Breakthrough matching


I have a simulated breakthrough curve with a general rate model for affinity chromatography by CADET software. I am now matching the simulated curve with the experimental breakthrough curve by CADETMatch software to estimate the parameters (Dax, Ds, Dp, Kf).

Facing following matching issues:

Problem 1: The fitted curve is above the experimental curve, although the simulated curve with initial guesses is aligning to the experimental curve.

Problem 2: Matching is not following the slope of the experimental breakthrough curve.

Matching details:

Search Method: ‘NSGA3’

Score: SSE

Can you please suggest a better matching strategy to get the best fit for the experimental curve?

Instead of using SSE use ShapeFront. Also if you can share the HDF5, CSV and JSON files I can look at them and try to figure out why it is not getting a better match.

Am I right that from a single breakthrough you want to calibrate the Axial Dispersion, Surface Diffusion, Pore Diffusion and Film Diffusion? Do you already know your column porosity and particle porosity?

1 Like

Thanks for reply.

Not able to upload .CSV file format here, so uploaded data in the text file.

Yes, I am calibrating that four mass transfer parameters with breakthrough curves at various operating conditions, and I do have column and particle porosity values. I am trying to match simulation results to real experimental data, but I think in CADETMatch, it fits with synthetic data generated by simulation. I am a new user of CADETMatch, so can you please advice in this regard.

Thank you.

(example.txt: simulation result data, EL43.txt: Experimental data)

EL43.txt (978 Bytes) example.txt (1.1 KB)

example.h5 (675.0 KB) example.json (1.2 KB)

I will look at your data and try to fit it as soon as I get a chance. CADETMatch itself doesn’t care if you fit to simulated or real data. All it sees is a CSV file to fit to. Most commonly we use CADETMatcht of it to indistrual data and not simulation results.

1 Like

Thank you so much for the information :smiley:, it will be great if it fits experimental data.

I have done with fitting with SSE and ShapeFront on both the experimental data and the simulated data. I can get a match on the simulated data without any problems. For the experimental data I think there is something missing from this model still. The most likely explanation is the porosity is wrong or tubing needs to be included since it looks like there is no combination of these 4 parameters can be explain the experimental data.

I have put all 3 results on my onedrive so you can download it.!Ar8xEUd14_wY9mXIuFqcn4cee3V5?e=UoHhvA

In each of the fit directories in the meta folder you will find the best results and what you will see for the experimental results is that none of them fit that well and it looks systemic.

1 Like

Thank you so much for your help and time :blush: :pray: