Accessing/exporting time derivatives from CADET simulations

Chromatography Modeling
1

Hello all,
I am currently working with CADET for generating simulation data in the context of physics-informed neural networks (PINNs). I would like to access the time derivatives of the concentration field (dc/dt). My goal is to obtain a reliable dc/dt field over the full column domain to build a probability density for adaptive sampling.

I already enabled the solution recorder option:

column.solution_recorder.write_soldot_bulk = True

From my understanding, this should store the time derivatives internally (soldot). However, I am not entirely sure:

  1. How can I properly access the recorded soldot data after the simulation?
  2. Does soldot_bulk correspond directly to dc/dt of the bulk concentration field, or are there any transformations/scaling applied internally?
  3. What would be the way to export the full spatio-temporal derivative field (t, x) → dc/dt ?
  4. Would you recommend computing dc/dt externally (e.g. via np.gradient) instead of relying on internal derivatives from CADET?

Thanks a lot in advance!
Konstantin

1 Like
2

Hello Konstantin and welcome to the CADET forum !

I only know how its done with the CADET-Python interface, but CADET-Process works probably pretty similar.

  1. In cadet-python there is a field under output/solution/unit_XXX/SOLDOT_BULK that you can load
  2. Yes, its \frac{\partial d c}{\partial t} , in SI units
  3. Loading output/solution/unit_XXX/SOLDOT_BULK gives you the full spatio-temporal field, Im not sure what you mean by export.
  4. I recommend using CADETs derivatives, since the only computational overhead comes from storing the data, no extra computations are conducted.

Best regards,
Jan

3

I opened an issue here.

1 Like