Add max capacity limits for renewable energies in Germany

[toniseibold]

Since our results showed too much renewable energies in 2020 and 2025, a maximum constraint is introduced. config["limits_capacity_max"] follows the same logic as `config["limits_capacity_min"]. Be aware that defining both upper and lower limits for renewables is possible but might lead to numerical issues.

[JulianGeis]

Nice work, the model runs through smoothly. :dancing_women:

Why are we limiting the capacity expansion and not the total capacity itself? If I see it right, only the newly installed capacity is restricted, but to the total value of capacity?

Comparing the results to the real system 2020 (365h): real: solar: 54 GW, onshore_wind: 54 GW, offshore: 8 GW model_2020: solar: 81 GW, onshore_wind: 61, offshore_wind: 0

real_2024: solar 88 GW, onhore_wind: 62 GW, offshore_wind: 8 GW model_2025: solar: 93 GW, onshore 61 GW, offshore: 16 GW

Maybe it makes sense to reintroduce the minimum limits at least in 2020 and just reproduce the system from the past. What is also weird is why is there no offshore wind in 2020. @toniseibold do you have the same results or is there some powerplantmatching weirdness on my side again? :D

[lindnemi]

• The way the capacity limits are implement at the moment, they only check if the carrier starts with solar. Hence, they would include solar thermal at the moment, which is probably not correct.
• @JulianGeis i can confirm 0 offshore wind in 2020. introducing the capacity min again could be a solution, but then we would basically introduce an equality constraint. i am not sure if this might lead to numerical errors, but we could try
• @toniseibold i think the constraint might have the wrong orientation and requires a <= instead of >= in line 80. It seems that at the moment the specified capacity_limits in 2020 are not applied in the model

[toniseibold]

• I set the limit to 0 in 2020 or when the config specifies 0. Deleting the extendable generators leads to warnings, since n.generators_t["p"] still includes time series for the deleted generators.

              if investment_year == 2020 or limit == 0:
                  # drop extendable_index
                  n.generators.drop(extendable_index, inplace=True)
                  n.generators_t["p"].drop(extendable_index, axis=1, inplace=True)
                  continue

• I leave the logic of restricting the expansion to limit - existing_capacity. This way, the boundary condition is only affecting a smaller set of generators/links.
• I had to set the maximum limit of onshore wind for 2025 to 69 GW since in 2020 there are already 61.052 GW installed which lead to infeasabilities.
• I filter out the carrier solar thermal

[lindnemi]

Existing solar capacities for the 2020 model are 81 GW, onshore wind 61 GW. Both are bigger than historical values. The reason is that the generator solar-2019 contains all capacities added in 2019 and later. Since our input data already contains solar added after 2020, we get excess capacities

[lindnemi]

we could throw out the extra capacities, but then they would be missing in later years (except that they probably get build by the model anyway)

[lindnemi]

My suggestion: Keep extra capacity, allow no further extension, issue a warning. This will imply that we have too much RES generation in 2020 and can not exactly validate against data for 2020.

[lindnemi]

@toniseibold changed the logic for Maximum limits slightly, please take a look

[toniseibold]

Logic looks good. Could add "residential rural solar thermal", "services rural solar thermal", "residential urban decentral solar thermal", "services urban decentral solar thermal", "urban central solar thermal" to the forbidden variable just to be on the safe side.