223031012
/
DAMASK_EICMD
mirror of https://github.com/achalhp/DAMASK_EICMD.git
1
0
Fork 0
Code Issues Projects Releases Wiki Activity
DAMASK_EICMD/src/DAMASK_marc.f90

419 lines
26 KiB
Fortran
Raw Normal View History

reorganized calculation modes for CPFEM, now having better readable and cleaner structure
2013-03-01 17:18:29 +05:30
!--------------------------------------------------------------------------------------------------
!> @author Philip Eisenlohr, Max-Planck-Institut für Eisenforschung GmbH
!> @author Franz Roters, Max-Planck-Institut für Eisenforschung GmbH
!> @author Luc Hantcherli, Max-Planck-Institut für Eisenforschung GmbH
!> @author W.A. Counts
!> @author Denny Tjahjanto, Max-Planck-Institut für Eisenforschung GmbH
!> @author Christoph Kords, Max-Planck-Institut für Eisenforschung GmbH
dont' store the working directory but set it centrally makes life easier, writing and reading always to CWD unless absolute path is given Spectral: using --wd argrument and C code MSC.Marc: using directory of input deck and Intel extension Abaqus: using function and Intel extension
2018-07-10 11:54:45 +05:30
!> @author Martin Diehl, Max-Planck-Institut für Eisenforschung GmbH
!> @brief Interfaces DAMASK with MSC.Marc
simplified interface of CPFEM_general (removed P and F, made cs and dcs/dE optional) commented and cleaned up the marc interface. !!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!! For marc simulations, run ./code/setup/setup_code.sh !!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!
2013-04-09 15:38:00 +05:30
!> @details Usage:
!> @details - choose material as hypela2
!> @details - set statevariable 2 to index of homogenization
!> @details - set statevariable 3 to index of microstructure
corrected linebreak in comment
2013-10-17 18:22:46 +05:30
!> @details - use nonsymmetric option for solver (e.g. direct profile or multifrontal sparse, the latter seems to be faster!)
!> @details - in case of ddm (domain decomposition) a SYMMETRIC solver has to be used, i.e uncheck "non-symmetric"
simplified interface of CPFEM_general (removed P and F, made cs and dcs/dE optional) commented and cleaned up the marc interface. !!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!! For marc simulations, run ./code/setup/setup_code.sh !!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!
2013-04-09 15:38:00 +05:30
!> @details Marc subroutines used:
!> @details - hypela2
commercial FEM solvers write results to HDF5
2019-05-05 15:36:55 +05:30
!> @details - uedinc
flux subroutine for Abaqus related comments
2019-02-27 13:53:05 +05:30
!> @details - flux
simplified interface of CPFEM_general (removed P and F, made cs and dcs/dE optional) commented and cleaned up the marc interface. !!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!! For marc simulations, run ./code/setup/setup_code.sh !!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!
2013-04-09 15:38:00 +05:30
!> @details - quit
!> @details Marc common blocks included:
!> @details - concom: lovl, inc
!> @details - creeps: timinc
reorganized calculation modes for CPFEM, now having better readable and cleaner structure
2013-03-01 17:18:29 +05:30
!--------------------------------------------------------------------------------------------------
dont' store the working directory but set it centrally makes life easier, writing and reading always to CWD unless absolute path is given Spectral: using --wd argrument and C code MSC.Marc: using directory of input deck and Intel extension Abaqus: using function and Intel extension
2018-07-10 11:54:45 +05:30
#define QUOTE(x) #x
#define PASTE(x,y) x ## y
#include "prec.f90"
added new, flexible debugging scheme. now all modules have their own debug specification. compiles and runs, I hope nothing is broken did a lot of polishing
2012-03-09 01:55:28 +05:30
module DAMASK_interface
clear separation between solver and DAMASK
2020-01-25 13:42:09 +05:30
use prec
preprocessor statements need to start at column 0
2019-06-07 11:15:54 +05:30
#if __INTEL_COMPILER >= 1800
clear separation between solver and DAMASK
2020-01-25 13:42:09 +05:30
use, intrinsic :: iso_fortran_env, only: &
compiler_version, &
compiler_options
phase out mesh_elem and theMesh
2019-06-07 09:48:42 +05:30
#endif
clear separation between solver and DAMASK
2020-01-25 13:42:09 +05:30
use ifport, only: &
CHDIR
adapted flux routine to current state of source_thermal_dissipation.f90
2016-07-20 12:14:12 +05:30
clear separation between solver and DAMASK
2020-01-25 13:42:09 +05:30
implicit none
private
avoid duplicated public/private statements variables should be private
2020-02-29 19:30:47 +05:30
logical, protected, public :: symmetricSolver
clear separation between solver and DAMASK
2020-01-25 13:42:09 +05:30
character(len=*), parameter, public :: INPUTFILEEXTENSION = '.dat'
public :: &
DAMASK_interface_init, &
getSolverJobName
# new interface for Abaqus # IO has some additional functionality for Abaqus parsing # ping pong scheme in FE interface now similar (and more human understandable) in both versions # mesh has better splitting of different tasks, plus operation on database whenever possible # FEsolver as new global var to indicate FEM solver type # computation mode reshuffling: 6 is now Marc special case of recycling...
2009-10-12 21:31:49 +05:30
added new, flexible debugging scheme. now all modules have their own debug specification. compiles and runs, I hope nothing is broken did a lot of polishing
2012-03-09 01:55:28 +05:30
contains
forgot to "use" the new init-routines
2009-08-31 20:51:15 +05:30
reorganized calculation modes for CPFEM, now having better readable and cleaner structure
2013-03-01 17:18:29 +05:30
!--------------------------------------------------------------------------------------------------
dont' store the working directory but set it centrally makes life easier, writing and reading always to CWD unless absolute path is given Spectral: using --wd argrument and C code MSC.Marc: using directory of input deck and Intel extension Abaqus: using function and Intel extension
2018-07-10 11:54:45 +05:30
!> @brief reports and sets working directory
reorganized calculation modes for CPFEM, now having better readable and cleaner structure
2013-03-01 17:18:29 +05:30
!--------------------------------------------------------------------------------------------------
added new, flexible debugging scheme. now all modules have their own debug specification. compiles and runs, I hope nothing is broken did a lot of polishing
2012-03-09 01:55:28 +05:30
subroutine DAMASK_interface_init
Introducing the capability to restart jobs that crashed in the middle of sims. At the moment, this feature is exclusive for Marc. Major changes: CPFEM.f90 => 1. Moving the initialization out of CPFEM_general into a separate subroutine, which is directly called by the hypela2 (Beware, the Abaqus version must also be modified in order to adapt with this change). 2. Restore primary state variables in CPFEM_init from binary files when requested (Marc flag: restart read). 3. Writing primary state variables into binary files (Marc flag: restart write). FEsolving.f90 => 1. Adding functions to recognize Marc restart flags: read and write and the corresponding restart file (parent job). 2. Change the initial value of cycleCounter = -1 in conjuction with the change made the ping-pong scheme homogenization_RGC.f90 => 1. Just syntax polishing. IO.f90 => 1. Adding functions/subroutines to open binary files for writing the primary state variables for restart purpose. mpie_cpfem_marc.f90 1. Modification of the general scheme for collection and calculation in order to accommodate the newly added restart feature.
