FAQs

Model

How do you forecast major and minor contingency events?

We don't forecast major and minor contingency events in the same way as way as some of our competitors. Volatility is fundamental to the battery business case so it is important to model it organically as part of the main power price model, instead of adding it in as post-processing. Our forecast is at 5-minute granularity and models the conditions that lead to volatility, as opposed to just adding volatility in. This has all been tuned with thorough backtesting as can been seen in our methodology. Volatility occurs in every single year to a certain extent, so to ensure we can capture volatility in the future, we have chosen a weather year to be used for solar/wind/demand as well as for generation and transmission outages. Our view is that a single weather year is the only way to ensure that the forecast is both realistic and transparent, but this does make it important to select the right weather year. After careful consideration of the past 10 years, we chose 2023 as our weather year, as this year was close to / slightly below average in terms of solar/wind traces, demand, temperature, and number of outages. This allows banks to have confidence in the level of volatility in our model as it is very realistic whilst being slightly conservative.

How do you forecast price spikes, and are these backtested to historical data?

Yes we have spent a lot of time backtesting at 5 minute granularity to ensure our forecast is as well-calibrated as possible, we go into our backtesting a lot more in our methodology. We have analysed our backtest in all possible cuts of the data, we have compared our average monthly prices and monthly spreads, daily price shapes, price distribution, and even spent considerable time looking at individual days to make sure we are capturing as many dynamics as possible at a settlement period level.

Does dispatch model assume the BESS is a price taker?

Our power price model co-optimises all generators (including batteries) in a linear program in order to calculate the price at each 5 minute settlement period. Our battery dispatch model then takes those prices as well as the chosen configuration of battery, and optimises to maximise its own revenue. For batteries 300 MW and under,the dispatch model is implicitly treating the battery as a price taker as it is given the fixed power prices, however the effect of this individual battery will already be accounted for in the power price model and the bidding logic we have there, based on our expectations of future BESS buildout. Historical analysis shows batteries of this scale can effectively capture the same returns as smaller systems.

Assumptions

What goes into your forward assumptions on Capex? What about Opex?

We have a house view on capex, formed from a combination of the latest GenCost inputs and our global BESS capex survey. This is validated against announced historic capex and third-party resources i.e. BNEF. For opex we use numbers from the ISP, which in turn is based on analysis by Aurecon.

How would an increase in DERs impact your forecast, especially if it increasingly participates in the market?

We expect DER and particularly DER storage to be a big sensitivity to the battery business case, but much less so than the capacity of utility-scale storage. We expect most DER batteries to be non-aggregated, and even those that are aggregated aren't going to be able to compete as aggresively as utility-scale storage. Higher capex and lower RTE means they will need a higher spread to justify cycling - therefore they will only cannibalise spreads when spreads are higher. The main effect they will have on the battery business case is a reduction in volatility events as we can expect the majority of aggregated DER batteries to be discharging at these times. As for the specific impact of DER price visibility suggested in the Nelson Review, this will allow the market to operate more effectively, but we don't see this significantly changing the dynamics already mentioned.

FCAS

Do you differentiate between global and local FCAS?

We don't currently differentiate between global and local FCAS. Increases in local FCAS requirement are small and with the rapid storage buildout we will quickly reach a point where even SA and QLD will be saturated on their own, meaning that frequency islanding will eventually not lead to local FCAS price spikes unless there is also an energy price spike. Additionally, we expect frequency islanding to get less and less frequent with further transmission buildout, such as Project EnergyConnect.

Could FCAS value grow long-term due to an increasingly unstable grid?

The massive increase in battery capacity far outweighs any potential increased FCAS requirements, which is ultimately linked to physical aspects of the network i.e. potential largest loss. This therefore is only going to continue FCAS saturation. We don’t see FCAS as providing any real additional value to large batteries in the long-term.

BESS investment case

How does near-term buildout of BESS have such an impact on spreads?

Existing batteries have been placing a lot of bids close to the price cap when the system margin is tight, as the level of capacity on the system is such that they know they will still be able to capture a lot of this value. If you can be confident that the price will at some point exceed $10,000 on a given day, then it makes sense to bid your battery close to $10,000 so that you avoid discharging when prices are lower, but make sure you can discharge and make money when prices are higher. This however only applies when you can be confident that prices will exceed $10,000 due to long periods of high residual demand combined with a battery fleet that is currently low in total rated power and also low in duration. Over the next few years we are seeing very few retirements until late 2027, but we are seeing huge numbers of batteries getting built with increasingly long durations. By the end of 2028 we expect to reach 18GW of batteries which is 6x what we have at the moment. With this number of batteries, there will be considerably more competition. In particular this will have a large suppressive effect on price spikes. Batteries will often no longer be able to bid as easily at $10,000, as the increased competition may lead to them not getting dispatched at all. This isn't a dynamic that is unique to batteries as hydro / pumped hydro also often bid in a similar way, and gas peakers (particular when part of portfolios) also often bid close to the price cap. Fundamentally this is a relationship with how tight system margin is, with a rapid increase in battery capacity leading to fewer instances of tight system margin, and therefore fewer price spikes.

Is BESS still investible if spreads do drop significantly? What role do government support schemes provide?

Our modelling highlights a risk of a short-term compression of price spreads caused by an overbuild of BESS (and transmission reinforcement) ahead of anticipated thermal retirements. The construction of Snowy 2.0 and Humelink has a similar effect in the 2030s. The largest impact from this is anticipated to be on one and two-hour batteries, as shorter-term volatility is supressed. There is a much more fundamental, long-term price shape for up to four-hours linked to the spread between low midday prices driven by solar and peak prices set by coal, and particularly gas. Based on location, and capex, this spread is still enough for four-hour BESS to reach a viable business case, which is supported by the shift in the market to this duration.

The presented spreads come from our central case, which represents what we consider to be the median outcome for revenues. Volatility in the NEM has an extremely long distribution tail, as prices can quickly increase from $300/MWh to over $10,000/MWh. This means that upside for projects is very real, especially with future uncertainty over transmission and renewable generation buildout, however this volatility needs certain conditions to occur to materialise, which becomes less likely overtime as the amount of storage increases. We'll be releasing a sensitivity analysis and scenarios for our forecast shortly.

We therefore see the role of government support schemes such as the CIS (or reforms through the Nelson Review), alongside full or partial contracting (e.g. virtual toll) as being increasingly important for new projects to provide downside protection in case of an overbuild scenario.

How much BESS does the system need? How does it impact midday prices, does it help to soak up additional solar?

More batteries will absolutely help to soak up additional solar. There is currently a lot of solar in the NEM which is cannibalising a lot of their own revenues. This is making it very difficult to form a viable business case for standalone solar. Over the next few years though, we will see much more battery capacity added than solar, which will help to bring up prices in the middle of the day and improve the solar business case again. Regardless, we believe it will be much easier to build solar co-located with batteries, rather than standalone solar. As for how much BESS is needed, this is a difficult question to answer as it depends on what timeframe you are talking about. By 2050 we have our forecast getting to roughly 30GW of batteries.