yade-dev team mailing list archive

[Christian Jakob <jakob@xxxxxxxxxxxxxxxxxxx>]

> I have the impression that Christian was getting the message in the
> initial confinement phase, when all particles are distant. In this case
> the message is irrelevant. Likewise, the message should not appear when
> the bridge between two distant spheres breaks. If we can't define a more
> restrictive condition for printing the warning only when there is really
> a problem, no message is better than 10k messages, I agree with the last
> revert.

I started testing with clumps (suction = 25 kPa) and voila, I get the  
ERROR again (this times once, at early consolidation stage). Without  
clumps I do not get errors at all (suction = 25 kPa).

I am using replacByClumps() method to create my model. So the grain  
size distribution does not change, but smaller clump members with  
radii < 0,063 mm could cause the problem.

I can live with the error message, but we should not call it an error.  
I suggest to leave the code as it is, but replace LOG_ERROR by  
LOG_WARN and maybe give more background informations in the message  
text.

btw: The error disappears, when suction is around 1 kPa in my model  
(tested 10, 5 and 1 kPa).


Zitat von Bruno Chareyre <bruno.chareyre@xxxxxxxxxxxxxxx>:

> Hello Jérôme,
>
> > This corresponds in both cases to smallest filling angle values around
> > 2 degrees (two last columns), whereas probably the theoretical maximum
> > value of suction here is infinite, and the theoretical minimum filling
> > angle should be 0.
> Yes. Did you check the value of saturation degree for a filling angle of
> 2%? It must be extremely small, isn't it?
>
> > If I am right, I guess the same issue arises for non contacting
> > spheres. For monosized spheres at an adimensionned distance of 0.1,
> > the maximum suction value in the capillary file is 8.7, corresponding
> > to a lowest filling angle value of 10.6 deg (l. 12876 of M(r=1)
> > file...). I guess these values only approximate the exact limit values
> > of maximum suction and minimum filling angle, but the exact suction
> > limit value is probably far from 1500 in this distant case.
> Well, as you pointed above there is no theoretical max value for
> touching spheres whereas there is one for distant spheres.
>
> > Do you share my understanding ?
> Yes, it seems.
> > If yes, there would be only one problem in our simulations : that the
> > code can not handle correctly the cases (existing in reality) falling
> > between the maximum suction value written in the files (that depends
> > on the radii ratio and distance), and the exact maximum suction value.
> > Is it really a big deal ?
> I think the code handles this correctly. Simply the maximum in the model
> is only an approximation of the exact maximum, but I think the
> approximation is relatively good. If the precision was not enough it
> could always be increased by refining the input data. The only thing
> that is not handled correctly is the case of touching spheres when
> pc->+?, since we can't approximate +? numerically. The only solution
> would be to derive an asymptotic expression for this case and to put it
> in the code.
>
> > (If not, it is finally maybe not catastrophic to run simulations with
> > uc* > e.g. 1500, and Christian was right to lower the "noise" of the
> > error message)
> I have the impression that Christian was getting the message in the
> initial confinement phase, when all particles are distant. In this case
> the message is irrelevant. Likewise, the message should not appear when
> the bridge between two distant spheres breaks. If we can't define a more
> restrictive condition for printing the warning only when there is really
> a problem, no message is better than 10k messages, I agree with the last
> revert.
>
> Bruno