2010-11-03 20:09:18 +05:30
avoiding hard-coded constants
2020-01-26 16:54:35 +05:30
integer, dimension(8) :: dateAndTime
integer :: ierr
character(len=pPathLen) :: wd
clear separation between solver and DAMASK
2020-01-25 13:42:09 +05:30
write(6,'(/,a)') ' <<<+- DAMASK_marc init -+>>>'
write(6,'(/,a)') ' Roters et al., Computational Materials Science 158:420–478, 2019'
write(6,'(a)') ' https://doi.org/10.1016/j.commatsci.2018.04.030'
write(6,'(/,a)') ' Version: '//DAMASKVERSION
! https://github.com/jeffhammond/HPCInfo/blob/master/docs/Preprocessor-Macros.md
avoid disturbing reporting
2019-02-16 03:24:38 +05:30
#if __INTEL_COMPILER >= 1800
clear separation between solver and DAMASK
2020-01-25 13:42:09 +05:30
write(6,'(/,a)') ' Compiled with: '//compiler_version()
write(6,'(a)') ' Compiler options: '//compiler_options()
avoid disturbing reporting
2019-02-16 03:24:38 +05:30
#else
clear separation between solver and DAMASK
2020-01-25 13:42:09 +05:30
write(6,'(/,a,i4.4,a,i8.8)') ' Compiled with Intel fortran version :', __INTEL_COMPILER,&
', build date :', __INTEL_COMPILER_BUILD_DATE
avoid disturbing reporting
2019-02-16 03:24:38 +05:30
#endif
clear separation between solver and DAMASK
2020-01-25 13:42:09 +05:30
write(6,'(/,a)') ' Compiled on: '//__DATE__//' at '//__TIME__
call date_and_time(values = dateAndTime)
write(6,'(/,a,2(i2.2,a),i4.4)') ' Date: ',dateAndTime(3),'/',dateAndTime(2),'/', dateAndTime(1)
write(6,'(a,2(i2.2,a),i2.2)') ' Time: ',dateAndTime(5),':', dateAndTime(6),':', dateAndTime(7)
inquire(5, name=wd)
wd = wd(1:scan(wd,'/',back=.true.))
ierr = CHDIR(wd)
if (ierr /= 0) then
write(6,'(a20,a,a16)') ' working directory "',trim(wd),'" does not exist'
call quit(1)
endif
symmetricSolver = solverIsSymmetric()
added new, flexible debugging scheme. now all modules have their own debug specification. compiles and runs, I hope nothing is broken did a lot of polishing
2012-03-09 01:55:28 +05:30
end subroutine DAMASK_interface_init
added version information to all files do NOT edit text like this: $Id: constitutive_phenopowerlaw.f90 406 2009-08-31 14:13:10Z MPIE\f.roters $
2009-08-31 20:39:15 +05:30
reorganized calculation modes for CPFEM, now having better readable and cleaner structure
2013-03-01 17:18:29 +05:30
!--------------------------------------------------------------------------------------------------
!> @brief solver job name (no extension) as combination of geometry and load case name
!--------------------------------------------------------------------------------------------------
independent of IO
2020-01-25 13:26:39 +05:30
function getSolverJobName()
changed fileopen routines in IO.f90 to use solver dependant functions defined in the interface routines DID NOT test ABAQUS versions yet, not shure whether (V)GETOUTDIR returns path with or without terminating slash
2010-05-10 20:32:59 +05:30
clear separation between solver and DAMASK
2020-01-25 13:42:09 +05:30
character(len=:), allocatable :: getSolverJobName
character(1024) :: inputName
character(len=*), parameter :: pathSep = achar(47)//achar(92) ! forward and backward slash
integer :: extPos
changed fileopen routines in IO.f90 to use solver dependant functions defined in the interface routines DID NOT test ABAQUS versions yet, not shure whether (V)GETOUTDIR returns path with or without terminating slash
2010-05-10 20:32:59 +05:30
clear separation between solver and DAMASK
2020-01-25 13:42:09 +05:30
inputName=''
inquire(5, name=inputName) ! determine inputfile
extPos = len_trim(inputName)-4
getSolverJobName=inputName(scan(inputName,pathSep,back=.true.)+1:extPos)
changed fileopen routines in IO.f90 to use solver dependant functions defined in the interface routines DID NOT test ABAQUS versions yet, not shure whether (V)GETOUTDIR returns path with or without terminating slash
2010-05-10 20:32:59 +05:30
added new, flexible debugging scheme. now all modules have their own debug specification. compiles and runs, I hope nothing is broken did a lot of polishing
2012-03-09 01:55:28 +05:30
end function getSolverJobName
reorganized calculation modes for CPFEM, now having better readable and cleaner structure
2013-03-01 17:18:29 +05:30
clear separation between solver and DAMASK
2020-01-25 13:42:09 +05:30
!--------------------------------------------------------------------------------------------------
!> @brief determines whether a symmetric solver is used
!--------------------------------------------------------------------------------------------------
logical function solverIsSymmetric()
character(len=pStringLen) :: line
integer :: myStat,fileUnit,s,e
open(newunit=fileUnit, file=getSolverJobName()//INPUTFILEEXTENSION, &
status='old', position='rewind', action='read',iostat=myStat)
do
read (fileUnit,'(A)',END=100) line
if(index(trim(lc(line)),'solver') == 1) then
read (fileUnit,'(A)',END=100) line ! next line
s = verify(line, ' ') ! start of first chunk
s = s + verify(line(s+1:),' ') ! start of second chunk
e = s + scan (line(s+1:),' ') ! end of second chunk
solverIsSymmetric = line(s:e) /= '1'
endif
enddo
100 close(fileUnit)
contains
!--------------------------------------------------------------------------------------------------
!> @brief changes characters in string to lower case
!> @details copied from IO_lc
!--------------------------------------------------------------------------------------------------
function lc(string)
character(len=*), intent(in) :: string !< string to convert
character(len=len(string)) :: lc
character(26), parameter :: LOWER = 'abcdefghijklmnopqrstuvwxyz'
character(26), parameter :: UPPER = 'ABCDEFGHIJKLMNOPQRSTUVWXYZ'
integer :: i,n
do i=1,len(string)
lc(i:i) = string(i:i)
n = index(UPPER,lc(i:i))
if (n/=0) lc(i:i) = LOWER(n:n)
enddo
end function lc
end function solverIsSymmetric
added new, flexible debugging scheme. now all modules have their own debug specification. compiles and runs, I hope nothing is broken did a lot of polishing
2012-03-09 01:55:28 +05:30
end module DAMASK_interface
forgot to "use" the new init-routines
2009-08-31 20:51:15 +05:30
dont' store the working directory but set it centrally makes life easier, writing and reading always to CWD unless absolute path is given Spectral: using --wd argrument and C code MSC.Marc: using directory of input deck and Intel extension Abaqus: using function and Intel extension
2018-07-10 11:54:45 +05:30
corrected bug: renamed file list for commercial FEM solvers to start with small letter
2014-12-08 22:06:22 +05:30
#include "commercialFEM_fileList.f90"
overhaul to allow for multiphase/homogenization setup.
