You might recall we’re trying to piece together a better understanding of sequence of events on Wednesday afternoon 5th August 2026, where there was a significant (but not ‘traditional’ frequency drop, below the NOFB).
There are five articles collated in the category ‘2026-08-05 Frequency Drop (a significant one due to data glitch)’ (with this now being the 6th article) that are us progressively exploring the event. Of these five articles:
- The fifth in the list is Linton’s ‘Initial review of unit output on Wednesday 5th August 2026’ …
- in which he reveals that there appears to be a range of issues with the unit-level SCADA data AEMO captured on a 4-second basis across a broad range of units just prior to, and during, the incident
- perhaps this had some bearing on NEMDE’s apparent hallucination that ~2,400MW of ‘Market Demand’ in NSW had disappeared at 16:20, for the 16:25 dispatch interval?
- the first in the list is Linton’s ‘Major change in NSW ‘Market Demand’ on Wed 5th August 2026 – ‘just’ a data glitch, or more?‘ …
- in which Linton probes whether the apparent aberration in metered flows on the two interconnectors joined to NSW was a factor
- including snapshots from ez2view ‘Interconnector Dashboard’ widget.
With apologies for a bit of repetition (in part to help me work through and better understand the sequence of events), I’m reposting a similar snapshot from a window in ez2view time-travelled back to 18:00 (NEM time) on Wednesday 5th August 2026:
- looking backwards just 3 hours,
- and with the tables on the right focused on the 16:20 dispatch interval
In the following table, we start to piece together a hypothesised sequence of events – with respect to four dispatch intervals in particular, on Wednesday 5th August 2026:
| DI ending 16:20 | DI ending 16:25 | DI ending 16:30 | DI ending 16:35 | |
|---|---|---|---|---|
|
NSW ‘Market Demand’, |
Looks normal …. |
In what seems to be a hallucination in the data fed into NEMDE, the ‘Market Demand’ in NSW drops by ~2,328MW. |
Demand jumps by +2,650MW … but this appears to just be largely a correction of the prior error. |
Looks normal…. |
|
Dispatch patterns, |
To be reviewed …. |
We’ve not looked directly yet, but many units were ‘ramped down’ to meet the hallucinated drop in Market Demand. |
We’ve not looked directly yet, but many units were ‘ramped up’ to meet the corrected (now ‘real’) demand. |
To be reviewed …. |
|
NEM Mainland Frequency, |
Significant (but linear) decline, to finish below the NOFB slightly after 16:25. |
Significant (and linear) return inside of the NOFB. |
More gradual return to within the PFR deadband (took until 16:40 to resolve). |
|
|
Dispatch Error, |
More to come …. |
Significant negative Dispatch Error for Coal and BESS (i.e. over-performance) representing efforts via PFR to resist the dropping frequency. |
Significant positive Dispatch Error for Coal (under-performance), still to be explored and understood. |
More to come …. |
|
Unit level 4-second SCADA data, |
For a large number of units there is an apparent aberation in the data at ~16:18. This is different for different units … |
… some units appear to be imagined at 0MW for several minutes, into the 16:20 dispatch interval. |
More to come …. |
More to come …. |
|
Interconnector flow data., |
The ‘metered MW flow’ data between 16:15 (start of this interval) and 16:20 (end of this interval) showed large changes:
Together this adds to an imagined ~2,400MW increased supply out of NSW region. |
Hallucination appears to disappear, and metered flow returns to ‘normal’ |
More to come …. |
More to come …. |
|
More to come …. |
More to come …. |
More to come …. |
More to come …. |
More to come …. |
We’ll return to this table, and progressively flesh out (in future articles) as more number crunching is done….
Without repeating in full a comment I made on Linton’s article last week (https://wattclarity.com.au/articles/2026/08/initial-review-of-unit-output-on-wednesday-5th-august-2026/#comment-822570), the real conundrum here is why Queensland and Victorian demand didn’t show corresponding “hallucinatory” jumps to offset the NSW fall. NEMDE’s demand targets start from a calculation of metered “Initial Supply”:
Initial Supply = local supply + imports – exports
If net interconnector flows out of NSW jump by 2,300 MW, then apparent NSW demand will drop, but apparent demands in the neighbouring regions should increase by about the same amount. But there were no material demand changes reported for either Qld or Vic.
@Allan O’Neil – if there was a fault on the NSW system for an extended time (> 4 s) this would lead to a reduction in load (and also generation) due to reduced voltage. After the fault cleared – both load and generation (which could have been solar PV) would recover as long as it wasn’t tripped. QLD and Vic demand would not necessarily be affected, and interconnector flows may have only been affected slightly. The possibility of a ‘data glitch’ can’t be ruled out on the evidence I have seen to date, but we should note that the fault hypothesis can also fit the known facts.
Thanks Bruce. My point is not about what might have happened physically on the power system – your hypothesis on that is certainly credible. It’s about how the market systems dealt with the SCADA data fed to them from AEMO’s Energy Management System – regardless of whether that data was physically correct or not.
The NSW market demand calculation for DI 16:25 unambiguously reflected the very large swing in interconnector flows described in the last row of Paul’s table above – an apparent change in flows of ~2,400 MW flowing out of NSW into Qld and Vic – and which is present in AEMO’s published market data. However the parallel market demand calculations for Qld and Vic, which should have used exactly the same interconnector flow data for QNI and VNI, equally unambiguously did not use that data.