Dusty gas model

[Ali Al-Matouq]

Hi all,

Previously there was a discussion about implementing the Dusty gas model in Matlab.  I've attached Matlab script for finding the required fluxes using Matlab for sharing.  To my knowledge it works as expected and was based on the literature.  See my comments inside the program for more detail.

Regards.

Ali.

[Ray Speth]

Hi Ali,

I'm sure that users of the Matlab toolbox will appreciate this contribution. I think in the longer term, the best route will be to expose the Dusty Gas model that is already implemented in the Cantera C++ core to Matlab in the same way as the interface to the other transport models is done, so that the model only needs to be implemented in one place.

Regards,

Ray

[Ali Al-Matouq]

Thanks Ray,

I totally agree and maybe we can lift the burden of keeping the Matlab toolbox up to date by having some instructions of how exactly can that be done in an efficient matter.  My experience with using Matlab Cantera with ode15s solver and Matlab optimization packages for a chemical reacting flow problem was a rewarding one.  Development time was much faster than I expected.  

Our previous discussion shed some light on how to do migration of C++ classes, but it wasn't really clear for me.  I will appreciate more details if you have the time.

Regards.

Ali.

[Steven DeCaluwe]

Hi Ali,

Sorry that I am just getting around to looking at this, but one thing caught my eye when I looked at the code.

A common problem in DGM implementations is that there is usually not a clear definition of what is meant when one says ‘tortuosity’ (some mean the tortuosity, tau_g, while some mean the “tortuosity factor”, (tau_g)^2).  

Your code seems to handle the tortuosity the same way that the Cantera DGM module does - i.e. the tortuosity used is actually the tortuosity factor, otherwise effective diffusion coefficients (binary and Knudsen) and permeability should be divided by (tau_g)^2, rather than tau_g.

I just wanted to point out that the tortuosity is also required to calculate property gradients.  The gradient we care about is the gradient with respect to pore length, and so the distance between discretized locations that is used to calculate property gradients should be the ‘axial length’ (i.e. in the direction of finite volume surface normal) multiplied by tau_g.  In your code, this would be coded as:

DISINV = 1/(PTXR-PTXL)/Tortuosity; 

Note that the tortuosity terms can still all be lumped into a “tortuosity factor” (and probably should, since it is computationally faster), and in this case the true physical meaning of the tortuosity factor for this implementation (and most DGM implementations) is (tau_g)^3.  

Lastly, I’ll note that for the Cantera DGM module, the user provides an input parameter ‘delta’ that specifies the “distance from state 1 to state 2” (i.e. the numerator for property gradient calcs).  The user could therefore choose to “pre-correct” delta for the tortuosity, such that the tortuosity factor provided to the DGM module is the common (tau_g)^2.  This seems somewhat risky, though, since the tortuosity provided to Cantera and that used to correct delta would be two different values.

Anyway, this is a longer comment than I originally intended to make, but are just some thoughts I wanted to share.  FYI - it is a medium-term goal of mine to make the underlying C++ module available from the Matlab interface.

Cheers,

Steven

  

--
You received this message because you are subscribed to the Google Groups "Cantera User's Group" group.
To unsubscribe from this group and stop receiving emails from it, send an email to cantera-user...@googlegroups.com.
To post to this group, send email to canter...@googlegroups.com.
Visit this group at http://groups.google.com/group/cantera-users?hl=en-US.
For more options, visit https://groups.google.com/groups/opt_out.
 
 
<DGMflux.m>