[Nek5000-users] Crash when implementing additional scalars

[nek5000-users at lists.mcs.anl.gov]

Dear Neks,


I get the following crash when implementing 3 scalar fields (temperature + 2 additional scalar fields):


**************


Step      4, t= 4.0000000E-03, DT= 1.0000000E-03, C=  0.331 7.1728E+01 2.2114E+01
             Solving for heat
 Solving for fluid F T T
       4   2.51148E-02   6.07580E-04   4.13357E+01   1 TEMP   PROJECT
          4    Hmholtz TEMP:      4   3.0551E-08   8.0302E-04   1.0000E-07
       4   2.51148E-02   1.60663E-08   1.56319E+06   2 PS 1   PROJECT
          4    Hmholtz PS 1:      0   3.0509E-08   3.0509E-08   1.0000E-07
       4   2.51148E-02   3.38605E-02   7.41712E-01   3 PS 2   PROJECT
     4   100 **ERROR**: Failed in HMHOLTZ: PS 2   1.6090E+10   5.4264E-02   1.0000E-07
        4proj_ortho:    3   4 PS 2   Detect rank deficiency:         NaN  1.0000E+00
        4proj_ortho:    2   4 PS 2   Detect rank deficiency:         NaN  1.0000E+00
        4proj_ortho:    1   4 PS 2   Detect rank deficiency:         NaN  1.0000E+00
          4   4.0000E-03  1.2605E+01 Heat done
 Reorthogonalize Basis:
       4   2.21548E-02   2.27058E-02   9.75731E-01   1 velx   PROJECT
          4    Hmholtz velx:      5   7.8903E-08   4.8255E-02   1.0000E-07
 Reorthogonalize Basis:
       4   2.21455E-02   2.26942E-02   9.75823E-01   1 vely   PROJECT
          4    Hmholtz vely:      5   7.8987E-08   4.8248E-02   1.0000E-07
 Reorthogonalize Basis:
       4   2.54684E-01   2.57237E-01   9.90076E-01   1 velz   PROJECT
          4    Hmholtz velz:      6   1.7695E-08   5.4216E-01   1.0000E-07
         4 U-PRES gmres:     13  9.8513E-07  1.0000E-06  3.0328E-03  1.1984E+00  4.7670E+00
          4   4.0000E-03  8.4136E+00 Fluid done
          4  4 qfilt:  6.2753E-08  1.4966E-07  4.9326E-10
  5.7232E-14  5.7288E-14  2.1974E+09


*****************


The strange thing is that this does not happen when I run with only temperature.

I develop my simulation in small steps, but at the step when I also simulate the scalars (that are treated identically as the temperature), I get this crash.

I wonder if there is somthing that I missed when increasing the number of scalars?


Below I describe how I set up the simulation.


Best,


Johan


*****************************


*** Only temperature case ***


I generate a mesh from six .box files using genbox. The corresponding .rea files were modified slightly, merged and extended to 3D using n2to3.

The 3D .rea files were merged using nekmerge2.



If I start from .box files with only temperature I get a .rea file with p023 = 0 (NPSCAL).

I set the Peclet number to -233, and the field settings to

 T T F F F F F F F F F  IFNAV & IFADVC (convection in P.S. fields)
 F F T T T T T T T T T T  IFTMSH (IF mesh for this field is T mesh)


end everything works excellently.


*** 3 scalar case ***


If modify the previously mentioned .box files by adding 2 to the number of fields, and adding the same BC:s to them.

I get a .rea file with p023 = 2 (NPSCAL).


I set the Peclet number for the new scalars to -233, so that they are the same as for the temperature.

I change the field settings to:

 T T T T F F F F F F F  IFNAV & IFADVC (convection in P.S. fields)
 F F F F T T T T T T T T  IFTMSH (IF mesh for this field is T mesh)


Now I get the crash described above.


*****************************


*** The .usr file***


The temperature is read out in userf as:


      ffz = -t(ix,iy,iz,e,1)/1.32


All the scalars have the same boundary and initial conditions (they enter useric and userf only as "temp").


*** The SIZE file ***


I use the maximum number of passive scalars (ldimt) = 4 for both cases.



-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.mcs.anl.gov/pipermail/nek5000-users/attachments/20170514/6f71bda4/attachment.html>