max number of variables in scalar solver section

[ea_Marc]

Hi All,

I try to calculate the flux linkage in an electrical machine in every stator slot. That means, I need to integrate the magnetic vector potential in every slot.

My idea was, to use the scalar solver, which already writes out the torque etc. However, I realized, that in the moment, I have more than 20 variables in the scalar solver, I get an error, complaining about the 21th variable.

Here is an example of the scalar solver:

Code: Select all

! scalar_solver
Solver 6
  Equation = SaveScalars
  Procedure = "SaveData" "SaveScalars"
  Filename = scalars.dat
  Exec Solver = After Timestep
  Variable 1 = Time
  Variable 2 = Position
  Operator 3 = cpu time
  Variable 4 = Magnetic Vector Potential 3
  Mask Name 4 = phase_A+_14.0
  Operator 4 = Body Int
  Variable 5 = Magnetic Vector Potential 3
  Mask Name 5 = phase_B-_14.0
  Operator 5 = Body Int
  Variable 6 = Magnetic Vector Potential 3
  Mask Name 6 = phase_C-_14.0
  Operator 6 = Body Int
  Variable 7 = Magnetic Vector Potential 3
  Mask Name 7 = phase_B+_14.0
  Operator 7 = Body Int
  Variable 8 = Magnetic Vector Potential 3
  Mask Name 8 = phase_A-_14.0
  Operator 8 = Body Int
  Variable 9 = Magnetic Vector Potential 3
  Mask Name 9 = phase_C+_14.0
  Operator 9 = Body Int
End

This example works, because I have "only" 9 variables defined. But in the moment, I have a bigger machine with a lot of stator slots and over 20 variables, I get an error:

Code: Select all

ERROR:: LoadInputFile: 
ERROR:: LoadInputFile:  Unknown specifier: [magnetic vector potential 3]
ERROR:: LoadInputFile:  In section: [solver 6]
ERROR:: LoadInputFile:  For property name:[variable 21]
STOP 1

It complains about the 21st variable, but it does no matter, which variable is the 21st.

So, now I think about creating multiple scalar solver sections to not exceed 20 variables per section. However, perhaps I am just totally wrong on how I try to get my integration results. Does anyone have a better idea?
Thx!

[raback]

Hi

This seems to be a problem when reading in the keyword. Probably it helps if you say

Code: Select all

Variable 21 = String "..."

Now this also suggest that your code is old since currently the keyword checker has no upper bound.

-Peter

[ea_Marc]

Thx yor your answer.
Changing the code like this:

Code: Select all

Variable 18 = String "Magnetic Vector Potential 3"
  Mask Name 18 = s_slot_27_B+
  Operator 18 = Body Int
  Variable 19 = String "Magnetic Vector Potential 3"
  Mask Name 19 = s_slot_28_C-
  Operator 19 = Body Int
  Variable 20 = String "Magnetic Vector Potential 3"
  Mask Name 20 = s_slot_29_B+
  Operator 20 = Body Int
  Variable 21 = String "Magnetic Vector Potential 3"
  Mask Name 21 = s_slot_30_B+
  Operator 21 = Body Int
  Variable 22 = String "Magnetic Vector Potential 3"
  Mask Name 22 = s_slot_31_A-
  Operator 22 = Body Int

It starts complaining

Code: Select all

ERROR:: LoadInputFile: 
ERROR:: LoadInputFile:  Unknown specifier: [s_slot_30_b+]
ERROR:: LoadInputFile:  In section: [solver 6]
ERROR:: LoadInputFile:  For property name:[mask name 21]

If you say, that my code is old, that is probably correct, I am learning from "old" examples, what is the modern way to write out the results?

[Rich_B]

Hello,

I think the reference to 'old code', means that you might be running an older version of Elmer. What is the Compile date listed in the ElmerSolver output to console?

Posting a minimal working example may be the fastest way for you to get an answer.

Rich.

