AUTOMATED FRAMEWORK FOR CONVERTING UNSTRUCTURED GEOSPATIAL ELECTRICAL DISTRIBUTION DATA INTO SIMULATION-READY POWER SYSTEM MODELS

Abstract

Modern distribution utilities increasingly require simulation-ready power system models to support voltage regulation, loss assessment, network expansion planning and Distributed Energy Resource (DER) integration. However, many utilities maintain feeder records as visualization-centric KMZ archives whose electrical attributes are embedded in inconsistent HTML-like metadata and whose line geometries lack reliable topological connectivity. This creates a data-to-model gap in which engineers manually interpret GIS records and rebuild feeder models before any power-flow study can be performed. This study addresses that gap by developing and validating an automated Python-based framework for converting unstructured geospatial distribution data into parameterized, simulation-ready power system models. The framework combines schema-on-read Regex parsing for metadata extraction, metadata-driven asset classification, Haversine-distance graph construction, Minimum Spanning Tree (MST) topological healing, Dijkstra-based bus-branch abstraction, conductor and transformer parameterization and capacity-based load allocation. Validation was conducted in two stages, a controlled 4-bus proof-of concept system and a real 11 kV KPLC feeder. The results show that the pipeline converts noisy KMZ records into solvable MATPOWER and OpenDSS models, restores feeder connectivity from fragmented geospatial records and produces cross platform voltage results that agree within the reported 0.1% benchmark. The real feeder case further demonstrates practical value by reducing modelling effort from approximately 40 man-hours to about five minutes and identifying 18 transformers absent from the legacy manual model. The main contribution is therefore a reproducible, low-cost methodology that bridges visual GIS archives and engineering simulation tools, advancing both academic knowledge on automated distribution modelling and utility practice in data-constrained contexts.