Impact of mixing model on predicted no formation in a nonpremixed, partially stirred reactor

Abstract

Stochastic simulations of non-premixed, partially stirred reactors (PaSRs) with three mixing models have been carried out for investigating the influence of unmixedness and differential diffusion on predicted NO levels. The three mixing models considered here include the modified Curl's model, the interaction-by-exchange-with-the-mean (IEM) model, and the linear-eddy model (LEM). The first two mixing models lack the ability in modeling the detailed diffusion process at the molecular level. The potential impact of differential diffusion on predicted NO levels is explored by using the LEM with/without differential diffusion. Comparisons of the predicted mean temperatures and NO levels indicate that at a given unmix-edness level, the IEM model gives highest NO and temperatures due to its deterministic nature. Differential diffusion is found to have little impact on the predicted mean temperatures by the LEM, but it increases the predicted NO levels by a factor of 2. The modified Curl's mixing model predicts temperatures and NO levels lying between those given by the IEM and the LEM. Comparisons of detailed statistics have been performed to gain better understanding of these mixing models, especially the importance of the subtle features of mixing processes embedded in the LEM. The results suggest that superequilibrium temperatures seen in the predictions by the LEM with differential diffusion are the cause of the increase in the predicted NO levels. The predicted differential diffusion effect is found to scale with Re−1/2 in agreement with the predictions for nonreacting flows. At a given unmixedness level, the mixing times computed by τ mix = C m 〈 ξ ″ ξ ″ 〉 / χ ˜ based on the LEM correlate reasonably well with the prescribed mixing times for the modified Curl's model or the IEM model.