Why does the solver crash?

[wangjinjing]

Hello，developers！
The solver crashes while running the study, does anyone know why?

[wangjinjing]

This is my bug report.Could you look at this for me, please?
combustioncase_2023-11-26T204747.588165.zip (4.5 MB)

[cgw]

I’ll take a look when I get a chance. But these kinds of “Access violations” usually occur when a particle has left the domain. This is typically due to either poor mesh quality, or walls/particles being too soft and passing through boundaries instead of bouncing off.

[cgw]

Another thing to try is to run in serial mode (no SMP) to see if this is an SMP issue

[cgw]

1. I note that you are running MFiX 22.3.1. When you get a chance, upgrade to the latest version (23.3.2)

2. While compiling your UDF code (usr_rates.f) I get the following warning:

Building Fortran object CMakeFiles/udfs.dir/tmp/combustioncase_2023-11-26T204747.588165/usr_rates.f.o
/tmp/combustioncase_2023-11-26T204747.588165/usr_rates.f:146:12:

  146 |       RATES(R3_Homogeneous) = c_CH4*c_H2O*KERT3
      |            1
Warning: Legacy Extension: REAL array index at (1)
/tmp/combustioncase_2023-11-26T204747.588165/usr_rates.f:147:12:

and similar warnings for lines 147 and 148

   107	      DOUBLE PRECISION ::  r3_homogeneous
   108	      DOUBLE PRECISION ::  r4_homogeneous
   109	      DOUBLE PRECISION ::  r5_homogeneous
   ...
   146	      RATES(R3_Homogeneous) = c_CH4*c_H2O*KERT3
   147	      RATES(R4_Homogeneous) = c_CO*c_H2O*KERT4
   148	      RATES(R5_Homogeneous) = c_CO2*c_H2*KERT5

You have declared r3_homogeneous as a double-precision value so it is incorrect to use it as an array index - the compiler is warning you about this, although allowing it. (It should probably be treated as a fatal error). Furthermore, this variable is uninitialized so you are essentially writing into a random location in the RATES array, triggering the access violation. (MFiX should show you more detail about where the error happened, perhaps if you upgrade to a more recent version you will get better error reports).

The array indices for RATES have to be integers corresponding to the reactions you have defined. These names are defined in a file species.inc which is created when you build the solver. Here are your reactions:

and here is the generated species.inc:

      INTEGER, PARAMETER :: CO = 1
      INTEGER, PARAMETER :: O2 = 2
      INTEGER, PARAMETER :: CO2 = 3
      INTEGER, PARAMETER :: H2 = 4
      INTEGER, PARAMETER :: H2O = 5
      INTEGER, PARAMETER :: CH4 = 6
      INTEGER, PARAMETER :: SiO2_2 = 1
      INTEGER, PARAMETER :: Moisture = 1
      INTEGER, PARAMETER :: Char = 2
      INTEGER, PARAMETER :: Ash = 3
      INTEGER, PARAMETER :: Volatile = 4
      INTEGER, PARAMETER :: R3 = 1
      INTEGER, PARAMETER :: R4 = 2
      INTEGER, PARAMETER :: R5 = 3
      INTEGER, PARAMETER :: Pyrolysis = 1
      INTEGER, PARAMETER :: Dring = 2
      INTEGER, PARAMETER :: R1 = 3
      INTEGER, PARAMETER :: R2 = 4

Note that there is no reaction called R3_Homogeneous. Take out the definitions of r[345]_homogeneous at lines 107-109 and use the correct indices at 146-148.

[wangjinjing]

Thanks！I try to take out the definitions of r[345]_homogeneous at lines 107-109 and use the correct indices at 146-148，but the problem remains unresolved.Is there anything else I can try? How can grid quality be improved?

[cgw]

Please upload your modified files

[wangjinjing]

Here’s my updated bug report, thanks!
combustioncase_2023-11-29T213105.947494.zip (4.5 MB)

[cgw]

Error message (are you seeing this?)

#0 c3m_usr_rates1
        at combustioncase_2023-11-29T213105.947494/usr_rates.f:148
#1 stiffchemclass_MOD_tsolverconstructor
        at chem/stiff_chem.f90:264
#2 stiff_chem_MOD_stiff_chem_solver
        at chem/stiff_chem_mod.f:88
#3 step_MOD_chem_mass
        at time_step.f:214
#4 run_fluid
        at mfix.f:205
#5 run_mfix
        at mfix.f:145
#6 main_MOD_run_mfix0
        at main.f:85

Line 148 of usr_rates.f

148	         ReactionRates(IJK,l) = RATES(L)         
[Current thread is 1 (Thread 0x7fdd41b616c0 (LWP 24845))]
(gdb) print ReactionRates
$1 = <not allocated>

You need to set ReactionRates array size if you are going to use that array

[wangjinjing]

Thank you! The previous problem has been solved, but I have a new problem: access conflicts cause the solver to crash. This is my error report, please help me to see how to solve it.

combustioncase_2023-12-03T163151.356831.zip (6.3 MB)

[cgw]

Some kind of memory corruption is happening here
The debugger tells me

63	         DES_RATES(Dring) = 5.13d6*exp(-8.79d7/(8314.d0*Tp))*MW_s(1,MOISTURE)
(gdb) p des_rates
value requires 944905285042854617 bytes, which is more than max-value-size

That number is absurdly large. I’m not sure what’s causing this, it may take a little while to debug.

– Charles

[cgw]

The problem is at line 32 of your usr_rates_des.f

Change

      INTEGER, INTENT(OUT) :: DES_RATES(:)

to

      INTEGER, INTENT(OUT) :: DES_RATES(NO_OF_DES_RXNS)

and the case runs.

[wangjinjing]

Thanks！ Problem solved！