[ea_Marc]

It's version 9:

Code: Select all

ELMER SOLVER (v 9.0) STARTED AT: 2023/08/08 08:23:03
ParCommInit:  Initialize #PEs:            1
MAIN: 
MAIN: =============================================================
MAIN: ElmerSolver finite element software, Welcome!
MAIN: This program is free software licensed under (L)GPL
MAIN: Copyright 1st April 1995 - , CSC - IT Center for Science Ltd.
MAIN: Webpage http://www.csc.fi/elmer, Email elmeradm@csc.fi
MAIN: Version: 9.0 (Rev: b0f5c763e, Compiled: 2023-02-09)
MAIN:  Running one task without MPI parallelization.
MAIN:  Running with just one thread per task.
MAIN: =============================================================
MAIN: 
MAIN: 
MAIN: -------------------------------------
MAIN: Reading Model: mSpW10.6_10-4_FU1.sif
LoadInputFile: Scanning input file: mSpW10.6_10-4_FU1.sif
LoadInputFile: Scanning only size info
LoadInputFile: First time visiting
LoadInputFile: Reading base load of sif file
ERROR:: LoadInputFile: 
ERROR:: LoadInputFile:  Unknown specifier: [s_slot_30_b+]
ERROR:: LoadInputFile:  In section: [solver 6]
ERROR:: LoadInputFile:  For property name:[mask name 21]

I attached the example. The file ending is bz2, but it is an xz, which I was not allowed to upload. xz was necessary due to file size restriction. Thx for your support!

[raback]

Hi

As said, the keyword checker has been enhanced. If you have an old version you can manipulate the SOLVER.KEYWORDS file or just add types for all keywords with index >20.

Code: Select all

  Variable 21 = String "Magnetic Vector Potential 3"
  Mask Name 21 = String "s_slot_30_B+"
  Operator 21 = String "body int"

Generally it is not very fruitful to chase problems in old code that may have been addressed in the current version.

-Peter

[Rich_B]

Hello,

From your post, the Elmer Compiled date is 2023-02-09, so the version is about 6 months old.

Running your example using Elmer with Compiled date of 2023-08-03 ran to completion with the attached solver log.

Rich.

[ea_Marc]

Thanky you very much Rich. I compiled ELMER in release branch and ELMER release-9.0 tag. Version from release branch throw the same error than before, version from release-9 tag throws: "elements.def not found".

I can not compile the actual develop branch, due to:

Code: Select all

/home/scma/devel/elmer/elmerfem/fem/src/SParIterComm.F90:5019:51:

      CALL MPI_ALLREDUCE( ssum, tsum, 1, MPI_COMPLEX, &
                                                   1
Error: Symbol ‘mpi_complex’ at (1) has no IMPLICIT type

even with WITH_MPI=OFF

But I understand, that the error has something to do with the compiled version, because it runs for you. Let's see, what the future brings.

@Peter: Well I was really not aware, that a 6 month old version with the same version number is "old code".

[Rich_B]

Hello,

I ran the Windows installer, which is based on the latest devel branch. Elmer has been on Version 9 for several years. That is why looking at the Compiled Date is of interest.

Which OS are you running?

Rich.

[ea_Marc]

Ah ok, I was not aware that version 9 is a LTS version

I use Rocky Linux 8.

Today I tried to compile the devel branch with ubuntu 20.04, but this does not work as well ... There he complains about not finding the mpif.h.

Code: Select all

[ 52%] Building Fortran object fem/src/CMakeFiles/elmersolver.dir/GeneralUtils.F90.o
/home/scma/devel/elmer/elmerfem/fem/src/GeneralUtils.F90:1994: Error: Can't open included file 'mpif.h'
make[2]: *** [fem/src/CMakeFiles/elmersolver.dir/build.make:245: fem/src/CMakeFiles/elmersolver.dir/GeneralUtils.F90.o] Error 1
make[1]: *** [CMakeFiles/Makefile2:15397: fem/src/CMakeFiles/elmersolver.dir/all] Error 2
make: *** [Makefile:163: all] Error 2

This applies for devel and release branch. It's curious, because the file is available and cmake set the MPI_Fortran_INCLUDE_PATH correctly.
However, I do not want to bother you with compilation issues.

I tested the windows version as well, there it works .

I installed the ppa Version on ubuntu 20.04, there it works as well, thx!