We have had our heads down working on our NEM price forecast update, and the results are in! We have updated all the usual variables: fuel price inputs, new supply forecasts and near-term constraints to name a few.
But this quarter we include a more fundamental enhancement to our analysis: the impact of large scale electricity storage in the NEM.
Here we present a summary of our main findings. While the results represent our first pass at the analysis, with plenty more refining to do, it produces a variety of thought-provoking conclusions.
– The introduction of storage into our model has caused us to significantly revise down our annual price forecasts. Prices are lower across all states, and the NEM weighted average flat load price has fallen by up to A$16/MWh.
– The impact of storage is most heavily felt in New South Wales. The future market for peaking generation is huge, and large-scale storage in the second half of the decade transforms the state’s requirement for peaking gas generation, significantly moderating evening prices.
– The main beneficiary from greater storage is coal generation – coal volumes through the day rise in all states with coal assets, but it is most pronounced in Queensland. Coal revenue however falls as a result of lower realised prices in the evening.
– Our analysis suggests that storage operators will need to bid aggressively to be dispatched outside New South Wales: sub A$80/MWh may be required given the strong competition from other sources. We therefore expect storage project developers will be banking on steep capital cost declines over the coming years.
– Our near term quarterly forecast prices have fallen compared to our previous forecast, largely as a result of cheaper gas. The trend remains downwards over time.
Gas price fall pulls down near-term prices
The shape of our quarterly price forecast remains very similar to our previous forecast: the trend is downwards with a few bumps along the way.
However, we have revised downwards our flat load prices for most states in most quarters over the next two years.
The dominant driver for our revised forecast this quarter is a lower gas input price assumption. We have revised our gas price assumption downwards in line with the recent downturn in east coast spot gas, and given the relatively low utilisation of the gas fleet on the average day, the gas generation price over the near-term falls in line with a lower short run marginal cost (SRMC).
But we note that this may not always be the case, and we remain cautious of arguments that suggest accessing cheaper sources of gas in Australia will lower electricity prices over the long term. The higher the utilisation of the gas fleet, the higher the margin will be between the SRMC of gas generation, and the prices that generators bid into the electricity market.
Over the longer term, and in the absence of significant electricity storage, gas generators could enjoy considerable market power during the evening in particular, reducing the relevance of the gas price as an influence on power prices.
Storage lowers average annual prices
Our annual price forecasts follow a now-familiar pattern. Sharp declines over the years to 2022 are followed by increases post-Liddell closure, and then steeper increases by 2030.
The main drivers of our forecast include the tension between growing variable renewable energy penetration, and coal-fired generator retirements; input fuel prices; and expansions in interconnector capacity between states.
But this quarter we introduce a major new driver, and key enhancement to our analysis: the impact of significant utility-scale storage volumes.
Our assumptions around storage – power, capacity, duration and bid/sell prices to name a few – will likely be fluid, particularly over the coming quarters, as we refine our views on both likely volumes entering the market, and potential bidding strategies for storage operations.
Storage, subject to all of our assumptions, has a clear impact on our price forecasts. We assume gradual growth in storage capacity, both from batteries and pumped hydro, from today throughout the forecast period.
Near term we make only a little storage capacity available to the energy market: between 100 and 200 MW (150 to 300 MWh) in most states. But from the middle of the decade we assume major growth in operating storage plants, most notably Snowy 2.0 pumped hydro in New South Wales. Note that total storage capacity in the market will be greater than our model assumptions, with some proportion committed to the FCAS markets as opposed to energy.
Of course our buy and sell price assumptions are key, and initially we assume that storage operators will attempt to shadow the short run marginal cost (SRMC) of open cycle gas generators. But in reality bidding will be more complex than this, and as we find in our analysis, storage may need to bid much lower to be dispatched.
In some regions – New South Wales and to a lesser extent South Australia – the impact is almost immediate.By the mid-2020s this is already taking the edge off gas-powered generation fleet utilisation, and therefore works to moderate prices indirectly by blunting gas scarcity pricing.
By 2030 in New South Wales – the highest cost region in our analysis – storage capacity is heavily and increasingly leaned on in the evening peak.
This results in storage setting the price at the margin for several intervals, while also reducing the use of gas and therefore the price, even when gas is the price setter.
