Issue with temperature dependent emissivity

[arved]

Hi,

I have a 2D axisymmetric model of a Silicon growth furnace with temperature-dependent emissivity (using the good old Gebhart factors model). It used to work with https://github.com/ElmerCSC/elmerfem/co ... d69bec1ab4 (Oct 7, 2022), now I updated the solver to https://github.com/ElmerCSC/elmerfem/co ... b6b7751b5e (last week) and get the following error message:

Code: Select all

(...)
HeatSolve: -------------------------------------
HeatSolve:  TEMPERATURE ITERATION           1
HeatSolve: -------------------------------------
HeatSolve: 
HeatSolve: Starting Assembly...

Program received signal SIGSEGV: Segmentation fault - invalid memory reference.

Backtrace for this error:
#0  0x149d90970d21 in ???
#1  0x149d9096fef5 in ???
#2  0x149d907a108f in ???
#3  0x149d90dcf918 in __bandmatrix_MOD_band_addtomatrixelement
	at /home/elmer/elmerfem/fem/src/BandMatrix.F90:115
#4  0x149d87dad229 in diffusegrayradiation
	at /home/elmer/elmerfem/fem/src/modules/HeatSolve.F90:1698
#5  0x149d87dad229 in addheatfluxbc
	at /home/elmer/elmerfem/fem/src/modules/HeatSolve.F90:1447
#6  0x149d87db0c70 in heatsolver_
	at /home/elmer/elmerfem/fem/src/modules/HeatSolve.F90:1096
#7  0x149d90eefaf9 in __mainutils_MOD_singlesolver
	at /home/elmer/elmerfem/fem/src/MainUtils.F90:5312
#8  0x149d90f05668 in __mainutils_MOD_solveractivate
	at /home/elmer/elmerfem/fem/src/MainUtils.F90:5580
#9  0x149d90f070b0 in solvecoupled
	at /home/elmer/elmerfem/fem/src/MainUtils.F90:3203
#10  0x149d90f084fe in __mainutils_MOD_solveequations
	at /home/elmer/elmerfem/fem/src/MainUtils.F90:2903
#11  0x149d911949c3 in execsimulation
	at /home/elmer/elmerfem/fem/src/ElmerSolver.F90:3186
#12  0x149d9119be83 in elmersolver_
	at /home/elmer/elmerfem/fem/src/ElmerSolver.F90:701
#13  0x563434ed53f5 in solver
	at /home/elmer/elmerfem/fem/src/Solver.F90:57
#14  0x563434ed515e in main
	at /home/elmer/elmerfem/fem/src/Solver.F90:34

If required I can setup a minimal test case.

Best regards
Arved

[raback]

Hi Arved,

For some reason the routine calls for a banded matrix. This is generally a bad idea. Probably it then exhausts the memory. I don't know why it would have started to use banded matrix. I could imagine using direct solver "banded" in case no other direct solver is available but I wonder how it could happen in this stage.

A test case would probably be good!

-Peter

[arved]

Hi Peter,

thanks for this information! I am trying to setup a simplified test case that reproduces this error, however, I haven't been successful yet. The problem seems to disappear once I removed 2 or 3 bodies from the setup. I'll let you know once I've got something.

This is the solver settings I used (also tried with BiCGStabl):

Code: Select all

Solver 1
  Equation = Heat Equation
  Procedure = "HeatSolve" "HeatSolver"
  Variable = "Temperature"
  Variable Dofs = 1
  Calculate Loads = True
  Exec Solver = Always
  Nonlinear System Convergence Tolerance = 1e-05
  Nonlinear System Max Iterations = 1000
  Nonlinear System Relaxation Factor = 0.7
  Steady State Convergence Tolerance = 1e-06
  Stabilize = True
  Optimize Bandwidth = True
  Linear System Solver = Iterative
  Linear System Iterative Method = Idrs
  Linear System Max Iterations = 100000
  Linear System Preconditioning = ILU
  Linear System Precondition Recompute = 1
  Linear System Convergence Tolerance = 1e-09
  Linear System Abort Not Converged = True
  Linear System Residual Output = 10
  Smart Heater Control After Tolerance = 0.01
  Update Gebhardt Factors = True
End

Best regards
Arved

[annier]

Dear Arved,
If you can upload a complete sif file with a simplified geometry/mesh file, then we can test it in our computers and provide the feedback.

Best Regards,
Anil