Running a Model¶

The practical workflow is:

1. create a project and generate templates;
2. populate and validate the inputs;
3. select the formulation and solver;
4. run the optimization;
5. inspect KPIs and detailed results.

From project to solved model¶

import microgridspy as mgp

mgp.validate_project("my_site")  # step 2
model = mgp.solve("my_site", solver="highs")  # steps 3–4
results = model.results()  # step 5

solve() reads the formulation from formulation.json, builds the appropriate model (TypicalYearModel or MultiYearModel), and solves it with the requested solver. It returns the solved model, from which you obtain a structured results object with model.results() or an xarray summary with model.results_summary().

Solver options¶

model = mgp.solve(
    "my_site",
    solver="highs",  # "highs" (open source) or "gurobi" (licensed)
    formulation=None,  # None → auto-detect; or "typical_year" / "multi_year"
)

Extra keyword arguments are forwarded to the model's solve_single_objective(...) (e.g. solver_params, problem_fn, log_file_path). At least one solver extra must be installed — see Installation.

Driving the model class directly¶

from microgridspy import TypicalYearModel

model = TypicalYearModel("my_site")
model.solve_single_objective(solver="highs")
results = model.results()

Use MultiYearModel the same way for dynamic studies.

Command line¶

The package also installs a microgridspy console command and a microgridspy-gui Streamlit workspace (with the [gui] extra). The CLI mirrors the create → validate → solve workflow; run microgridspy --help for the available subcommands.

Once solved, see Results for what the outputs mean.