Missing functionality in unit commitment

[sstroemer]

The current UC formulation is missing a few points that should be properly integrated:

• Cyclic behaviour, similar to that of stateful Nodes, for the parameters that support initial pre-modeling-period settings (e.g., startup which already takes unit.is_on_before into account, but does not allow choosing that)
• Fixes to min/max on/off time formulations (with different Snapshot durations, unit counts, ...)
• Properly implemented/tested endogenous investment decisions (from an external Decision) combined with active UC
• ...?

A lot of that is actually due to the current/old issues with representative Snapshots. Since they conflict with various things, and complicate a lot of different formulations, we discussed removing them on multiple occasions (and switching to a different formulation, e.g., Kotzur et al.). Therefore, this might be best done as part of a larger rework to prevent future mess/dept in the whole UC implementation.

[GerhardTotschnig]

@sstroemer
Title: Change part load calculation to avoid incremental efficiency:
In the distric heat model we changed form using the incremental efficiency to a more readable form:

The old form with GenAboveMinLoad_MW:

Elec_generation_MW[u,t]=UnitsOnline[t,u]* unitCap_MinLoadMW[u,t,f]+GenAboveMinLoad_MW[t,u]
Constraint: Elec_generation_MW[u,t]<=UnitsOnline[t,u]*unit_cap_MW[t,g]
FuelConsumption_MW[u,t]=UnitsOnline[t,u]* unitCap_MinLoadMW[u,t,f]/eff_minLoad[u,t] + GenAboveMinLoad_MW[t,g]/incremental_eff[u,t]

So here you need to calculate the incremental_efficiency. Which is a little complicated to explain.

With a change form GenAboveMinLoad_MW -> Units_At_Full_Load (see below), the calculation of the input and output flows is much more easy to understand. There is no need to calculate incremental efficiencies anymore:

The new formulation for unit type „u“, which may have mutliple identical units installed, and has multiple inputs and putputs:
Var:
• UnitsOn[u,t]: 0<=, up and running units of type „u“; linear, binary or integer
• AvaiableUnits[u,t]: 0<= Number of units available in a certain time step ; linear
Costraint: 0<=UnitsOn[u,t] <= AvaiableUnits[t]
Costraint: AvaiableUnits[u,t] <= exiting_units[u,year[t]]+new_installed_units[u,year[t]]-decomissioned_units[u,year[t]]
-temporarilty_nonavailable_units[u,t]
• Units_At_Full_Load[u,t]: 0<=, Number of Units at full load; linear
Constraint: 0<= Units_At_Full_Load[u,t] <= UnitsOn[u,t]

NEW

Input_Output_MW_Expression[u in Units, t in Steps, f in In_Out_Flows]:=
           UnitsOn[u,t]* unitCap_MinLoadMW[u,t,f] 
           + Units_At_Full_Load[u,t]*(unitCap_FullLoad_MW[u,t,f] -unitCap_MinLoadMW[u,t,f])

There is no need to calculate incremental efficiencies anymore and easy to understand.

[sstroemer]

Note: The "mention" above from issue "# 25" is not intentional and just GitHub messing with the auto-mention ...