It’s happened before … but it’s not often that WattClarity becomes a reference for discussions in Federal Parliament. One of our readers flagged to us that this was the case in today’s session (which is recorded on Youtube here) of the Senate Committee on Energy Planning and Regulation in Australia:
For those who want to replay, suggest you watch around the ~04:48:00 mark.
The topic of discussion at that time was what units in NSW were offline through Wednesday 27th November 2024 … which:
1) was a time of tight supply-demand balance in NSW – including:
(a) Actual LOR2 and
(b) forecast LOR3; and
(c) triggering of Reserve Trader (a.k.a. RERT)
2) But also a period where I was attempting to ignore* the NEM and take a short break, leaving the articles on the day (collated here) to others**.
* with only partial success as those alerts were very persistent!
** 100% success rate with me not posting articles, at least.
3) It’s worth singling out two particular articles relevant to this one:
(a) Adam’s article ‘A chronological record of events in NSW on Wednesday 27th November 2024’ is a useful starting point for putting it all in context; and
(b) Dan’s article ‘Which units are still offline in NSW and QLD, as of 3:30pm on 27th Nov 2024’ is a useful reference … and might even be the article Matt Canavan referred to in the recording with respect to what happened on the day.
Earlier today I’d written about ‘The extended outage at Eraring unit 3 (as at Thu morning 5th Dec 2024)’ (incidentally an article that Matt Canavan had referred to in the recording above) and had noted an intent to take a look back at what I’d missed.
Given the topic above was related to which units in NSW were offline*, (including whether planned or unplanned) I thought it would be useful to explore here…
* and also given the growing number of questions we’ve been receiving on the broader topic of supply-side unit outages (another article is coming on that one)
(A) What Daniel Westerman noted today in the Senate Committee
At ~04:38:40 in the video above, Daniel Westerman notes his understanding was that, on Wednesday 27th November 2024, there were 5 significant units out in NSW. He says:
1) Three units were on planned outage:
… ER03, BW02 and TALWA1
2) Two units were on unplanned outage:
… BW03 and VP6
(B) What is shown in the ‘Generator Outages’ widget in ez2view
Opening my own copy of the ‘Generator Outages’ widget in ez2view I have chosen to:
1) time-travel back to 15:30 (NEM time) on Wednesday 27th November 2024:
…. which was the same dispatch interval used by Linton in his article written on the day;
2) but I’ve chosen to filter differently (i.e. zooming into just NSW units, and bringing the cap down to 100MW to include some smaller ones) time-travelling back to 15:30 (NEM time) on Wednesday 27th November 2024:
As a result of this, we see the following:
… for ease of reference, I’ve highlighted the 5 units nominated.
(C) It’s not just black and white (or grey and red), in relation to Planned or Unplanned
Without falling into that rabbit hole, readers should understand that there is considerable nuance required in understanding whether a unit is on a ‘planned’ or an ‘unplanned’ outage … and even that two equally capable people might describe the same outage in that grey area between black and white differently.
From 9th October 2023, the data set published by the AEMO in the MT PASA DUID Availability data space was significantly* enhanced.
* alas not for Semi-Scheduled units, but that’s only one aspect of remaining invisibility in the market more broadly.
This stemmed from an AEMC Rule Change … part of which the AEMO was required to produce Guidelines for how generators were to classify their outages. These AEMO’s guidelines (subsequently upgraded) drew on various international practice (including IEEE standard 762-2006 … which has since been superseded by IEEE standard 762-2023).
Without readers falling down their own rabbit hole, a key take-away is that there’s some subjectivity involved in trying to squeeze every outage permutation into two broad buckets, if that’s all that’s allowed in terms of categories. That’s further complicated by the fact that the ez2view software classifies (i.e. colours) outages utilising logic embedded in the software interpreting what each individual participant has submitted (and the AEMO has hence published) as data:
1) Planned outages are (mostly) shown as ‘grey’ in the ‘Generator Outages’ widget.
2) Unplanned outages are shown as ‘red’ in the ‘Generator Outages’ widget.
Let’s look at the 5 cases noted above…
(D) What about the 5 units specifically mentioned?