2009-03-04 17:18:54 +05:30
reorganized calculation modes for CPFEM, now having better readable and cleaner structure
2013-03-01 17:18:29 +05:30
!--------------------------------------------------------------------------------------------------
!> @brief This is the MSC.Marc user subroutine for defining material behavior
!> @details (1) F,R,U are only available for continuum and membrane elements (not for
!> @details shells and beams).
!> @details
simplified interface of CPFEM_general (removed P and F, made cs and dcs/dE optional) commented and cleaned up the marc interface. !!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!! For marc simulations, run ./code/setup/setup_code.sh !!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!
2013-04-09 15:38:00 +05:30
!> @details (2) Use the -> 'Plasticity,3' card(=update+finite+large disp+constant d)
!> @details in the parameter section of input deck (updated Lagrangian formulation).
reorganized calculation modes for CPFEM, now having better readable and cleaner structure
2013-03-01 17:18:29 +05:30
!--------------------------------------------------------------------------------------------------
simplified interface of CPFEM_general (removed P and F, made cs and dcs/dE optional) commented and cleaned up the marc interface. !!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!! For marc simulations, run ./code/setup/setup_code.sh !!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!
2013-04-09 15:38:00 +05:30
subroutine hypela2(d,g,e,de,s,t,dt,ngens,m,nn,kcus,matus,ndi,nshear,disp, &
dispt,coord,ffn,frotn,strechn,eigvn,ffn1,frotn1, &
strechn1,eigvn1,ncrd,itel,ndeg,ndm,nnode, &
jtype,lclass,ifr,ifu)
pInt not needed anymore
2019-05-15 02:14:38 +05:30
use prec
clear separation between solver and DAMASK
2020-01-25 13:42:09 +05:30
use DAMASK_interface
phase out mesh_elem and theMesh
2019-06-07 09:48:42 +05:30
use numerics
use FEsolving
use debug
use mesh
use CPFEM
major restructuring of code. homogenization as well as constitutive are now free to choose. the runtime got somewhat longer (25% on simple tests) compared to a hardcoded isostrain homogenization. this might be a point of further optimization at a later stage... please use homogenization_isostrain.f90 as starting point / example for future developments of homog-schemes. the homogenization scheme now can additionally output certain results. hence, the userdata structure at each integration point now looks like this: - sizeHomogPostResults - block of that size containing homogPostResults then for each grain: - sizeGrainPostResults - block of that size containing crystallitePostResults, which consist of: + phaseID + volFrac + Eulers (3) + any constitutive post results requested
2009-05-07 21:57:36 +05:30
simplified interface of CPFEM_general (removed P and F, made cs and dcs/dE optional) commented and cleaned up the marc interface. !!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!! For marc simulations, run ./code/setup/setup_code.sh !!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!
2013-04-09 15:38:00 +05:30
implicit none
!$ include "omp_lib.h" ! the openMP function library
pInt not needed anymore
2019-05-15 02:14:38 +05:30
integer, intent(in) :: & ! according to MSC.Marc 2012 Manual D
simplified interface of CPFEM_general (removed P and F, made cs and dcs/dE optional) commented and cleaned up the marc interface. !!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!! For marc simulations, run ./code/setup/setup_code.sh !!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!
2013-04-09 15:38:00 +05:30
ngens, & !< size of stress-strain law
adapted flux routine to current state of source_thermal_dissipation.f90
2016-07-20 12:14:12 +05:30
nn, & !< integration point number
simplified interface of CPFEM_general (removed P and F, made cs and dcs/dE optional) commented and cleaned up the marc interface. !!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!! For marc simulations, run ./code/setup/setup_code.sh !!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!
2013-04-09 15:38:00 +05:30
ndi, & !< number of direct components
nshear, & !< number of shear components
ncrd, & !< number of coordinates
itel, & !< dimension of F and R, either 2 or 3
ndeg, & !< number of degrees of freedom
ndm, & !< not specified in MSC.Marc 2012 Manual D
nnode, & !< number of nodes per element
jtype, & !< element type
ifr, & !< set to 1 if R has been calculated
ifu !< set to 1 if stretch has been calculated
pInt not needed anymore
2019-05-15 02:14:38 +05:30
integer, dimension(2), intent(in) :: & ! according to MSC.Marc 2012 Manual D
simplified interface of CPFEM_general (removed P and F, made cs and dcs/dE optional) commented and cleaned up the marc interface. !!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!! For marc simulations, run ./code/setup/setup_code.sh !!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!
2013-04-09 15:38:00 +05:30
m, & !< (1) user element number, (2) internal element number
matus, & !< (1) user material identification number, (2) internal material identification number
adapted flux routine to current state of source_thermal_dissipation.f90
2016-07-20 12:14:12 +05:30
kcus, & !< (1) layer number, (2) internal layer number
simplified interface of CPFEM_general (removed P and F, made cs and dcs/dE optional) commented and cleaned up the marc interface. !!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!! For marc simulations, run ./code/setup/setup_code.sh !!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!
