Nonpremixed Methane Diffusion Flames

While our long-term focus is directed to understanding the coupling of fluid turbulence with NOx and soot production, much of our current research is directed towards detailed simulations of simple laboratory configurations where measurements of certain flame observables are readily available. A standard laboratory experiment for analysis of diffusion flames is a fuel jet with an annular coflow, as shown schematically in the figure. Typically, these are modelled in two space dimensions assuming cylindrical symmetry. The plane in the figure, centered at the nozzle origin, illustrates the typical computational domain.

The basic low Mach number discretization algorithm was first tested in the IAMR framework for unity Lewis number and global reduced mechanisms. Global reduced mechanisms for methane combustion typically contain 2-6 species, and 3-6 reactions. The species may represent "real" chemical components (such as oxygen, carbon-dioxide, etc) or chemical mixtures (such as "fuel"). Global mechanisms are valuable for modelling the location and behavior of the primary heat release zone in nonpremixed flames. In this example, a coannular laboratory flame is fueled with comparable fuel and oxidizer velocities, producing a natural flickering that results from buoyancy effects due to density differences between fuel and air. The example illustrates a three-level dynamically adapting grid. The boxes in the figure, each representing hundreds of grid points, track regions of high temperature. Red blocks are refined a factor of two from the base grid, and orange blocks yet another factor of two. These profiles demonstrate excellent agreement with experimental results, matching flame length and liftoff height, and profile structures for the dominant species. The natural flickering frequency was recovered to less than 0.5% error. Furthermore, the simulation data showed that adaptive refinement saved more than an order of magnitude in both execution time and memory relative to a comparable uniform computation.

Methane Combustion with Detailed Chemistry

Our next example uses the adaptive low Mach number combustion methodology with detailed reaction kinetics and preferential diffusion effects and was applied to an axisymmetric jet diffusion flame where time-dependent measurements of various flame radicals allows for a more detailed validation of the methodology. A 26-species, 83-reaction methane mechanism was employed, together with the applicable portions of the ChemKin thermodynamic and mixture-averaged transport data. For this case, small concentric nozzles issue nitrogen-diluted methane fuel and co-flowing air. When the inflow of fuel is modulated at 20Hz with 25% and 50% perturbations, the entire flame undergoes quasi-periodic oscillations. We initialize simulations using a resolved steady solution (computed with the fuel flow rate fixed at the mean value), and then allow the flame to evolve through 4 cycles of the perturbation. The animation illustrates the CH mole fraction profile during the last two of the four oscillation periods that results from the 50% perturbation level. The Temperature for this case is also available in MPEG form, as well as the Temperature, and XCH for the case with 25% perturbation. As discussed in the related paper, the reported experimental observations are consistent with a perturbation level between 25% and 50%.

For more about the adaptive methodology for combustion applications, or about these calculations, contact Marc Day or John Bell of CCSE.

CCSE Combustion
CCSE Research