[Fedecola]

Hi,

I would like to optimize the kinetic parameter (pre-exponential factor) for the Sage kinetic solver in mech.dat. I need to reach a certain level of emissions and they are influenced from this parameter. Is it possible to obtain my target value by the CONGO algorithm? Is it possible to activate it on the case setup of Converge Studio?

[dprobst]

Originally Posted by Fedecola

Hi,

I would like to optimize the kinetic parameter (pre-exponential factor) for the Sage kinetic solver in mech.dat. I need to reach a certain level of emissions and they are influenced from this parameter. Is it possible to obtain my target value by the CONGO algorithm? Is it possible to activate it on the case setup of Converge Studio?

Yes, you can do that but optimization requires a lot of simulations.
You would set up parameters in the mech.dat of important reactions and then run an optimization, targeting the emissions parameters. It could be setup with the help of Studio.
Assuming you could achieve your targets, it would be computationally costly. Our experience with mechanism tuning is that appx 2000 generations are required with a population of 9. This would be about 16,000 individual Converge simulations. For this reason, most mechanism tuning focuses on ign delay and flamespeed, although it is has been done with full Converge engine simulations.

[Fedecola]

Thanks Dan. I appreciate your answer.

[Fedecola]

I am trying to develop the CONGO algorithm for 3 pre exponential factor in mech.dat with a population of 5 and 100 generation. The case seems to fail when generating the merit function. It creates the ga_output.out but the written value of the performance variables inside the file are zero. I need to get the value from the 'species_mass_fraction_region6.out' but I don't understand how the CONGO manage to get these value from that file.

This is the merit.in :

ga_output output_file_name
2 num_output_vars
CH4 output_var_name
NH3 output_var_name
0 merit_lib_flag
2 num_performance_vars
CH4 performance_var_name
maximize type
0.00016 value
0.5 weight
NH3 performance_var_name
maximize type
5.6E-05 value
0.5 weight
0 num_constraint_vars

I have no parameter to insert the output file from when the merit function get the values. Any suggestion?

[dprobst]

Originally Posted by Fedecola

I am trying to develop the CONGO algorithm for 3 pre exponential factor in mech.dat with a population of 5 and 100 generation. The case seems to fail when generating the merit function. It creates the ga_output.out but the written value of the performance variables inside the file are zero. I need to get the value from the 'species_mass_fraction_region6.out' but I don't understand how the CONGO manage to get these value from that file.

This is the merit.in :

ga_output output_file_name
2 num_output_vars
CH4 output_var_name
NH3 output_var_name
0 merit_lib_flag
2 num_performance_vars
CH4 performance_var_name
maximize type
0.00016 value
0.5 weight
NH3 performance_var_name
maximize type
5.6E-05 value
0.5 weight
0 num_constraint_vars

I have no parameter to insert the output file from when the merit function get the values. Any suggestion?

The results file for each individual run is of the form ga_output.3-14 (where 3 is the individual and 14 is the generation). Typical file contents include ignition delay and/or flamespeed. An example of the file contents is shown below. When the GA folder is created by mechtune, this file should be generated when Converge finishes. THere could be a number of issues causing your case to not work. Feel free to contact support if you continue to have problems.

1 runcompleteflag
1.1289231e-01 ignitiondelay1
5.3893531e-02 ignitiondelay2
2.7772379e-02 ignitiondelay3
1.5350810e-02 ignitiondelay4
9.0360312e-03 ignitiondelay5
5.6253114e-03 ignitiondelay6
3.6826836e-03 ignitiondelay7
2.5260741e-03 ignitiondelay8
1.8151106e-03 ignitiondelay9
1.3714352e-03 ignitiondelay10
1.1101857e-03 ignitiondelay11
1.1099758e-01 ignitiondelay12
5.2817240e-02 ignitiondelay13
2.7121570e-02 ignitiondelay14
1.4935417e-02 ignitiondelay15
8.7584322e-03 ignitiondelay16
5.4313256e-03 ignitiondelay17
3.5421039e-03 ignitiondelay18
2.4190868e-03 ignitiondelay19
1.7293353e-03 ignitiondelay20
1.2981063e-03 ignitiondelay21
1.0381123e-03 ignitiondelay22

[khaledhmz]

Hello Dan,
can you help me with the conversation of pre exponential factor in chemkin file, I use reduced mechanism of propane (chemkin file) to simulate premixed flame and wheen I import it to fluent the units of pre exponential factor changed I m reallz looked in this...

[arajendran]

Originally Posted by khaledhmz

Hello Dan,
can you help me with the conversation of pre exponential factor in chemkin file, I use reduced mechanism of propane (chemkin file) to simulate premixed flame and wheen I import it to fluent the units of pre exponential factor changed I m reallz looked in this...

Hi,



This sounds like more of a Fluent-import issue, and you might be served better asking it on a Fluent forum.

[khaledhmz]

okay thank you