Implementing multicomponent diffusion in numerical combustion studies is computationally expensive due to the challenges involved in computing diffusion coefficients. As a result, mixture-averaged diffusion treatments or simpler methods are used to avoid these costs. However, the accuracy and appropriateness of the mixture-averaged diffusion model has not been verified for three-dimensional turbulent premixed flames. This study evaluates the role of multicomponent mass diffusion for a range of premixed, laminar and turbulent hydrogen, n-heptane, and toluene flames, representing a range of fuel Lewis numbers. Secondary Soret and Dufour effects are neglected to isolate the impact of mass diffusion from thermal diffusion effects. Direct numerical simulation (DNS) of these flames is performed by implementing the Stefan--Maxwell equations in the DNS code NGA. A low-memory, semi-implicit algorithm decreases the computational expense of inverting the full multicomponent ordinary diffusion array while maintaining simulation accuracy and stability. The algorithm is demonstrated to be stable, and verified against one-dimensional premixed hydrogen flames in Cantera. A priori analysis shows significant errors in the diffusion flux vectors for mixture-averaged diffusion in regions of high flame curvature. These errors seem to distort local transport, impacting viscous dissipation and altering global flame statistics such as the normalized turbulent flame speed by modifying the average internal flame structure. In general, these results demonstrate that mixture-averaged diffusion may not fully capture the complexity of full multicomponent diffusion.
The Global Consumption Speeds of Premixed Large-Hydrocarbon Fuel/Air Turbulent Bunsen Flames
Niemeyer Research Group
Funding Statement (additional comments about funding)
This research used resources of the Extreme Science and Engineering Discovery Environment (XSEDE), which is supported by National Science Foundation grant number ACI-1548562.
This material is based upon work supported by the National Science Foundation under Grant Nos.~1314109-DGE and CBET-1761683.
This research used resources of the National Energy Research Scientific Computing Center, a DOE Office of Science User Facility supported by the Office of Science of the U.S. Department of Energy under Contract No. DE-AC02-05CH11231