[FLASH-BUGS] bug report

[Congyao Zhang]

To whom it may concern,

      I report a bug to flash-bugs at flash.uchicago.edu, but the mail is
automatically rejected.  So I copy the bug report to here.

      I am modeling a hydrostatic gas distribution in the gravity field by
using 2D FLASH simulations. The gravity parallels to the y-axis. Both of
the gas distribution and the gravity field have a constant velocity along
the y-axis. The model is shown as the following figure. The gas density
profile is moving to the left with a constant velocity, but the shape of
the profiles is stable because of the balance between the gravity and the
gas pressure.

     I find that the FLASH code produces an artificial velocity pattern
when using multiple processors (shown as the figures below). In my test,
when the number of processors is N <= 15, everything works ok. The
hydrostatic gas profiles are quite stable in the gravity field. However,
when using more processors, the artificial velocity pattern appears (all
other simulation settings are the same). The simulated velocity fields
become quite unstable.

     I have tested the following points:

            1. the artificial velocity pattern does not related with the
spatial resolution and the type of the boundary condition.

            2. I find that if the motion of the gas distribution is
perpendicular to the gravity direction, there is no any artificial
velocity. It seems that the problem occurs only when the velocity and
gravity have the same direction.

            3. I have tested the "Split" and "Unsplit" Hydro solvers,
"Adaptive Mesh" and "Uniform Grid" units. They all show the artificial
velocity pattern.

            4. With regard to the gravity, I tried the "PlanePar" unit
(directly calculate the gravity force based on the analytical equations)
and "UserDefined" (introduce a time-dependent gravitational potential in
the simulation, e.g., g(x,t) = (Phi(x+dx,t)-Phi(x,t))/dx), they both show
the same artificial velocity pattern.

            5. I have tested the FLASH4.0, FLASH4.3 and FLASH4.4. They all
have the same problem.

       I have attached my Simulation unit, if you want to do some tests.
Hope we could find which parts cause this problem. Thank you so much for
your help.


Best regards,
Congyao Zhang




​
​
-------------- next part --------------
An HTML attachment was scrubbed...
URL: http://flash.uchicago.edu/pipermail/flash-bugs/attachments/20170220/f1358dcd/attachment.html 
-------------- next part --------------
A non-text attachment was scrubbed...
Name: 24.png
Type: image/png
Size: 71289 bytes
Desc: not available
Url : http://flash.uchicago.edu/pipermail/flash-bugs/attachments/20170220/f1358dcd/attachment.png 
-------------- next part --------------
A non-text attachment was scrubbed...
Name: 4.png
Type: image/png
Size: 54104 bytes
Desc: not available
Url : http://flash.uchicago.edu/pipermail/flash-bugs/attachments/20170220/f1358dcd/attachment-0001.png 
-------------- next part --------------
A non-text attachment was scrubbed...
Name: 40.png
Type: image/png
Size: 95606 bytes
Desc: not available
Url : http://flash.uchicago.edu/pipermail/flash-bugs/attachments/20170220/f1358dcd/attachment-0002.png 
-------------- next part --------------
A non-text attachment was scrubbed...
Name: density profiles.png
Type: image/png
Size: 89833 bytes
Desc: not available
Url : http://flash.uchicago.edu/pipermail/flash-bugs/attachments/20170220/f1358dcd/attachment-0003.png 
-------------- next part --------------
A non-text attachment was scrubbed...
Name: 2D_test.tar.gz
Type: application/x-gzip
Size: 18604 bytes
Desc: not available
Url : http://flash.uchicago.edu/pipermail/flash-bugs/attachments/20170220/f1358dcd/attachment.gz