OptiGrid: BESS in Australia’s NEM ‘rarely go fully contracted’

This approach features a high-risk profile with no protection against losses if market conditions deteriorate or if the asset’s performance is affected by unfavourable pricing spreads. The high-risk profile is reflected in proportionally higher revenue share for the operator/owner.

Meanwhile, a fully contracted BESS operates under long-term agreements that provide guaranteed revenue streams. This model offers plannable, predictable revenue through fixed payments regardless of market conditions.

Contracted systems offer complete protection against market downside risks, with the counterparty (often a utility) assuming market risk.

The trade-off for this security is a cap on potential returns, as the battery owner/operator cannot capture additional value during extreme market events beyond contract terms. The counterparty typically controls dispatch decisions, determining when the battery charges and discharges.

When prompted on the best optimisation strategies for BESS operating in the NEM, Karimi quickly emphasises that it’s not all about maximising revenue; some are more interested in providing grid stability services.

“Not every operator wants to maximise revenue. Some operators want to manage risk. For example, for gentailers, their commercial objective is not they would like to make some revenue,” Karimi says.

“Each operator would probably have their own preferred risk level. Someone wants to go aggressive with the optimisation strategy and maximise every little bit of the revenue they can capture, but some want to be more risk-averse.

“I think what we’ve seen, and what really resonates with the industry, is that the optimisation strategy should be tailored for each customer, each developer.”

Delving deeper into optimisation strategies, Karimi references the power and opportunity that AI and machine learning can have on maximising revenue, particularly with how the NEM operates.

Indeed, one of the NEM’s most important aspects is its five-minute settlement period. Under this rule, both dispatch and settlement occur at five-minute intervals.

This means the price set every five minutes is the actual price paid by retailers and received by generators, creating a direct financial incentive to respond to real-time market conditions. Five-minute settlement (‘5MS’) was introduced in 2021, eliminating the averaging effect that previously diluted price signals and sometimes resulted in generators receiving payments lower than the actual dispatch price.

Karimi believes the five-minute market and the volatility in the NEM requires asset owners to update their forecasts quickly.

“Since the NEM operates on a five-minute trading cycle, it is highly volatile. To effectively update forecasts and predict electricity prices rapidly, it is essential to use machine learning and AI models. This approach optimises battery operation based on these price forecasts,” Karimi tells ESN Premium.

“These should be designed for the specific operator in a way that suits their kind of commercial objectives and their risk management strategy.”

Karimi adds that the NEM’s five-minute settlement structure makes AI and machine learning particularly valuable, as traditional methods struggle to process and respond to such rapid market changes.

Beyond trading optimisation, Karimi notes that fault detection is another important application area. Machine learning models are increasingly being deployed for asset diagnostics and predictive maintenance, analysing vast datasets to identify potential battery cell issues before they become critical problems.

Karimi will present a case study on the deployment of a 5MWh BESS at the 4.1MW Hepburn wind project in Victoria at Solar Media’s upcoming Battery Asset Management Summit Australia 2025, which will be held from 26-27 August in Sydney. You can get 20% off your ticket using the code ESN20 at checkout.