What helped (and what didn’t help) to contain the frequency drop with this morning’s Kogan trip?

For today (Thursday 2nd April 2026) we’ve already noted:

1)  the ‘(Close to) Full Load trip of Kogan Creek early on Thursday 2nd April 2026’, which showed:

(a)  the trip is visible in the 03:05 Dispatch Interval (NEM time) … remembering the building blocks of time;

(b)  with a Dispatch Error for the KPP_1 unit being +720MW … remembering:

i.  that a positive dispatch Dispatch Error represents underperformance, relative to Target

ii.  so acting to drag the frequency down…

2)  … which we saw in this ‘first look at the frequency performance, with today’s trip at Kogan Creek’.

Because of the 04:00 end-of-market-day, we already have access to the ‘next day public’ data set – so I quickly opened the pre-configured ‘Trends Engine’ query in ez2view looking at ‘Dispatch Error (NEMwide) by Fuel Type’ and extended to look at yesterday and 4 hours of today:

With this image, we can see:

1)  Specifically for the 03:05 Dispatch Interval (NEM time):

(a)  The effect of the Dispatch Error for the KPP_1 unit (+720MW) is clearly visible in the Fuel Type aggregates

(b)  In terms of what helped to arrest the frequency drop:

i.  Largest beneficial contribution came from ‘all Large BESS’ (Aggregate Dispatch Error = -375MW)

ii.  Second largest beneficial contribution came from ‘all the other Coal Units’ (Aggregate Dispatch Error = –246MW net of KPP)

iii.  with notable positive contributions from ‘all hydro’ (Aggregate Dispatch Error = -51MW) and ‘all gas’ (Aggregate Dispatch Error = -22MW)

(c)  Harmful contributions came from ‘all wind’ (Aggregate Dispatch Error = +44MW)

(d)  The ez2view ‘Notification’ widget snapshot in the first article for 03:05 showed some of these at a unit level.

2)  But remembering that the subsequent 03:10 Dispatch Interval (NEM time) also saw frequency on the low side of the PFR deadband for much of* the dispatch interval:

A caveat for readers here is:
>  that, whilst the frequency trace shows the transition of frequency through the 03:10 Dispatch Interval (i.e. 3,000 data points through the interval) the Dispatch Error calculation is an end-of-interval calculation (i.e. only 1 data point, and at the end, by which point frequency was back in the deadband.
>  So be careful in reading too much into this!

(a)  a simplifying assumption is that (at least through most of the dispatch interval, though perhaps not right at the end) a positive negative Dispatch Error would still have been beneficial

… in which case we see that helpful at an end-of-interval basis were:

i.  ‘all BESS’

ii.  and ‘all wind’

iii.  and ‘all hydro’ helped

(b)  and positive Dispatch Error still harmful through the dispatch interval (but perhaps not at the end):

… in which case we see:

i.  ‘all coal’ was actually pulling the wrong way (Aggregate Dispatch Error = +84MW) … which is a curiosity we might explore more as another layer of the onion …

c)  Here’s the zoomed-in view:

(d)  The ez2view ‘Notification’ widget snapshot in the first article for 03:10 also showed some of these at a unit level.

3)  The 28-hour view above:

(a)  is useful to highlight the scale of the collective unders from Large Solar through the day on Wednesday 1st April 2026;

(b)  to help reinforce that we were probably fortunate the Kogan trip happened during night-time hours, or the frequency performance would probably have been worse

(c)  and so wonder when it might be that AEMO procures and enables more Regulation FCAS during daylight hours to address the foibles of solar PV

d)  as a reminder:

i.  of the ongoing challenges of forecasting VRE with sufficient accuracy;

ii.  and the one-sided nature of the Semi-Dispatch Cap.