Let’s work through each in turn (in each case utilising the ‘Forecast Convergence’ widget* in ez2view (also time-travelled to 15:30 (NEM time) on Wednesday 27th November 2024), in the order they have been highlighted in the image above:
* Remember that this widget allows one to ‘look up a vertical’ in order to see ‘that other dimension of time’.
BW02 … shown as ‘Basic Planned Outage’
With this image we see that the outage was planned over a year in advance and, at the time of the snapshot, appeared to be running to schedule:
BW03 … shown as ‘Unplanned Forced Outage’
In contrast, here, this is clearly an unplanned outage:
ER03 … shown as ‘Extended Planned Outage’
Remember that this morning we subsequently* wrote about ‘The extended outage at Eraring unit 3 (as at Thu morning 5th Dec 2024)’
* i.e. we now know that the outage plans changed somewhat.
But back at 15:30 (NEM time) on Wednesday 27th November 2024, we already saw that there’d been two instances of slippage of expected return to service dates.
I can see why the participant has classified this as a planned outage … but important to keep in mind that this outage has been extended (for reasons unknown at this point).
VP6 … shown as ‘Extended Planned Outage’
The picture above (for ER03) has some similarities for VP6:
In this case there’s been one slippage of expected return to service date … but not to the same extent as for ER03 above:
1) Only one slippage for VP6.
2) Not as much delay.
I can also see why the participant has classified this as a planned outage … but important to keep in mind that this outage has been extended (for reasons unknown at this point). Note that (unless I misheard?) Daniel Westerman referred to this unit as an unplanned outage, but I’d not have classified it as such (especially if ER03 was planned)?
Zooming into this display (image not shown here) we see that:
1) Until the end of the day on 15th November 2024, the participant had the expected return to service date as Monday 25th November 2024
2) From that point forwards, the expected return to service date was 1st December 2024 or 2nd December 2024 (it varied a few times).
Given we used ‘Bids & Offers’ widget for ER03 in this morning’s article, we take a similar approach with this image … albeit that this window is also time-travelled to 15:30 (NEM time) on Wednesday 27th November 2024:
As highlighted there, there were submissions made of (new) ‘Daily Bids’ on the early evening of 16th November 2024 to take account of the extended outage. There does not need to be a rebid reason for these (i.e. they are Daily Bids, not Rebids after Gate Closure #1), so we’re none the wiser as to the reason for the extension to the outage.
TALWA1 … shown as ‘Basic Planned Outage’
Here’s a view of the same widget for Tallawarra A:
In this case a planned outage, with (at this point) no slippage in expected return to service.
(E) From a Reliability (i.e. capital ‘R’) perspective
Whilst this example could easily lead to readers falling down the rabbit hole in terms of defining outages, at the end of the day one perspective is to ask oneself whether each of these 5 units were expected to be available and operational (say) N days/months prior to Wednesday 27th November 2024. If using this line of questioning:
1) BW02 and TALWA1 behaved ‘according to the plan’ produced many weeks (months!) beforehand;
2) BW03 clearly did not
3) For ER03 and VP6, it really depends on ones frame of reference.
… especially with Wednesday 27th November 2024 being a very sensitive day (with extra cost incurred for RERT, high prices in the spot market, public calls for conservation, and ‘blackouts’ in the news headlines etc…).
… and finally, with respect to ‘according to plan’ here’s something (obliquely relevant?) that comes to mind for me.
That’s all for now.
Nice work Paul. It’s also notable that despite the work done to provide guidance on how participants should classify outages in MTPASA, some of them routinely seem to ignore that guidance.
Without naming individual generators, there are cases we’ve seen of various units extending planned outages (sometimes for significant numbers of days and/or multiple times) but continue to label the additional outage period as a ‘Basic Planned Outage’, when the correct classification would obviously be ‘Extended Planned Outage’.
Other NEM participants and spectators could perhaps infer these generators are:
– unaware of AEMO’s guidance
– aware but unable to interpret or implement it,
– aware but choosing to ignore it, or
– trying to avoid scrutiny of maintenance slippages.