Overview
PyPSA-AR is an open model of Argentina's SADI (Sistema Argentino de Interconexión), built on the
500 kV transmission backbone by Fundación Torcuato Di Tella (FTDT). It models generation, demand
and transmission for 2024 and solves capacity-expansion problems for future years. We rebuild it
through the upstream pipeline and host a reduced, free set of products in both Convexity .db and
PyPSA .nc formats. Open the
Convexity demo
to explore it in the browser.
The SADI network
The topology is a CAMMESA PSS/E base case of the 500 kV grid, geolocated against GeoSADI (substation and line coordinates). The full nodal network has 343 buses and 575 dispatchable generators, each resolved to its real power plant (e.g. ATUCHA II, Central Genelba, Nihuil 3) and placed at its true site — 273 distinct locations on the map, rather than scattered around an electrical bus. The 500 kV corridors (e.g. Comahue → Buenos Aires) carry the long-distance hydro-to- load flows that drive congestion. Marginal costs and efficiencies are CAMMESA-derived; carriers span hydro, CCGT, OCGT, steam, diesel, nuclear, wind, solar, pumped-hydro and biofuels.
What we build
| 2024 dispatch | 2035 expansion | |
|---|---|---|
| Network | 343-bus nodal (500 kV) | 10-cluster (k-means) |
| Horizon | full-year 2024, hourly (8784 h, solved in monthly chunks) | 16 typical days (TSAM-reduced) |
| Mode | DC economic dispatch (fixed fleet) | capacity expansion + dispatch |
| Costs | CAMMESA 2024 | NREL ATB-2035 |
The expansion scales demand to ~188 TWh (+3 %/yr) and adds ~50 expandable wind/solar/thermal generators with per-cluster renewable caps. Brazilian imports (~6 % of 2024 energy) are not dispatched, so domestic thermal/hydro pick up that share — a documented limitation. The 2024 dispatch is backtested against CAMMESA's metered generation; see the Validation tab for the per-technology detail.
Scenarios
Every .db opens with at least one what-if you can re-solve in Convexity (only the changed inputs are
stored; everything else is shared with the base case):
- Gas-price shock (×1.5) — on the 2024 dispatch
.dbs: every gas generator's marginal cost is raised 50%, so CCGT/OCGT cede the merit order to hydro, nuclear and renewables and prices climb in the hours gas would have set them. - 2035 BAU and 2035 Gas-Cap-60 — on the 2035 expansion
.db: business-as-usual versus a 60 MMm³/day winter natural-gas cap, both over the shared 10-cluster topology, at ATB-2035 costs.
Map & border layer
The .db carries Argentina's national border as a model-wide GeoJSON layer, shown in the Convexity
app's network-tree "GeoJSON" section and on the map — the orienting country outline behind the
located plants and 500 kV lines. (Buses are electrical nodes with no provincial geometry, so this is a
single national border, not per-bus polygons.)
Downloads & access
All files are free (sign-in): the solved 2024 dispatch and the 2035 expansion networks as PyPSA
.nc, Convexity .db exports for both (the 2035 .db carries both scenarios over one topology), and
a fast 1-week demo .db for a quick browser solve.
Sources
- Model and pipeline: PyPSA-AR (FTDT).
- CAMMESA: market/network data via Argentina's national open-data framework (Decreto 117/2016; topology, generation, demand, costs, GeoSADI coordinates).
- NREL ATB: 2035 technology costs (CC-BY).
No non-commercial clause applies. Please attribute CAMMESA, NREL ATB and FTDT.