Title:
Turbulence-chemistry interactions for lean premixed flames

Thumbnail Image
Author(s)
Dasgupta, Debolina
Authors
Advisor(s)
Lieuwen, Timothy C.
Advisor(s)
Editor(s)
Associated Organization(s)
Supplementary to
Abstract
Turbulent combustion, particularly premixed combustion has great practical importance due to their extensive industrial usage in gas turbines, internal combustion engines etc. However, the physics governing the inherent multi- scale interactions of turbulence, flow-field and chemistry is not yet well established. A complete understanding of each of these interactions and their coupling is essential for the development of models that can aid simulations of realistic engines (using Large Eddy Simulations (LES) or Reynolds averaged Navier-Stokes equations (RANS). Particularly, understanding the flame structure and its stabilization requires an understanding of the turbulence-chemistry interactions. This can manifest itself in many different forms. For example, flame wrinkling gives rise to flame stretch that can modify the local temperature and species concentrations in turn altering the local chemistry. Also, the smaller eddies in a turbulent flow can penetrate into the preheat and reaction zones changing the species’ gradients within the flame. The influence of turbulence on chemistry can be analyzed in two different ways: firstly, a “global” analysis which investigates the direct impact of turbulence on the chemical pathways (a series of elementary reactions involved in the fuel oxidation process) and secondly, a “local” analysis which investigates the influence of turbulence on the chemical flame structure (i.e. species and reaction rate profiles). To understand these influences of turbulence, this work performs Direct Numerical Simulations (DNS) for lean premixed flames involving three fuels: hydrogen, methane and n-dodecane. A “global” analysis using different metrics such as heat release and species consumption/production is performed to quantify the changes in the chemical pathways. This analysis is performed for the metrics averaged over the entire flame and conditioned on local flame features such as fuel consumption, curvature etc. The results are also compared and contrasted with simple laminar flame models such as unstretched flames, stretched flames and perfectly stirred reactors. In general, the laminar models provide a good estimate for the chemical pathways for these key metrics suggesting turbulence does not have a significant impact on the fuel oxidation pathways. However, this is not true for the reaction rate and species profiles across the flame. Conditional means of these quantities are calculated to identify the “local” influence of turbulence on chemistry. These conditional means are also compared with laminar unstretched and stretched flames to identify regions of good agreement and deviation. The laminar calculations are performed using two different transport models; firstly, the mixture-averaged transport wherein every species diffuses into the mixture with its molecular diffusivity and secondly, Le=1 transport wherein the mass diffusivity of every species is equal to the thermal diffusivity of the mixture eliminating effects of preferential and differential diffusion. Le=1 is considered the theoretical limit of transport where turbulent mixing governs the transport process opposed to molecular diffusivity. For lean hydrogen/air flames (Le<1), the behavior of the profiles is similar to the evolution of the laminar profiles with increasing stretch. However, for the lean methane/air flames (Le~1), with increasing turbulence intensity, the flame profiles deviate from the evolution of laminar profiles with increasing stretch and align more closely with the Le=1 transport model laminar flame profiles. For n-dodecane/air flames (Le>1), the evolution of the turbulent flame profiles, with slight increase in turbulence intensity, replicates the behavior of stretched flames. However, with a further increase, a deviation is seen from the stretched flame profiles. Additionally, these profiles significantly deviate from the Le=1 transport model suggesting the inadequacy of stretched flames and a simple Le=1 model to replicate the behavior of stretched flames. In order, to identify the effect of increased diffusivity due to turbulence, a new transport model is implemented for unstretched and stretched flames wherein a constant is added to the mass diffusivity of the species obtained from the mixture-averaged transport. This constant covers multiple orders of magnitude mimicking the effect of increased turbulence diffusivity. For the lean hydrogen flames(Le<1), the turbulent flame profiles are seen to evolve similar to the laminar profiles with increasing stretch and not similar to the laminar profiles with increasing diffusivity. This suggests mixtures containing a highly diffusive fuel does not need the aid of turbulence to enhance transport. For the lean methane flames (Le~1), the turbulent flame profile evolution is similar to the effect of increasing diffusivity for unstretched flames suggesting a significant effect of diffusivity on the flame structure. For the lean n-dodecane flames (Le>1), the turbulent flame profiles evolve similar to the effect of increased diffusivity on stretched flames. This further emphasizes the necessity to include diffusivity in laminar models used to replicate turbulent flame structure. Overall, this work helps identify the key players in turbulence-chemistry interactions which need to be considered for modeling real combustors.
Sponsor
Date Issued
2018-10-09
Extent
Resource Type
Text
Resource Subtype
Dissertation
Rights Statement
Rights URI