Zeta potential not working?

[sebastiansturm]

First of all, many thanks to the Elmer team for giving away such a fine piece of software and for the large amount of work that has obviously been invested in providing high-quality documentation! I'd like to use Elmer for some electrokinetic calculations. Although it seems to work in principle, I'm having a number of minor problems, namely
(1) VTK export doesn't export all available fields, only the electrostatic quantities. This is unimportant, though; switching to VTU solves that issue.
(2) The FluidicForce solver may calculate something (?), but I'm unable to find the resulting value. Tried searching the .results-file and invoking a SaveScalars solver, to no avail.
(3) The helmholtz_smoluchowski boundary condition seems to ignore the "Zeta Potential" setting. Even insanely high values will not produce a flow field; setting the EO Mobility, however, will do.
Am I doing something wrong? I have attached my .sif file and the accompanying mesh.
best regards,
Sebastian

[tzwinger]

Dear Sebastian,
on a first look I had I think there's something not OK with your mesh or I have issues in understanding what you are after.

Three issues I found:

1) the left side boundary for me appears not to be defined all over the height (just load the mesh into ElmerPost and display all lines and klick out the surf1 in "Groups" and you see it)
2) being part of the interupted left side boundary you DNA Side boundary falls on the y=0 position, which physically makes no sense to me (as this is a singular line)
3) there is an unasigned boundary (occuring as Condition 0 in your result file) in the upper part of the domain

All in all I have the impression your mesh is corrupted. If you could fix your mesh (or alternatively explain the reason for it being as is) and rerun and then we could try to fix the other issues (should they remain to be issues)

Best regards,

Thomas

[raback]

Hi

I had quickly a look at your case. Some comments:

2) The FluidicForce computes integral forces i.e. the result is not a field. You can save the things with SaveScalars. That being said, the fluidic force cannot work correctly with these BCs because no viscous boundary layer is solved.

3) There was a bug. I corrected it and also simplified the solver quite a bit. The new solver is in svn.

Whether the results are correct or not, I don't know. I see Thomas just sent a mail on this. Anaways, a simple minimalistic test case on this would help to ensure consistancy of results in future.

-Peter

[sebastiansturm]

Thanks alot for your quick replies! The reason why the "DNA side" boundary only spans part of the left side is that the DNA is held fixed along the capillary, but doesn't reach from one end to the other. The rest of the capillary is supposed to be filled with water. I don't quite understand what's singular about y=0. Maybe I didn't define the coordinate system correctly; I intended it to be cylindrically symmetric with the x-coordinate corresponding to r and y corresponding to z.
I'd like to take a look at my mesh using ElmerPost, but so far I'm unable to compile it (on Mac OS X 10.6, with Apple-supplied Tcl/Tk libraries, complains about undefined symbols like "_glXQueryServerString"). Is there a working binary for me to download?
Also, I'm a bit baffled by the remark that no viscous boundary layer is solved. I had assumed that all I had to do was to make sure that the mesh is finely subdivided close to the boundaries and then run a standard NS-simulation. Do I have to tell Elmer somehow that it's supposed to consider boundary layers?
Best regards,
Sebastian

[Juha]

hI,

o the coordiantes (x=r, y=z) are OK
o you need to enforce u_r=0 (Velocity 1 = 0) for the symmetry axis always
for the whole flow domain, this is not done automatically...
o didn't quite understand the viscous boundary layer comment either...

Juha

[raback]

Hi Sebastian

Viscous boundary layer is not solved because these BCs assume that its ridicilously small compared to the other dimensions, and analytical slip velocity is used instead. There is the PoissonBoltzmann solver that can be used when the EDL is in the same range as the width of the geometry. Then one would use no-slip conditions for velocity and apply body forces on the EDL.

-Peter

[sebastiansturm]

Aha, I see. Thanks alot! I had assumed that I still needed a higher mesh resolution close to the slip boundaries in order for the calculated FluidicForce to be accurate. Maybe that was a misconception. I'll add the u_r = 0 constraint and see what comes out. Concerning the FluidicForce, however, I still have problems getting hold of the result. Adding a SaveScalars solver only yields one single value, namely the potential difference of 0.1 [V].
I have attached my current .sif file (still without u_r = 0) and the corresponding SaveScalars output.
Best regards,
Sebastian

[raback]

Well, you should give the "Calculate Fluidic Force" flag at the boundary of interest. However, I can ensure you that you cannot get anything sensible out of it in this case. -Peter

[sebastiansturm]

ok, thanks! I'll try to use it in conjunction with the Poisson-Boltzmann solver.
Best regards,
Sebastian

[raback]

Hi

I realized that there was no public example of the combined PoissonBoltzmann + Navier-Stokes equations. Now there is one under the trunk/fem/tests/PoissonBoltzmann directory in svn

-Peter