2013-04-09 15:38:00 +05:30
lclass !< (1) element class, (2) 0: displacement, 1: low order Herrmann, 2: high order Herrmann
real(pReal), dimension(*), intent(in) :: & ! has dimension(1) according to MSC.Marc 2012 Manual D, but according to example hypela2.f dimension(*)
e, & !< total elastic strain
de, & !< increment of strain
dt !< increment of state variables
real(pReal), dimension(itel), intent(in) :: & ! according to MSC.Marc 2012 Manual D
strechn, & !< square of principal stretch ratios, lambda(i) at t=n
strechn1 !< square of principal stretch ratios, lambda(i) at t=n+1
real(pReal), dimension(3,3), intent(in) :: & ! has dimension(itel,*) according to MSC.Marc 2012 Manual D, but we alway assume dimension(3,3)
ffn, & !< deformation gradient at t=n
ffn1 !< deformation gradient at t=n+1
real(pReal), dimension(itel,*), intent(in) :: & ! according to MSC.Marc 2012 Manual D
frotn, & !< rotation tensor at t=n
eigvn, & !< i principal direction components for j eigenvalues at t=n
frotn1, & !< rotation tensor at t=n+1
eigvn1 !< i principal direction components for j eigenvalues at t=n+1
real(pReal), dimension(ndeg,*), intent(in) :: & ! according to MSC.Marc 2012 Manual D
disp, & !< incremental displacements
dispt !< displacements at t=n (at assembly, lovl=4) and displacements at t=n+1 (at stress recovery, lovl=6)
real(pReal), dimension(ncrd,*), intent(in) :: & ! according to MSC.Marc 2012 Manual D
coord !< coordinates
real(pReal), dimension(*), intent(inout) :: & ! according to MSC.Marc 2012 Manual D
t !< state variables (comes in at t=n, must be updated to have state variables at t=n+1)
real(pReal), dimension(ndi+nshear), intent(out) :: & ! has dimension(*) according to MSC.Marc 2012 Manual D, but we need to loop over it
s, & !< stress - should be updated by user
g !< change in stress due to temperature effects
real(pReal), dimension(ngens,ngens), intent(out) :: & ! according to MSC.Marc 2012 Manual D, but according to example hypela2.f dimension(ngens,*)
d !< stress-strain law to be formed
CPFEM_general now independed of the wrapper code (UMAT, hypela2, etc.) renamed CPFEM_stressIP to CPFEM_MaterialPoint renamed CPFEM_stressCrystallite to CPFEM_Crystallite introduced new global variables to keep track of FE state within module FEsolving
2008-03-15 03:02:57 +05:30
simplified interface of CPFEM_general (removed P and F, made cs and dcs/dE optional) commented and cleaned up the marc interface. !!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!! For marc simulations, run ./code/setup/setup_code.sh !!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!
2013-04-09 15:38:00 +05:30
!--------------------------------------------------------------------------------------------------
overhaul to allow for multiphase/homogenization setup.
2009-03-04 17:18:54 +05:30
! Marc common blocks are in fixed format so they have to be reformated to free format (f90)
! Beware of changes in newer Marc versions
marc doesn't need error prone setup_code any more, moved Marc includes from code/include to lib/includeMarc DAMASK_marcXXXX.f90 now is not a copy any more but contains one definition and an include statement setup_code of spectral solver is now replaced by the standard 1) configure 2) make 3) make install
2013-04-22 20:11:33 +05:30
include files are not libraries
2018-07-10 21:55:42 +05:30
#include QUOTE(PASTE(./MarcInclude/concom,Marc4DAMASK)) ! concom is needed for inc, lovl
#include QUOTE(PASTE(./MarcInclude/creeps,Marc4DAMASK)) ! creeps is needed for timinc (time increment)
overhaul to allow for multiphase/homogenization setup.
2009-03-04 17:18:54 +05:30
simplified interface of CPFEM_general (removed P and F, made cs and dcs/dE optional) commented and cleaned up the marc interface. !!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!! For marc simulations, run ./code/setup/setup_code.sh !!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!
2013-04-09 15:38:00 +05:30
logical :: cutBack
fixed some dangerous constructs in computation mode assignments introduction of separate stress and stiffness variables to pass on to CPFEM_general (of dimension 6)
2010-02-18 15:53:02 +05:30
real(pReal), dimension(6) :: stress
real(pReal), dimension(6,6) :: ddsdde
pInt not needed anymore
2019-05-15 02:14:38 +05:30
integer :: computationMode, i, cp_en, node, CPnodeID
set OMP related variables to 4 byte integer as expected when compiling marc
2016-12-23 18:50:29 +05:30
!$ integer(4) :: defaultNumThreadsInt !< default value set by Marc
simplified interface of CPFEM_general (removed P and F, made cs and dcs/dE optional) commented and cleaned up the marc interface. !!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!! For marc simulations, run ./code/setup/setup_code.sh !!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!
2013-04-09 15:38:00 +05:30
pInt not needed anymore
2019-05-15 02:14:38 +05:30
if (iand(debug_level(debug_MARC),debug_LEVELBASIC) /= 0) then
simplified interface of CPFEM_general (removed P and F, made cs and dcs/dE optional) commented and cleaned up the marc interface. !!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!! For marc simulations, run ./code/setup/setup_code.sh !!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!
2013-04-09 15:38:00 +05:30
write(6,'(a,/,i8,i8,i2)') ' MSC.MARC information on shape of element(2), IP:', m, nn
Merge branch 'MarcThermoMech' into development
2017-02-08 15:16:39 +05:30
write(6,'(a,2(i1))') ' Jacobian: ', ngens,ngens
write(6,'(a,i1)') ' Direct stress: ', ndi
write(6,'(a,i1)') ' Shear stress: ', nshear
added support for Marc2016
2016-10-28 12:46:20 +05:30
write(6,'(a,i2)') ' DoF: ', ndeg
write(6,'(a,i2)') ' Coordinates: ', ncrd
Merge branch 'MarcThermoMech' into development
2017-02-08 15:16:39 +05:30
write(6,'(a,i12)') ' Nodes: ', nnode
write(6,'(a,i1)') ' Deformation gradient: ', itel
simplified interface of CPFEM_general (removed P and F, made cs and dcs/dE optional) commented and cleaned up the marc interface. !!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!! For marc simulations, run ./code/setup/setup_code.sh !!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!
2013-04-09 15:38:00 +05:30
write(6,'(/,a,/,3(3(f12.7,1x)/))',advance='no') ' Deformation gradient at t=n:', &
avoid converstions and use of global variables
2019-01-14 17:15:07 +05:30
transpose(ffn)
simplified interface of CPFEM_general (removed P and F, made cs and dcs/dE optional) commented and cleaned up the marc interface. !!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!! For marc simulations, run ./code/setup/setup_code.sh !!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!
