this process. Also, a comparison is made between the axisymmetric configuration and a two-dimensional (line source) configuration.