An analysis of turbulent mixing effects on the soot formation in high pressure n-dodecane sprays

An n-dodecane spray flame, known as Spray A, is simulated under the diesel engine conditions. The simulations are based on the well-mixed assumption where the turbulence-chemistry interactions are ignored, and employ the semi-empirical multi-step Moss-Brookes soot model coupled with the Reynolds-averaged turbulence model and a Lagrangian discrete phase spray model. A 54-species reduced n-dodecane chemical mechanism is employed in all simulations to evaluate the reaction rates. The importance of the turbulent mixing on the soot formation is analysed using three different turbulent Schmidt numbers; Sc_t = 0.7, 1.1 and 1.4. The non-reacting case is first validated using the mixture fraction and the velocity fields. It is found that the jet velocity and penetration length are unaffected by Sc_t for the inert case, however, the mixture fraction is sensitive to Sc_t where Sc_t = 1.1 leads to an excellent agreement with the measurements. Reacting simulations are compared with experimental data in terms of the ignition delay and the flame lift-off length. The results confirm the ignition delay time is marginally affected by changes in Sc_t. At the baseline condition, Sc_t = 0.7 and 1.1 result in a very similar value for the lift-off length which is in good agreement with the experiment although at higher ambient temperatures, only Sc_t = 0.7 agrees well with the measurements. It is found that the formation of formaldehyde and acetylene increases as the level of mixing decreases while the trend is opposite for the OH mass fraction. Consequently, with increasing Sc_t the soot volume fraction increases and the soot-containing region is extended. The results show that the development of the soot mass is not well captured, regardless of the value of the turbulent Schmidt number.