Experiments and simulations of lean methane combustion

Computational fluid dynamics of methane combustion using chemical kinetics for description of the chemistry is studied. Linear least squares data fit to measured concentrations and temperatures is used to modify reaction rate parameters in the Arrhenius rate equation for combustion of methane. The modification of reaction rate parameters influences the result of CFD-simulations to predict combustion at experimental conditions where the Fluent rate equation failed. This first test show promising results but to further develop a global reaction model for combustion of methane and more complex fuels, a more extensive experimental study is required.The reaction rate for combustion of methane is rapid making ordinary sampling techniques for measuring to crude to collect sufficient amount of data for modification. Numerical simulation of fast chemistry using the Arrhenius theory often suffer from convergence difficulties as a result of the stiffness in the chemistry formulation. An alternative method for assessing the space discretization error is proposed. Richardson extrapolation is the most common model used for assessment of solution accuracy but the rigidity of the method allow little variation in the results. For engineering purposes qualitative methods can be sufficient for error assessment. The space discretization error of a two-dimensional axisymmetric simulation of combustion of methane in turbulent flow is studied. Profiles of temperature and carbon dioxide concentration is investigated and a second order polynomial fit is compared to the Richardson extrapolation. The profiles indicate grid independency of the solution but the Richardson method do not. The second order polynomial fit gives a better goodness of fit than obtained using Richardson, and by studying the first and second order term of the solution an estimation of the reaction order can be obtained and used to evaluate the accuracy of solution where the rigidity of the Richardson method provide unrealistic results.Godkänd; 2004; 20070131 (ysko