The objective of the present study is to analyze soot formation in diesel engine combustion by using multi-dimensional combustion simulations with a parallelized explicit ODE solver. Parallelized CHEMEQ2 was used to perform detailed chemical kinetics in KIVA-4 code. CHEMEQ2 is an explicit stiff ODE solver developed by Mott et al. which is known to be faster than traditional implicit ODE solvers, e.g., DVODE. In the present study, about eight times faster computation was achieved with CHEMEQ2 compared to DVODE when using a single thread. Further, by parallelizing CHEMEQ2 using OpenMP, the simulations could be run not only on calculation servers but also on desktop machines. The computation time decreases with the number of threads used. The parallelized CHEMEQ2 enabled combustion and emission characteristics, including detailed soot formation processes, to be predicted using KIVA-4 code with detailed chemical kinetics without the need for reducing the reaction mechanism.
After validating the code, diesel engine combustion was simulated to investigate combustion and emission characteristics, focusing on soot formation, growth and oxidation at different EGR ratios. To predict soot formation, a gas-phase polycyclic aromatic hydrocarbons (PAH) precursor formation model was coupled with a detailed phenomenological particle formation model, which included soot nucleation from precursors, surface growth/oxidation and particle coagulation. The results indicate that increased soot emission at high EGR ratios is mainly caused by decreased oxidation by oxygen and OH radicals because mixing fuel and gases (including oxygen and OH) has significant effects on reducing the mass of soot.