Victoria to NSW Interconnector, a functionally impaired asset?

In recent times daytime northern flow limits on the Victoria to NSW Interconnector (VNI) have dramatically reduced.  Figure 1 compares calendar 2015 average flow limits by time of day with those for the year from 20 February 2022 to 19 February 2023. As can be seen in 2015 the flow limits peak at around 850 MW in the evenings and then reduce to between 650 MW and 700 MW in daylight hours. In recent times the export limits on VNI have decreased such that it now resembles an inverted solar generation profile. In the early morning hours (in darkness) the average limits are around 700 MW and in the evening the limits range between 700-800 MW. In contrast, during daylight hours limits reduce rapidly and range between 300-400 MW over the course of the day until daylight wanes when they steadily increase.

If utilization and dispatch efficiency (ie, lower priced regions export to higher priced regions) were the income metrics for the asset (as opposed to regulated monopoly returns), the asset’s value would be likely to be impaired under the accounting standard AASB 136. What has changed since 2015 that has caused VNI capacity to drop when the sun is shining?

Figure 1: VNI average export limits (MW) by time of day – 20 February 2022 to 19 February 2023

Source: NEOExpress and AEC analysis.

The cause

It has been acknowledged both by the Australian Energy Regulator (AER) and the Australian Energy Market Operator (AEMO) that VNI is being regularly constrained during the day when certain solar farms are generating. This is often attributable to the infamous ‘X5’ 220kV line constraint in NSW that runs from Balranald to Darlington Point (see Figure 2). The actual name of this constraint is N^^N_NIL_3 but will be referred to hereafter as the X5 constraint. This constraint is a voltage stability limit. The X5 constraint is the most prominent of a group of constraints protecting the network in South-Western NSW from simultaneously carrying the output of solar farms in that area with flows from Victoria.

Figure 2 Partial TransGrid network map

Source: TransGrid 2022 Transmission Annual Planning Report 2022

With respect to VNI flows to NSW this constraint equation is somewhat brutal in that two very large solar farms on the left-hand side (LHS) of the equation have coefficients of one: 220 MW Limondale 1 and 29 MW Limondale 2 Solar Farms which started generating July 2020 and December 2019 respectively; and the 200 MW Sunraysia Solar Farm which started generating November 2020. Another generator has a coefficient of 0.6665, the 53 MW Broken Hill Solar Farm, which started generating September 2015. In contrast VIC1-NSW1 flows have a coefficient of 0.09327.

In summary, to achieve a 1 MW reduction on the LHS requires VIC1-NSW1 to reduce its flow north by 10.72 MW. In crude terms, all other things being equal, and the Sunraysia solar farm is generating at 200 MW and the constraint binds it effectively reduces northward flow to zero or even less, i.e. it can drive the flow southwards, against the prevailing regional prices.

It is also worth noting that if a parameter has a coefficient of less than .075 it is moved to the right-hand side (RHS) of the constraint. The VIC1-NSW1 coefficient is very close to this and if it were reduced to that level then it would result in it being allowed to flow according to the prevailing regional prices and other generators on the LHS would have to reduce output instead.

One of the key problems with the VNI is the interconnection point to Victoria at Red Cliffs, a 220 kV line which connects directly into the X5 line and is often the trigger for the X5 constraint to bind.

Figure 3 displays the distribution across the day when the X5 constraint applies. ‘Impact’ is where the NSW price is higher than Victoria’s and the VNI north flow limit is below 700 MW. ‘No Impact’ is where the NSW price is below Victoria’s price. The calculations do not include loss factors.

One could be forgiven for thinking it is the daily generation profile of a solar farm. This is not a coincidence – it is the output of solar farms constructed around its route that directly cause it. Over the one-year X5 applies 17 per cent of the time and it has an adverse dispatch efficiency impact 14 per cent of the time.

Figure 3: Count of 5min intervals when X5 constraint equation applies

Source: NEOExpress and AEC analysis

From 15 March 2023 AEMO split the X5 constraint into two constraint equations. Although as noted by Watt Clarity the coefficients of the LHS parameters of the two new equations are identical to the old X5 constraint equation. Whether this improves VNI northward flow limits is yet to be determined.

