Models / ISPyPSA — AEMO ISP (NEM)

Overview

ISPyPSA is the open-source implementation of AEMO's Integrated System Plan — the 20-year roadmap for the National Electricity Market (NEM). This model ships it as a ready-to-explore Convexity database: a multi-period (FY2025–2050) zonal capacity-expansion of the NEM, solved for the three canonical ISP scenarios and packaged as one switchable database.

Scenarios (switchable in one database)

• Step Change — the ISP central scenario: fast, coordinated transition in line with global climate commitments.
• Progressive Change — a slower, less coordinated transition.
• Green Energy Exports — strong electrification and a hydrogen / green-export-led expansion.

All three share one candidate topology (NEM sub-regions + renewable energy zones), so you can switch scenarios and compare optimal build-out, dispatch and costs in place.

How it works

Built directly from AEMO's IASR workbook + demand/VRE traces through ISPyPSA's public API (unmodified, pinned as a dependency), translated to a PyPSA network, and solved as a multi-period capacity expansion with the MOSEK interior-point solver. Each modelled year is represented by one demand-characteristic week (the residual-peak-demand week) at 30-minute resolution across the six investment periods FY2025 → FY2050 — a representative-period reduction that keeps the national multi-period optimisation tractable. Zonal buses are placed from a committed coordinate table so the network renders geographically.

Coverage

Status & validation

This is an indicative, planning-scale model — an open re-implementation of the AEMO ISP, not a settlement- or dispatch-accurate reproduction — and it is published without a quality tier. The simplified temporal resolution (one representative week per modelled year) produces known artifacts, so we have not yet published a validation we would stand behind as a quality rating. The system scale and decarbonisation trajectory are realistic (total generation ≈ NEM actuals, coal retires across the pathway, a renewables-dominated ~150 GW 2050 fleet), but the detailed mix diverges from full-chronology models and observed outturn: the optimiser under-builds wind (the single residual-peak-demand week under-weights wind's year-round contribution) and over-dispatches hydro (a single week carries no annual hydro energy limit). A higher-fidelity, validated build — more representative weeks, an annual hydro constraint, and comparison against AEMO ISP / OpenNEM outturn — is planned.

Access & downloads

• Three solved PyPSA .nc networks (free) — one per ISP scenario (Step Change, Progressive Change, Green Energy Exports).
• One Convexity .db (licence-gated) holding all three scenarios over the shared topology, switchable in the UI.

Provenance & licensing

Methodology + code: ISPyPSA / Open-ISP (GPL-3.0-or-later). Data: AEMO IASR workbook + traces, used under AEMO's Copyright Permissions (free use, with attribution). Please attribute AEMO and ISPyPSA / Open-ISP. A produced network is data output, not a derivative of the GPL code.

Limitations

• Zonal (sub-region + REZ) resolution, not nodal — interconnection is modelled as links between zones.
• One representative week per year (residual-peak-demand, 30-min): under-builds wind and over-dispatches hydro versus full-chronology models (see Status & validation).
• The .db is a faithful frozen export, not yet re-solvable in Convexity — the multi-period investment formulation isn't supported by the in-app solver today (single-period dispatch models are).
• Renewable-energy-zone transmission expansion is represented by per-REZ expansion generators (~48 GW of *_exp capacity), which appear without a fuel carrier in the network explorer.
• No historical-outturn validation published yet — comparison vs AEMO/OpenNEM actuals is planned.