OCR Text |
Show conditions were found We did find that comparisons with the entire modelable data on methane combustion do uncover occasional disagreements with results of various investigations. The optimization procedure can in some cases demonstrate that there is no practical means within the uncertainty constraints of the mechanism to reconcile such data with the bulk of other results. Hence we are led to conclude that these disagreements are due to some systematic errors in the experiments. Only a small number of rate parameter adjustments were required as a result of the optimization process. We are selective in this procedure in that we decline to adjust rate constants when only a small improvement in the minimization of the objective function would result from its inclusion. Details of the adjustments and parameters can be retrieved from the Internet addresses. In the final GRI-Mech 1.2 optimization, we adjusted only four rate constants. Most significantly, the value of the H + C~ -+ CH4 recombination rate constant remained the most significant parameter, even after our reevaluation of its value as a result of the initial GRI-Mech 1.1 optimization. It should not be inferred that the mechanism represents a unique way of obtaining a good fit to the experimental data base; its role is to provide a reliable method for predicting combustion properties for interpolated and extrapolated conditions. Thus periodic reoptimizations (as version 1.2 is to 1.1) are planned which will provide improved mechanisms reflecting new kinetics results, new targets, user feedback, and knowledge gained from exploring other conditions with the current mechanism. There are limitations to the applicability of the mechanism that was derived in this study, and users of our model should consider them when adopting it for applications to describe combustion conditions different from those which were included in the optimization. Firs4 our model was optimized as a whole; modification of any component of its thermochemistry or rate coefficient sets can lead to unpredictable results. Second, our model does not consider the well known low temperature chemistry of hydrocarbon combustion, i.e., below 1000 K it may be expected to have significant shortcomings. Finally, there is only brief consideration of the chemistry of very rich methane combustion, i.e., of the processes leading to formation of poly aromatic compounds and soot We have however performed some validation tests against data for other related fuels also of interest in natural gas combustion, such as hydrogen and ethane. (A few shock tube results on ethane were included among the GRI-Mech 1.2 optimization targets.) The satisfactory flame speed and shock tube agreements may be examined at the World Wide Web site. 5 |