2013-04-09 15:38:00 +05:30
write(6,'(/,a,/,3(3(f12.7,1x)/))',advance='no') ' Deformation gradient at t=n+1:', &
avoid converstions and use of global variables
2019-01-14 17:15:07 +05:30
transpose(ffn1)
simplified interface of CPFEM_general (removed P and F, made cs and dcs/dE optional) commented and cleaned up the marc interface. !!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!! For marc simulations, run ./code/setup/setup_code.sh !!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!
2013-04-09 15:38:00 +05:30
endif
fixed bug: initialize computation mode to zero, used to be unset for lovl=4 and not(timinc < theDelta .and. theInc == inc), which unpredictably caused trouble with marc
2013-04-23 20:12:22 +05:30
!$ defaultNumThreadsInt = omp_get_num_threads() ! remember number of threads set by Marc
remove deprecated spectral load case definition of temperature
2015-07-24 20:27:29 +05:30
if (.not. CPFEM_init_done) call CPFEM_initAll(m(1),nn)
# new interface for Abaqus # IO has some additional functionality for Abaqus parsing # ping pong scheme in FE interface now similar (and more human understandable) in both versions # mesh has better splitting of different tasks, plus operation on database whenever possible # FEsolver as new global var to indicate FEM solver type # computation mode reshuffling: 6 is now Marc special case of recycling...
2009-10-12 21:31:49 +05:30
simplified interface of CPFEM_general (removed P and F, made cs and dcs/dE optional) commented and cleaned up the marc interface. !!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!! For marc simulations, run ./code/setup/setup_code.sh !!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!
2013-04-09 15:38:00 +05:30
!$ call omp_set_num_threads(DAMASK_NumThreadsInt) ! set number of threads for parallel execution set by DAMASK_NUM_THREADS
some polishing concerning openMP initialization number of threads is now written to *.out file
2010-12-02 16:34:29 +05:30
pInt not needed anymore
2019-05-15 02:14:38 +05:30
computationMode = 0 ! save initialization value, since it does not result in any calculation
parallelExecution flag now passed as input variable by CPFEM_general; flag is set on the solver level (DAMASK_abaqus_exp.f, DAMASK_marc.f90, etc.)
2013-08-02 16:50:11 +05:30
if (lovl == 4 ) then ! jacobian requested by marc
plain mode (no ping pong) now seems to work for marc, further testing needed regular sequence of computation modes is inc 0: 1,1,1,… inc 1: 0,0,0,… 7,1,1,… 0,0,0,… 1,1,1,… and so on after a cutback the computation modes follow as inc 5: 8,0,0,… 1,1,1,… 0,0,0,… 1,1,1,…
2013-08-02 21:49:45 +05:30
if (timinc < theDelta .and. theInc == inc .and. lastLovl /= lovl) & ! first after cutback
computationMode = CPFEM_RESTOREJACOBIAN
removed variable lastMode from FE_solving, used to be used for detection of first call after ping pong; now this is done by checking for a change in the lovl (macro) or a change in the calculation step (abaqus)
2013-08-02 18:58:50 +05:30
elseif (lovl == 6) then ! stress requested by marc
store mapping MARC/FEM2DAMASK mapping do not calculate the mapping for elements and nodes per call on the fly, rather store it. Not memory efficient in the case that numbers are not consequtive (order does not matter, but missing nodes/elements would waste some 2 integers per missing number). However, this seem to cause problems anyway when range indicators like '1 to 10' are used.
2020-01-31 04:03:33 +05:30
cp_en = mesh_FEM2DAMASK_elem(m(1))
simplified interface of CPFEM_general (removed P and F, made cs and dcs/dE optional) commented and cleaned up the marc interface. !!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!! For marc simulations, run ./code/setup/setup_code.sh !!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!
2013-04-09 15:38:00 +05:30
if (cptim > theTime .or. inc /= theInc) then ! reached "convergence"
now with first draft of nonlocal constitutive law debugging memory leak closed debugging counters corrected center of gravity stored in mesh state updated is now split into a collecting loop and an execution updateState and updateTemperature fill sequentially separate logicals and evaluate afterwards to converged added 3x3 transposition function, norm for 3x1 matrix and 33x3 matrix multiplication in math non-converged crystallite triggers materialpoint cutback (used to respond elastically) non-converged materialpoint raises terminal illness which in turn renders whole FE increment useless by means of odd stress/stiffness and thus waits for FE cutback
2009-08-11 22:01:57 +05:30
terminallyIll = .false.
simplified interface of CPFEM_general (removed P and F, made cs and dcs/dE optional) commented and cleaned up the marc interface. !!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!! For marc simulations, run ./code/setup/setup_code.sh !!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!
2013-04-09 15:38:00 +05:30
cycleCounter = -1 ! first calc step increments this to cycle = 0
if (inc == 0) then ! >> start of analysis <<
lastIncConverged = .false. ! no Jacobian backup
outdatedByNewInc = .false. ! no aging of state
calcMode = .false. ! pretend last step was collection
removed variable lastMode from FE_solving, used to be used for detection of first call after ping pong; now this is done by checking for a change in the lovl (macro) or a change in the calculation step (abaqus)
2013-08-02 18:58:50 +05:30
lastLovl = lovl ! pretend that this is NOT the first after a lovl change
commercial FEM solvers write results to HDF5
2019-05-05 15:36:55 +05:30
write(6,'(a,i6,1x,i2)') '<< HYPELA2 >> start of analysis..! ',m(1),nn
flush(6)
simplified interface of CPFEM_general (removed P and F, made cs and dcs/dE optional) commented and cleaned up the marc interface. !!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!! For marc simulations, run ./code/setup/setup_code.sh !!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!
2013-04-09 15:38:00 +05:30
else if (inc - theInc > 1) then ! >> restart of broken analysis <<
lastIncConverged = .false. ! no Jacobian backup
outdatedByNewInc = .false. ! no aging of state
calcMode = .true. ! pretend last step was calculation
commercial FEM solvers write results to HDF5
2019-05-05 15:36:55 +05:30
write(6,'(a,i6,1x,i2)') '<< HYPELA2 >> restart of analysis..! ',m(1),nn
flush(6)
simplified interface of CPFEM_general (removed P and F, made cs and dcs/dE optional) commented and cleaned up the marc interface. !!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!! For marc simulations, run ./code/setup/setup_code.sh !!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!
