I am just getting familiar with CADET-MATCH and run into some troubles
I would like to determine tubing dispersion with a puls injection, mimicking a acetone injection from a sample loop. The system would be inlet -> sample tubing -> tubing -> outlet
While it is working fine with artificial data as soon as I use data obtained from the AKTA system I get the Error “runExperiment 318 The simulation failed”. So I suppose something is wrong the way I prepare the .csv file. However, I dont know what causes the problem.
I attached the error log and other files.
Another thing I want to ask is in regard of match.plot_best(). It seems, it plots the chromatogram of the reference model as the simulation output. The results of the fitting parameters are correct. Is there anything I am doing incorrectly ?
Thanks a lot for your help.
match_inject.ipynb.txt (6.8 KB)
error_log.txt (13.8 KB)
data_csv.csv (59.0 KB) model_csv.csv (15.0 KB)
Can you attach the h5 and json files? The error indicates that for all combinations of parameters tried the actual simulation ended with an error.
thanks for your help.
reference_model.h5 (548.1 KB) match_peak.json (752 Bytes)
I can’t replicate the error but what I can tell you is that there is no good match for your system.
From what I can see your setup has some tubing with 1000 mol/m^3 (1000 mM) 1.1 meters long with an area of 4.4e-7 m^2 and flow that into a tube that is 6.3 meters long and 7.9e-7 m^2 area.
The simulation has far more mass in it than the experimental data that it is fitting against. You have a 3-component system but are only using one of them and only one of them even has any mass.
In terms of the fitting don’t use the null transform. You are searching across 4 orders of magnitude and a null transform just can’t span that space effectively. Use the auto transform, it will automatically adjust between a linear and log transform to improve searching. I would also normally recommend using Shape instead of SSE
If I lower the initial concentration on the tubing to 600 mol/m^3 and I allow the dispersion to be much higher I can get a reasonable fit with a dispersion of 1e-3.
thanks a lot for help and testing.
Just want to double check, you can run CADET match with both .csv files, data_csv and model_csv ?
Anyways, at least I know now that there is nothing wrong with my dataset, and there must a problem on my system.
One more quick questions:
In the error message: “util.py runExperiment 318 The simulation failed None” What does the “None” stands for ?