PyPSA to SMS++ conversion step by step¶

This notebook describes how to perform the conversion step by step: first convert the PyPSA model into a SMS++ object, then optimize it and finally read back a solved PyPSA network with the results from SMS++.

In particular:

1. Create a PyPSA network
2. Convert the PyPSA network to a SMS++ object and explore the SMS++ object
3. Optimize the SMS++ object
4. Read back the results into a PyPSA network

1. Creation of the PyPSA model (from UCBlock)¶

We first create a simple PyPSA model.

In [1]:

import pypsa
import pandas as pd
import pypsa2smspp

import pypsa import pandas as pd import pypsa2smspp

The following code creates the desired pypsa model.¶

In [2]:

n = pypsa.Network()
n.set_snapshots(pd.date_range("2024-01-01T00:00", "2024-01-01T23:00", freq="h"))

# Add carriers
n.add("Carrier", "AC")

# Add buses
n_buses = 2
for b in range(n_buses):
    n.add("Bus", f"bus{b}", carrier="AC")

# Add lines in a radial topology using bidirectional links
n_lines = n_buses - 1
for l in range(n_lines):
    n.add(
        "Link",
        f"line{l}",
        bus0=f"bus{l}",
        bus1=f"bus{l+1}",
        length=1,
        capital_cost=1000,
        p_min_pu=-1,
        p_nom_extendable=True,
    )

# Add a load to each bus
n_loads = n_buses
for l in range(n_loads):
    n.add("Load", f"load{l}", bus=f"bus{l}", p_set=pd.Series(100, index=n.snapshots))

# Add a generator to the first bus
n.add(
    "Generator",
    "gen0",
    bus="bus0",
    p_nom_extendable=True,
    capital_cost=1000,
    marginal_cost=1,
)

n_clean = n.copy()

n = pypsa.Network() n.set_snapshots(pd.date_range("2024-01-01T00:00", "2024-01-01T23:00", freq="h")) # Add carriers n.add("Carrier", "AC") # Add buses n_buses = 2 for b in range(n_buses): n.add("Bus", f"bus{b}", carrier="AC") # Add lines in a radial topology using bidirectional links n_lines = n_buses - 1 for l in range(n_lines): n.add( "Link", f"line{l}", bus0=f"bus{l}", bus1=f"bus{l+1}", length=1, capital_cost=1000, p_min_pu=-1, p_nom_extendable=True, ) # Add a load to each bus n_loads = n_buses for l in range(n_loads): n.add("Load", f"load{l}", bus=f"bus{l}", p_set=pd.Series(100, index=n.snapshots)) # Add a generator to the first bus n.add( "Generator", "gen0", bus="bus0", p_nom_extendable=True, capital_cost=1000, marginal_cost=1, ) n_clean = n.copy()

In [3]:

n.optimize(solver_name="highs")

n.optimize(solver_name="highs")

/tmp/ipykernel_1900/3144390136.py:1: FutureWarning: The default value of `include_objective_constant` will change from True to False in version 2.0. Set `include_objective_constant` explicitly to suppress this warning. Using False improves LP numerical conditioning by not including the objective constant as a variable.
  n.optimize(solver_name="highs")

INFO:linopy.model: Solve problem using Highs solver

INFO:linopy.io: Writing time: 0.04s

INFO:linopy.constants: Optimization successful: 
Status: ok
Termination condition: optimal
Solution: 50 primals, 146 duals
Objective: 3.05e+05
Solver model: available
Solver message: Optimal

INFO:pypsa.optimization.optimize:The shadow-prices of the constraints Generator-ext-p-lower, Generator-ext-p-upper, Link-ext-p-lower, Link-ext-p-upper were not assigned to the network.