2013-04-09 15:38:00 +05:30
else ! >> just the next inc <<
lastIncConverged = .true. ! request Jacobian backup
outdatedByNewInc = .true. ! request aging of state
calcMode = .true. ! assure last step was calculation
commercial FEM solvers write results to HDF5
2019-05-05 15:36:55 +05:30
write(6,'(a,i6,1x,i2)') '<< HYPELA2 >> new increment..! ',m(1),nn
flush(6)
Introducing the capability to restart jobs that crashed in the middle of sims. At the moment, this feature is exclusive for Marc. Major changes: CPFEM.f90 => 1. Moving the initialization out of CPFEM_general into a separate subroutine, which is directly called by the hypela2 (Beware, the Abaqus version must also be modified in order to adapt with this change). 2. Restore primary state variables in CPFEM_init from binary files when requested (Marc flag: restart read). 3. Writing primary state variables into binary files (Marc flag: restart write). FEsolving.f90 => 1. Adding functions to recognize Marc restart flags: read and write and the corresponding restart file (parent job). 2. Change the initial value of cycleCounter = -1 in conjuction with the change made the ping-pong scheme homogenization_RGC.f90 => 1. Just syntax polishing. IO.f90 => 1. Adding functions/subroutines to open binary files for writing the primary state variables for restart purpose. mpie_cpfem_marc.f90 1. Modification of the general scheme for collection and calculation in order to accommodate the newly added restart feature.
2010-11-03 20:09:18 +05:30
endif
simplified interface of CPFEM_general (removed P and F, made cs and dcs/dE optional) commented and cleaned up the marc interface. !!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!! For marc simulations, run ./code/setup/setup_code.sh !!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!
2013-04-09 15:38:00 +05:30
else if ( timinc < theDelta ) then ! >> cutBack <<
plain mode (no ping pong) now seems to work for marc, further testing needed regular sequence of computation modes is inc 0: 1,1,1,… inc 1: 0,0,0,… 7,1,1,… 0,0,0,… 1,1,1,… and so on after a cutback the computation modes follow as inc 5: 8,0,0,… 1,1,1,… 0,0,0,… 1,1,1,…
2013-08-02 21:49:45 +05:30
lastIncConverged = .false. ! no Jacobian backup
outdatedByNewInc = .false. ! no aging of state
# new interface for Abaqus # IO has some additional functionality for Abaqus parsing # ping pong scheme in FE interface now similar (and more human understandable) in both versions # mesh has better splitting of different tasks, plus operation on database whenever possible # FEsolver as new global var to indicate FEM solver type # computation mode reshuffling: 6 is now Marc special case of recycling...
2009-10-12 21:31:49 +05:30
terminallyIll = .false.
simplified interface of CPFEM_general (removed P and F, made cs and dcs/dE optional) commented and cleaned up the marc interface. !!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!! For marc simulations, run ./code/setup/setup_code.sh !!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!
2013-04-09 15:38:00 +05:30
cycleCounter = -1 ! first calc step increments this to cycle = 0
calcMode = .true. ! pretend last step was calculation
commercial FEM solvers write results to HDF5
2019-05-05 15:36:55 +05:30
write(6,'(a,i6,1x,i2)') '<< HYPELA2 >> cutback detected..! ',m(1),nn
flush(6)
simplified interface of CPFEM_general (removed P and F, made cs and dcs/dE optional) commented and cleaned up the marc interface. !!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!! For marc simulations, run ./code/setup/setup_code.sh !!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!
2013-04-09 15:38:00 +05:30
endif ! convergence treatment end
# new interface for Abaqus # IO has some additional functionality for Abaqus parsing # ping pong scheme in FE interface now similar (and more human understandable) in both versions # mesh has better splitting of different tasks, plus operation on database whenever possible # FEsolver as new global var to indicate FEM solver type # computation mode reshuffling: 6 is now Marc special case of recycling...
2009-10-12 21:31:49 +05:30
adapted flux routine to current state of source_thermal_dissipation.f90
2016-07-20 12:14:12 +05:30
removed variable lastMode from FE_solving, used to be used for detection of first call after ping pong; now this is done by checking for a change in the lovl (macro) or a change in the calculation step (abaqus)
2013-08-02 18:58:50 +05:30
if (usePingPong) then
calcMode(nn,cp_en) = .not. calcMode(nn,cp_en) ! ping pong (calc <--> collect)
if (calcMode(nn,cp_en)) then ! now --- CALC ---
plain mode (no ping pong) now seems to work for marc, further testing needed regular sequence of computation modes is inc 0: 1,1,1,… inc 1: 0,0,0,… 7,1,1,… 0,0,0,… 1,1,1,… and so on after a cutback the computation modes follow as inc 5: 8,0,0,… 1,1,1,… 0,0,0,… 1,1,1,…
2013-08-02 21:49:45 +05:30
computationMode = CPFEM_CALCRESULTS
removed variable lastMode from FE_solving, used to be used for detection of first call after ping pong; now this is done by checking for a change in the lovl (macro) or a change in the calculation step (abaqus)
2013-08-02 18:58:50 +05:30
if (lastLovl /= lovl) then ! first after ping pong
call debug_reset() ! resets debugging
outdatedFFN1 = .false.