Estimating the cost of inefficient dispatch

To obtain simple estimates of the cost of the pricing inefficiencies created by VNI flows north being limited, the following analysis has been conducted. For each five-minute dispatch interval (DI) over the one-year period 20 February 2022 to 19 February 2023 (105,120 DIs or 8,760 hours) the following criteria will rule a DI into the cost calculation:

• NSW price is higher than Victoria’s price. VNI flows to NSW should be at the VNI limit when this occurs.
• VNI Export Limit (ie, flow north to NSW) is less than 700 MW. As shown in figure 1, VNI limits typically sat at about 700MW in 2015 prior to the construction of the solar farms. This is an estimate of the amount of generation from Victoria that could have been available to reduce prices in NSW but were not due to the export capacity being reduced.
• Loss factors have not been accounted for.

The price differences are then multiplied by (700 minus VNI Export Limit) for that DI. These values are summed and then divided by 12. The total dispatch impact cost estimate is $145 million. In total this event occurs in 37,483 DIs (3,124 hours) which represents 36 per cent of total DIs in the sample period. The average volume weighted price difference is $116/DI and the maximum is $15,555/DI.

Of the $145 million cost, $56 million is recorded when the X5 constraint applies, noting that most of the rest of the time VNI is limited by constraints with similar characteristics.

Figure 4 displays the distribution of the price differences between NSW and Victoria when the event occurs noting that under the first criteria (above) they will all be positive.

Figure 4: Distribution of price differences NSW minus VIC when VNI north limit less than 700 MW (log scale y axis)

Source: NEOExpress and AEC analysis

This approach implicitly assumes that if VNI was able to flow at 700 MW to NSW, prices in NSW and Victoria are unchanged. While this is an unrealistic assumption because it would be expected that the spread between NSW and Victoria’s prices might narrow, the purpose of this paper is to highlight the issue and provide an upper bound cost estimate for inefficient dispatch.

To provide some sensitivity assessment, the 700 MW assumption is reduced to 600 MW. This reduces the dispatch efficiency cost to $118 million and occurrences to 28 per cent of the DIs in the sample period. Reducing it further to 500 MW reduces the cost to $96 million and occurrences to 21 per cent of DIs. Even with these reductions to the assumption of a ‘normal’ limit on VNI the estimated dispatch inefficiency cost is close to or well above $100 million over the course of a year.

Negative Residues

These constraints are so severe that they frequently force VNI to flow “Counter-Price”, which means that the VNI northern limit becomes negative and power is forced from the NSW into a lower priced Victoria. In other words, to safely accommodate the output of the solar farms north of the border, AEMO must flow an even larger quantity of power southwards across the Murray River and incur a trading loss known as a “Negative Settlement Residue”.

Figure 5: Counter-Price Flows on VNI

Source: AEMO Quarterly Energy Dynamics Q4 2022

That trading loss has to be recovered from somewhere of course, and AEMO charges the importing transmission company who then recovers this from customers in network fees. Victorian customers are therefore paying for the costs of protecting the NSW network following its recent connection of some very poorly located generators.

Figure 6: Victoria funds the vast majority of recent NEM Negative Residues

Source: AEMO Quarterly Energy Dynamics Q4 2022

Those negative residues would have been even worse except that AEMO “clamps” VNI to zero flow once a daily accumulation exceeds $100,000. When this occurs, the solar farms are constrained back instead of forcing VNI to flow backwards. This explains the downward sloping shapes in figure 5 – the effect of AEMO accumulating negative residues in the morning followed by clamping in the afternoon.

Meanwhile, the solar farms’ incentive to continue to locate and generate in this area remains strong, as they are paid the NSW price, which is determined from the conditions at the Sydney West node, many transmission links and hundreds of kilometers away. In a strange twist of dispatch mathematics, by blocking flows from Victoria to NSW in this way, these solar farms marginally increase the NSW price that they are paid.

Conclusion

While this analysis is relatively simplistic it does illustrate the magnitude of the costs to the market caused by northward flows on VNI being limited. Furthermore, it illustrates the perverse outcomes that can be created in the National Electricity Market’s open access and state-based pricing regime ie, the locations of a handful of solar farms can significantly and regularly impair the operation of a major interconnector that customers pay for.

Considering the apparent scale of inefficiency, it is surprising how little attention is paid to the impact of constraints on VNI northward flows by the market bodies and how it came to be. AEMO has been mentioning the constraint in its Quarterly Energy Dynamics report for several years, and figures 5 and 6 above are drawn from these. Meanwhile, the AER’s equivalent quarterly report finally noted poor VNI performance on page 37 of its Q4 2022 report, but did not explain its causes.

The AER is responsible for regulating networks and seeing that they support the energy market efficiently.  Perhaps in conjunction with the market operator, the AER should conduct a more thorough analysis to determine a more accurate measure of the costs to the market and what learnings can be drawn to lessen such inefficiencies in the future.