The Energy Security Board (ESB), under the direction of the National Cabinet, is currently working on a number of changes that will impact how generators connect to and operate within the National Electricity Market (NEM). The Congestion Relief Market (CRM) model is one such change to how the NEM currently operates that could result in new revenue streams for battery operators providing load in constrained network areas. A detailed design of the specifics of the CRM is to be provided by the ESB in mid-2023[1].
A new market adds complexity to battery decision-making – particularly when to charge
Whilst offering another revenue source for battery storage is a great way to encourage investment into the storage market, it will likely add further complexity to the decision on when to charge batteries from the grid. Within the current market, and assuming batteries are attempting to maximise arbitrage revenues, the battery operator (or software) will seek to charge from the grid at the lowest prices and discharge at the peak. In the figure below, this would mean charging between 11am-1pm and discharging from 6pm.
With the introduction of a Congestion Relief Market, this decision will likely change as there will be an opportunity to act as a load at different times of the day and receive payments from the CRM market. For example, if the price at 10am was $115/MWh and the estimate was for a price of $100/MWh at midday, then the battery operator would be willing to charge at 10am as long as there was a difference in CRM payments between the two periods of at least $15MWh[2].
A key consideration in this calculation will be the price forecast used for future prices when the battery operator (or software) is making its decision. With this in mind, this Chart of the week looks at AEMO’s pre-dispatch forecasts[3] and how they have differed from actual NEM prices over the past year.
The chart below details the differences between the forecast of NSW prices from AEMO’s 10am pre-dispatch forecasts compared with actual prices at 10:30am, 12pm, and 2pm. Data between April 2022 and April 2023 has been used.
Figure 2 shows that around 57% of the time, the difference between actual and forecast prices at 10:30am (dark blue line) is negative, meaning the price that occurred was lower than forecast 30 minutes prior. The most common outcome for the 2hr and 4hr ahead forecasts (green and yellow lines) is a positive difference, i.e. a higher realised price than forecast. Also of interest is how rarely the pre-dispatch forecast price is close to the actual price. For example, if we use a difference of $10 between the forecast and actual price as a measure of “closeness”, then the 30-minute ahead forecast would be considered close 25% of the time, with the 2hr and 4hr forecasts being close 21% and 22% of the time respectively.
For those using pre-dispatch prices as an input into battery charge (or other load) decision-making, there may be important implications if these results persist over time. For example, if making the decision at 10am and looking at the price difference between 10:30am and 12pm, the results over the last year would indicate that the 10:30am price is likely to be lower than forecast whilst the price later in the day is likely to be higher than forecast. This may mean the operator would be better off charging earlier in the day than a “perfect foresight” model would suggest, particularly if there are CRM payments available.
Cornwall Insight Australia is hosting a webinar on 9 May 2023 discussing the proposed Congestion Relief Market and potential outcomes for those participating in the market. Registration and attendance are free; we would love to see you there.
Register here or contact enquiries@cornwall-insight.com.au.
[2] Note this is a very simplified example. In reality there will also be tradeoffs between FCAS participation and having sufficient energy stored to respond to unanticipated price spikes etc.
[3] Data is taken from nemweb.com.au – /Reports/Archive/PredispatchIS_Reports/
