Obvious spurious velocity for axisymmetric surface tension

227 views
Skip to first unread message

lj.f...@gmail.com

unread,
Aug 25, 2021, 6:13:06 AM8/25/21
to basilisk-fr
Dear all, 

There are some unreal currents in spots near the free surface when I run a droplet equilibrium test.
The weirdest thing is that the spot can change its location if we set the sphere much lower and only a cap on top of the boundary, which means the contact angle model is used.
I guess it's a bug or something.
Files are attached.

Regards,

Jason

cap.eps
equilibrium.c
sphere.eps

Stephane Zaleski

unread,
Aug 25, 2021, 6:50:54 AM8/25/21
to basilisk-fr
Dear Jason

  Your sphere.eps file is 7Mb, it is too large. 

   Best

Stéphane

Stephane Zaleski

unread,
Aug 25, 2021, 7:10:36 AM8/25/21
to basilisk-fr
Hi again

  I was finally able to open your second file. Attached is a much smaller file with *I think* the same image. It is not clear what you want to show. Which field is displayed ? Where are the spurious velocities ? You say you are surprised that their location changes, but that is a well known feature, not a bug. 

  Best

S. 

Message has been deleted

lj.f...@gmail.com

unread,
Aug 25, 2021, 7:29:17 AM8/25/21
to basilisk-fr
Dear Prof. Zaleski,

Thanks, I re-uploaded the images. The field is velocity and I hope the two new pics may tell it much clearer.
These spots which are not observed in 2D cases may ruin the equilibrium at last.

Best,

lj.f...@gmail.com

unread,
Aug 25, 2021, 7:29:59 AM8/25/21
to basilisk-fr
cap.jpgsphere.jpg

Stephane Zaleski

unread,
Aug 25, 2021, 7:36:19 AM8/25/21
to basilisk-fr
Now it is smaller but it is also different from mine ! Please also send a jpg of cap.eps . First of all you should read Popinet JCP 2009. You will see that spurious velocities are transient, ie they decrease in time, so you should plot your spurious velocities as a function of time. Stéphane P. has made a lot of tests and analyses of this (see Popinet ARFM) so I doubt there is an obvious error or bug, but if you think there is one even after plotting the time dependence you should give much more details and analyses.



lj.f...@gmail.com

unread,
Aug 25, 2021, 7:54:28 AM8/25/21
to basilisk-fr
The spurious velocities are indeed transient but can end up like really large ones and the shape of the circle also deforms for some cases.
They are not distributed on the interface but near it, especially for cells without one component of the height function vector.
People who are interested in this issue can rerun the c code attached above, and I hope I got it wrong though.

Thanks

lj.f...@gmail.com

unread,
Aug 25, 2021, 11:58:09 PM8/25/21
to basilisk-fr
Well, if nobody has experienced this problem, probably my fault then.
So, could anyone give me some hints?

A deformed sphere is like this.
Screenshot from 2021-08-26 11-56-32.png
Thanks

Vignesh T G

unread,
Aug 26, 2021, 3:49:42 AM8/26/21
to basil...@googlegroups.com

Hi Jason,

Can you explain what problem are you trying to solve?     Usually, droplet equilibrium test will not involve contact angle !

Regards,

Vignesh

--
You received this message because you are subscribed to the Google Groups "basilisk-fr" group.
To unsubscribe from this group and stop receiving emails from it, send an email to basilisk-fr...@googlegroups.com.
To view this discussion on the web visit https://groups.google.com/d/msgid/basilisk-fr/f62ecf83-ee2b-4777-a2eb-21257f5c0d15n%40googlegroups.com.

lj.f...@gmail.com

unread,
Aug 26, 2021, 4:42:01 AM8/26/21
to basilisk-fr
Hi Vignesh, 

I tried to verify a contact angle model to see if it can maintain the equilibrium, but it's not the point right now.
The issue given above has nothing to do with the contact line. As you can see, the droplet sphere deforms without touching any boundary in the absence of gravity.
I modified the *rising* test and no such problem.
a[] and alpha[]*(p[]-p[-1])/Delta show me no balance on the interface.
Totally confused.
I guess there must be something I got wrong with.

Best,
Jason

lj.f...@gmail.com

unread,
Aug 26, 2021, 5:58:04 AM8/26/21
to basilisk-fr
To make it clear, I plot the error points between tension and pressure gradient.
Values in these spots are much higher than in other locations.
Again, this is an axisymmetric case, and a half-circle is enough to represent a sphere.
Screenshot from 2021-08-26 17-51-43.png
Reply all
Reply to author
Forward
0 new messages