Running HiGHS 1.11.0 (git hash: 364c83a): Copyright (c) 2025 HiGHS under MIT licence terms
LP   linopy-problem-ant48rj_ has 146 rows; 50 cols; 242 nonzeros
Coefficient ranges:
  Matrix [1e+00, 1e+00]
  Cost   [1e+00, 1e+03]
  Bound  [0e+00, 0e+00]
  RHS    [1e+02, 1e+02]
Presolving model
72 rows, 2 cols, 72 nonzeros  0s
0 rows, 0 cols, 0 nonzeros  0s
Presolve : Reductions: rows 0(-146); columns 0(-50); elements 0(-242) - Reduced to empty
Solving the original LP from the solution after postsolve
Model name          : linopy-problem-ant48rj_
Model status        : Optimal
Objective value     :  3.0480000000e+05
P-D objective error :  0.0000000000e+00
HiGHS run time      :          0.00

Out[3]:

('ok', 'optimal')

2. Create a SMS++ object from the PyPSA model¶

The following code creates the SMS++ object from the PyPSA model, as a pysmspp.SMSNetwork object, and explores it.

If the user wants to use only this object or edit it before the processing, they can do so at this point. Be aware that editing the SMS++ object may lead to inconsistencies with the original PyPSA model and hence may cause issues when reading back the results into a PyPSA network.

In [4]:

tran = pypsa2smspp.Transformation()
verbose = True
sms_network = tran.create_model(n_clean, verbose=verbose)

tran = pypsa2smspp.Transformation() verbose = True sms_network = tran.create_model(n_clean, verbose=verbose)

[START] consistency_check
[ OK  ] consistency_check (0.000s)
[START] direct
[ OK  ] direct (0.115s)
[START] prepare_tssb_interface
[ OK  ] prepare_tssb_interface (0.000s)
[START] convert_to_blocks
[ OK  ] convert_to_blocks (0.096s)

Showcase the pysmspp object that is created by the conversion process.

In [5]:

sms_network

sms_network

Out[5]:

SMSNetwork Object
Block object
Attributes (1): SMS++_file_type
Dimensions (0): None
Variables (0): None
Blocks (1): Block_0

and its inner block structure using the print_tree function from pysmspp.

In [6]:

sms_network.print_tree(show_all=True)

sms_network.print_tree(show_all=True)

SMSNetwork
  Attributes (1): SMS++_file_type=1
└── Block_0 [UCBlock]
    Dimensions (6): TimeHorizon=24, NumberUnits=1, NumberElectricalGenerators=1, NumberNodes=2, NumberLines=1, NumberNetworks=1
    Variables (11): ActivePowerDemand, NodeName, LineName, GeneratorNode, StartLine, ... (11 total)
    Attributes (1): id=0
    ├── UnitBlock_0 [IntermittentUnitBlock]
    │   Variables (8): MaxPower, MinPower, ActivePowerCost, MinGeneration, MaxGeneration, ... (8 total)
    │   Attributes (1): name=gen0
    └── NetworkBlock_0 [DesignNetworkBlock]
        Dimensions (2): NumberDesignLines=1, NumberSubNetwork=24
        Variables (4): InvestmentCost, MinCapacityDesign, MaxCapacityDesign, DesignLines

3. Execute the optimization on the transformation object¶

In [7]:

tran.optimize(verbose=verbose)

tran.optimize(verbose=verbose)

[START] optimize
Executing command:
ucblock_solver temp_test_case.nc -S uc_solverconfig.txt -c /home/docs/checkouts/readthedocs.org/user_builds/pypsa2smspp/conda/latest/lib/python3.13/site-packages/pysmspp/data/configs/UCBlock/ -p /home/docs/checkouts/readthedocs.org/user_builds/pypsa2smspp/checkouts/latest/docs/examples/output/ -O /home/docs/checkouts/readthedocs.org/user_builds/pypsa2smspp/checkouts/latest/docs/examples/output/solution_test_case.nc

Using a default Block configuration
Running HiGHS 1.14.0 (git hash: n/a): Copyright (c) 2026 under MIT licence terms
Solver: HiGHSMILPSolver
LP has 240 rows; 122 cols; 432 nonzeros
Coefficient ranges:
  Matrix  [1e+00, 1e+00]
  Cost    [1e+00, 1e+03]
  Bound   [1e+07, 1e+09]
  RHS     [1e+02, 1e+02]
WARNING: Problem has some excessively large column bounds
WARNING:    Consider scaling the    bounds by 1e-3, or setting the user_bound_scale option to -10
Presolving model
0 rows, 0 cols, 0 nonzeros 0s
0 rows, 0 cols, 0 nonzeros 0s
Presolve reductions: rows 0(-240); columns 0(-122); nonzeros 0(-432) - Reduced to empty
Performed postsolve
Solving the original LP from the solution after postsolve