And the influence of storage in New South Wales extends across the NEM, bringing prices down across the market given the increasing interconnector capacity between states.
The result of our Q1 update on annual flat load prices is a significant reduction for most years. The dispatch weighted NEM average price is lower for all periods expect FY20, and by the end of the forecast period the reduction reaches a sizeable A$16/MWh, predominantly as a result of storage.
The direct impact of our storage assumptions is much less in other states. By 2030 the growth in renewable generation, particularly wind, and the continuing strong coal generation capacity reduces the demand for storage dispatch on the ‘average day’ outside New South Wales. South Australia utilizes some of the storage we make available in our assumptions through the middle part of the decade.
But by 2030, while prices are low enough during the day to incentivise storage charging and/or pumping, the scale of VRE output is so large that our assumed storage sell price is too high to allow for any storage dispatch.
This is also broadly true for Victoria and Queensland throughout our forecast period, with the addition of course of considerable coal generation in the mix.
This suggests that storage in many instances will be reserved for specific high price events rather than daily balancing. But this also implies that the scale of required storage for states outside New South Wales will be much smaller.
Alternatively, our results may support the idea that storage operators will bid into the market at considerably lower prices than we have assumed for this analysis. In order to be dispatched in Queensland and South Australia during the evening peak, we think storage will need to bid into the pool at levels lower than combined cycle gas.
In Victoria, storage will need to directly compete with local coal by 2030. This is heavily dependent on the assumptions made around brown coal generator retirements.
In the event of storage bidding into the pool at prices competitive with combined cycle gas or even coal, storage will play a much bigger role in all states in the NEM than we illustrate here. But our concern then becomes project economics – it is hard to see storage investors earning a reasonable (or any?) return on equity with such suppressed sell prices.
The hope must be for rapidly falling capital costs before the facilities are constructed.
Winners and losers
It is clear that large scale storage will erode a significant chunk of the market for gas-fired generation in the NEM. But what about coal? Our results suggest that large scale storage in the NEM is a positive for coal generation volumes, although it is a negative for coal generation revenues.
Queensland coal gets a particular volume benefit, as a cheaper source of generation than New South Wales coal, and as there is plenty of spare interconnector capacity in the middle of the day between these states by 2030.
To an extent this may depend on whether storage is grid connected or integrated with a VRE plant behind-the-meter. But we tend to feel it will come out in the wash: if batteries charge directly from wind and solar farms, this reduces daytime supply (all else equal) and would result in more coal generation to satisfy operational demand.
As always, we are loath to draw concrete conclusions from such a complicated data set, and we consider the following risks to be the most important to understand when interpreting our price forecasts.
VRE growth. A common feature on our lists of risks, a faster or slower roll out of wind and solar can have major implications for our forecasts. We have pushed back our forecasts for near term project commissioning in response to connection difficulties and commissioning delays, especially in Victoria. We will keep a close watch on these developments, and further problems will cause our wholesale price forecasts to be higher than they otherwise would be. Conversely, a more aggressive long-term expansion of VRE than we have catered for will put downward pressure on prices.
Coal retirements. A key risk is that coal plants exit the market earlier than we allow for in our assumptions. We consider the most likely candidates to be an early closure of Vales Point in New South Wales, and Yallourn in Victoria. While we may model these scenarios in a future note, our initial view is that this will push up wholesale prices, and encourage greater investment in both VRE and utility storage.
Transmission. We assume that the relevant transmission will be constructed in time to support the VRE capacity we anticipate coming on lone over the forecast period. This includes upgrades to the QNI, VNI, and construction of the EnergyConenct link between NSW and SA. We consider the risks to these projects lay almost exclusively on the downside: delays are far more likely than faster completion. In which case prices for New South Wales will rise. Prices may fall for other states during intervals where they are exporting at maximum interconnector capacity, which may be a negative for new generation investment.
ITK provides detailed analysis of the Australian power market. We offer:
– One off price forecast updates
– Subscription forecast products
– Tailored consulting on prices, PPA reviews, asset transaction analysis and more.
David Leitch is a regular contributor to Renew Economy. He is principal at ITK, specialising in analysis of electricity, gas and decarbonisation drawn from 33 years experience in stockbroking research & analysis for UBS, JPMorgan and predecessor firms.