Solar curtailment in Australia is at record levels. A RenewEconomy analysis put utility-scale solar and wind curtailment in the NEM at roughly 7.9 TWh for the year to 10 April 2026, with about 82% of it occurring while prices were negative. Most of that is economic curtailment, where the plant chooses to back off, rather than network curtailment, where AEMO's constraints force it down, and the two need different fixes.
For an asset owner, the distinction matters more than the headline. A plant losing energy to negative prices has a commercial and storage problem. A plant losing energy to a binding constraint has a network problem. Treating one as the other wastes capital.
This article sets out the current numbers, explains the split, and walks through five engineering fixes for solar farms that are already operating.
The Numbers
The figures below come from industry analysis of AEMO data published during 2025 and 2026. Treat them as indicative. They move month to month, and different analysts measure curtailment in slightly different ways.
| Measure | Reported figure | Source and context |
|---|---|---|
| Utility solar and wind energy curtailed | About 7.9 TWh | Year to 10 April 2026, RenewEconomy analysis |
| Share of curtailment during negative prices | About 82% | Same period and source |
| Solar curtailment as a share of available output | About 22% | September 2026, Modo Energy analysis |
| Average solar capture price | About $17/MWh | September 2026, Modo Energy analysis |
| Regions with the most negative midday prices | South Australia and Victoria | September 2026, Modo Energy analysis |
| Negative-price hours in a single month | SA 351, VIC 290, QLD 189, NSW 159 | December 2025, PV Tech analysis |
Two things stand out. First, curtailment is concentrated in the middle of the day in spring, when rooftop and utility solar together exceed what the market can absorb. Second, the regions with the most negative-price hours are also the regions with the highest curtailment. That is not a coincidence. Most curtailment is a price response.
The other change is the value of certificates. Large-scale generation certificate prices have fallen a long way from their 2022 levels. When a certificate was worth tens of dollars, a solar farm could keep running through moderately negative prices and still make money. With LGC prices far below their 2022 levels, that cushion is much smaller, so plants self-curtail at prices much closer to zero.
Economic vs Network Curtailment: Why the Fix Differs
Economic curtailment happens when the plant's own bid, or its trader, takes it offline because the price it would receive is below what it earns by generating. Network curtailment happens when a constraint equation in AEMO's dispatch engine binds and caps the plant's output, whatever the price. We covered the theory, and how both interact with marginal loss factors, in our guide to MLF and curtailment risk.
The practical consequence is simple. Energy lost to economic curtailment can be recovered by moving it in time, which points to storage and smarter bidding. Energy lost to network curtailment can only be recovered by changing the network, the plant's position in the constraint, or the load next to it. Before choosing a fix, you need to know which kind of curtailment you have and in what proportion.
Fix 1: BESS Retrofit to Shift Energy
Adding a battery is the most direct response to economic curtailment. Energy that would have been spilled at midday is stored and exported into the evening peak, when prices are typically positive and often high. Our article on hybrid solar and BESS in Australia covers the new-build case. A retrofit has extra constraints.
A retrofit can be AC-coupled or DC-coupled, and either way it alters a connected generating system, so expect updated power system models, an amended performance standard and new commissioning tests. We cover the coupling choice, sizing and the connection process in adding a battery to an existing solar farm.
Fix 2: Bidding and Self-Curtailment Set-Points
Many plants curtail at the wrong price. The break-even point for a semi-scheduled solar farm is roughly where the regional price, multiplied by the plant's MLF, plus the value of the certificate and any PPA top-up, falls below zero. If that calculation was set when certificates were expensive and has not been updated, the plant may be generating into prices that lose money, or curtailing when it should run.
Practical steps:
- Recalculate the break-even price with the current MLF, current certificate value and the actual PPA terms. Some PPAs pay a fixed price regardless of spot. Others include negative-price clauses that change the answer completely.
- Bid in price bands that reflect the real break-even, rather than a single floor-price band that keeps the plant running through every negative interval.
- Check that the plant actually follows its dispatch target when it rebids or self-curtails. Slow or uneven response turns a correct commercial decision into lost energy or conformance issues.
This is the cheapest fix on the list, and it is often where an assessment finds the quickest gain.
Fix 3: Plant Controller and Ramp Settings
When a constraint binds, AEMO's dispatch engine sets a cap on the plant's output, and the plant controller has to hold that cap. How well it does so affects how much energy the plant keeps.
- Response accuracy. A controller that undershoots its dispatch target spills energy every interval the cap applies. Tuning the active power loop so the plant tracks the target closely, without overshoot, recovers a small but steady share.
- Ramp rates. Conservative ramp limits in the controller or the bid slow the plant's return to full output when a constraint relaxes. Check that the ramp settings match what the performance standard allows, not a default left over from commissioning.
