Modeling of Soot Formation in Turbulent Diffusion Flames Impinging on a Cold Surface

Abstract

Soot is well known to be hazardous to both health and the environment. Over 40% of the world's population relies on biomass-fueled stoves for cooking, and these stoves are a significant source of soot. Especially problematic is the direct exposure to soot experienced by the user during cooking. The biomass-fired cookstove application often involves the impingement of a diffusion flame on the relatively cold cookware surface. Earlier experimental work showed that soot emissions increased substantially during flame impingement. This study seeks an improved understanding of the mechanisms responsible for this observation. The work primarily uses modeling as the tool to identify the mechanism. The approach focuses on two areas. The first is the development of an improved model for the oxidation of soot by O2 that includes a more realistic temperature dependence than has heretofore been presented in the literature. This is hoped to address the problem of overprediction of soot oxidation that appears in many modeling results. The second part consists of a broader study of reactive soot dynamics during the impingement process. Detailed soot models reflect the four processes governing soot dynamics: nucleation, surface growth, coagulation, and oxidation. Several studies have reported that the modeling of soot oxidation by O2 is inaccurate as it tends to overpredict the oxidation rate resulting in an underprediction of emissions. One of the challenges with modeling the oxidation process is that as the particle ages in the flame, it becomes more crystalline in structure and becomes less reactive to oxidation. The parameter α is widely used to represent this aging process wherein the soot surface reactivity ranges from α = 0 (nonreactive) to α = 1 (fully reactive). The surface deactivation process has been modeled primarily by using in-flame soot measurements for developing functional forms and parameters to describe the decay of surface reactivity, α. Among these, the model developed by Appel et al. has been widely used. Several studies indicate that the aging process should depend on both temperature and exposure time. Nevertheless, the time-temperature history of the particle is not directly taken into account in the model of Appel et al. A more recent model proposed by Khosousi et al. introduces a new parameter called the thermal age which incorporates the temperature-time history of the particle into the aging process. The model is, however, empirically based and does not reflect the non-linear temperature dependence one would expect of the physics of such a process. The present work presents a new soot aging model that is based on the framework of the well-studied problem of coal char oxidation deactivation. With a more fundamental framework, including the more realistic Arrhenius temperature dependence for the deactivation rate, the present model provides a better basis for approaching a wider range of flames. The new model is configured to provide the same parameter α so that it can be easily applied to existing soot models. The new approach is implemented in the fixed-pivot sectional soot model, with the model showing improved agreement with literature flame data. As mentioned above, soot emissions were found to increase substantially in cookstove experiments when the flame impinged on the cooking pot. Pundle et al. studied this phenomenon by examining emissions when an ethylene/air turbulent diffusion flame impinged on a cold pot surface compared to the system without the pot. The ethylene flame was used because it is a sooting flame with an extensive free flame database. (This particular study was the first in the literature to examine the problem of soot emissions resulting from flame impingement.) The present study computationally examines this configuration to identify the mechanisms leading to emission enhancement. The study uses a computational fluid dynamics code, a detailed chemical kinetic mechanism that includes soot precursor species, the laminar flamelet model to provide the turbulence/chemistry coupling problem, and the sectional model to simulate the soot dynamics. To reflect the experimental setup, the computational flame impinges on a surface at 373 K. The presence of the surface potentially affects soot behavior via (1) heat extraction by the surface leading to quenching of the chemical reactions, and (2) distortion of the flow field/mixing rates caused by the presence of the surface. To test this, runs were made in which the surface was assumed to be adiabatic, thus eliminating the heat extraction component. Very broadly, the insertion of the adiabatic surface leads to (1) an enhancement of local soot concentrations upstream of the surface, and (2) a decrease in soot concentrations as the flow diverges around the surface. In both cases, this results primarily from the alteration of the flow field. The initial increase in soot concentrations results from the surface causing an enhancement of the local turbulent scalar dissipation rate. The increased mixing rate leads to enhanced C2H2 concentrations which increase soot concentrations by promoting the surface growth process. The later decrease in soot concentrations occurs when the surface forces the flame to spread around itself forming a thinner concentration boundary layer which allows more O2 to diffuse in. This process enhances the O2 oxidation process at the location of the surface causing a sharp decrease in the in-flame integrated soot flux. The net result, however, is an enhanced emission. One concern in the cookstove literature is that adopting emission control strategies may lead to a reduced particle size for the remaining emissions, this potentially increasing the health impacts. This was experimentally examined by Pundle et al. who obtained size distribution measurements as part of the surface impingement work. The particle number density grows in all size ranges as the surface moves away from the fuel nozzle. This behavior occurs when the surface is located near the fuel nozzle (< 245 mm). However, there is no difference in the shape of the size distribution nor in the peak diameter. As the surface moves further away, the 245-mm case has fewer small-sized particles than the 300-mm case. The coagulation process is found to be responsible for the observed behavior as the 300-mm case has more space for particles to disperse resulting in lower concentration and lower coagulation process.