T. Mantel et Rw. Bilger, SOME CONDITIONAL STATISTICS IN A TURBULENT PREMIXED FLAME DERIVED FROM DIRECT NUMERICAL SIMULATIONS, Combustion science and technology, 111, 1995, pp. 393-417
The conditional moment closure (CMC) method constitutes an alternative
approach in the modelling of turbulent reacting systems. In the prese
nt study, the CMC formulation is introduced for turbulent premixed com
bustion and the transport equation for the conditional mass fraction i
s derived. It appears that the key points for these methods consist in
the modelling of the conditional scalar dissipation rate and mean vel
ocity. In order to provide information concerning these two quantities
, an existing data base for direct numerical simulations (DNS) of thre
e-dimensional turbulent premixed flames is used. From the DNS, the eff
ects of the Damkohler and Lewis numbers on the conditional means are p
resented and discussed. It appeal's that the conditional mean scalar d
issipation remains unaffected by the turbulent mixing in the reaction
zone. However, at low Damkohler numbers, the gradients of the mean pro
gress variable are steeper in the preheat zone of the instantaneous fl
ame fronts leading to a higher dissipation rate. The conditional mean
velocity is found to evolve almost linearly through the turbulent flam
e brush, as expected from the mean pressure gradient that arise from t
he overall heat release. A noticeable negative slip velocity associate
d with the wrinkling of the turbulent flame is observed in the simulat
ions. It is also found that there is no significant change in the axia
l velocity through the instantaneous flame front. This suggests that t
he flame adjusts itself so that it can provide dilatation without axia
l acceleration in the instantaneous flame front. The probability densi
ty function of the progress variable exhibits a strongly bimodal shape
, but the peak corresponding to the fresh gas side is broader. Turbule
nt fluxes for the mean progress variable are derived from the DNS and,
contrary to previous studies, no counter-gradient diffusion is observ
ed in these simulations.