Development of a microgrid design tool: consumer selection, electrical sizing, and economic feasibility analysis

Abstract

This thesis presents an open and modular workflow for rural microgrid planning that com- bines geospatial processing in QGIS with MGDOt—a research-stage MILP optimiser. MGDOt is run at hourly resolution in both economic-dispatch and DC-OPF modes to generate sizing, operational, and economic outputs, as well as to test network-aware iterations. The workflow is applied to a Malawi case study under two demand scenarios: “Electrified Buildings” (206 connections) and “All Buildings” (897 connections). The load profiles used in designing the original project were reconstructed and applied in MGDOt across clusters to derive daily and annual demands for both scenarios. Core geospatial steps include generation of a minimum spanning tree (MST) as a radial template, manual obstacle-aware redrawing, pole placement, and service-drop allocation; infrastructure indicators (LV/MV line lengths, pole counts, and service-drop lengths) are computed from exported layer attributes. Reported MST totals are 10.48 km (51 m/building) vs. 18.01 km (20 m/building). MGDOt iterations indicate least-cost cases around 0.13 $/kWh (tool output tables), with network post-processing showing trans- former loading up to ∼0.605 p.u. under the full-demand scenario. The distribution grid bill of quantities yields a shift in infrastructure composition and a reduction in cost per connection from 1 520 USD to ∼632 USD (58% decrease) when all buildings are connected. The document also outlines a future rasterization-based automation of routing to avoid high-cost zones and reserve building peripheries for later service-drop optimization.