Earlier this week it was brought to our attention that the latest AEMO Integrated System Plan (ISP) modelling includes materially higher assumed wind capacity factors compared to previous iterations — an input that carries significant weight in modelled outcomes.

In this short article, we’ll outline what the changes are, and what prompted the AEMO to update their assumptions. We’ll also clarify some of the terminology surrounding availability and production. Finally, we’ll use unit-level data from our recently released Generator Statistical Digest 2025 report — to see how close (or far away) those assumptions are from the current reality.

The changed assumptions

A LinkedIn post by David Leitch earlier this week drew attention to the fact that wind capacity factor assumptions had shifted markedly between the 2024 and draft 2026 versions of the ISP. The table from David, reproduced below, shows the scale of the changes across several Renewable Energy Zones (REZs), with a number of zones now assigned capacity factors exceeding 40%.

Assumed wind capacity factors were revised sharply across several REZs in the latest ISP, with multiple zones now modelled with assumed wind capacity factors above 40%.

Source: David Leitch

The full set of updated assumptions is available in the AEMO’s 2025 Inputs and Assumptions Workbook, and the wind traces are accessible at the bottom of the Draft 2026 ISP consultation webpage.

Why did the assumptions change?

Our team spends most of our time in the weeds of the day-to-day operations of the NEM, so we don’t always have the time to follow and track each round of ISP consultations closely. But seeing David’s table above raised a few eyebrows in our office, especially coming so soon after our recent review of VRE curtailment and production across last calendar year.

To get myself up to speed, I spent this morning reading through each of the submissions to the consultation process related to this change.

A small number of submissions advocated for increased wind capacity factor assumptions for the South-West NSW REZ, (a.k.a Zone N5) in particular, arguing that previous assumptions were too low, compared to on-site wind measurements and independent energy-yield assessments that they were seeing, with at least one submission arguing that the the AEMO should use different power curves in their modelling. For reference to the earlier table, below is a map of all current REZ candidates from the Draft 2026 ISP.

Source: AEMO Draft 2026 ISP Appendix A3

In the consultation summary report, AEMO acknowledged these concerns and summarised the feedback as follows:

(Stakeholders shared) concerns that the ISP remains highly sensitive to wind capacity factor assumptions – to improve accuracy and reduce this sensitivity, recommend allowing confidential submissions of high-quality, on-site wind measurement data to calibrate AEMO’s wind resource maps, conducting targeted sensitivity analysis on key REZs, and refining turbine power curve selection based on modelled wind speeds.

– 2025 IASR Consultation Summary Report, Pg. 21

They concluded that a methodology revision was warranted:

AEMO has now reviewed the basis and inputs for the wind trace data used for the market modelling, and the results show that the wind quality data for the South-West NSW REZ were lower compared to what they would have been if higher hub height for new entrant wind farms was assumed and higher granularity data set was used. This updated methodology and data source has also now been applied to other REZs to ensure consistency.

– 2025 IASR Consultation Summary Report, Pg. 74

Understanding capacity, availability, and production

Before comparing those new ISP assumptions against real-world data,  it’s worth reinforcing three distinct concepts about unit operations that are not as widely understood as they perhaps should be.

Due to historical norms, many in the industry tend to describe energy assets in terms of nameplate capacity. In the old world of coal and gas plants, that framing made more sense, but for VRE it’s often an irrelevant — and misleading — descriptor for the asset. A more useful way to think about things is through three nested concepts:

