Invalid Plog in Klippenstein_Glarborg2016

[sevyharris]

Problem Description

I got this error when I tried to convert my RMG mechanism to a Cantera yaml:

*******************************************************************************
CanteraError thrown by addReactions:

*******************************************************************************
CanteraError thrown by PlogRate::validate:

Invalid rate coefficient for reaction 'CH2CHOO(58) <=> CH3(35) + CO2(7)'
at P = 1.0132e+07, T = 500.0
*******************************************************************************
*******************************************************************************

The reaction CH2CHOO = CO2 + CH3 has a negative rate at T=500K and P=100atm.

It's part of the Klippenstein_Glarborg2016 reaction library:

https://github.com/sevyharris/RMG-database/blob/e883a4c47eb25fac07510888dab1e861bd314b94/input/kinetics/libraries/Klippenstein_Glarborg2016/reactions.py#L10490

but originates from this 2015 paper: https://pubs.acs.org/doi/10.1021/acs.jpca.5b01088.

I don't see any errors in copying the parameters, but the MAE is pretty large for this range.

CH2CHOO=CO2+CH3                                  6.03E-03        3.46    17420.0
  PLOG/1.000E-02    5.09E+33       -7.95    31290.0/
  PLOG/1.000E-02    4.20E+122      -39.75    43640.0/                           ! fit btw. 400 and 1250 K with MAE of 1.5%, 4.0%
  PLOG/1.000E-01    1.21E+118      -33.13    73790.0/
  PLOG/1.000E-01    1.96E+29       -6.29    30920.0/                            ! fit btw. 400 and 1350 K with MAE of 0.4%, 1.1%
  PLOG/3.160E-01    8.56E+32       -7.21    33550.0/
  PLOG/3.160E-01    5.10E-66       21.37    -11110.0/                           ! fit btw. 400 and 1450 K with MAE of 1.2%, 3.4%
  PLOG/1.000E+00    3.27E+33       -7.22    34990.0/
  PLOG/1.000E+00    1.76E-47       15.85    -5283.0/                            ! fit btw. 400 and 1550 K with MAE of 0.9%, 3.6%
  PLOG/3.160E+00    3.49E-79       25.01    -21020.0/
  PLOG/3.160E+00    3.82E+32       -6.80    35690.0/                            ! fit btw. 550 and 1650 K with MAE of 0.9%, 3.9%
  PLOG/1.000E+01    8.16E+32       -6.76    37270.0/
  PLOG/1.000E+01    4.62E+00        2.10    17170.0/                            ! fit btw. 450 and 1750 K with MAE of 1.8%, 7.5%
  PLOG/3.160E+01    7.01E+37       -8.06    42200.0/
  PLOG/3.160E+01    3.49E+14       -1.58    26470.0/                            ! fit btw. 500 and 1900 K with MAE of 2.2%, 10.5%
  PLOG/1.000E+02    -2.51E+03        1.41    14420.0/
  PLOG/1.000E+02    4.05E-09        5.14    10480.0/                            ! fit btw. 400 and 2100 K with MAE of 40.7%, 150.7%

Question

I'm wondering what the best course of action is here. For my particular mechanism, I'm tempted to delete the 100atm parameters since I don't need that high of pressure anyways. That seems easier than trying to suppress the downstream Cantera errors.

But what about for the RMG library? I don't know if Franklin still has the data, but I imagine refitting the parameters with a constraint that the rate never be negative would be extremely involved. Would it be easier to comment out the last two lines with a note about the negative rate at ~500K and 100atm? Or what if we do a fit of the fit? Take the current plot at 100atm and fit a new Plog that never goes negative?