Abstract
A strategy is described for the computational prediction of strongly sooting turbulent non-premixed flames in which are coupled a comparatively detailed model of soot formation and oxidation together with significant radiative heat loss. The considerable computational simplifications introduced by mixture fraction-centred flamelet modelling to address turbulence interaction effects are extended to embrace soot source term modelling through the introduction of a family of flamelet profiles, each characterised by varying radiative loss. Encouraging comparisons with experimental measurement are reported for single and multiple flamelet representations in turbulent jet flames of ethylene and pre-vaporised kerosine.