(Current research in premixed turbulent reacting flows by Ed Knudsen and Dirk Veenema, and non-premixed turbulent reacting flows by Matthias Ihme)

Large-eddy simulation (LES) is an approach to solving turbulent flows in which the larger scales of turbulent motion are exactly resolved while the smaller scales are modeled. This approach, as compared to the traditional Reynolds Averaged Navier-Stokes (RANS) method in which all turbulent scales are modeled, leads to significantly improved results in many real world flows. Such improvement is due to the resolution of the large turbulent scales. The chemical reactions that control combustion, however, occur on the smallest scales of the flow and hence can almost never be fully resolved. As such, modeling approaches are needed in order to accurately predict the chemical behavior of reacting flows. Furthermore, different modeling approaches are required in premixed and non-premixed combustion.

In premixed combustion, fuel and oxidizer enter reaction zones already mixed. Such flows occur in a host of industrially relevant applications, ranging from kitchen stove top burners to spark ignition internal combustion engines and gas turbines. Chemistry models for this type of combustion most often involve tracking the location of the flame front, which is typically very thin relative to the geometry of the burner. Such tracking can be accomplished by solving a levelset equation, the G-equation, that describes the front. This equation requires information about the flame's burning velocity, which must be modeled. Research in this area focuses on improving models for the turbulent burning velocity and on improving the accuracy of the numerical methods used to solve the levelset equation.

Non-premixed combustion occurs in applications such as diesel engines and aircraft engines. In such cases, where diffusive processes are controlling, chemistry can only be exactly determined by solving a large multidimensional space involving hundreds of chemical species. Current models overcome this difficulty by assuming that this space can be characterized by a much smaller number of dimensions. Specifically, two dimensional manifolds consisting of a mixture fraction variable and the scalar dissipation rate or a reaction progress variable are often used. Such models can be developed by using the laminar flamelet concept. To account for partially premixed fuel, local extinction, re-ignition, and better pollution modeling, a more detailed description of the chemistry is needed. Several novel combustion models are being developed to tackle this complex phenomena.

	Large Eddy Simulation of F3 Flame Color: Temperature Black Line: Flame front Click image to see animation. Temperature contours in a large eddy simulation of the F3 flame studied by Chen, et al (1996). F3 is a stoichiometric premixed methane / air flame where Re = 23,400. A pilot flame is used for stabilization. The black contour line represents the flame front, which was tracked using the G-equation. (Provided by Ed Knudsen)
	Large Eddy Simulation of a Reacting Spray Experiment Large Eddy Simulation (LES) of a reacting spray experiment. Temperature and axial velocity contours, drop position (~1.6 millions droplets). Click image to see full size. (Provided by Olivier Desjardins)
	DNS of Isotropic Decaying Turbulence DNS of isotropic decaying turbulence, with vorticity contours plotted; Re_lambda=67. Click image to see movie. (Provided by Olivier Desjardins)
	Color: Temperature Distribution Black Line: Contour of Stoichiometric Mixture Fraction Click image to see animation (Scroll down to see zoom in near field region and scalar dissipation rate)
	Zoom in Near Field Region Color: Temperature Distribution Black Line: Contour of Stoichiometric Mixture Fraction Click image to see animation.
	Scalar Dissipation Rate Color: Scalar Dissipation Rate Black Line: Contour of Stoichiometric Mixture Fraction Click image to see animation.