Methodology Overview¶

MicroGridsPy Planning is a bottom-up, open-source optimization tool for the planning of energy systems in remote and underserved areas. The model is implemented in Linopy, a Python-based optimization framework, and provides an open and transparent approach to the sizing and dispatch of mini-grids. It explicitly addresses the challenges of energy-system scaling, technology selection, and operational planning in contexts characterized by limited data availability, high uncertainty, and strong economic constraints.

An integrated modelling platform¶

MicroGridsPy is part of a broader integrated modelling platform designed to address the interrelated challenges of rural electrification within the Comprehensive Energy System Planning (CESP) methodology — a collection of interoperable, open-source Python tools, each targeting a stage of the electrification-planning process:

Distribution-grid topology     GIS-based network design; who is grid-connected vs stand-alone
        ↓
Load demand assessment         RAMP: stochastic, seasonal, evolving demand; rural archetypes
        ↓
Resource assessment            PVGIS API: solar irradiation and wind speed time series
        ↓
MicroGridsPy optimization      optimal sizing and least-cost energy mix (this tool)
        ↓
Dispatch simulation            load-following / cycle-charging; degradation, control strategies
        ↓
Power-flow analysis            congestion and energy flows once nodes are defined

This ecosystem framing is preserved because it defines the well-specified data interfaces that let each stage remain modular — see the internal data contract.

Reference energy system and technology scheme¶

The modelled system exchanges power through a common AC system bus, the central balance point of the model. Conversion efficiencies are applied between each technology and the bus to represent losses in power electronics, electrical conversion, or system interfaces. The main components are:

• Renewable generation technologies — generic renewable units driven by resource- availability profiles and techno-economic parameters. This flexible representation covers photovoltaics, wind turbines (with optional power-curve representation), small hydro, or other renewable resources describable through a production time series.
• Battery energy storage system (BESS) — a single aggregated battery bank characterized by state-of-charge dynamics, charge/discharge limits, depth of discharge, degradation, and round-trip efficiency. Storage shifts renewable generation in time and supports reliability.
• Backup generator — a single dispatchable unit (diesel, biomass, or other controllable technology) defined through efficiency and fuel-consumption parameters.
• External grid connection (optional) — import and optional export, subject to grid availability (outages) and limited connection capacity.
• Aggregated electrical demand — the total system load, modelled as an aggregated time series rather than node-level loads.

Conceptual technology interaction scheme used in MicroGridsPy. The diagram represents the modelled connections between technologies and the common AC system bus rather than a physical electrical layout. Dashed elements (grid connection) are optional components activated through parameter settings.

Technology activation and model flexibility¶

The scheme is highly modular: individual components are activated or deactivated through parameter settings rather than structural model changes. In particular, a technology can be effectively disabled by assigning:

• Zero nominal (or maximum installable) capacity — removes the component from the system (e.g. zero generator capacity forces operation without backup).
• Unity conversion efficiencies — neutralizes conversion losses: the step remains in the formulation but introduces no losses.

Stochastic structure¶

The optimization runs at hourly resolution and can account for multiple scenarios \(\omega \in \Omega\) representing alternative realizations of time-dependent parameters (demand, renewable availability, grid conditions), each with a probability \(p_\omega\). The formulation is a two-stage stochastic optimization problem with recourse, solved in deterministic-equivalent form:

• Here-and-now decisions — system sizing and investment are shared across all scenarios, ensuring a robust design.
• Recourse decisions — operational (dispatch) decisions adapt separately to each scenario, minimizing expected cost.

Both planning modes can represent a fully off-grid (isolated) system or a weakly grid-connected one, governed by a grid-availability matrix.

Planning workflow¶

Demand + resource inputs
          ↓
Technology characterization
          ↓
Investment decisions        (here-and-now, shared across scenarios)
          ↓
Hourly operational decisions (recourse, scenario-specific)
          ↓
Economic accounting
          ↓
Least-cost system configuration

Planning modes¶

Two complementary formulations are available:

• Typical-Year Planning — a steady-state approximation described by a single representative operating year.
• Multi-Year Planning — a dynamic formulation over a horizon \(y = 1,\dots,H\) where decisions and exogenous parameters evolve over time.

Both are investment-oriented and rely on an annuity-based cost formulation.

How the methodology is organized¶