Planning Modes¶

MicroGridsPy supports two planning formulations. The choice is recorded in formulation.json at project creation and drives which model class is used.

Mode	`create_project(formulation=...)`	Main use	Time representation
Typical-year	`"typical_year"`	screening / steady-state studies	one representative year
Multi-year	`"multi_year"`	long-term planning	explicit horizon and investment steps

Renamed in 0.4 — breaking change

These two formulations were previously called steady_state and dynamic, and the typical-year model class was SteadyStateModel. The names above are now the only ones accepted, in formulation.json, in the API and in the CLI.

Projects created before 0.4 must be updated. Change core_formulation in the project's formulation.json from steady_state to typical_year, or from dynamic to multi_year. Passing an old name now raises InputValidationError rather than being silently translated. In code, replace SteadyStateModel with TypicalYearModel.

Practical differences¶

Typical-year minimizes an equivalent annual cost for a single representative year that is assumed to repeat. It is the fastest option and is ideal for feasibility screening, technology comparisons, and multi-scenario studies where tractability matters. Intertemporal discounting does not affect sizing in this mode.

Multi-year represents an explicit horizon \(y = 1,\dots,H\) with year- and scenario-dependent inputs. It supports capacity expansion across predefined investment steps (with non-decreasing installed capacity), technology-specific WACC-based annuities, a social discount rate, and cohort-based annuity persistence (implicit like-for-like replacement) over the modelled horizon.

Selecting a mode in code¶

import microgridspy as mgp

# typical-year
mgp.create_project("screening", formulation="typical_year", resources=["solar"])

# multi-year with a 20-year horizon and four 5-year investment steps
mgp.create_project(
    "expansion",
    formulation="multi_year",
    horizon_years=20,
    capacity_expansion=True,
    investment_steps_years=[5, 5, 5, 5],
    resources=["solar", "wind"],
)

solve(), load_results() and load_inputs() auto-detect the formulation from formulation.json, so you don't need to repeat it. For the mathematical distinction, see Methodology → Planning Modes.