some rate rules don't account for degeneracy

[goldmanm]

trafficstars

It appears that some rate rules from Sharma et al. 2010, which are mentioned in the rate rules just with the text 'Sumathy CBS-Q calculations' did not have the degeneracy modified before they were input as rate rules. This leads RMG to mis-predict many of these reactions by a factor of 2 or 3.

This was confirmed for entry 798 which is methyl peroxyl isomerization. Ideally, these should be removed as rate rules and input as training reactions.

[zjburas]

This issue appears to be more widespread than just the intra_H_migration family: it has also come up for Disproportionation. For example, the iso-propyl self-disproportionation rate rule is written on a per-molecule basis, as determined from Michael Harper's (MRH) extensive comments, but RMG assumes that it is on a per-H basis and multiplies by the degeneracy of 6 when calculating the overall rate coefficient. Therefore, the overall rate is over-estimated by a factor of 6, and as a result the disproportionation rate is predicted to be faster than even radical recombination. This led to a mismatch with our experimental MS results because RMG was predicting much higher concentrations of disproportionation products than actually observed.

Within the disproportionation family, it appears that almost all of MRH's rate rules have this issue. It also seems likely that rate rules for other families have this issue too. In the case of the MRH rate rules it's at least easy to identify problematic reactions because the comments are exhaustive, but for less well-document rate rules it will be harder to know. I guess migrating all rate rules to training reactions is the long term solution?