- Availability forecasting. Semi-scheduled plants are dispatched against a forecast of their available output. A poor forecast can lead to targets lower than the plant could deliver. Some owners have improved this with a self-forecast, which AEMO allows once it meets accuracy criteria.
- Reactive power. Where curtailment is driven by voltage rather than thermal limits, the plant's reactive capability, used properly, can sometimes reduce how often the constraint binds. This needs studies to confirm.
None of these removes a constraint. Together they reduce the energy given up while one is binding.
Fix 4: DC Oversizing and Clipping for New Stages
For an operating plant, the DC/AC ratio is fixed. For a second stage or an expansion behind the same connection, it is a design choice worth revisiting.
Australian utility solar has typically been built with DC/AC ratios around 1.2 to 1.35. Higher ratios increase clipping at midday, but midday is exactly when prices are lowest and curtailment is highest. The extra DC capacity adds output in the morning and late afternoon shoulders, when the energy is worth more.
The trade-off is now different from when most existing plants were designed. Clipped midday energy costs little if it would have been sold at a negative price. The added shoulder energy is worth more than it used to be. Combined with a DC-coupled battery that can absorb what would otherwise be clipped, a higher ratio can lift revenue per MW of connection capacity. It needs an hourly price and curtailment model, not a yield model alone.
Fix 5: Network-Side Options
Where the data shows network curtailment is a large share of losses, the fix sits outside the fence.
- Review the constraint equations. AEMO publishes the equations that bind on your plant. Check which ones bind, how often, and what coefficient your plant has in each. Sometimes a constraint is driven by an outage or a conservative limit that can be raised with the network service provider.
- Connection and network upgrades. A larger transformer, a line upgrade or a reconfigured connection can lift a local limit. These are expensive and slow, and the benefit can be shared with neighbouring generators, so the economics need care.
- Control schemes. In some cases a runback or generator tripping scheme lets the network run closer to its limit by acting only after a contingency, rather than holding plants down beforehand. This depends entirely on the NSP's appetite and the network.
- Co-located load. A load connected behind the same connection point, such as an electrolyser, data centre or industrial process, absorbs energy before it reaches the constrained network. This changes the plant's commercial model and its connection arrangements, so it is a long-term option.
Network fixes take the longest. They are also the only fixes that address the root cause of network curtailment.
How to Decide: A Curtailment Assessment Workflow
A sensible assessment follows the energy before it chooses the fix.
Curtailment Assessment Steps
- Collect data: 12 to 24 months of plant SCADA, inverter availability and AEMO 5-minute dispatch data for the DUID
- Rebuild available energy: model what the plant could have produced in each interval from irradiance and availability
- Split the losses: allocate the gap to economic curtailment, network constraints, clipping, outages and other losses
- Value each loss: price each interval at the regional price, the plant's MLF and current certificate value
- Screen the fixes: test bidding changes first, then controller tuning, then a BESS retrofit, then network options
- Model the shortlist: run hourly revenue models and connection studies for the options that survive screening
Done well, this tells you how much curtailment is costing, where it comes from and which fix pays. It also gives lenders and the board a defensible basis for any capital spend.
Request a Curtailment Assessment
ElectraGlobe works with owners of operating solar farms to quantify curtailment, separate its causes and test the fixes, from bidding set-points to BESS retrofits. Where a fix changes the connection, our grid connection studies team carries it through the NSP and AEMO process, including models, studies and the amended performance standard.
If your plant is losing more output than the financial model assumed, contact our team and we can tell you quickly where the energy is going.
Frequently Asked Questions
How much solar is curtailed in Australia?
A RenewEconomy analysis reported about 7.9 TWh of utility solar and wind curtailment in the NEM in the year to 10 April 2026, about 82% of it during negative prices. Modo Energy's analysis put solar curtailment at around 22% of available output in September 2026, with South Australia and Victoria seeing the most negative midday prices. Figures vary by month, region and how each analyst measures it.
What is the difference between economic and network curtailment?
Economic curtailment is when a plant reduces output because the price it would receive is below its break-even, usually during negative prices. Network curtailment is when an AEMO constraint equation caps the plant's output to keep the network within limits, whatever the price. Economic curtailment can be addressed with storage and bidding. Network curtailment needs network, controller or load-side changes.
Does adding a battery reduce solar curtailment?
Yes, for economic curtailment. A battery stores energy that would have been spilled at negative prices and exports it later, when prices are higher. A DC-coupled battery can also capture clipped energy. A battery helps less with network curtailment if the evening export is limited by the same constraint, so the curtailment split should be understood first.
Are solar farms compensated for curtailment?
Generally, no. Generators in the NEM do not hold firm access rights to the network, so output lost to a binding constraint is not compensated. Economic curtailment is a commercial choice by the plant, so there is nothing to compensate. Some PPAs allocate curtailment risk between the parties, so the contract terms matter.