by jasper » 08 February 2014, 14:42
In the last couple of updates, a lot has changed in the reactors in COUSCOUS. The Gibbs reactor has also been completely revised as far as the solution algorithm goes. You may have noticed with your old solution that the concentration of oxygen and methane were practically zero. So close to zero in fact that within the specified relative tolerance (on the reaction extents) specified for the reactor, it may as well be zero, slightly negative, or a few orders of magnitude higher. In this case the result was slightly negative, and the reactor failed.
I have changed the reactor to detect this solution, to reset the concentrations to just within the feasible region and continue. This does not mean that the concentrations for these compounds are well predicted; the reaction should not be considered in equilibrium in this case, but rather as irreversible (within numeric precision of the unit). The unit will now solve and post an appropriate error message. Thanks for reporting the issue.
The update is available via CUP.
I could not reproduce the dimensionality mismatch; at this point I have to assume that that problem has already been resolved. Let me know if it pops up again please.