Model status        : Optimal
Objective value     :  3.0480000000e+05
P-D objective error :  0.0000000000e+00
HiGHS run time      :          0.00
Elapsed time: 9.86211000e-04 s
Status = 10 (Success)
Upper bound = 3.04800000e+05
Lower bound = 3.04800000e+05

----- IntermittentUnitBlock 0 -----
Function value = 2.04800000e+05
Capacity       = 2.00000000e+11
Active power   = [      2.00000000e+02      2.00000000e+02      2.00000000e+02      2.00000000e+02      2.00000000e+02      2.00000000e+02      2.00000000e+02      2.00000000e+02      2.00000000e+02      2.00000000e+02      2.00000000e+02      2.00000000e+02      2.00000000e+02      2.00000000e+02      2.00000000e+02      2.00000000e+02      2.00000000e+02      2.00000000e+02      2.00000000e+02      2.00000000e+02      2.00000000e+02      2.00000000e+02      2.00000000e+02      2.00000000e+02 ]

----- DesignNetworkBlock 0 -----
Function value   = 1.00000000e+05
Design variables = [
  line    0 : 100
]

  --- DCNetworkBlock 0 ---
  Power flow       = [      1.00000000e+02 ]
  Auxiliary variable     = [      1.00000000e+02 ]

  --- DCNetworkBlock 1 ---
  Power flow       = [      1.00000000e+02 ]
  Auxiliary variable     = [      1.00000000e+02 ]

  --- DCNetworkBlock 2 ---
  Power flow       = [      1.00000000e+02 ]
  Auxiliary variable     = [      1.00000000e+02 ]

  --- DCNetworkBlock 3 ---
  Power flow       = [      1.00000000e+02 ]
  Auxiliary variable     = [      1.00000000e+02 ]

  --- DCNetworkBlock 4 ---
  Power flow       = [      1.00000000e+02 ]
  Auxiliary variable     = [      1.00000000e+02 ]

  --- DCNetworkBlock 5 ---
  Power flow       = [      1.00000000e+02 ]
  Auxiliary variable     = [      1.00000000e+02 ]

  --- DCNetworkBlock 6 ---
  Power flow       = [      1.00000000e+02 ]
  Auxiliary variable     = [      1.00000000e+02 ]

  --- DCNetworkBlock 7 ---
  Power flow       = [      1.00000000e+02 ]
  Auxiliary variable     = [      1.00000000e+02 ]

  --- DCNetworkBlock 8 ---
  Power flow       = [      1.00000000e+02 ]
  Auxiliary variable     = [      1.00000000e+02 ]

  --- DCNetworkBlock 9 ---
  Power flow       = [      1.00000000e+02 ]
  Auxiliary variable     = [      1.00000000e+02 ]

  --- DCNetworkBlock 10 ---
  Power flow       = [      1.00000000e+02 ]
  Auxiliary variable     = [      1.00000000e+02 ]

  --- DCNetworkBlock 11 ---
  Power flow       = [      1.00000000e+02 ]
  Auxiliary variable     = [      1.00000000e+02 ]

  --- DCNetworkBlock 12 ---
  Power flow       = [      1.00000000e+02 ]
  Auxiliary variable     = [      1.00000000e+02 ]

  --- DCNetworkBlock 13 ---
  Power flow       = [      1.00000000e+02 ]
  Auxiliary variable     = [      1.00000000e+02 ]

  --- DCNetworkBlock 14 ---
  Power flow       = [      1.00000000e+02 ]
  Auxiliary variable     = [      1.00000000e+02 ]

  --- DCNetworkBlock 15 ---
  Power flow       = [      1.00000000e+02 ]
  Auxiliary variable     = [      1.00000000e+02 ]

