Modeling NO and CO Emissions in Lean Natural Gas Spark Ignition Engines

Legislation limiting NO x and CO emissions from production natural gas engines dictate a need for greater understanding of the mechanisms governing these pollutants' formation. This work concerns the formulation of a one dimensional, multizoned, thermochemical model incorporating detailed kinetics for calculating NO and CO emissions from small, lean burning natural gas engines. The model calculates the temperature and pressure histories for each part of the working fluid throughout the combustion, compression and expansion phases. The gradients in the burned gases' properties resulting from the finite combustion time are maintained by the segmentation of the product field. The burn rate is input as either an harmonic function approximating standard bum rates or a burn rate discretized from experimental measurements. Because of the potential for significant non equilibrium effects from the lean operation, high engine speeds and small combustion chamber sizes, full kinetic mechanisms are used to integrate the reaction rates over the duration of the product segments' histories. The bulk gas properties are calculated at the end of every crank angle via an energy equation which balances the energy budget during that crank angle. The model is used to predict NO emissions from experimental engines over a small range of lean conditions: 0 = 0.91 and 0 = 0.66. The richer case is more easily modeled than the leaner cases and further investigation into this difference helps to illuminate the dominant mechanisms for NO formation in these engines. The richer mixtures produce NO by processes which are formation limited in the colder, later burned region, and decomposition limited in the hotter, earlier burned region. In the lean engines, however, NO is formed solely via a formation limited process. The balanced process in the richer engines renders modeling easier as the calculations are not as sensitive to inputs as are the unbalanced lean conditions. CO emissions in lean engines have been seen to result from flame quenching via wall interactions. Further experimental data is reviewed here to substantiate that observation and to emphasize the need for a quenching mechanism in any CO modeling effort. The lack of spatial dimension in the model leaves this phenomena hard to model and thus CO predictions using the fundamental model differ greatly from experiment. Adding a mechanism to approximate imperfect mixing helps to illustrate the possible significance it could have in real engine and combustor environments. The advantages of building a model with detailed kinetic …