pInt not needed anymore
2019-05-15 02:14:38 +05:30
cycleCounter = cycleCounter + 1
untangling - avoid public variables - openMP in initialization hardly useful - structure of init should reflect tasks: 1) reading 2) discretization 3) nonlocal stuff
2019-10-08 13:00:32 +05:30
!mesh_cellnode = mesh_build_cellnodes() ! update cell node coordinates
more cleaning
2019-10-13 23:06:30 +05:30
!call mesh_build_ipCoordinates() ! update ip coordinates
removed variable lastMode from FE_solving, used to be used for detection of first call after ping pong; now this is done by checking for a change in the lovl (macro) or a change in the calculation step (abaqus)
2013-08-02 18:58:50 +05:30
endif
if (outdatedByNewInc) then
improved comments
2013-10-14 20:05:41 +05:30
computationMode = ior(computationMode,CPFEM_AGERESULTS) ! calc and age results
removed variable lastMode from FE_solving, used to be used for detection of first call after ping pong; now this is done by checking for a change in the lovl (macro) or a change in the calculation step (abaqus)
2013-08-02 18:58:50 +05:30
outdatedByNewInc = .false. ! reset flag
endif
else ! now --- COLLECT ---
plain mode (no ping pong) now seems to work for marc, further testing needed regular sequence of computation modes is inc 0: 1,1,1,… inc 1: 0,0,0,… 7,1,1,… 0,0,0,… 1,1,1,… and so on after a cutback the computation modes follow as inc 5: 8,0,0,… 1,1,1,… 0,0,0,… 1,1,1,…
2013-08-02 21:49:45 +05:30
computationMode = CPFEM_COLLECT ! plain collect
removed variable lastMode from FE_solving, used to be used for detection of first call after ping pong; now this is done by checking for a change in the lovl (macro) or a change in the calculation step (abaqus)
2013-08-02 18:58:50 +05:30
if (lastLovl /= lovl .and. & .not. terminallyIll) &
call debug_info() ! first after ping pong reports (meaningful) debugging
if (lastIncConverged) then
plain mode (no ping pong) now seems to work for marc, further testing needed regular sequence of computation modes is inc 0: 1,1,1,… inc 1: 0,0,0,… 7,1,1,… 0,0,0,… 1,1,1,… and so on after a cutback the computation modes follow as inc 5: 8,0,0,… 1,1,1,… 0,0,0,… 1,1,1,…
2013-08-02 21:49:45 +05:30
computationMode = ior(computationMode,CPFEM_BACKUPJACOBIAN) ! collect and backup Jacobian after convergence
removed variable lastMode from FE_solving, used to be used for detection of first call after ping pong; now this is done by checking for a change in the lovl (macro) or a change in the calculation step (abaqus)
2013-08-02 18:58:50 +05:30
lastIncConverged = .false. ! reset flag
endif
preprocessor statements need to start at column 0
2019-06-07 11:15:54 +05:30
!do node = 1,theMesh%elem%nNodes
phase out mesh_elem and theMesh
2019-06-07 09:48:42 +05:30
!CPnodeID = mesh_element(4+node,cp_en)
disable coordinate update for the moment
2019-06-06 16:43:22 +05:30
!mesh_node(1:ndeg,CPnodeID) = mesh_node0(1:ndeg,CPnodeID) + numerics_unitlength * dispt(1:ndeg,node)
preprocessor statements need to start at column 0
2019-06-07 11:15:54 +05:30
!enddo
# new interface for Abaqus # IO has some additional functionality for Abaqus parsing # ping pong scheme in FE interface now similar (and more human understandable) in both versions # mesh has better splitting of different tasks, plus operation on database whenever possible # FEsolver as new global var to indicate FEM solver type # computation mode reshuffling: 6 is now Marc special case of recycling...
2009-10-12 21:31:49 +05:30
endif
adapted flux routine to current state of source_thermal_dissipation.f90
2016-07-20 12:14:12 +05:30
plain mode (no ping pong) now seems to work for marc, further testing needed regular sequence of computation modes is inc 0: 1,1,1,… inc 1: 0,0,0,… 7,1,1,… 0,0,0,… 1,1,1,… and so on after a cutback the computation modes follow as inc 5: 8,0,0,… 1,1,1,… 0,0,0,… 1,1,1,…
2013-08-02 21:49:45 +05:30
else ! --- PLAIN MODE ---
computationMode = CPFEM_CALCRESULTS ! always calc
if (lastLovl /= lovl) then
if (.not. terminallyIll) &
call debug_info() ! first reports (meaningful) debugging
call debug_reset() ! and resets debugging
outdatedFFN1 = .false.
pInt not needed anymore
2019-05-15 02:14:38 +05:30
cycleCounter = cycleCounter + 1
disable coordinate update for the moment
2019-06-06 16:43:22 +05:30
!mesh_cellnode = mesh_build_cellnodes() ! update cell node coordinates
!call mesh_build_ipCoordinates() ! update ip coordinates
plain mode (no ping pong) now seems to work for marc, further testing needed regular sequence of computation modes is inc 0: 1,1,1,… inc 1: 0,0,0,… 7,1,1,… 0,0,0,… 1,1,1,… and so on after a cutback the computation modes follow as inc 5: 8,0,0,… 1,1,1,… 0,0,0,… 1,1,1,…
2013-08-02 21:49:45 +05:30
endif
if (outdatedByNewInc) then
computationMode = ior(computationMode,CPFEM_AGERESULTS)
outdatedByNewInc = .false. ! reset flag
endif
if (lastIncConverged) then
computationMode = ior(computationMode,CPFEM_BACKUPJACOBIAN) ! backup Jacobian after convergence
lastIncConverged = .false. ! reset flag
endif
corrected determination and evaluation of updateJaco introduced check whether FFN1 changed and force recollect in that case
2008-07-14 20:08:19 +05:30
endif
CPFEM_general now independed of the wrapper code (UMAT, hypela2, etc.) renamed CPFEM_stressIP to CPFEM_MaterialPoint renamed CPFEM_stressCrystallite to CPFEM_Crystallite introduced new global variables to keep track of FE state within module FEsolving
2008-03-15 03:02:57 +05:30
simplified interface of CPFEM_general (removed P and F, made cs and dcs/dE optional) commented and cleaned up the marc interface. !!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!! For marc simulations, run ./code/setup/setup_code.sh !!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!
2013-04-09 15:38:00 +05:30
theTime = cptim ! record current starting time
theDelta = timinc ! record current time increment
theInc = inc ! record current increment number
removed variable lastMode from FE_solving, used to be used for detection of first call after ping pong; now this is done by checking for a change in the lovl (macro) or a change in the calculation step (abaqus)
2013-08-02 18:58:50 +05:30
# new interface for Abaqus # IO has some additional functionality for Abaqus parsing # ping pong scheme in FE interface now similar (and more human understandable) in both versions # mesh has better splitting of different tasks, plus operation on database whenever possible # FEsolver as new global var to indicate FEM solver type # computation mode reshuffling: 6 is now Marc special case of recycling...
2009-10-12 21:31:49 +05:30
endif
corrected lastLovl introduced in rev2593
2013-08-02 21:14:28 +05:30
lastLovl = lovl ! record lovl
2007 shortened stress_recovery = ... added (crucial) deMandeling for return values of CPFEM_general 2005 matched to 2007 version
2008-02-19 00:19:06 +05:30
parallelExecution flag now passed as input variable by CPFEM_general; flag is set on the solver level (DAMASK_abaqus_exp.f, DAMASK_marc.f90, etc.)
