The Reaction Mechanism Generator RMG

We have developed a software package, RMG, for exactly this purpose. The software is designed to take the hierarchical database trees as its main input, and to return a kinetic model as its main output. This kinetic model can be in the conventional List-of-Reactions format suitable for use in CHEMKIN or other integrators, or, if desired, the RMG program package can perform the integration itself and return product yield/selectivity profiles. 

which was implemented by Jing Song (2004), uses a version of the rate-based model-construction algorithm described by Susnow et al. (1997), but it treats non-thermalized chemically-activated reaction paths on an equal basis with ordinary thermal reactions, as done by Matheu et al. (Matheu, 2003 Matheu et al., 2003a, b). Byproducts and activated reaction intermediates are 

Unlike most model-generation software described in the literature, RMG correctly handles pressure and temperature variations it does this by using k(T,P) computed for the chemically-activated reactions at discrete (T,P) to determine coefficients in a Chebyshev form (Venkatesh et al., 1997) suitable for use in the differential equation solver. 

As a simple example, consider the predicted yield vs. conversion data for butane oxidation at 715 K, Figs. 12 and 13, plotted on top of experimental data points measured by Cernansky s group at Drexel (Wilk et al., 1995). As is typical, most of the discrepancies between the model predictions and the experimental data can be safely attributed to the uncertainties in the rate 

Many detailed chemical kinetic models show discrepancies at high conversions similar to what is seen in Fig. 12, in large part due to the time-integrated effects of even relatively small errors in the estimated rate parameters, and in some cases also due to the omission of side reactions, which become important as products build up. 

---