A Skeletal Mechanism for the Reactive Flow Simulation of Methane Combustion

Abstract

A skeletal mechanism for the prediction of NOx emissions from methane combustion at gas turbine conditions is developed in the present work. The goal is a mechanism that can be used in computational fluid dynamic modeling of lean premixed (LPM) combustors. A database of solutions from 0-D, adiabatic, homogeneous reactors (PSRs) is computed using CHEMKINPRO [1] over a parameter space chosen to include pressures from 1 to 30 atm, equivalence ratios from 0.4 to 1.0, and mean PSR residence times from slightly greater than blowout to 3ms. A resisidence time of 3 ms represents a useful maximum for the super-equilibrium flame zone where most of the NOx forms in LPM combustors. Fuel oxidation and NOx formation are treated separately in the reduction process. The method of Directed Relation Graph (DRG) is applied for methane oxidation and its extension, DRG-aided sensitivity analysis (DRGASA), is used to determine the skeletal NOx mechanism to append to the methane mechanism. Post-processing of the PSR solution database and implementation of the reduction algorithm are accomplished in SAGE [2], a Python based, open-source mathematics software package. The skeletal oxidation and NOx mechanisms are validated against full GRI 3.0 [3] in both PSR and laminar flame speed calculations. When compared with the detailed GRI 3.0 mechanism, NOx emissions are predicted within 7% near blowout and 3% at 3ms, and laminar flame speeds are predicted within 20% over the range of equivalence ratios and pressures. The skeletal mechanism is presented here and it should be noted that all reactions of the H2/CO submechanism are retained. The skeletal mechanism consists of 22 species and 122 reactions for methane oxidation and an additional 8 species and 55 reactions to describe NOx formation (30 species, 177 reactions total). The final skeletal mechanism with NOx chemistry is available for download here [4]. To demonstrate the predictive capability of the validated mechanism in a reactive flow system, it is implemented in an ANSYS Fluent model of a single jet stirred reactor, the results of which are compared to experimental reactor data presented in [5] and [6]. Predicted and measured profiles of temperature and NOx emissions are shown. Temperature and NOx emissions compare well in the recirculation zone of the JSR, although both NOx emissions and temperature are under-predicted in the jet region. Finally, the contribution of each chemical pathway for NOx formation is evaluated.