research/cantera
1
0
Fork 0
Code Issues Pull requests Projects Releases Packages Wiki Activity Actions

Fixed an error that occured when components of multispecies

Browse source 
phases were  zeroed because other phases were being zeroed.
Then the noncomponents of that same multispecies phase could 
comprise the entire phase and the Hessian diagonal would then be
zero. The fix to zero the species seems to work.
This commit is contained in:
Harry Moffat 2010-10-23 01:35:10 +00:00
parent f26ff94525
commit 1bdee90289
1 changed files with 27 additions and 0 deletions
Show stats Download patch file Download diff file Expand all files Collapse all files

27
Cantera/src/equil/vcs_rxnadj.cpp
View file

@ -306,6 +306,33 @@ namespace VCSnonideal {
* added back into the component species.
*/
if (dss != 0.0) {
if ((k == kspec) && (m_SSPhase[kspec] != 1)) {
/*
* Found out that we can be in this spot, when components of multispecies phases
* are zeroed, leaving noncomponent species of the same phase having all of the
* mole numbers of that phases. it seems that we can suggest a zero of the species
* and the code will recover.
*/
#ifdef DEBUG_MODE
sprintf(ANOTE, "Delta damped from %g to %g due to delete %s",
m_deltaMolNumSpecies[kspec],
-m_molNumSpecies_old[kspec], m_speciesName[kspec].c_str());
#endif
m_deltaMolNumSpecies[kspec] = -m_molNumSpecies_old[kspec];
#ifdef DEBUG_MODE
if (m_debug_print_lvl >= 2) {
plogf(" --- %-12.12s", m_speciesName[kspec].c_str());
plogf(" %12.4E %12.4E %12.4E | %s\n",
m_molNumSpecies_old[kspec], m_deltaMolNumSpecies[kspec],
m_deltaGRxn_new[irxn], ANOTE);
}
#endif
continue;
}
/*
* Delete the single species phase
*/
m_molNumSpecies_old[kspec] += dss;
m_tPhaseMoles_old[m_phaseID[kspec]] += dss;
for (j = 0; j < m_numComponents; ++j) {