Abstract
In this paper, we computationally study the effects of large Polycyclic Aromatic Hydrocarbons (PAH) (more than two aromatic rings) on soot yield and distribution in turbulent flames, when they are considered as nucleating species for soot formation. This is examined in two turbulent non-premixed sooting jet flames using ethylene and a jet fuel (JP-8) surrogate as fuels. For each flame, two Large-Eddy Simulations (LES) are performed with two different soot nucleation strategies. In the first strategy, a range of PAH from naphthalene to cyclopenda[cd]pyrene are considered as nucleating species, while in the second strategy, naphthalene is considered as the only soot nucleating species and the effects of larger PAH are represented entirely by naphthalene. Flamelet-based chemistry-tabulation is used for the major thermochemical quantities, such as density, temperature, and major species mass fractions. Turbulence-chemistry interactions for PAH are accounted for by transporting their mass fractions and using a recently developed PAH relaxation model for their source term closure. The effects of large PAH on soot are highlighted by comparing the PAH profiles, soot nucleation rate, and soot volume fraction distributions obtained from both simulations for each test flame. These results are also compared against experimental measurements, when available. From these comparisons, it is first shown that naphthalene is predominantly formed along the flame centerline, and larger PAH species with more than two aromatic rings are primarily formed away from the centerline. Further, it is found that these larger PAH species only contribute to around of the overall soot nucleation rate along the flame centerline but their contributions are more substantial (up to ) away from the centerline. Finally, it is shown that representing the effects of these large PAH species by naphthalene leads to an under-prediction of around for the soot volume fraction magnitude along the flame centerline. Away from the centerline, this under-prediction can be as much as 100
Acknowledgments
The authors would also like to acknowledge the Institute of CyberInfrastructure (ICS) at the Pennsylvania State University for the computational resources to perform the simulations shown in this study.
Disclosure statement
No potential conflict of interest was reported by the authors.
ORCID
Abhishek Jain http://orcid.org/0000-0003-1418-9425
Yuan Xuan http://orcid.org/0000-0001-9326-2197