2013-08-02 16:50:11 +05:30
call CPFEM_general(computationMode,usePingPong,ffn,ffn1,t(1),timinc,m(1),nn,stress,ddsdde)
fixed some dangerous constructs in computation mode assignments introduction of separate stress and stiffness variables to pass on to CPFEM_general (of dimension 6)
2010-02-18 15:53:02 +05:30
avoid converstions and use of global variables
2019-01-14 17:15:07 +05:30
d = ddsdde(1:ngens,1:ngens)
s = stress(1:ndi+nshear)
simplified interface of CPFEM_general (removed P and F, made cs and dcs/dE optional) commented and cleaned up the marc interface. !!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!! For marc simulations, run ./code/setup/setup_code.sh !!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!
2013-04-09 15:38:00 +05:30
g = 0.0_pReal
if(symmetricSolver) d = 0.5_pReal*(d+transpose(d))
adapted flux routine to current state of source_thermal_dissipation.f90
2016-07-20 12:14:12 +05:30
simplified interface of CPFEM_general (removed P and F, made cs and dcs/dE optional) commented and cleaned up the marc interface. !!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!! For marc simulations, run ./code/setup/setup_code.sh !!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!
2013-04-09 15:38:00 +05:30
!$ call omp_set_num_threads(defaultNumThreadsInt) ! reset number of threads to stored default value
* environment variable MPIE_NUM_THREADS now determines number of threads used in parallel regions * default value of the OMP_NUM_THREADS variable has to be restored at the end of mpie subroutine, since marc also seems to use and change(!) this * usage: "export MPIE_NUM_THREADS=<number of threads>" to set variable in shell, then restart mentat and compile with option 3 (at the moment this does only work on ws 6, since all other workstations use compiler option "-save"; this puts all local variables by default in static memory, which is a killer for parallelization!)
2010-11-19 23:15:27 +05:30
added new, flexible debugging scheme. now all modules have their own debug specification. compiles and runs, I hope nothing is broken did a lot of polishing
2012-03-09 01:55:28 +05:30
end subroutine hypela2
major restructuring of code. homogenization as well as constitutive are now free to choose. the runtime got somewhat longer (25% on simple tests) compared to a hardcoded isostrain homogenization. this might be a point of further optimization at a later stage... please use homogenization_isostrain.f90 as starting point / example for future developments of homog-schemes. the homogenization scheme now can additionally output certain results. hence, the userdata structure at each integration point now looks like this: - sizeHomogPostResults - block of that size containing homogPostResults then for each grain: - sizeGrainPostResults - block of that size containing crystallitePostResults, which consist of: + phaseID + volFrac + Eulers (3) + any constitutive post results requested
2009-05-07 21:57:36 +05:30
# separated CPFEM_Crystallite from CPFEM. Now in separate module crystallite.f90 as "SingleCrystallite" # improved SingleCrystallite to advance by true cutbacking (instead of improving guess and integrating always from t_0) # module "crystal" renamed to "lattice" together with its prefix for variables # extension of "computationMode" to deal with cutbacks (CPFEM_general). # cutback and new inc detection for MARC is based on common block variable cptim (and inc), not incsub anymore! # generalized GrainInterAction as new homogenization scheme # two symbolic links are required: constitutive.f90 and CPFEM.f90
2008-04-07 20:24:29 +05:30
added code for solving heat equation
2016-05-03 20:36:55 +05:30
!--------------------------------------------------------------------------------------------------
cleaning up Marc subroutine flux can be used for specifying volume flux boundary condition (dissipated plastic work) in thermomechanically coupled simulations
2017-02-08 14:32:56 +05:30
!> @brief calculate internal heat generated due to inelastic energy dissipation
added code for solving heat equation
2016-05-03 20:36:55 +05:30
!--------------------------------------------------------------------------------------------------
subroutine flux(f,ts,n,time)
clear separation between solver and DAMASK
2020-01-25 13:42:09 +05:30
use prec
use thermal_conduction
use mesh
adapted flux routine to current state of source_thermal_dissipation.f90
2016-07-20 12:14:12 +05:30
clear separation between solver and DAMASK
2020-01-25 13:42:09 +05:30
implicit none
real(pReal), dimension(6), intent(in) :: &
ts
integer, dimension(10), intent(in) :: &
n
real(pReal), intent(in) :: &
time
real(pReal), dimension(2), intent(out) :: &
f
store mapping MARC/FEM2DAMASK mapping do not calculate the mapping for elements and nodes per call on the fly, rather store it. Not memory efficient in the case that numbers are not consequtive (order does not matter, but missing nodes/elements would waste some 2 integers per missing number). However, this seem to cause problems anyway when range indicators like '1 to 10' are used.
2020-01-31 04:03:33 +05:30
call thermal_conduction_getSourceAndItsTangent(f(1), f(2), ts(3), n(3),mesh_FEM2DAMASK_elem(n(1)))
cleaning up Marc subroutine flux can be used for specifying volume flux boundary condition (dissipated plastic work) in thermomechanically coupled simulations
2017-02-08 14:32:56 +05:30
end subroutine flux
added code for solving heat equation
2016-05-03 20:36:55 +05:30
commercial FEM solvers write results to HDF5
2019-05-05 15:36:55 +05:30
!--------------------------------------------------------------------------------------------------
polishing
2019-06-15 19:51:15 +05:30
!> @brief trigger writing of results
avoid HDF5 error when creating link twice
2020-01-12 05:25:00 +05:30
!> @details uedinc is called before each new increment, not at the end of a converged one.
!> Therefore, storing the last written inc with an 'save' variable is required to avoid writing the
! same increment multiple times.
commercial FEM solvers write results to HDF5
2019-05-05 15:36:55 +05:30
!--------------------------------------------------------------------------------------------------
subroutine uedinc(inc,incsub)
pInt not needed anymore
2019-05-15 02:14:38 +05:30
use prec
phase out mesh_elem and theMesh
2019-06-07 09:48:42 +05:30
use CPFEM
commercial FEM solvers write results to HDF5
2019-05-05 15:36:55 +05:30
implicit none
integer, intent(in) :: inc, incsub
avoid HDF5 error when creating link twice
2020-01-12 05:25:00 +05:30
integer, save :: inc_written
commercial FEM solvers write results to HDF5
2019-05-05 15:36:55 +05:30
#include QUOTE(PASTE(./MarcInclude/creeps,Marc4DAMASK)) ! creeps is needed for timinc (time increment)
avoid HDF5 error when creating link twice
2020-01-12 05:25:00 +05:30
if (inc > inc_written) then
call CPFEM_results(inc,cptim)
inc_written = inc
endif
commercial FEM solvers write results to HDF5
2019-05-05 15:36:55 +05:30
end subroutine uedinc