Aerosol-dynamics-based soot modeling of flames

Open Access
Author:
Roy, Somesh Prasad
Graduate Program:
Mechanical Engineering
Degree:
Doctor of Philosophy
Document Type:
Dissertation
Date of Defense:
February 24, 2014
Committee Members:
  • Daniel Connell Haworth, Dissertation Advisor
  • Daniel Connell Haworth, Committee Chair
  • Stephen R Turns, Committee Member
  • Robert John Santoro, Committee Member
  • Padma Raghavan, Committee Member
Keywords:
  • soot modeling
  • discrete sectional model
  • method of moments
  • aerosol dynamics
  • laminar flame
  • direct numerical simulation
Abstract:
Modeling of soot formation and destruction in combustion systems involves modeling of fluid dynamics, chemistry, and radiative transfer. Each of these sub-problems are highly complex in nature and computationally very intensive. Considering complexity of these inter-connected processes of real-world combustion systems and computational cost associated with their modeling, a systematic comparative study of soot models is needed to identify an affordable, yet comprehensive and accurate model of soot prediction in device-scale and real-world combustion systems. Such a systematic study of two detailed soot models is performed in the current work. The models used in this study are a discrete sectional method (DSM) and a method of moments with interpolative closure (MOMIC). A semi-empirical soot model is also included in the study for comparison. Several gas-phase chemical mechanism were also tested with the soot models to identify the relative importance of gas-phase chemical mechanisms in the outcome of the soot simulations. Results showed the importance of the surface growth and nucleation schemes in predicting the soot volume fraction. Even though the surface growth contributed most to the soot mass growth, nucleation plays a critical role in final soot volume fraction by way of influencing the soot number density. Therefore, accurate prediction of nucleating species is key to the success of detailed soot model. The comparison of DSM and MOMIC showed very similar prediction trend in global quantities. The semi-empirical model, with proper tuning, was found to perform well in all the flames studied. DSM was found computationally most costly, while the semi-empirical model was computationally least expensive. The study in laminar flames was complemented by a direct numerical simulation (DNS) of a two-dimensional turbulent flame using MOMIC. A robust numerical scheme was developed and tested for MOMIC in DNS. The effects of turbulence on gas-phase chemistry and soot dynamics were explored from the data generated. Relative importance of soot surface reactions were found to be affected by both the scalar dissipation rate and the curvature of the instantaneous flame surface.