Research & Papers

GIST releases public synthetic model of Korean power grid with 2,064 buses

First open-source, geographically accurate model of Korea's islanded grid for research.

Deep Dive

Researchers at the Gwangju Institute of Science and Technology (GIST) have released the GIST 2064-Bus Test System, the first open-source, geographically grounded synthetic model of the Korean power grid built entirely from public data. Unlike fully synthetic test cases that ignore real corridors or aggregated models that lose native resolution, this new system reconstructs the 345 kV and 154 kV backbone from OpenStreetMap and OpenInfraMap power-layer data using a multi-source shortest-path algorithm. It fills gaps with a minimum-spanning-tree approach and calibrates circuit lengths against published national statistics, achieving 108%, 107%, and 97% accuracy at 765 kV, 345 kV, and 154 kV respectively.

The final model consists of 2,064 buses, 512 generation and renewable sources totaling 144 GW, 3,044 AC line circuits plus HVDC equivalents, 3,073 transformers, and reactive resources including 11 FACTS devices. The researchers validated it by running a general-purpose pandapower Newton-Raphson solver on an 85 GW high-demand snapshot, reaching a converged operating point with mean voltage of 0.996 per unit, 2.3% losses, and no undervoltage buses. The structure is consistent with the independent KPG-193 model. The entire dataset, maps, and tooling are released as a citable platform for power flow analysis, transmission planning, and decarbonization studies—a critical resource for an islanded grid with extreme separation between generation and the Seoul Metropolitan Area load center.

Key Points
  • Model built from public OpenStreetMap/OpenInfraMap data using shortest-path reassembly and minimum-spanning-tree gap-filling.
  • Includes 2,064 buses, 512 generators (144 GW), 3,044 AC circuits, 3,073 transformers, and 11 FACTS devices.
  • Validated against an 85 GW snapshot: converges with 2.3% losses and mean voltage 0.996 pu, matching independent KPG-193 model.

Why It Matters

Enables reproducible, open-source research on one of the world's most distinctive islanded power grids for decarbonization planning.

📬 Get the top 10 AI stories daily