How can I solve this problem? Thank you very much
silane-cgp_2023-11-16T230519.557125.zip (7.2 MB)
This looks like a normal SMS (Segregated Mesher/Solver) run. In this mode, the processing stops after the mesh has been computed, you need to inspect and accept the mesh to proceed. The intention is to ensure the mesh is good before wasting time running the simulation over a poor quality mesh.
So, go into the Mesh pane and click “Accept”. Or else disable SMS mode under Settings in th inte main menu,
Is this the option to enter the mesh pane and click ‘Accept’? I still cannot run after choosing.
In addition, the mesher workflow in the main menu settings is in standard mode, so I still don’t know what to do to run this case. Please help me double check. Thank you very much.
silane-cgp_2023-11-17T191622.778731.zip (31.6 MB)
Sorry, the message in your first screenshot above (“PRE_PROCESSING COMPLETE”) led me to believe that SMS mode was active.
In SMS mode, the controls to generate/accept mesh are here:
But if you are not running in SMS mode this does not apply. The option to enable array re-indexing is unrelated - but we have seen errors related to re-indexing so you may want to turn this off.
Running your case in MFiX 23.3.2 I find there is an error which causes the keyword des_interp_width to be clobbered when the file is loaded. The strike-through indicates a keyword being unset. The startup message looks like this:
Welcome to MFiX - https://mfix.netl.doe.gov MFiX-GUI version 23.3.2 Ready Loading silane-CGp.mfx from /tmp/silane-cgp_2023-11-17T191622.778731 Solver: MFiX-CGPdes_interp_width
When I run the case, I immediately get an error popup:
I have fixed this keyword bug and we will release an updated version (23.3.3) shortly.
Once we get past that problem, I have the following:
Program received signal SIGFPE: Floating-point exception - erroneous arithmetic operation.
Backtrace for this error:
#0 calc_resid_mod_MOD_calc_resid_u
at calc_resid.f:800
#1 u_m_star
at solve_vel_star.f:322
#2 solve_vel_star_mod_MOD_solve_vel_star
at solve_vel_star.f:134
#3 iterate_MOD_do_iteration
at iterate.f:255
#4 run_fluid
at mfix.f:191
#5 run_mfix
at mfix.f:145
#6 main_MOD_run_mfix0
at main.f:85
Going in with the debugger:
Program terminated with signal SIGFPE, Arithmetic exception.
#0 0x00007f90e20d31a6 in calc_resid_mod::calc_resid_u (u_m=..., v_m=..., w_m=..., a_m=..., b_m=..., m=0, num=nan(0xfffffffffffff),
den=559.65250896583655, resid=0, max_resid=5.3124284493905395e-07, ijk_resid=3247)
at /home/cgw/mambaforge/envs/mfix-git/share/mfix/src/model/calc_resid.f:800
800 IF (RESID_IJK( IJK ) > MAX_RESID) THEN
(gdb) print(ijk)
$1 = 3524
(gdb) print resid_ijk(ijk)
$2 = nan(0xfffffffffffff)
(gdb) print max_resid
$3 = 5.3124284493905395e-07
Note that resid(ijk) is NaN (Not a Number) and floating-point comparisons of NaN vs any other number will trigger an FPE. So we have to figure out where this NaN residual is coming from…
– Charles
First, there is some issue with the mesh. It is always a good idea to take a look at the mesh before running the simulation:
Here the mesh doesn’t deem to see the stl file near the bottom and top planes.
This comes from the way the wall region is defined. Instead of using a selection shape of “Box”, use “All”:
Then you will get a better mesh:
That didn’t solve the crash though. This seems to be related to the SMP solver. I runs fine with the serial solver.