A priori analysis of a presumed subfilter probability density function model for soot modeling in large eddy simulation

This study proposes and evaluates a presumed subfilter probability density function (PDF) model for soot–turbulence–chemistry interactions in large eddy simulations (LES). Subfilter-scale variability of soot moments is modeled using an exponential distribution, whose peak near zero allows the main features of soot intermittency to be represented without introducing an explicit intermittency factor. Goodness-of-fit metrics show that the exponential form outperforms alternative distributions in most cases and closely matches soot statistics extracted from direct numerical simulation (DNS) data. The model is assessed through a series of a priori tests using filtered DNS data of turbulent flames, focusing on the reconstruction of filtered source terms, including coagulation and oxidation. For the coagulation source term, which involves nonlinear coupling among soot moments and thermochemical quantities, the exponential PDF performs better than traditional delta-function-based closures. For soot oxidation, the exponential PDF conditioned on the moment M0,1 provides the most accurate predictions. Incorporating a mixture-fraction-dependent conditional exponential PDF via a Heaviside function further improves agreement with DNS by suppressing unphysical oxidation in fuel-lean regions. Overall, the results establish the exponential distribution as a robust subfilter PDF model for soot, enabling the main effects of intermittency and soot–thermochemistry coupling to be represented in LES of turbulent combustion without additional intermittency parameters.