Policy targets constraints

[abhishek0208]

Policy targets are added as custom linopy constraints in a PyPSA model.
This was first done in the tza-apg repo and has now been brought into this repo and generalised.

The steps to add custom constraints are:

from tz_pypsa.model import Model
from tz_pypsa.constraints import (constr_bus_self_sufficiency, 
                                  constr_cumulative_p_nom, 
                                  constr_policy_targets, 
                                  constr_max_annual_utilisation)

# Load a stock model with user-defined configuration
network = (
    Model.load_model(
        'ASEAN', 
        frequency = '72h',
        select_nodes = ['MYSPE', 'MYSSH', 'MYSSK', 'SGPXX'], 
        years = [2023, 2050],
        backstop = False,
        set_global_constraints = False,
    )
)

# Instantiate the linopy model
lp_model = network.optimize.create_model()

# Add policy targets as custom linopy constraints
constr_policy_targets(network,
                      lp_model,
                      'ASEAN') # This requires that a CSV with the policy targets are provided within the stock model data

[Handriyanti]

All functions can be called smoothly. But, I tried to set max_utilisation_rate = 1 and compare it with max_utilisation_rate = 0.85 in constr_max_annual_utilisation The results of p_nom, p_nom_opt and p are the same. Did I check the wrong variable on this?
Others are okay. Once this function is good, we can merge this branch to main and go on with brownfield.py in CFE.

[Handriyanti]

Coming back to this... All constraints including policy targets are actually able to be called and applied nicely. Just don't forget to solve the model first before defining a constraint then resolve it after defining a constraint. Please correct me here.

from tz_pypsa.model import Model
from tz_pypsa.constraints import (constr_bus_self_sufficiency, 
                                  constr_cumulative_p_nom, 
                                  constr_policy_targets, 
                                  constr_max_annual_utilisation)

# Load a stock model with user-defined configuration
network = (
    Model.load_model(
        'ASEAN', 
        frequency = '72h',
        select_nodes = ['IDNSM','IDNJW','IDNKA'], 
        years = [2023, 2030, 2050],
        backstop = False,
        set_global_constraints = False,
    )
)

# solve the model
network.optimize()

# Instantiate the linopy model
lp_model = network.optimize.create_model()

# Add policy targets as custom linopy constraints
constr_policy_targets(network,
                      lp_model,
                      'ASEAN') # This requires that a CSV with the policy targets are provided within the stock model data

# Let's solve the model again
network.optimize.solve_model()

I tried to run three IDN nodes because coal are super intensively used so the results with and without constraints are easily distinguished. I documented the results comparison here. ONe thing to be remembered is: The model is infeasible if it's heavily constrained.