  --- DCNetworkBlock 16 ---
  Power flow       = [      1.00000000e+02 ]
  Auxiliary variable     = [      1.00000000e+02 ]

  --- DCNetworkBlock 17 ---
  Power flow       = [      1.00000000e+02 ]
  Auxiliary variable     = [      1.00000000e+02 ]

  --- DCNetworkBlock 18 ---
  Power flow       = [      1.00000000e+02 ]
  Auxiliary variable     = [      1.00000000e+02 ]

  --- DCNetworkBlock 19 ---
  Power flow       = [      1.00000000e+02 ]
  Auxiliary variable     = [      1.00000000e+02 ]

  --- DCNetworkBlock 20 ---
  Power flow       = [      1.00000000e+02 ]
  Auxiliary variable     = [      1.00000000e+02 ]

  --- DCNetworkBlock 21 ---
  Power flow       = [      1.00000000e+02 ]
  Auxiliary variable     = [      1.00000000e+02 ]

  --- DCNetworkBlock 22 ---
  Power flow       = [      1.00000000e+02 ]
  Auxiliary variable     = [      1.00000000e+02 ]

  --- DCNetworkBlock 23 ---
  Power flow       = [      1.00000000e+02 ]
  Auxiliary variable     = [      1.00000000e+02 ]

Peak CPU Usage: 0.00 %
Peak Memory Usage: 0.72 MB
Total Time: 0.20 seconds

[ OK  ] optimize (0.226s)

Out[7]:

UCBlockSolver
	
	solver_name=ucblock_solver
	status=10 (Success)
	configfile=/home/docs/checkouts/readthedocs.org/user_builds/pypsa2smspp/conda/latest/lib/python3.13/site-packages/pysmspp/data/configs/UCBlock/uc_solverconfig.txt
	fp_network=/home/docs/checkouts/readthedocs.org/user_builds/pypsa2smspp/checkouts/latest/docs/examples/output/temp_test_case.nc
	fp_solution=/home/docs/checkouts/readthedocs.org/user_builds/pypsa2smspp/checkouts/latest/docs/examples/output/solution_test_case.nc

4. Perform the conversion back to a PyPSA network and explore the results¶

In [8]:

n_smspp = tran.retrieve_solution(n_clean, verbose=verbose)
n_smspp

n_smspp = tran.retrieve_solution(n_clean, verbose=verbose) n_smspp

[START] parse_solution_to_unitblocks
[split] No merged DCNetworkBlock entries found; nothing to do.
[ OK  ] parse_solution_to_unitblocks (0.000s)
[START] inverse_transformation
[ OK  ] inverse_transformation (0.017s)

Out[8]:

PyPSA Network 'Unnamed Network'
-------------------------------
Components:
 - Bus: 2
 - Carrier: 1
 - Generator: 1
 - Link: 1
 - Load: 2
Snapshots: 24

5. Check the results¶

Retrieve PyPSA statistics

In [9]:

n_stat = n.statistics.energy_balance(comps=["Generator"]).droplevel([0, 2])
n_stat

n_stat = n.statistics.energy_balance(comps=["Generator"]).droplevel([0, 2]) n_stat

Out[9]:

carrier
-    4800.0
dtype: float64

Retrieve SMS++ statistics

In [10]:

n_parsed_stat = n_smspp.statistics.energy_balance(comps=["Generator"]).droplevel([0, 2])
n_parsed_stat

n_parsed_stat = n_smspp.statistics.energy_balance(comps=["Generator"]).droplevel([0, 2]) n_parsed_stat

Out[10]:

carrier
-    4800.0
dtype: float64

Calculate difference between the two

In [11]:

error_stat = n_stat - n_parsed_stat

merged_stat = pd.concat([n_stat, n_parsed_stat, error_stat], axis=1)
merged_stat.columns = ["pypsa", "smspp", "difference"]
merged_stat

error_stat = n_stat - n_parsed_stat merged_stat = pd.concat([n_stat, n_parsed_stat, error_stat], axis=1) merged_stat.columns = ["pypsa", "smspp", "difference"] merged_stat

Out[11]:

	pypsa	smspp	difference
carrier
-	4800.0	4800.0	0.0