Towards DNS of Statistically Stationary Turbulent Premixed Flames in Lean Methane/Air Mixtures

A computational framework for direct numerical simulations (DNS) of statistically stationary turbulent premixed flames in lean methane/air mixtures is presented. The mixture conditions are selected to be representative of those in lean-burn natural gas engines. DNS of propagating turbulent premixed flames are straightforward but computationally expensive: fine resolution is required for resolving the Kolmogorov scale and even finer resolution for resolving the reaction zone while the domain needs to be sufficiently large to capture multiples of integral length scales; in addition, several realizations may be required to achieve statistical convergence on the estimation of the flame speed, resulting in prohibitively expensive computations. In this work, this challenge is addressed by computing statistically stationary turbulent premixed flames in an inflow/outflow configuration. The inflow speed is dynamically adjusted at run-time to stabilize the location of the flame within the computational domain. Homogeneous and isotropic turbulent fluctuations, generated by an auxiliary simulation, are superimposed on the inlet velocity. The steady turbulent burning speed is obtained once the flame front becomes statistically stationary inside the domain. The simulations explore the dependence of the turbulent flame speed on turbulent field properties and on mixture conditions near the lean limit. Results are shown in a two-dimensional setup only. The extension to three dimensions is straightforward and currently underway.