Bug Reports for CFDEM®coupling, LIGGGHTS®, and ParScale

(Possible) bugs / suspicious behavior should be reported as a new thread here, not in the user forum. This should give both you and the developers a clear idea about the status of your bug report (submitted/assigned/fixed)

engineSearchIB Bugs (w/ fixes) - cfdemSolverIB

Submitted by Nucleophobe on

Hi,

First, thank you to everyone involved with the CFDEM®project for your hard work. Dr. Hager, your fictitious domain solver has been particularly helpful for my Ph.D. research.

There are a couple of bugs in the locate model 'engineSearchIB' that lead to an incorrect generation of the satellite points. Thus, when a particle resides on two or more processors, part of the particle may be 'missed'. In my experience, this can lead to large errors in the particle pathlines in some cases.

read_data using fix keyword

Submitted by TobRas on

Hi,

in the documentation of the method read_data (http://www.cfdem.com/media/DEM/docu/read_data.html) it is mentioned how the fix keyword can be used to process a specific portion of the data file through a pre-defined fix.

Trying to apply this functionality, however, fails. When I look through the source code I end up at the function ReadData::fix :

Compilation abort for latest github version

Submitted by jenslaube on

Hi,

I tried to compile the latest Github version from today (Version 3.3.1, 29th of February) and receive the following Warning:

fix_check_timestep_gran.cpp(67): warning #47: incompatible redefinition of macro "BIG" (declared at line 49 of "../global_properties.h")
#define BIG 1000000.

with a subsequent compilation abort.
In comparison to the former versions, the line "#define BIG 1.0e20" was inserted into the file "global_properties.h".
Can you fix this or tell me how to avoid the Error? Thanks a lot!

Best wishes
Jens

wrong(?) equation for Capillary Force in gran cohesion easo/capillary/viscous

Submitted by giuraso on

Hi,
I have checked the capillary force model in cohesion easo/capillary/viscous.
I think the equation you give as reference is

B = (-0.34*LN(volBondScaled)-0.96)*contactAngleEff*contactAngleEff - 0.019*LN(volBondScaled) + 0.48
C = 0.0042LN(volBondScaled)+0.078;

but in the code you use LOG instead of LN, and converted LN into LOG.
So you divided the coefficient 0.34, 0.0019 and 0.0042 by 2.303.

B = (-0.148*log(volBondScaled)-0.96)*contactAngleEff*contactAngleEff - 0.0082*log(volBondScaled) + 0.48
C = 0.0018*log(volBondScaled)+0.078;

very minor bug in unfix and restart when using same name for a fix

Submitted by pfalkingham on

I have a two-part simulation with a set-up file and a run file.

I happen to name the contact fix 'cont0' in both. The set-up input creates fix cont0, runs what needs to run, then unfixes cont0 and writes the restart file. The run input then reads the restart file and creates a new fix cont0 (with the same properties).

On Cygwin, this is fine and causes no issues (compiled as fedora). But when compiled on XC30, this results in a seg fault when running the run input.

Bug in compute nparticles/tracer/region

Submitted by Christoph on

Hi,

It seems to me that the mass calculation of marked particles with the compute nparticles/Tracer/Region is not working correct.
The reason therefore may be that the marker of the Particle will be reset by the first call of compute_vector_eval.
As the Method compute_vector calls the compute_vector_eval twice, it have to be ensured that the marker will not be reset by the first call.

The following modification worked for me:
compute_nparticles_tracer_region.cpp 253
replace if(reset_marker_)
by if(countMass && reset_marker_)

Bug in contact handling

Submitted by richti83 on

I'm trying to simulate a sieving stack. The Particles are 6% larger than the holes.
I reduced the problem to one single hole and one particle.
When the particle is exactly centered to tho hole everything works as expected. But if it is a little bit excentric the particles jumps arround unexpectedly.
See attached script and images for details.
I think something is wrong in handling edge contacts.
I tried different youngsmodulus and timesteps allready.

thanks for having a look,
Christian.

Segfault with fix deform

Submitted by Guesnet Etienne on

Hi everyone,

I have found a segfault in liggghts (with atom_style granular). Files to reproduce are joined. To reproduce, use mpirun with at least 2 processors (mpirun -np 2 liggghts -in agg.in_.txt); segfault will occur between 15'000 and 20'000 time step.

About system: Linux; result of uname -a command: Linux PC-GUESNET 3.16.0-38-generic #52~14.04.1-Ubuntu SMP Fri May 8 09:43:57 UTC 2015 x86_64 x86_64 x86_64 GNU/Linux; mpirun version is 1.6.5.

About Liggghts: last version in github (3.3.0), compiling with default option (make fedora), no any modifications.

Washino and easo capillary wall adhesion

Submitted by abrahamp on

The wall interaction in the Washino and Easo model is not implemented for what I can see. Below is a cut from the original file, it is comented and there is hence no visible cohesion in the simulations either.
if(sidata.is_wall) {
/* !!!!!!!!!!!!!!!!!!!!!!!!!!!!!!TODO HERE!!!!!!!!!!!
const double area_ratio = sidata.area_ratio;
i_forces.delta_F[0] += Ft1 * area_ratio;
i_forces.delta_F[1] += Ft2 * area_ratio;
i_forces.delta_F[2] += Ft3 * area_ratio;
i_forces.delta_torque[0] = -sidata.cri * tor1 * area_ratio;

region mesh/tet

Submitted by Daniel Queteschiner on

You should keep an eye not only on the compiler errors but also on the warnings.
The following lines are missing parentheses or should use !=, the logical not is only applied to the left hand side of comparison
https://github.com/CFDEMproject/LIGGGHTS-PUBLIC/blob/3.3.0/src/region_m…
https://github.com/CFDEMproject/LIGGGHTS-PUBLIC/blob/3.3.0/src/region_m…