Liquid phase with RMG4 and RMG3 #297
Comments
|
Thanks for bringing this up. Looks like there are many issues here! For the kinetic changes, it looks like all the big differences are in H abstraction - the R_recombination and intramolecular_... look unchanged. The large T exponential terms (n=2.55, for example) makes comparing A and Ea values difficult/fruitless/misleading. I guess either the family values in the database changed, or RMG-4 contains the explicit solvent corrections for H-abstraction (but not other families). @ajalan - can you confirm? For the thermo differences, could you evaluate the polynomials and see if/how they actually vary with temperature? (or at 298K). Looking at the diff screenshots, in all cases it looks like the first 5 coefficients in each temperature range are the same across versions, and only paramateres 6 and 7 vary. The difference in the 6th should be the constant difference in Hf/R and the difference in the 7th the constant difference in S/R. http://www.me.berkeley.edu/gri-mech/data/nasa_plnm.html (the first 5 are the temperature-dependence). If all we claimed to fix between RMG3 and RMG4 is the temperature-dependence of solvation, then it looks like something is wrong. Please ensure @bslakman understands what's required in terms of constantconcentration in RMG-Py (is there an issue open in the RMG-Py project for this?) so we can fix and get that working for you. Also, please could you clarify what you mean "obtained the same RMG4 behavior from octane to hexadecane..."? |
|
Can you also post your chemkin files for us to look at what the changes are. Thanks! |
|
@rwest and @bslakman, I will start working on fixing the constant concentration issue on RMGPY soon (there is an issue open), we thought about a way to avoid the problem. I will update the issue when i will have some progress on it. @rwest there is more information on kinetic parameters determination below. I still have the log file. I may have more information in my log file for the thermo part. This is a diff on the 2 chem.inp files (cleaned up because some species number changed). As you said the 5 first parameters are constant on all species. And by obtained the same RMG 4 behavior, I mean: @ajalan so the test presented here were generated at 298K. All the other test were preformed between 350-550K Just to compare properly Species 1,2,4,5,6 are the same in both mecanisms but SPC 31 become 29, 49 ->45, 32->30, 50->46 from RMG4 to RMG3. this is the RMG4 chem.inp file ! This chemkin file was generated by RMG - Reaction Mechanism Generator (http://rmg.mit.edu) ELEMENTS H C O N Ne Ar He Si S Cl END !Estimated by RMG using Group Additivity !Primary Thermo Library: GRIMech3.0 (Species ID: s00010295) !Primary Thermo Library: GRIMech3.0 (Species ID: s00010103) !Estimated by RMG using Group Additivity !Estimated by RMG using Group Additivity !Estimated by RMG using Group Additivity !Estimated by RMG using Group Additivity !Estimated by RMG using Group Additivity !Estimated by RMG using Group Additivity END REACTIONS KCAL/MOL MOLES Here start the chem.inp file from RMG3 ELEMENTS H C O N Ne Ar He Si S END !Estimated by RMG using Group Additivity !Primary Thermo Library: GRIMech3.0 (Species ID: s00010295) !Primary Thermo Library: GRIMech3.0 (Species ID: s00010103) !Estimated by RMG using Group Additivity !Estimated by RMG using Group Additivity !Estimated by RMG using Group Additivity !Estimated by RMG using Group Additivity !Estimated by RMG using Group Additivity !Estimated by RMG using Group Additivity !Estimated by RMG using Group Additivity !Estimated by RMG using Group Additivity !Estimated by RMG using Group Additivity !Estimated by RMG using Group Additivity !Primary Thermo Library: GRIMech3.0 (Species ID: s00009882) !Estimated by RMG using Group Additivity !Estimated by RMG using Group Additivity !Estimated by RMG using Group Additivity END REACTIONS KCAL/MOL MOLES C8H17J(5)+O2(2)=SPC(29) 7.540e+12 0.00 0.00 !R_Recombination estimate: (Average:) [ C_rad/H/NonDeC , O2_birad ] Diffusion factor=0.479, rate=3.61e+12, Keq=1.24e+23, at T=298K. Octane(1)+C8H17J(6)=C8H17J(7)+Octane(1) 6.080e+01 3.19 10.31 !H_Abstraction exact: [ C/H2/NonDeC , C_rad/H/NonDeC ] Diffusion factor=0.00, rate=130, Keq=1.00, at T=298K. C8H17J(6)+C8H17J(7)=C8H16(28)+Octane(1) 1.026e+14 -0.35 0.00 !Disproportionation exact: [ C/H2/Nd_Csrad , C_rad/H/NonDeC ] Diffusion factor=0.475, rate=3.31e+12, Keq=9.34e+44, at T=298K. |
Hello,
I am starting my work on liquid phase and I have started with RMG-Java instead of Python (because of the constant concentration issue). However, I found something strange between RMG4 and RMG3.
As you mentioned on the RMG website, calculation of the enthalpy of solvation with RMG3 is incorrect.This problem should appear with temperatures different than 298K. However, I did a simple comparison of the liquid phase example (octane autoxidation) at 298K with RMG3 and RMG4, and at the end I don't have the same thermochemistry with the same species. See the following differences of the 2 chem.inp files.
Just cheking species coming from the GRI-MECH library :
you can see that they are different (but they shoudn't at 298K for the reasons explained before. Correct me if I am wrong).
So I made a diff between some directories of my 2 RMG versions to check that my libraries and groups have the same values. This is True for :
thermo_libraries/SolventLibraries/Library.txt
thermo_librairies/GRI-MECH3.0/Library.txt
thermo_librairies/GRI-MECH3.0/Dictionnary.txt
thermo_librairies/primaryAbrahamLibrary/Dictionary.txt and library.txt
thermo_group/Group_library.txt (except sulfur containing groups)
Then when you just look at the kinetic parameters obtained for the first reactions they are very different. Energy barrier heights drastically change from E+10 to E-10 for reactions involving reactants. Are those changes in agreement with RMG4 improvements?
I tried to check some kinetic libraries and groups but there is too many changes (new families, new groups...).
This has a huge impact on the mechanism obtained with RMG4.
I checked with several solvents and I obtained the same RMG4 behavior from octane to hexadecane at 5 different temperatures.
In addition, RMG3, which is supposed to not work properly at temperatures other than 298K, is able to reproduce more realistic fuel consumption delays compare to RMG4 mechanisms which are several orders of magnitude too long then too short after a given temperature. It seems there is something wrong with the solvation in RMG4. What do you think?
Did you ever report this issue before. Do you think that it is possible to find it in the RMG-Py version after the constant concentration issue is solved?
Thanks
The text was updated successfully, but these errors were encountered: