Has anyone tried using positive advection schemes for scalar transport in LES. While use of positive schemes is somewhat imperative in reacting flow simulations, the truncation error in such schemes smoothes out any steep gradients in the scalar field. This is at odds with the requirement(s) in the LES that schemes should have minimal dissipation and dispersion.

(1). You either get an oscillating solution, which might diverge later on, or eliminate it first so that you can continue on. (2). If there was an ideal algorithm, then people would not be using such scheme I guess.

There are schemes which retain sharpness of the fronts without the undershoots and overshoots and excessive damping. These are smart ways of removing unwanted behavior of the solution with minimal loss of accuracy.

Typically a monotone (upwind-type) scheme is used to predict the solution and then a corrector step (often called the anti-diffusion step) is used to remove the excessive diffusion caused by the monotone scheme (in the predictor). This works well for conserved scalar fields since the advection-diffusion equation is linear (for a given velocity field) and thus allows for this split predictor-corrector approach. I was just wondering if anyone has used these methods in LES of reacting flows.