Models / PyPSA-AR

Overview

PyPSA-AR is a reproducible model of Argentina's SADI interconnected system (Sistema Argentino de Interconexión), built on the 500 kV transmission backbone by Fundación Torcuato Di Tella (FTDT). We ship two ready-to-run artifacts derived from upstream's own pipeline — no Snakemake, no data wrangling:

• a solved full-year 2024 dispatch of the full 343-bus nodal network (all 8784 hours, validated vs CAMMESA), and
• a solved 2035 capacity-expansion scenario set (10-cluster), in two policy variants.

Use Cases

• Dispatch & generation mix — hourly DC economic dispatch of the Argentine grid, by technology and node.
• Grid & congestion analysis — per-line loading on the 500 kV corridors (e.g. the Comahue→Buenos Aires axis).
• Policy & scenario analysis — 2035 business-as-usual vs a winter gas-supply restriction.
• Decarbonisation & investment — optimal new wind/solar/thermal build under ATB-2035 costs.

How it works

The 500 kV network topology comes from a CAMMESA PSS/E base case, geolocated against GeoSADI (substation/line coordinates). 2024 hourly generation and demand, plus CAMMESA-derived marginal costs and efficiencies, are mapped onto it. The 2024 dispatch solves a linear DC OPF on HiGHS over all 8784 hours of 2024 (in monthly chunks). The 2035 scenarios simplify and k-means-cluster the network to 10 regions, scale demand +3 %/yr (→ ~188 TWh), add expandable generators at NREL-ATB-2035 costs with per-cluster RES caps, and solve a joint capacity-expansion + dispatch LP (TSAM-reduced to 16 typical days).

Coverage

Scenarios

• 2024 historical dispatch — the hero artifact; the model's native validation case.
• 2035 BAU — business-as-usual, unconstrained fuel; ~188 TWh demand, 35 % renewable, ~$5.75 B/yr, ~42 MtCO₂.
• 2035 Gas-Cap-60 — same topology with a 60 MMm³/day natural-gas cap (winter restriction); renewable share rises to 36 %, cost to ~$5.96 B/yr.

Validation

See the Validation tab. The 2024 dispatch reproduces the CAMMESA generation mix within a few percentage points per technology (hydro, thermal, nuclear, wind, solar), and the 2035 BAU scenario reproduces FTDT's published headline figures (demand, cost, CO₂). Bronze — first validated cut; Brazilian imports (~5.7 % of energy) are not modelled, a documented limitation.

Access & downloads

This page covers both the 2024 dispatch and the 2035 expansion — four artifacts in total.

• PyPSA .nc (free): the solved full-year 2024 dispatch and the 2035 BAU expansion network — open in PyPSA to inspect dispatch, optimal build-out, generation mix and nodal prices.
• Convexity .db (licence-gated): a 2024-dispatch database and a 2035 capacity-expansion database (BAU + Gas-Cap-60 over the shared 10-cluster topology), for interactive exploration in Convexity.

In the 2024-dispatch database, all 575 generators are resolved to their real power plants — each unit carries its CAMMESA/GeoSADI plant name (e.g. ATUCHA II, Central Genelba, Nihuil 3) and sits at its true geographic location on the map (273 distinct sites), rather than being scattered around its electrical bus. The 2035 database is the 10-cluster abstraction, so generators there sit at cluster centroids.

Provenance & licensing

Model and pipeline: FTDT — PyPSA-AR (repo). Data is public and redistributable with attribution: CAMMESA market/network data via Argentina's national open-data framework (Decreto 117/2016; topology, generation, demand, costs, GeoSADI coordinates) and NREL ATB 2035 technology costs (CC-BY). No non-commercial clause applies. Please attribute CAMMESA, NREL ATB and FTDT.

Limitations

• 500 kV backbone only — 220/132 kV networks are a future upstream stage; secondary buses are collapsed to their 500 kV parents.
• Brazilian imports are not dispatched (≈5.7 % of 2024 energy), so thermal/hydro pick up that share.
• 2035 scenarios are 10-cluster and time-aggregated (16 typical days) — directional capacity signals, not nodal operational detail.