We use a chart to see how semi-scheduled unit availabilities changed their origin. With interruptions to SCADA across NSW, many of the inputs that an availability forecast depends on could have been impacted.
Adding to the list of articles we've collated about the frequency disturbance on Tuesday 19th August 2025, we're copying in what AEMO published in its 2025 Q3 monitoring report about the glitch.
The (concerning) frequency spike across the NEM in the middle of the day (Tuesday 19th August 2025) has triggered many proximate questions - understandable - but also has raised these 3 Broader Questions.
Yesterday's release of the draft report of the Nelson Review has prompted me to re-frame a question I've asked before about the Semi-Scheduled category.
It was now just over 5 years ago when Marcelle wrote ‘Not as simple as it appears – estimating curtailment of renewable generation’. This week's invocation of the 'N-DPWG_63_X5' constraint set provides a great case study to illustrate some of the complexities!
A sequence of afternoon intervals stand out because the forecast appeared to be biased low – self-forecasts suddenly dropped roughly 30-40 percentage points and then increased a short time later.
Theoretically, if a self-forecasting system never offers forecasts for more than 60% of intervals it may perpetually skip the performance assessment and the system could continue for use unsuppressed.
Taking a guess at frequency need to earn a positive causer-pays factor through self-forecast biasing appears at-best uncertain in the intervals we review.
In another style of biasing a self-forecast, "lunar megawatts" represent an expectation of solar farm generation at night when it really should be zero.
The forecast differences would contribute to improved lower RMSE and MAE scores, relative to AWEFS_ASEFS, in the weekly performance assessment.
In today’s article (part 1 in this series) we present an example of biasing (at an unnamed solar farm), which we find aligns with FCAS cost mitigation.
We summarise how wind units are using self-forecasting to-date. The analysis leads us to consider where upcoming market change may lead the industry.
I've promised to write this article for some time, but a few recent events (and looming start of FPP) have caused me to publish this now.
Two weeks after the AER published its updated Compliance Bulletin (and Compliance Checklist) for Semi-Scheduled units, we've finally found time to note about it.
Semi-scheduled generators shouldn't treat self-forecasts as commercial parameters. The regulator holds significant concerns with the biasing of self-forecasts for commercial benefit.
A case study, drawing heavily on various widgets in ez2view, looking at some large (and unexpected) oscillations in output at Darlington Point Solar Farm through a ~2 hour period on Thursday 29th August 2024.
Gate Closure #3 represents the cut-off time for solar and wind availability forecasts to be used in dispatch. Since 2019 the timings are reducing and narrowing.
Our GSD2023 was released just over 4 weeks ago. This is the second sequential article utilising the GSD2023 Data Extract to take a look across all the Semi-Scheduled units (i.e. Wind and Solar). In this case looking at use of Self-Forecasting, and revealing more questions...
In early February 2024, we (at Global-Roam Pty Ltd, providers of this WattClarity service) were pleased to be able to come onboard as a shareholder of ETSI.
This morning I was fortunate to listen into an ESIG Webinar with respect to their report 'Weather Dataset Needs for Planning and Analyzing Modern Power Systems'