getCapillaryStress() returns Matrix::Zero()

[Christian Jakob]

Hi,

I noticed, that getCapillaryStress() function returns

Matrix3(0,0,0, 0,0,0, 0,0,0)

whatever the input is (getCapillaryStress(), getCapillaryStress(1), getCapillaryStress(0), getCapillaryStress(vol), ...)

In a yade version from july of last year it was working while a newer version (march 2014) is not

Can someone confirm, please?

[Bruno Chareyre (bruno-chareyre)]

Hi,
The code looks ok.
Are you sure that there is at least one capillary force in the
simulation when computing the stress?
Bruno

[Christian Jakob (jakob-ifgt)]

Yes, the code looks fine and there was no change in the related files during that time except splitting Shop.cpp into two files ...

also there was a change in parameter names about one year ago (Fcap <-> fCap)

Yes, I am sure, there are contacts, there are meniscii, there are capillary forces, ...

btw getStress() is working fine at the same model

chr.

[Christian Jakob (jakob-ifgt)]

for reproducing see script in the attachment

btw, the link for getting capillary files is no more working:

https://yade-dem.org/index.php/CapillaryTriaxialTest

(links to https://yade-dem.org/doc/index.html)

[Christian Jakob (jakob-ifgt)]

This is really strange.

I definitely got output from getCapillaryStress() with Yade 2013-07-18.git-f0b5fd0 when running scripts last year. I can proof that!
The code of getCapillaryStress() did not change between july and december 2013.
Today (or since beginning of this year) I get always

Matrix3(0,0,0, 0,0,0, 0,0,0)

whatever I do.
I ran the script at my second machine but still the output is zero.
Even when running the script with the old version ( Yade 2013-07-18.git-f0b5fd0) ...

I am helpless, can anyone confirm this behavior please?

christian

[Bruno Chareyre (bruno-chareyre)]

The problem is getStress is trying to static_cast MindlinCapillaryPhys as a CapillaryPhys while they are independant classes.
I don't think it ever worked before. I'm surprised that it doesn't simply crash. Maybe last year you were not using Mindlin law.
Or, you were very lucky and somehow the wrong cast was giving the right value since both classes have the member fCap.

Anyway, it is fixed now [1].
You can use getCapillaryStress(mindlin=True)
Note that you should give the volume argument to.

[1] https://github.com/yade/trunk/commit/84a28bebea5a8c2a772a11ccd88934e9e6ae88fe