1. Capacity: Generally speaking, this is the maximum amount of power the facility can generate. For a dispatchable plant such as a coal, gas, hydro or a battery unit, this is a meaningful operational ceiling as they can (and often do) operate at their full capacity when instructed. For VRE, this measure is arguably far less relevant, as they reach their full capacity more rarely. As a side note, there are a couple of different measures of capacity used by the AEMO, and Paul has previous written on WattClarity about when each should be used.
2. Availability: Conceptually refers to how much power the unit could theoretically produce given actual on-site resource conditions — for example, wind speed in the case of a wind farm. The availability factor is availability divided by capacity over time. In order to dispatch the market every 5-minutes the AEMO needs an availability value for each semi-scheduled wind and solar farm, and Linton has previously written on WattClarity about how this availability value is determined.
3. Production (or output): How much power the unit actually generated, after accounting for economic and network curtailment. The term capacity factor is therefore a bit of a misnomer: it’s simply, actual production divided by capacity over time. If anything, it should probably be referred to as a “production factor” — but the energy sector has no shortage of terminology that doesn’t match to what is being described.

Below is a screenshot taken from our GSD2025 report which illustrates the distinction between these three concepts for a single wind farm.

The 2025 operating profile for the Murra Warra Wind Farm (Stage 1) — illustrating a common gap that exists between capacity (the horizonal lines in each chart), availability (left chart), and actual production (right chart).

Source: GSD2025

In the example above, most people would describe Murra Warra 1 as a 225 MW wind farm — referring to its maximum capacity. But if we instead framed things in terms of its average availability, it would be an 84 MW wind farm (based on 2025 data). And if we framed things by its average actual production last year, we’d call it a 53 MW wind farm.

Each layer of that reduction — from capacity to availability to output — reflects different types of losses, each of which matters for comparing operational data with model assumptions.

Comparing those ISP assumptions to 2025 real-world data

With the terminology out of the way, we can now look at how those updated ISP assumptions stack up against how wind farms in the NEM have actually been operating in recent times.

The map below draws from our GSD2025 data extract and shows the distribution of availability factors (availability-based) and capacity factors (output-based) across all semi-scheduled wind units operating in the NEM in 2025.

The median capacity factor across all 84 active semi-scheduled wind units in the NEM in 2025 was 31.18%. The median availability factor was 35.35%.

Source: GSD2025 Data Extract

Confusingly, the ISP uses the term ‘capacity factor’ to refer to the renewable resource quality — because the model is trying to solve for an ‘optimal development path’ — in that sense their definition of ‘capacity factor’ is more akin to an availability factor.

But even with that interpretation, the ISP assumptions still don’t resemble what existing operating wind farms delivered on an availability basis in 2025.

In fact, of the 84 semi-scheduled wind units that were operating in the NEM in 2025, only 8 had an availability factor greater than 40% for 2025. The median was 35.35%.

In that earlier table, we saw that there are at least four REZs with modelled capacity factors in excess of 43%.

As we noted in our recent curtailment review, combined network and economic curtailment has been growing at a faster rate than underlying VRE availability or production over the past couple of years — which also means that there is a growing gap between capacity and availability factors in the real-world.

Final remarks

The AEMO seems willing to incorporate feedback in the ISP process — and these revisions look like a clear example of that feedback loop at work.

There may be plausible arguments for why wind capacity or availability factors could lift above today’s levels over time. Those handful of submissions to the AEMO pointed to higher hub heights, better matching of rotor diameters to sites, etc. — and perhaps there is some wishful thinking surrounding upcoming network enhancements such as PEC Stage 2.

That said, evidence to date isn’t as clear that we are on an upward trajectory of increasing efficiency across the wind fleet. A little over two years ago I wrote an article for WattClarity titled “Bigger or better: Are newer wind farms outperforming older ones?“. The TL;DR for that analysis was that, up to that point in time, there was no conclusive evidence that newer wind farms in the NEM were delivering materially higher capacity factors than those built 5, 10, or even 15 years earlier. This implied that any technological gains in wind development appeared to have been offset by growing curtailment, site scarcity, and/or other forces over the long-term.

The ISP inevitably has to simplify the world, but if it’s intended as a practical roadmap of where we want to go, it’s difficult for me to see why these assumptions should be so far removed from where we are now.

What are your thoughts? Do you agree or disagree? Leave a comment below.