fixed_new_capacity sets CAP_NEW but not CAP about message_ix HOT 4 OPEN

This behaviour actually makes sense - currently, there is no mechanism forcing the newly installed capacity to be maintained in the first period, i.e. CAP_NEW(node, tec, vintage) does not have to equal CAP(node, tec, vintage, vintage). In an optimization problem, the model wouldn't build new capacity that it doesn't maintain.

In this particular case, I guess that the model just doesn't need the capacity, so it builds it (as defined) and immediately throws it away.

@ClaraLuisa, can you confirm?

More generally, one could add a new constraint like

CAPACITY_MAINTENANCE(node,inv_tec,vintage,vintage)$( map_tec_lifetime(node,inv_tec,vintage, vintage) AND NOT historical(vintage) )..
    CAP(node,inv_tec,vintage,vintage) =E=
    remaining_capacity(node,inv_tec,vintage, vintage) * (
* new capacity built in the current period (vintage == year)
        + ( duration_period(vintage) * CAP_NEW(node,inv_tec,vintage) )
    ) ;

from message_ix.

It does seem like @danielhuppmann is right on that the model just doesn't really need it and thus "throws it away". However, i would very much appreciate an extension to the maintenance equation as suggested by @danielhuppmann as this would be more realistic (also in the cost accounting).

from message_ix.

@danielhuppmann are you going to, or should I try to implement the equation suggested by you into the gams code.

from message_ix.

@ClaraLuisa it would be great if you could give it a try!

Thinking about it, you most likely wouldn't even need a new equation, just refine the existing CAPACITY_MAINTENANCE to apply only if vintage < year (removing the section with CAP_NEW) and then adding the blurb I suggested earlier.

And you don't even need a new test for it, the 'Westeros' and 'Austria' tutorials should catch it if anything goes awry...

from message_ix.